Exp AstronDOI 101007s10686-008-9126-5

ORIGINAL ARTICLE

Einstein Gravity Explorerndasha medium-classfundamental physics mission

S Schiller middot G M Tino middot P Gill middot C Salomon middot U Sterr middot E Peik middotA Nevsky middot A Goumlrlitz middot D Svehla middot G Ferrari middot N Poli middotL Lusanna middot H Klein middot H Margolis middot P Lemonde middot P Laurent middotG Santarelli middot A Clairon middot W Ertmer middot E Rasel middot J Muumlller middot L Iorio middotC Laumlmmerzahl middot H Dittus middot E Gill middot M Rothacher middot F Flechner middotU Schreiber middot V Flambaum middot Wei-Tou Ni middot Liang Liu middot Xuzong Chen middotJingbiao Chen middot Kelin Gao middot L Cacciapuoti middot R Holzwarth middotM P Heszlig middot W Schaumlfer

Received 3 December 2007 Accepted 6 October 2008copy Springer Science + Business Media BV 2008

S Schiller (B) middot A Nevsky middot A GoumlrlitzInstitut fuumlr ExperimentalphysikHeinrich-Heine-Universitaumlt 40225 Duumlsseldorf Germanye-mail stepschilleruni-duesseldorfde

G M Tino middot G Ferrari middot N PoliDipartimento di Fisica and LENS LaboratoryUniversitagrave di Firenze CNRINFM INFN - Sez di Firenze50019 Sesto Fiorentino Italy

P Gill middot H Klein middot H MargolisNational Physical Laboratory Teddington TW11 0LW UK

C SalomonLaboratoire Kastler Brossel Ecole Normale Supeacuterieure24 rue Lhomond 75231 Paris France

U Sterr middot E PeikPhysikalisch-Technische Bundesanstalt Bundesallee 10038116 Braunschweig Germany

D SvehlaInstitute of Astronomical and Physical Geodesy Technische Universitaumlt MuumlnchenArcisstrasse 21 80333 Munich Germany

L LusannaINFN Sez di Firenze Polo Scientifico 50019 Sesto Fiorentino Italy

P Lemonde middot P Laurent middot G Santarelli middot A ClaironLNE-SYRTE Observatoire de Paris 61 Avenue de lrsquoobservatoire 75014 Paris France

W Ertmer middot E RaselInstitut fuumlr Quantenoptik Leibniz Universitaet HannoverWelfengarten 1 30167 Hannover Germany

Exp Astron

Abstract The Einstein Gravity Explorer mission (EGE) is devoted to a precisemeasurement of the properties of space-time using atomic clocks It tests oneof the most fundamental predictions of Einsteinrsquos Theory of General Relativ-ity the gravitational redshift and thereby searches for hints of quantum effectsin gravity exploring one of the most important and challenging frontiers infundamental physics The primary mission goal is the measurement of thegravitational redshift with an accuracy up to a factor 104 higher than thebest current result The mission is based on a satellite carrying cold atom-based clocks The payload includes a cesium microwave clock (PHARAO)an optical clock a femtosecond frequency comb as well as precise microwavetime transfer systems between space and ground The tick rates of the clocksare continuously compared with each other and nearly continuously withclocks on earth during the course of the 3-year mission The highly ellipticorbit of the satellite is optimized for the scientific goals providing a largevariation in the gravitational potential between perigee and apogee Besidesthe fundamental physics results as secondary goals EGE will establish a global

J MuumlllerInstitut Fuumlr Erdmessung Leibniz Universitaumlt HannoverSchneiderberg 50 30167 Hannover Germany

L IorioSezione di Pisa INFN Viale Unitagrave di Italia 68 70125 Bari Italy

C Laumlmmerzahl middot H DittusCentre of Applied Space technology and Microgravity (ZARM)University of Bremen Am Fallturm 28359 Bremen Germany

E GillDepartment of Earth Observation and Space SystemsDelft University of Technology Kluyverweg 12629 HS Delft The Netherlands

M Rothacher middot F FlechnerGeoForschungsZentrum Potsdam Germany

U SchreiberFundamentalstation Wettzell 93444 Bad Koetzting Germany

V FlambaumUniversity of New South Wales Sydney Australia

W-T NiPurple Mountain Observatory Chinese Academy of Sciences2 Beijing W Road Nanjing 210008 China

L LiuShanghai Institute of Optics and Fine MechanicsChinese Academy of Sciences Jiading Shanghai 201800 China

Exp Astron

reference frame for the Earthrsquos gravitational potential and will allow a new ap-proach to mapping Earthrsquos gravity field with very high spatial resolution Themission was proposed as a class-M mission to ESArsquos Cosmic Vision Program2015ndash2025

Keywords Gravitational redshift middot Gravitational frequency shift middot Atomicclock middot Optical clock middot Equivalence principle middot Local position invariance middotSpecial relativity middot Lorentz invariance middot General relativity middot Geoid middotFrequency transfer middot Microwave link middot Gravitational potential

Abbreviations

AOM acoustooptic modulatorshifterACES atomic clock ensemble in spaceCNES centre national etudes spatialesCCR corner cube reflectorEEP Einstein Equivalence PrincipleEGE Einstein Gravity ExplorerEO electro-optic modulatorESA European space agencyFCDP frequency comparison and distribution unitGGOS global geodetic observing systemGNSS global navigation satellite systemGPS global positioning systemGS ground station(s)

X Chen middot J ChenInstitute of Quantum Electronics School of Electronics Engineeringand Computer Science Peking University Beijing 100871 China

K GaoWuhan Institute of Physics and Mathematics The Chinese Academy of SciencesPO Box 71010 Wuhan 430071 China

L CacciapuotiESA Research and Scientific Support Department ESTEC2200 AG Noordwijk ZH The Netherlands

R HolzwarthMax-Planck-Institut fuumlr Quantenoptik85748 Garching Germany

R HolzwarthMenlo Systems GmbH 82152 Martinsried Germany

M P HeszligASTRIUM Space Transportation D-88093 Friedrichshafen Germany

W SchaumlferTimeTech GmbH 70563 Stuttgart Germany

Exp Astron

GR General RelativityGRACE gravity recovery and climate experimentISS international space stationLPI Local Position InvarianceMWL microwave linkMOLO microwave-optical local oscillatorMampC monitoring and controlOFC optical frequency combPHARAO Projet drsquohorloge atomique par refroidissement drsquoatomes en

orbitePLL phase-locked loopPTB Physikalisch-Technische BundesanstaltPRARE Precise Range And Range-Rate EquipmentQCD quantum chromodynamicsRF radiofrequencyRx receiverSLR satellite laser rangingSME standard model extensionTC telecommandTM telemetryTampF time and frequencyULE ultralow-expansion glassUHV ultra-high vacuumUTC universal time coordinatedUSO ultrastable quartz oscillatorXPLC onboard computercontrol system

1 Introduction

Einsteinrsquos theory of General Relativity is a cornerstone of our current descrip-tion of the physical world It is used to describe the motion of bodies rangingin size from satellites to galaxy clusters the propagation of electromagneticwaves in the presence of massive bodies and the dynamics of the Universeas a whole In general the measurement of general relativistic effects is verychallenging due to their small size [1 2] Thus only few of its predictions havebeen tested with high accuracy for example the time delay of electromagneticwaves via the Cassini mission [3] and the existence of gravitational waves byobservation of binary star systems The accuracy of the determination of theseeffects is at the 20 ppm level

The situation is similar for one of the most fascinating effects predictedby General Relativity and other metric theories the slowing of clocks ina gravitational field (gravitational time dilation gravitational redshift) Thegravitational redshift was measured with 7 times 10minus5 relative inaccuracy in the1976 Gravity Probe-A experiment [4] by comparing a ground clock with asimilar clock on a rocket as the height changed The most performing clocks

Exp Astron

available at the time hydrogen masers were used for this experiment TheACES (Atomic Clock Ensemble in Space) mission planned to fly on the ISS in2014 seeks to improve the determination by a factor 25 by using a cold atomclock (PHARAO) and a hydrogen maser [5 6] Recent progress on cold atomclocks in the optical domain and in optical technology enable performing thisfundamental test with orders of magnitude better sensitivity

Although it has been very successful so far General Relativity as well asnumerous other alternative or more general theories proposed in the courseof the development of theoretical physics are classical theories As such theyare fundamentally incomplete because they do not include quantum effectsIn fact it has not yet been possible to develop a theory of gravity that includesquantum mechanics in a satisfactory way although enormous effort has beendevoted to this goal Such a theory would represent a crucial step towards aunified theory of all fundamental forces of nature Several approaches havebeen proposed and are currently under investigation (eg string theory) Afull understanding of gravity will require observations or experiments thatdetermine the relationship of gravity with the quantum world This topic isa prominent field of activity and includes the current studies of dark energy

The Einstein Gravity Explorer is a fundamental physics mission dedicatedto this question its primary task is to measure the gravitational frequency shiftwith unprecedented accuracy The EGE mission uses a satellite on a highlyelliptic orbit (see Fig 1) The ratio of the frequencies of two on-board clocksand the ratios between satellite and ground clocks are the main observables

Fig 1 General concept of the EGE mission Clocks on the satellite and on ground are intercom-pared as the satellite orbits the Earth on a highly elliptic orbit

Exp Astron

These can be interpreted toward tests of General Relativity (GR) and ofcompeting metric theories of gravity and as tests for the existence of newfields associated to matter (see Section 3) The outcome of the mission will beeither a confirmation of GR and metric theories within the accuracy providedby the instruments or the discovery of a deviation In the latter case themission would provide a first indication of the breakdown of current (classical)gravitational physics theories and could pave the way towards a unified theoryof all forces

2 Scientific objectives

21 Summary of science objectives

211 Fundamental physics science objectives (1 ppm = 1 part in 10 61 ppb = 1 part in 10 9)

Primary objectives

High-precision measurement of the earth gravitational frequency shift at25 ppb accuracy a factor 3000 improvementFirst precise measurement of the sun gravitational frequency shift at 1 ppmlevel a factor 2 times 104 improvementTest of space and time variability of fundamental constants with accuracy30 ppbFirst search for neutron scalar charge with accuracy 30 ppbTest of anomalous coupling of matter to the Standard Model quantumfields

Secondary objectives

Tests of Lorentz Invariance in the matter and photon sector (factor 20improvement)Contribution towards establishing a new definition of the unit of time

212 Spin-off to other fields (outside fundamental physics)

Establishment of a new approach to the determination of the geopotentialwith geoid height resolution of a few cmComparison of distant terrestrial clocks at the 10minus18 levelDemonstration of clock and link technology for future applications eg inprecision spacecraft navigationDemonstration on high performance real-time range determination withprecision 25 times (based on code phase) to 1000 times (based on carrier

Exp Astron

phase) higher compared to systems onboard previous missions (ERS-2TOPEX POSEIDON)Comparison of three different orbit determination systems Laser-Ranging μm-precision microwave ranging GPS-based orbit deter-minationAtmospheric signal propagation study precise determination of thirdorder ionospheric termMonitoring stability of the GalileoGPSGlonass time scales

22 Gravitational frequency shift measurement and search for new physics

The primary science goal of EGE is a measurement and test of the gravita-tional frequency shift in the weak-field limit as predicted by metric theoriesof gravitation These theories also predict that the shift is independent of thetype of clock used to measure it The comparison of the frequencies ν oftwo identical clocks 1 2 operating at different locations x1 x2 yields a fre-quency ratio

v2(xprime)v1

(xprime) sim= 1 + (U (x2) minus U (x1))

c2 (1)

Here νi(xprime) is the frequency of clock i located at xi as observed (measured) ata particular location xprime where the comparisons are performed The comparisonlocation can be arbitrary but will often be x1 or x2 or both U is thegravitational potential U(x) = minusGM|x| in case of a spherically symmetricbody of mass M In short time runs (or clocks tick) more slowly the closer thesystemclock is to a massive body Stationary clocks and observers and weakgravitational fields have been assumed in Eq 1

Through its measurement EGE searches for a possible violation of the grav-itational frequency shift A violation may be described phenomenologicallyby a dependence of one or more fundamental constants on the gravitationalpotential X = X(Uc2) where X is a generic dimensionless fundamentalconstant or a dimensionless combination of fundamental constants Such adependence is a violation of the principle of Local Position Invariance (LPI)which states that the outcome of local nongravitational experiments is inde-pendent of where and when in the Universe they are performed If a constantdepends on the gravitational potential then the results of those laboratory-type (ie local) experiments that depend on that constant will also depend onthe location namely on the location of the laboratory with respect to the massdistribution generating the potential and will violate LPI

LPI is one of the three cornerstones of the Einstein Equivalence Principle(EEP) the others being the Weak Equivalence Principle and Local LorentzInvariance It is a common belief that the EEP may not be absolutely valid -therefore spatial dependence and time dependence of fundamental constantshave become a very intense research topic in the last decade with experimentsundertaken worldwide to search for violations of various aspects of EEP For

Exp Astron

example extensive studies of atomic and molecular spectra of gas clouds in thedistant Universe are currently undertaken to search for a difference of theseparameters compared to todayrsquos values (see eg [7ndash11] and references therein)Also some differences in Big Bang nucleosynthesis data and calculations canbe naturally explained by a variation of fundamental constants [12 13]

The EGE mission is thus a probe of possible space-dependence of fun-damental constants The search may help to point the way toward a unifiedtheory of all forces reveal hypothetical extra dimensions in our Universe or theexistence of many sub-Universes with different physics chemistry and biologyIndeed theories unifying all forces applied to cosmology suggest space andtime dependence of the coupling constants [14ndash16] Another argument forspatial variation of fundamental constants comes from the anthropic principleThere must be a fine tuning of fundamental constants which allows humans andany life to appear Life in its present form is not consistent with fundamentalconstants whose values only slightly differ from the actual ones This finetuning can be naturally explained by the spatial variation of the fundamentalconstants we appeared in the area of the Universe where the values of thefundamental constants are consistent with our existence

In the Standard Model there are three main fundamental dimensionless pa-rameters determining the structure and energy levels of stable matter (atomsmolecules) the electromagnetic fine structure constant α and the two ratiosof electron mass me and light quark mass mq to the strong QCD energy scaleQCD Xeq = meqQCD The fundamental masses me and mq are proportionalto the vacuum Higgs field which determines the electroweak unification scaleTherefore the constants meqQCD can also be viewed as the ratio of the weakenergy scale to the strong energy scale

How can a space-time variation of the fundamental constants and a depen-dence on the gravitational potential occur Light scalar fields very naturallyappear in modern cosmological models affecting α and meqQCD Oneof these scalar fields is the famous ldquodark energyrdquo which causes acceleratedexpansion of the Universe Another hypothetical scalar field is the dilatonwhich appears in string theories together with the graviton Cosmologicalvariation of these scalar fields should occur because of drastic changes ofthe matter composition of the Universe During the Big Bang nucleosyn-thesis the Universe was dominated by radiation then by cold dark matterand now by dark energy Changes of the cosmic scalar field ϕ0(t) lead tothe variation of the fundamental constants X(ϕ0) Massive bodies (galaxiesstars planets) can also affect physical constants They have a large scalarcharge S = sp Z + se Z + sn N where Z is the number of protons and elec-trons and N is the number of neutrons sp se sn are the scalar charges ofthe proton electron and neutron respectively There is also a contributionof the nuclear binding energy (scalar charge of virtual mesons mediatingnuclear forces)

The scalar charge produces a Coulomb-like scalar field ϕs = SR where Ris the distance from the charge The total scalar field is ϕ = ϕ0 + ϕs therefore

Exp Astron

we may have a variation of the fundamental constants inversely proportionalto the distance from massive bodies

X (ϕ) = X (ϕ0 + ϕs) = X (ϕ0) + δX (R)

δX (R) = (dX

dϕ

)S

R

where a nonzero δX violates LPI and a small ϕs was assumedThe gravitational potential U is essentially proportional to the number of

baryons Z + N and inversely proportional to the distance R U = -GMR

Therefore the change of the fundamental constants near massive bodies canbe written

δX

X = K (X i) δ(Ui

c2

) (2)

where i refers to a particular composition of the mass M and where thecoefficients K(X i ) = minus(dXdϕ) c2SiGMi are not universal but are expectedto be different for eg the Sun and the Earth K(X Earth) = K(X Sun)Indeed the Sun consists mostly of hydrogen therefore the Sunrsquos scalar chargeis SSun Z (sp + se) The Earth contains heavier elements where the numberof neutrons exceeds the number of protons N 11 Z Therefore the scalarcharge of the Earth SEarth Z (11 sn + sp + se) and is sensitive to the neutronscalar charge (and also to the contribution of the nuclear binding) in contrastto the Sun

The most sensitive experiments suitable for a search of a space-time de-pendence in local or near-local experiments (such as satellite experiments) areclockndashclock comparisons In one type of terrestrial local experiment dissimilarclocks located nearby are compared for a duration spanning at least one yearDuring such periods the earth moves periodically closer and further awayfrom the sun so that the solar gravitational potential on Earth is modulatedThis enables a search for a dependence of the fundamental constants on thesunrsquos scalar charge [17 18] These experiments are able to set limits to K(XSun) for X = α meqQCD at the level of 10minus6 [12 13]

The proposed EGE satellite experiments are complementary to terrestrialexperiments allowing to set limits to K(X Earth) and combining with theresults on K(X Sun) from terrestrial comparisons to set limits on the depen-dence of the fundamental constants on the neutron scalar charge sn

In summary when the frequencies of two identical clocks (of nominalfrequency νa) that are located at different positions x1 x2 are comparedby measuring their frequency ratio at some position xrsquo the result may bewritten as

va2(xprime)

va1

(xprime) = 1 +

[1 +

sum

XAXa K (X i)

](Ui (x2) minus Ui (x1))

c2 (3)

The sensitivity factor AXa = X (d(ln νa)dX) is the sensitivity of the clockfrequency to a particular constant X and can be calculated using atomic and

Exp Astron

nuclear theory (see eg [12 13]) The measurement of this ratio which issensitive to K represents a test of General Relativityrsquos redshift

When the frequencies of two co-located but dissimilar clocks a b (ofnominal frequencies νa0 and νb 0) are compared in-situ at two locations x1 x2

(LPI test) the result is

[va1 (x1)

vb 1 (x1)

] [va2(x2)

vb 2 (x2)

] = 1 +[X

(AXa minus AXb

)K (X i)

](Ui (x2) minus Ui (x1))

c2 (4)

The result of the EGE mission may be given as a limits for (or nonzerovalues of) K(X Earth) for the three fundamental constants X = α meQCDand mqQCD

23 Measurement approaches for the gravitational frequency shift

The gravitational potential experienced by the ground and satellite clocksincludes the contributions from the earth and the sun (the contribution fromthe moon is an order smaller and may be neglected for the purposes of thediscussion but will be taken into account in data analysis)

Nonzero gravitational frequency shift signals (Eq 3) can be obtained inthree ways Method (1) below is the most sensitive and represents a strongreason to employ an optical clock which on short averaging times is ableto provide a higher frequency stability than a microwave clock Method (2)and (3) are less accurate but are complementary in terms of measurementapproach Finally method (3) addresses the solar gravitational frequencyshift while (1) and (2) measure the terrestrial effect The performance of theinstruments assumed in the following is shown in Fig 2

(1) Repeated measurements of satellite clock frequency variation betweenapogee and perigeeIn this measurement approach the difference (or ratio) between earthndashsatellite clock frequency ratio at perigee and at apogee is consideredThis difference is measured at each perigee and apogee passage andaveraged over the mission duration Systematic shifts (if not correlatedwith orbital motion) are expected to average out leading to a gain insensitivity of up to n12 where n = number of orbits This mode usesthe stability properties of the satellite clocks on the timescale fromperigee to apogee rather than their accuracy Assuming a specificationof le 3 times 10minus16 satellite optical clock fractional frequency instability forintegration times between 1000 and 25000 s n sim= 1000 comparisons(ie most orbits of the 3-year mission) and a perigee-apogee potentialdifference (U(apogee) minus U(perigee))c2 = Uc2 sim= 4 times 10minus10 allows a

Exp Astron

Fig 2 Performance of clocks and links on EGE Curve ldquooptical clocks EGErdquo is the specificationCurve ldquooptical clocks EGE (goal)rdquo refers to expected near-future performance of an opticalclock based on neutral atoms ldquoUpgraded PHARAOrdquo refers to the PHARAO microwave clockusing the high-quality microwave-optical local oscillator The microwave link (MWL) performanceexhibits a discontinuity since common-view contact time of ground stations with EGE is limited toseveral h The different slopes arise because on short time scales only white phase noise is relevantwhile on long time scales flicker noise also contributes The reproducibility is taken into account

measurement of the redshift at 25 ppb accuracy The ground clocks towhich the satellite clocks are compared to at perigee and at apogee willbe located in different ground stations The inaccuracy of these groundclocks must therefore be sufficiently smaller than 1 times 10minus17 in order notto contribute to the 25 ppb error Current state-of-the art inaccuracy is2 times 10minus17 and is likely to decrease further

(2) Absolute comparison between ground and satellite clocks with the satel-lite at apogeeHere the terrestrial gravitational potential difference is maximumU(apogee) minus U(perigee)c2 sim= 65 times 10minus10 Assuming a space optical clockinaccuracy of 1 times 10minus16 and a ground clock inaccuracy a factor 3lower (current state-of-the-art) allows a measurement of the redshift at150 ppb This measurement takes advantage of the repetitive nature ofthe orbit only to a small extent and relies on the accuracy of the opticalclocks in particular of the space clocks It is thus complementary to (1)albeit less accurateIn (1) and (2) the dominant contribution to the redshift is from Earthrsquosgravitational potential The contribution from the solar potential isstrongly reduced due to the fact that satellite and earth are an interactingsystem in free-fall toward the Sun [19] and can be modeled

(3) Comparisons between terrestrial clocksThe frequency comparison link to the EGE satellite allows terrestrialclock comparisons by a common-view method whereby the frequency

Exp Astron

signals from two distant ground clocks are sent to the satellite andcompared there The frequency difference between two identical groundclocks is determined if systematic instrumental errors are negligible bythe difference in gravitational potential at the two clock locations Thispotential difference includes a terrestrial and a solar contributionThe solar contribution varies according to the orientation of the clockpair with respect to the Sun with a 1 day-period For two clocks at asurface distance of sim6000 km and suitable orientation with respect tothe sun the normalized modulation amplitude is sim= 4 times 10minus13 This is thesame as the earth potential difference for 4 km height difference It ispossible that ground clocks will have reached an inaccuracy at the levelof 1 times 10minus18 at the time of the mission in 2015 Using the microwave linka comparison of ground clocks over sim3 h (typical common-view duration)with a resolution of 1 times 10minus17 can be performed limited by the instabilityof the link Repeating the clock comparison at each common-view win-dow (n = 1000 comparisons) will allow averaging and a measurementof the solar gravitational shift with fractional inaccuracy of 1 ppm Incomparison the best current results for the solar gravitational frequencyshift are at the few-percent level [20 21] A moderate averaging out ofground clock systematic shifts is assumed as well as modeling of the earthgravitational potential variations on the daily timescale at the appropriatelevel

(4) Comparisons of on-board clocksThe two on-board clocks are continuously compared these data areanalyzed for violation of LPI (Eq 4) Since the measurement is a nulltest the required accuracy of the gravitational potential at the satellitelocation is low and easily satisfied

24 Test of local position invariance and search for neutron scalar charge

The test of LPI is performed in conjunction with the measurement of thegravitational frequency shift

With approach 23 (1) the optical clock will be able to measure K (α Earth)with accuracy 30 ppb (spec)

Combining the results of the microwave and optical clocks will determineK(meQCD Earth) + εK(mqQCD Earth) with accuracy 85 ppb (spec)where ε asymp minus01

Simultaneously the on-board comparison 23 (4) yields the combination19 K(α Earth) +K(meQCD Earth) + εK(mqQCD Earth) with accuracy85 ppb (spec)

Note that these measurements are fully complementary to current andfuture terrestrial clock-clock comparisons which are sensitive to K(X Sun)and can be combined with the latter to determine or set limits to the neutronscalar charge

Exp Astron

25 Theoretical analysis of frequency comparison signals

The on-board clock-clock comparison signals can be analyzed straightfor-wardly in terms of a violation of LPI since signal propagation effects areabsent

The determination of the gravitational frequency dilation via ground-satellite clock comparisons is more complex (Eq 1 is strongly simplified) andrequires an accurate treatment of all relevant relativistic effects Gravitationin the solar system is described in the barycentric (BCRS) and geocentric(GCRS) non-rotating celestial reference systems by means of post-Newtoniansolutions of Einsteinrsquos equations for the metric tensor in harmonic gaugescodified in the conventions IAU2000 [22] In these conventions the relativisticstructure of Newton potential (order 1c2) and gravitomagnetic (order 1c3)

potentials are given For the Newton potential the multipolar expansion ofthe geopotential determined by gravity mapping is used The post-Newtonianeffects of the metric (described in terms of dimensionless parameters β and γ )

can be included Orbits of satellites are evaluated in the GCRS at the levelof a few cm by means of the Einstein-Infeld-Hoffmann equations at the order1c2 [23] using the relativistic Newton potential Instead the trajectories of theground clocks in the GCRS are evaluated from their positions fixed onthe Earth crust (ITRS International Terrestrial Reference System) by usingthe non-relativistic IERS2003 conventions [24 25]

For the propagation of light rays (here radio signals) between the satelliteand the ground stations one uses the null geodesics of the post-Newtoniansolution in the GCRS The timefrequency transfer properties have beentheoretically evaluated for an axisymmetric rotating body [26ndash29] at the order1c4 (ie with Newton and gravitomagnetic potentials developed in multipolarexpansions) These results have been developed for use in the ACES missionThe relative frequency dilation can be expressed as

ν

ν= 1 + 1

cL1 +

4sum

n = 2

1

cn(Ln + Gn)

where Ln describes special-relativistic Doppler effects and Gn the general-relativistic effects These quantities are time-dependent due to the orbitalmotion For concreteness we report the orders of the terms The kinematicaleffects are |L1c| lt 1 times 10minus5 |L2c2| lt 2 times 10minus10 |L3c3| lt 3 times 10minus15 |L4c4| lt

7 times 10minus20 The Newtonian contributions are G2 = G(M)2 + G(J2)

2 + G(J4)2 + G(J6)

2(M is the mass monopole Jn are the lowest multipoles) with estimates|G(M)

2 c2| lt 3 times 10minus11 |G(J2)2 c2| lt 2 times 10minus13 |G(J4)

2 c2| lt 3 times 10minus16 |G(J6)2 c2| lt

1 times 10minus16 For G3 = G(M)3 + G(J2)

3 the estimates are |G(M)3 c3| lt 2 times 10minus14

|G(J2)3 c3| lt 1 times 10minus18 Finally the G4 = G(M)

4 + G(S)4 contributions (G(S)

4 is thelowest gravitomagnetic effect) are of order |G(M)

4 c4| lt 1 times 10minus19 |G(S)4 c4| lt

1 times 10minus22 and thus not relevant for EGE

Exp Astron

The available theory is fairly complete for the data analysis in mode 23 (1)described below For modes 23(2 3) the theory will need to be generalizedto include the effect of the gravitational potential at the location of the groundclock(s) (including time-varying effects such as the tides) and of the solar andlunar potentials

The uncertainties of the various contributions above are minimized by ausing precise EGE orbit data obtained by satellite tracking and by usingcurrent and future accurate earth gravity information Required orbit positionknowledge is at 1 cm level near perigee and sim50 cm near apogee reachablealready today with the proposed orbitography approach described below Thevalidity of the special-relativistic Doppler shift has been and will continue to beindependently verified with increased accuracy by spectroscopy of relativisticatomic ion beams [30] Potential violations of this aspect of Lorentz Invari-ance are expected to be sufficiently bounded so as not to affect the signalinterpretation

26 Local Lorentz invariance tests

There has been an explosion of interest in Local Lorentz invariance in re-cent years [31 32] with numerous astronomical studies and high-precisionexperiments applied to search for violations of Lorentz invariance Signifi-cant developments have also occurred on the theoretical side and a theory(standard model extension SME [33]) has been worked out that provides aunified framework for describing and analyzing Lorentz Invariance violationsin a variety of systems Two aspects of Lorentz Invariance can be tested withEGE

261 Independence of the speed of light from the laboratory velocity

A possible dependence of the speed of light on the speed v of the laboratorycan be modeled according to the MansourindashSexl test theory [34] by

c (v) = c0(1 + Bv2

c2

)

Here v 377 kms is the velocity with respect to the cosmic microwavebackground the cosmologically preferred frame and B is a combinationof parameters describing deviations from the usual Lorentz transformationformulas B = 0 if Lorentz Invariance holds For a terrestrial experiment therotation of the Earth around its axis modulates v with a 300 ms amplitude Adependence c(v) can be searched for by measuring the frequency differencebetween a resonance frequency of a highly stable optical cavity (cavity fre-quencies are proportional to c) and an optical clock (Kennedy-Thorndike-typeexperiment [35]) and determining the modulation of the frequency differencecorrelated with the modulation of v The advantages of a space experimentare the high orbital velocity and strongly reduced cavity deformation thanksto microgravity [36 37] On the proposed elliptic orbit v varies between+4 and minus4 kms over approximately 1 h This variation is 13 times larger

Exp Astron

than for the case of a terrestrial experiment The shorter time scale is alsoadvantageous since the drift of the cavity is more predictable In the EGEmission the reference cavity of the clock laser is used The large numberof orbits improves the statistics An improvement of the accuracy of B by afactor 20 compared to the best current terrestrial results is realistic The dataanalysis can also be performed in the framework of the SME theory takinginto account complementary bounds obtained from terrestrial experimentsand astrophysical observations

262 Independence of Zeeman splitting frequency on the directionof the magnetic field

One class of tests of Lorentz Invariance consists in measuring the frequencysplitting between two levels of a quantum system induced by a static magneticfield as a function of orientation of the field direction with respect to thestars (HughesndashDrever-type experiments [31]) This class addresses the so-called matter sector and is complementary to combined-photonmatter-sectortests such as the above The most precise experiments are performed withatomic clocks eg masers or cold atom clocks [38] The EGE instruments allowsuch experiments since magnetic fields are applied continuously in the clockor repeatedly for calibration Compared to terrestrial experiments the highorbital velocity of EGE could lead to an improvement by a factor sim20

27 Application to geophysics

Using GPS coordinates of a point on the Earth can be obtained with aninaccuracy well below 1 cm in a well defined international terrestrial referencesystem The coordinates are purely geometrical and do not contain any gravityinformation A local gravitational potential UB is determined with respect toa reference gravitational potential U A by ldquolevellingrdquo the surface of the Earthie sequentially measuring gravity g and height differences every ca 50 m

UB minus U A = minusBint

A

g (r) middot d n

where n is the difference vector between subsequent locations The disadvan-tage of measuring height differences by the levelling method is a random walkeffect of accumulated errors

In the classical definition a geoid is defined as the particular equipotentialsurface nearest to mean sea level In these terms the geoid serves as areference surface for measuring the height and also to define a referencefor the gravitational potential Such a vertical reference frame historicallywas realized for a country or several countries by determining the mean ofsea level observations at tide gauge stations taken over long period of timeHowever modern satellite altimetry missions such as TOPEXPOSEIDON orENVISAT show that departure of the mean sea level from an equipotential

Exp Astron

surface may reach up to several meters on the global scale see eg [39]Therefore height systems based on different tide gauge stations may differin the realisation of the geoid and may differ with respect to the reference byseveral meters

The best gravity field models obtained from satellite data (eg missionGRACE) have reached a precision of about 1 cm over sim250 km half wave-length [40] Satellite data also reveals changes in time [41] However for typicalEarth topography with height variations of eg 1000 m over 30 km horizontaldistance one may expect variations in the geoid of about 80 cm Such ahigh-spatial-frequency signal in geoid variations cannot be detected by spacegravity missions and requires a combination of satellite and terrestrial gravitymeasurements like gravity anomalies deflections of vertical and GPSlevellingpoints The best combined global gravity field models are provided up to de-gree and order 360 in terms of spherical harmonic representation correspond-ing to a half-wavelength of about 55 km However in combining the satelliteand terrestrial gravity field measurements the problem of terrestrial data givenin different height systems between continents and different countries remainsand has to be tied to satellite measurements

The geodetic scientific community is currently establishing a Global Geo-detic Observing System (GGOS) [42] Its objectives are the early detectionof natural hazards the measurement of temporal changes of land ice andocean surfaces as well as the monitoring of mass transport processes in theEarth system Global change processes are small and therefore difficult toquantify Therefore the required precision relative to the Earthrsquos dimensionis 1 ppb GGOS will be established by the combination of geodetic spacetechniques (GPS Laser-Ranging very large baseline interferometry) andrealized by a very large number of terrestrial and space-borne observatoriesThe purely geometric terrestrial 3D coordinate system is in good shape andfully operational It has to be complemented however with a globally uniformheight system of similar precision The current precision level of regionalheight systems in terms of gravity potential differences is in the order of1 m2s2 (10 cm) with inconsistencies between these various systems up toseveral 10 m2s2 (several meters) The actual requirement in the context ofGGOS is 01 m2s2 (1 cm) with the need of a permanent ie dynamical controlThis requirement of high precision height control comes from the need tounderstand on a global scale processes such as sea level change global andcoastal dynamics of ocean circulation ice melting glacial isostatic adjustmentand land subsidence as well the interaction of these processes Only by meansof monitoring in terms of gravity potential changes at the above level ofprecision the change of ocean level can be understood as a global phenomenonand purely geometric height changes be complemented by information aboutthe associated density or mass changes

The geoid can also be defined in a relativistic way [43] as the surface whereaccurate clocks run with the same rate and where the surface is nearest tomean sea level The relation between the differences in the clock frequenciesand the gravitational potential is given in simplified form by Eq 1 At present

Exp Astron

there is no operational way to compare frequencies of the already availableoptical clocks on the global scale at the same level as their accuracy wouldallow With EGE and by then improved ground optical clocks it will becomepossible to obtain gravitational potential differences on a global scale by com-paring frequencies This would be a significant new dimension to gravity fielddetermination since such observables are given on the global scale and providein situ local gravity information at the same time EGE will allow clock-basedgravitational potential mapping with the same payload instruments used forthe fundamental physics experiments The on-board clocks are actually notrequired for this purpose MOLO MWL and FCDP suffice Thus EGE willallow establishing a global reference frame for the Earth gravitational fieldwith accuracy in the order of a few cm in terms of the geoid heights Thisassumes that by the time EGE is operational (mobile) ground clocks withfractional inaccuracy at the level of 10minus18 are available the measurementsare limited by MWL noise and thus require a long integration times to matchthe expected ground clock performance Such a reference frame will serve asreference for all gravity field modeling and clocks will be used to define theinternational atomic time scale and the Earth gravitational potential at thesame time

In summary with optical clocks geodesy will undergo a second revolutionafter geometry now being measured by clock-based GNSS systems also physi-cal heights and potential will be measured by (mobile) optical clocks using thegravitational redshift effect

28 Application to frequency standards and terrestrial fundamentalphysics studies

As has been described repeating ground clock comparisons and integratingover several months will provide global ground clock frequency comparisonsat the level of 1 times 10minus18 more than two orders of magnitude below the currentGPS and two-way satellite time and frequency transfer methods In the timeand frequency community the availability of such a global high performancemicrowave link will accelerate the process leading to a redefinition of the SIsecond based on clocks operating in the optical domain

The global character of the ground clock comparisons made possible byEGE can also contribute to the search for a time-variation of the fundamentalconstants

The comparisons do not require availability of the on-board clocks The riskof this type of measurements is therefore reduced Moreover the lifetime ofmany subunits required for this measurement (MWL and FCDP see below)can exceed 10 years (from previous experience with similar devices) andthus offers a motivation to increase the mission duration beyond the nominal3 years

Further applications to metrology are listed under the secondary goals inSection 21 above

Exp Astron

3 Mission profile

31 Orbit

The orbit must satisfy the following requirements

Primary 1 Large gravitational potential difference between apogee andperigee

2 sufficient contact time with at least two ground stations atperigee

Secondary Simultaneous visibility of satellite from distant (inter-continental) ground stations for several hours

To satisfy the primary objective the orbit must be highly elliptic The perigeeshould be low A very low perigee means a short contact time which decreasesthe accuracy of the frequency measurement A compromise is a perigeealtitude of sim2500 km For a sufficiently high apogee this yields a U of 60of the theoretical maximum between the earth surface and infinity

The baseline scenario is an orbit of approx 12 h period with the followingproperties

Apogee Altitude 37856 km (minus 10 deg latitude one over Atlantic andPacific resp)

Perigee Altitude 2500 km (+ 10 deg latitude one over Guatemala andMalaysia resp)

Inclination 634 argument of perigee 170RAAN 25 True Anomaly 220Apogee drift rate 70 kmday perigee drift rate sim 14 kmday

True anomaly and right ascension of the ascending node (RAAN) wereoptimized to minimize orbital precession Figure 3 shows the ground track andthe ground coverage at one of the perigees Common-view contact time forstations in Europe is about 6 h while between a station in Europe and in theUSA it is about 35 h Frequency comparison windows between a given groundstation and EGE repeat every 24 h

A final selection of the orbit requires an in-depth study of the costs of thepropellant and maneuvers as well as of the setting up of new ground stations

32 Ground segment

The mission requires two separate ground segments the science groundsegment and the standard satellite control segment which supports telemetryand telecommand (TMTC)

Exp Astron

Fig 3 Baseline EGE orbit One (out of two) apogee and one (out of two) perigee ground pointsare shown The ground coverage at one of the perigees is shown with circles at elevation anglesteps of 10 starting from 10

321 Science ground segment

As described above one part of the mission is to perform frequency com-parisons between ground-to-space and ground-to-ground (common-view andnon-common view) The use of a two-way microwave link (MWL) systemhas the advantage that a precise orbit or the precise relative velocity of thesatellite must not be known [4] The orbit determination requirements fora two-way system are significantly relaxed compared to the requirements tomodel the relativistic effects on the clocks The frequency comparison resultsare available both on ground and on the satellite

For ground-to-satellite clock comparisons five ground stations arefavourable two capable of viewing perigee three for viewing apogee Thisallows for the classic three-clock configuration where the performance of eachclock can be obtained individually The perigee stations will be observing thesatellite for sim25 min every perigee passage for the duration of the missionThe apogee stations will observe for sim3 h every apogee passage Of coursea larger number of stations as well as longer observation time are usefulfor robustness redundancy and orbit determination The stations could beinstalled on a temporary basis for the duration of the mission

The stations must have access to optical clocks of performance at least ashigh as todayrsquos state-of-the-art ground clocks but preferably higher Theywill include an optical frequency comb by means of which the optical clockfrequency is transformed to a radio frequency for comparison with the EGEclocks High-accuracy fiber-optic link technology already existing today allowsdistances of sim100 km between the ground station and the correspondingoptical clock allowing a certain freedom in the choice of the clock locationWith further increase of the fiber-optic link range currently being investigated

Exp Astron

the area over which the ground station can be positioned relative to the opticalclock will increase This option is of importance for the implementation of ageophysics measurement campaign for which a large number of clock locationsis of interest

The above ground stations will be complemented by some 15 additionalground terminals provided by the world-wide laboratories equipped with highperformance primary clocks at the time of the mission These will participate inorbit determination to 1 cm accuracy along the full orbit which can be achievedwith non-optical atomic clocks The EGE MWL is fully compatible to plannedACES ground terminals which might reduce the number of required stations

With a larger number of transportable optical clock ground stations de-ployed and moved across the continents a geophysics campaign could beimplemented to map the gravitational potential

Ground station equipment MWL uses dedicated small transportable termi-nals which can be easily installed at any location of interest Each terminalconsists of a steerable dish antenna of 1 m diameter with pointing accuracy of01 using a transmitted power of 5 W per band The high directivity antenna(beam aperture Ka-band 07 Ku-band 14 S-band 20) avoids multi-pathfrom nearby objects Antenna noise temperatures are Ka-band 450 K Ku-band 340 K Operation is at elevations above 10 and measurements areperformed above 20 For ranging purposes terminal location is determined byGPS surveying Each ground station computes its own ionospheric correctionsbased on a triple-band receiver The terminal has a built-in delay monitorwhich allows calibrated ranging during several months once its bias has beendetermined by laser ranging Its M and C interface is via local area networkto either a userrsquos computer or directly to the science network control centreby remote MampC The stationrsquos operation is fully automatic and unattendedSchedule and orbital data are received via LAN once per day Signal acquisi-tion is fully autonomous Data transmission is after each pass The terminal hasan uplink telecommand capability of 1 kBits and a downlink data capacity ofup to 10 kBitss For the local user there is a visual interface providing quick-look data from real-time comparison to the spacecraft clocks to verify theirgood health The final full performance product is calculated at the sciencedata centre

Local metrological data (temperature pressure humidity) are recorded tocorrect for tropospheric delay of the ranging data The main advantage ofMWL terminals compared to laser based systems is their all-weather capabil-ity although microwaves suffer higher propagation errors in the ionosphereand troposphere

The EGE ground terminal will profit from the ERS-2 mission PRARE(precise range and range rate equipment) [45] experiments whose terminalshad similar design and have been operated under extreme climatic conditionsincl Antarctica and high temperature regions from existing designs fromthe space segment (mission ACES) and no further special developments aredeemed necessary Cost per EGE master station is modest

Exp Astron

Satellite tracking via laser ranging Laser ranging is alternative to MWL rang-ing and also serves for absolute calibration of the latter For this purpose anon-board corner cube reflector (CCR) is provided Ranging will be done withthe International Laser Ranging Service Ideally the satellite will be rangedat every perigee passage and every apogee passage simultaneously with thefrequency comparison procedures Todayrsquos ranging precision is sufficient tosatisfy the science goals For apogee observations in the northern hemispherea large number of laser ranging stations is available The number of stationscapable of ranging at or near perigee in the southern hemisphere is six andthus is sufficient even considering unfavourable weather conditions at someof these The MWL ground stations are able to range even under thoseconditions

322 Control ground segment

Average data transfer rate is estimated at 300 MBday (approx 40 kBs onaverage) Data will be stored on board (2 GB capacity required) and will betransferred to a receiving station on the Earth once a day

XPLC

incl Data StorageIF to TMTC Link

HK amp TMTC

to ground

Opt Clock

AOM

MOLO

PHARAO

FCDP

10 GHz

100 MHz

250 MHz

GNSS Rx

MWLTampF Link

uplink Data

UTC(k)

TampF Space vs Ground

TM from GS

Tim

e S

cale

Orbit GNSS Time UTC

TampF Space vs Ground

MampC Time Scale

MampC

MampC

MampC

MampC

Mamp

C

PHARAO vs Opt Clock

PHARAO vs Opt Clock

10 GHz amp Epoch

10 MHz

amp Epoch

Epoch

10 MHz

amp Epoch

Opt Link

Fig 4 Schematic of the payload showing subunits and their signals The results of the frequencycomparison between the two on-board clocks (optical clock and microwave clock PHARAO) areobtained from a combination of the frequency values applied to the frequency shifters AOM (seeFigs 7 and 8) and those output from the frequency control and distribution package (FCDP)as indicated by the arrows ldquoPHARAO vs Opt Clockrdquo The satellite clock to ground clockcomparison results are output from the MWL XPLC is the on-board computer and is involvedin generation of signals for instrument control The optical link shown dashed is a possible add-onproviding an alternative to the MWL GS ground stations MampC monitoring and control HKhousekeeping TampF time and frequency

Exp Astron

4 Main payload instruments

41 Overview

The satellite payload consists of a pair of cold-atom clocks a microwave-optical local oscillator (MOLO) an on-board frequency distribution unit(FCDP) a satellite-to-ground clock comparison unit (MWL) and the onboardcomputercontrol system (XPLC) see Fig 4 Auxiliary units are the posi-tion determination GPSGalileo receiver (GNSS Rx) attitude monitor (startracker not shown) and the corner cube reflector array (CCR not shown)

The science goals require clocks of very high frequency stability on thetimescales of half the orbital period (6 h) as well as perigee passage (sim1000 s)see Fig 2 Only clocks based on cold atoms are suited for the science goalsTable 1 reports values for a subset of atomic clocks that have been investigatednot including clocks that require cryogenic operation or are too complex foruse in an medium-class mission The choice of a clock pair rather than a singleone fulfils several requirements it addresses more science goals improves thevalidity of some science results and it enables calibration accuracy evaluationand redundancy The validity of the science results is enhanced since theclocks differ not only in the employed atoms but also in terms of havingdifferent sensitivity to perturbations by electric and magnetic fields electronicserrors and misalignments A comparison of the results arising from each clockallows for an understanding of measurement errors Both the clocksrsquo and thefrequency comparison unitrsquos performances will be determined after launchand regularly verified during the course of the mission The only reliable

Table 1 Properties of different clock candidates for EGE

Clock Clock Fractional σ = Fractional Sensitivity Figure of Sensitivityatom frequency inaccuracy instability to α Aα merit |Aα| to meq

at 1000 s σ [1016] QCD

Sr+ 445 THz 3 times 10minus17 (p) 5 times 10minus17 (p) 042 08 (p) ndashYb+ (1) 688 THz 3 times 10minus16 (d) 3 times 10minus16 (d) 088 029 (d) ndash

3 times 10minus17 (p) 1 (p)Yb+ (2) 642 THz lt 1 times 10minus17 (p) 1 times 10minus17 (p) minus53 53 (p) ndashSr 435 THz 1 times 10minus16 (d) 1 times 10minus16 (d) 009 009 (d) ndash

lt 1 times 10minus17 (p) 09 (p)Yb 519 THz 7 times 10minus16 (d) 1 times 10minus16 (d) 031 03 (d) ndash

lt 1 times 10minus17 (p) 1 times 10minus17 (p) 31 (p)Cs 91 GHz 2 times 10minus16 (d) 1 times 10minus15 (d) 275 027 (d) 1Yb+ 126 GHz 1 times 10minus15 (p) 1 times 10minus15 (p) 34 034 (p) 1

A relevant figure of merit is the ratio of |AX | and fractional frequency instability σ on the perigeevisibility timescale The Cs clock is the only high performance clock with adequate sensitivity tothe particle masses me mq The indicated Cs clock performance was achieved using a microwavelocal oscillator based on a cryogenic oscillator On EGE the latter is to be replaced by an opticallocal oscillator-based device (MOLO) with similar short-term performance but without the needto operate at cryogenic temperature The clock instability at 10000 s is typically similar to theindicated inaccuracy(p) projected (d) demonstrated (1) quadropole (2) octupole

Exp Astron

way to achieve this is by having at least two clocks on board The on-boardclock-clock comparison provides performance evaluation independent of theground-satellite comparison unit Once the satellite clock performances areestablished the link performance can be established as well

The clocks are similar in basic structure but differ in the detailed implemen-tation They consist of

a source of atomsa preparation subunit which cools the atoms to near-standstill (millikelvintemperature or below)a trapping subunit that confines the atom(s) to a tiny volumean interrogation subunita clock control electronics package

The clocks share a common central unit the optical-microwave local oscil-lator which serves as a flywheel oscillator with outstanding stability on shorttimescales (01ndash10 s) for both clocks It consists of a laser stabilized to an ultra-low loss optical resonator made of ultra-low-expansion glass A microwavesignal is derived from the laser by phase-locking a frequency comb to the laserwave and using a multiple of the repetition rate

While the subunits of any optical clock have a similar technological basis(optics electrooptics acoustooptics thermal control vacuum systems) theydiffer in their specifications Depending on the choice of the atom species de-livery confinement and interrogation methods clocks of different perfor-mance result A number of laboratory optical clock demonstrators have beendeveloped worldwide This ensures availability of the know-how for the indus-trial implementation of the flight models

The first clock is a microwave clock based on the interrogation of slowlymoving ensembles of cold Cs atoms It is essentially based on the PHARAOclock [5 6] developed to engineering model level by CNES in the frame-work of the ESA ISS mission ACES In EGE this clock is upgraded usingthe microwave-optical local oscillator This improves significantly the clockperformance The specification is a relative instability (Allan deviation) of3 times 10minus14 (τ s)minus12 ie at the integration time τ = 1000 s it drops to 1 times 10minus15and to 3 times 10minus16 at 10000 s While this specification is at least a factor 3 lessstringent than for the optical clocks the significantly lower cost of obtaining aflight model and the sensitivity of its frequency to the particle masses makes ita suitable choice

The optical clock baseline instrument is a single-ion optical clock Amongvarious suitable ion species (Hg+ Al+ Sr+ Yb+) [44ndash47] the Yb+ ion hasbeen selected Its breadboard performance is established at the instability levelof 3 times 10minus16 over 1000 s it exhibits a large sensitivity coefficient Aα hassmall sensitivity to magnetic fields and exhibits significant potential for furtherimprovements (use of the octupole transition with sim6 times larger Aα andlower overall instability [48] see below) A trapped Yb+ ion is also a veryrobust system it has been demonstrated that a single Yb+ ion can remaintrapped uninterruptedly for many months Finally Yb+ is the only ion with

Exp Astron

which optical clocks demonstrators have been operated in not a single but twomajor metrology laboratories

Another attractive optical clock is based on ensembles of cold neutral Sror Yb atoms trapped in an optical lattice produced by laser beams This clocktype has a lower instability than a single-ion clock thanks to the large numberof atoms (sim105) used Its development is in progress in several laboratoriesin the world (in part supported by space agencies) and has to date reached6 times 10minus17 instability at 5000 s and 1 times 10minus16 uncertainty [49 50] with potentialfor even higher performance However it is more complex and costly thanthe Yb+ ion clock since the cold atom preparation and trapping subunitsinclude more components require higher laser powers and the respectiveoperating procedures are more complex It is therefore not considered for thepresent mission designed to be developed in the very near future with limitedresources but is an interesting candidate for missions without such constraintsIts estimated physical parameters are indicated in Table 2 for comparison

42 Main instruments

421 Microwave clock

The PHARAO cold atom clock is one of the two atomic clocks of the spacemission ACES managed by ESA It is proposed here to include a secondversion of this instrument as its costperformance ratio is very attractive forthe EGE scientific objectives Its concept is very similar to ground basedatomic fountains but with the major difference of zero-g operation Atomsslowly launched in free flight cross two microwave fields tuned to the transitionbetween the two hyperfine levels of the cesium ground state The interrogationmethod based on two separate oscillating fields (Ramsey scheme) allowsthe detection of an atomic line whose width is inversely proportional to thetransit time between the two microwave cavities The resonant microwavefield at 9192631770 GHz (SI definition of the second) is synthesized startingfrom an ultrastable quartz oscillator (USO) in the FCDP phaselocked to themicrowave output of the microwave-optical local oscillator and stabilized tothe clock line using the error signal generated by the cesium resonator In thisway the intrinsic qualities of the cesium hyperfine transition both in terms ofaccuracy and frequency stability are transferred to the USO In a microgravityenvironment the velocity of atoms along the ballistic trajectories is constantand can be changed continuously over almost two orders of magnitude (5ndash500 cms) Differently from atomic fountain clocks presently operated onground very long interaction times (up to few seconds) will be possible whilekeeping reasonable the size of the instrument

The instrument consists of 4 main elements a laser source which providesthe light required to cool and detect the cesium atoms (Fig 5 left) a frequencychain which generates the 92 GHz microwave signal a cesium vacuum tubewhere the interaction between the microwave field and the atoms occurs

Exp Astron

Table 2 Resources and technology readiness level (TRL) of the EGE spacecraft

Unit Volume Mass Power TRL Comment[l] [kg] [W]

Ion clock 100 70 60Atom source 4Atom preparation 6Atom trap 4Atom interrogation 6Clock control 6

Microwave clock 200 91 110 8 TRL by end 2008Local oscillator

Stabilized laser 18 15 23 6Frequency comb 20 16 38 6

Microwave link 18 18 55 8FCDP 7 8 12 8GPS receiver 2 5 7 9CCR 2 2 0 9Payload sum 367 225 305Thermal control 14 17 30 8OBDH 20 24 38 9Star tracker 1 1 8 9Radiation shield 10 40 0 9Support structure 170 204 0 9Power incl solar arr 50 58 10 9Propulsion 21 25 20 9AOCS 36 43 82 9TMTC 8 10 15 9Satellite bus sum 330 422 203 Flexbus conceptPayload plus bus incl 20 margin 8364 7764 6096Propellant 40Launch adapter 100SC wet masspower incl Margins 9164Lattice clock 135 105 152 Alternative clock

Atom source 4Atom preparation 6Atom trap 4Atom interrogation 6Clock control 6

Also included are the resources for an optical lattice clock based on neutral Strontium atoms

(Fig 5 right) and a computer The engineering model of PHARAO is fullyassembled at CNES Toulouse and undergoes functional and performancetests until the end of 2008 The operating temperature range is 10ndash33 andnon-operating temperatures are minus45 +60 The cesium tube consists in aUHV chamber pumped by getters and a 3 ls ion pump a cesium reservoira microwave cavity coils to provide a uniform magnetic field and three layersof magnetic shields Vacuum windows with fiber-optics collimators enable thetransmission of the cooling and detection light onto the cesium atoms

In autonomous mode PHARAO uses the USO as interrogation oscillatorand will provide a clock signal with fractional frequency stability below1 times 10minus13 (τs) minus12 and inaccuracy near 1 times 10minus16 When using the microwave-

Exp Astron

Fig 5 Left The PHARAO laser source engineering model with cover removed Dimensionsare 530 times 350 times 150 mm3 and the mass is 200 kg Ten polarization maintaining optical fibers(yellow) guide the laser beams to the cesium tube All diode lasers are mounted on Peltier coolersfor temperature regulation within 2 mK (courtesy EADS SODERN) Right The PHARAOengineering model cesium tube without the two external magnetic shields The volume is 990 times336 times 444 mm3 and the total mass is 44 kg

optical local oscillator PHARAO will provide a reduced frequency instabilityof 3 times 10minus14 (τs) minus12 This corresponds to 1 times 10minus15 at 1000 s and 3 times 10minus16

at 104 s

422 Ion optical clock

The 171Yb+ ion clock shown in Figs 6 and 7 provides an atomic frequencyreference based on a single 171Yb+ ion confined within an RF electric end-captrap (similar to a Paul trap) within an ultra high vacuum (UHV) chamber andlaser cooled to sim1 mK close to the Doppler cooling limit As a result the ionrsquosfirst order Doppler motion is completely removed and the ion experienceslittle collisional perturbation from its environment Under these conditionsthe narrow linewidth 2S12 (F = 0 mF = 0) minus 2 D32 (F = 2 mF = 0) quadrupole

Fig 6 Left Simplified level scheme of 171Yb+ showing relevant cooling (blue) auxiliary (red)quadrupole clock (green) and octupole clock (dark blue) transitions δνnat is the natural linewidthof the clock transition Middle close-up of Paul-type RF trap (PTB Braunschweig) the ion istrapped in the hole (diameter 14 mm) Right Complete ion trap with vacuum flange height 10 cm(NPL Teddington)

Exp Astron

935 nm D32-state repumper laser

638 nm F72-state repumper laser

739 nm extended cavity diode laser

AOM

PMT

Ion trap UHV chamber

vacuum pumps

Fibre-free space interface

Mu-metal shielding

EC

T

B

Magnetic field

Temperature

Ion fluorescence Processor

Clock laser frequency step and control

4355 nm

Frequengy doubling II

3695 nm cooling wave

Low-drift cavity

Low-drift cavity 171Yb+ 435 nm ion clock unit

876 nm clock laser wave from MOLO

Frequengy doubling

Fig 7 Simplified schematic of the single-ion optical clock unit AOM acousto-optic frequencyshifter PMT photomultiplier tube EC endcap T temperature stabilization system B magneticfield

clock transition at 4355 nm can be interrogated by spectrally narrow and stableclock laser light A diagram of the simplified energy level scheme for 171Yb+showing the relevant cooling auxiliary and clock transitions is given in Fig 6All these wavelengths are provided by diode lasers operated in fundamentalor frequency doubled mode The clock laser light is obtained from the 871 nmwave provided by the MOLO unit (see Section 423) after frequency-doublingin single pass by a KNbO3 nonlinear-optical crystal to 4355 nm The clocktransition spectral profile is observed by stepping the clock laser frequency bymeans of an acousto-optic frequency shifter (AOM in Fig 7) and recordingthe statistics of the quantum jumps in cooling laser induced fluorescence asa function of this frequency Under clock operation the AOM repeatedlysteps back and forward between the half-intensity points on the transitionprofile monitoring the signal imbalance between these points Any detectedimbalance is servo-corrected to zero by feedback to the AOM

In detail the optical clock architecture will comprise (Fig 7)

bull an RF end-cap trap for ionising and confining a single 171Yb+ ion withinan ultra-high vacuum chamber pumped by a small ion pump and non-evaporable getter pump The trap is driven by means of an ac voltage ofa few hundred volts at a drive frequency of sim10 MHz A small oven withseveral mg of 171Yb isotope is heated to provide a low flux of Yb atomsfrom which single ions can be ionised by electron bombardment from ahot wire filament within the trap potential well Low voltage dc is applied

Exp Astron

to additional electrodes to position the ion precisely at trap potentialcentre in order to minimise the ionrsquos micromotion at the drive frequencyThis needs automatic periodic monitoring and correction especially in theperiod after re-loading an ion into the trap External magnetic field coilsin three orthogonal axes allow the nulling of external fields and settingof a fixed field of sim1 μT The trap and coils are surrounded by mu-metalshielding to minimise external field changes during the orbit A level oftemperature control of 1 K of the region surrounding the trap is requiredfor maintaining the blackbody-induced frequency shift uncertainty withtemperature at the 10minus17 level

bull a laser platform to provide frequency doubling of an extended cavity739 nm diode laser for Doppler cooling of the ion on the 2S12(F = 1) minus2 P12 (F = 0) dipole transition at 3695 nm Light from an amplified diodelaser device frequency-doubled in a periodically poled LiTaO3 crystal insingle-pass will be used Typical laser powers used for driving the coolingtransition below saturation are about 2 μW for a beam waist in the trapof about 50 μm By modulation of the injection current of the 739 nmdiode laser at 147 GHz a sideband is generated that excites the 2S12 (F =0) rarr2 P12 (F = 1) ldquorepumperrdquo transition in order to avoid optical pump-ing between the ground hyperfine states

bull The platform also houses the auxiliary lasers at 935 and 638 nm The935 nm diode provides repumping of the ion from the 2 D32(F = 1)metastable level after occasional branching decays to this level during thecooling sequence The 638 nm diode allows fast recovery of the ion fromthe very-long-lived 2F72 metastable state after very occasional collisionaldecay to that state Currently extended cavity lasers are forseen butdistributed-feedback lasers may be available in the near future

bull a high NA lens imaging system and photomultiplier detection system torecord the statistics of 369 nm fluorescence quantum jumps as a functionof clock laser frequency step

bull a fibre system to deliver the various cooling auxiliary and clock light fromsource to trap making use of achromatic doublets where necessary at thefibre-free space interface for launching into the trap

bull a monitoring and control processor which provides the primary coolingand clock laser pulse magnetic field and detection sequencing to observeand lock to the ion clock transition frequency The processor also monitorsfrequency and amplitude data necessary to determine normal laser andion operational conditions and initiate resetting and recovery algorithmswhere necessary and laser unit failure The actual magnetic field presenton the atomic volume is determined and actively stabilized so that aninaccuracy at the level of 1 times 10minus17 can be achieved

bull a redundancy level of at least two units for cooling clock and repumperlasers plus similar for frequency doubling crystals All redundancy unitsfor each wavelength will be fibre multiplexed as standard allowing redun-dant unit activation on determination of prior unit failure

Exp Astron

For an optimized quantum-projection-noise-limited performance of thesingle-ion clock Ramsey interrogation with a cycle time equal to the lifetimeof the metastable D32 level of 50 ms will be performed leading to an Allandeviation of about 27 times 10minus15 (τ s)minus12 This is sim45 times 10minus17 at 1 h satisfyingrequired stabilities for the science objectives Current laboratory results are astability of 1 times 10minus15 at 30 s averaging down to 3 times 10minus16 at 1000 s (PTBGermany) Also comparisons between two independent traps showed agree-ment at the level of several parts in 10minus16 over a series of eight measurementruns

The 171Yb+ ion clock offers a second option namely the 2S12ndash2 F72 octu-pole transition at 476 nm (see Table 1 and Fig 6) [48] It has an extremelylong upper state lifetime low electric quadrupole and second order Zeemansystematic shifts and therefore is a candidate for an even higher performanceclock if a clock laser system with lower instability than the above is providedApart from the clock laser it shares the laser system of the quadrupoleclock Thus with moderate additional resources the Yb+ ion apparatus couldprovide two optical clocks in a single package and specifically perform directfrequency comparisons between the quadrupole and the octupole transitionwith a large sensitivity Aα

The ion clock will operate under microprocessor control providing thecontrol algorithms for the integrated pulse sequencing for cooling repumpingclock interrogation magnetic field switching cooling fluorescence monitoringThe controller will also provide error flags for critical processes and systemreset and correction where necessary This will include ion re-loading andmicromotion reduction algorithms Automatic monitoring of laser power andspectral quality with signal re-optimisation or laser failure determination isrequired Redundant laser units will be pre-aligned and multiplexed into thefibre delivery to the trap so that redundant units can be readily activated in thecase of diode laser failure

423 Microwave-optical local oscillator (MOLO)

This oscillator provides two ultrastable frequency outputs one in the opticalregion the other in the microwave region the two frequencies being coher-ently related The MOLO is composed of three subunits the clock laser thereference cavity and the frequency comb see Fig 8

The clock laser subunit delivers the light for the excitation of the opticalclock transition It is generated by an extended-cavity diode laser at a wave-length of 871 nm Most of the output wave of this laser is sent to the ionclock The laser subunit containing two lasers for redundancy will require3 kg 3 l 15 W The frequency instability of the clock laser must be superiorto the minimum clock instability 27 times 10minus15 (τ s)minus12 for τ up to the timeconstant (sim10 s) of the servo system that locks the laser to the ionrsquos resonancesignal In order to achieve this the laser frequency is stabilized to the reference

Exp Astron

871 nm clock laser

EO

Detector and servo

AOM Femtosecond

frequency comb

To optical clock 10 GHz To FCDP

Microwave-optical local oscillator

ULE cavity

Fig 8 Microwave-optical local oscillator (MOLO) consisting of clock laser reference ULE cavityand femtosecond frequency comb (OFC) Radiofrequency inputs and outputs are not shown EOelectro-optic modulator ULE ultra-low expansion AOM frequency shifter

cavity where the PoundndashDreverndashHall method is used About 10 μW of thelaser output are used for this

The system will exhibit sim04 Hz (sim1 times 10minus15 relative) linewidth and afrequency drift lt01 Hzs The cavity consists of two high-reflectivity mirrorsoptically contacted to a cylinder of diameter 70 mm length 150 mm made ofULE (a glass exhibiting ultra low thermal expansion coefficient) with a cavityfinesse of about 300000 Residual vibrations of the satellite (arms lt 1 times 10minus6 g)should not be a limiting factor to the desired linewidth for safety marginan optimized cavity support can reduce the sensitivity to accelerations Foracoustic and thermal isolation the cavity resides inside an aluminium vacuumchamber equipped with a small (3 ls) ion-getter pump Two stages of polishedaluminium shields are implemented around the cavity which are activelytemperature stabilized so that the cavity is precisely kept at the zero crossingtemperature of the ULE thermal expansion coefficient Heat application andremoval is by thermoelectric elements between the shield and the vacuumchamber Additionally the entire vacuum chamber is temperature-stabilized

During transportation and launch the cavity will be rigidly fixed insidethe vacuum chamber by pressing on it from the sides by piezo-mechanicalactuators allowing accelerations of several g After bringing the satellite onthe orbit the actuators will be released

The optical setup for frequency stabilization will be a classic Pound-Drever-Hall scheme in reflection using an electro-optic modulator (EO) Couplingof the clock laser radiation to the cavity is via single-mode polarization main-taining optical fiber The fiber in- and out-couplers are mounted on miniaturepiezo-motor driven multiaxis translation stages Microprocessor control ofthe translation stages performs a laser-to-fiber incoupling stabilization andlaser-to-cavity mode-matching optimization (positioning inaccuracy lt10 μmalignment inaccuracy lt10 μrad)

The frequency stability of the clock laser will be limited by thermal 1f noisein the mirrors with an fractional Allan deviation slightly below 1 times 10minus15 from1 s to a few 10 s Resources including control electronics are 15 l 12 kg 8 W

Exp Astron

A femtosecond fiber laser based optical frequency comb generator (OFCFig 8) will be used for the clock frequency comparisons and to obtain aspectrally ultrapure microwave local oscillator output for the operation ofPHARAO for the microwave link and for the GNSS receiver The fem-tosecond laser is based on erbium-doped fiber pumped by fiber-coupled laserdiodes It emits 100 fs pulses at a repetition rate of 250 MHz with an averagepower of 100 mW The emission spectrum is centered around 1560 nm Spec-tral broadening to more than one octave is obtained via nonlinear interactionin a microstructure fibre so that the range from about 1050 to 2100 nm iscovered

Second harmonic generation of a portion of the broadened spectrum allowsobtaining the required light for the measurement of the clock laser frequencyat 871 nm The OFC will be referenced to the clock laser frequency (afterintroduction a controllable shift via the frequency shifter AOM) by makingthe carrier envelope offset frequency vanish so that the repetition rate fr isa high-order subharmonic of the optical clock reference frequency f1 = m1 frThe link between the optical and the microwave domain is then simply thedistribution of the repetition rate frequency or more conveniently a high(order 40) harmonic of the comb repetition rate 10 GHz

It has been demonstrated by several groups that frequency comparison andfrequency division with an OFC can be performed to very high accuracy andstability beyond the limitations imposed by the optical clocks themselves Forexample a fiber laser OFC with an instability (Allan deviation) of 6 times 10minus17

in 1 s has been demonstrated [51] OFCs based on fiber technology possessgood long-term reliability and high power efficiency The projection for a spacequalified OFC based on current industrial terrestrial technology is 16 kg mass20 l volume and 38 W power requirement

424 Frequency comparison and distribution package (FCDP)

This package distributes the ultra-stable 10 GHz reference frequency outputof the MOLO to the MWL performs frequency comparisons between MOLOand PHARAO and generates a variety of different reference signals for theindividual instruments including the 250 MHz reference clock to the OFCAll signals are coherent to the MOLO In the FCDP a USO is locked to theMOLO signal by a digital PLL A divider chain generates timing pulses (1 pps1000 pps etc) and keeps a continuous local time scale From the digital phasecomparison data the phase noise density spectrum between the USO and theMOLO is available by telemetry (TM) to assess the overall frequency purityin-flight

As the FCDP distributes the full performance reference from the MOLOdirectly to the MWL there is minimum impact of the FCDP on overall clockcomparison performance The USO-phase lock loop uses a similar designas for ACES-FCDP The digital PLL with linear phase detector allows theUSO to be slightly offset from its reference which is selectable by groundcommand This results in an accurate on-board time scale despite any offset

Exp Astron

of the MOLO reference frequency which can be synchronised to UTC Thefrequency synthesis chain re-uses modules developed for ACES-FCDP andtheir frequency inaccuracy Expected relative uncertainty (Allan deviation)of the frequency comparison between two signals provided to the FCDP is47 times 10minus17 at 1000 s 95 times 10minus18 at 10000 s

During non-availability of the MOLO ie during system start-up andcontingency conditions the USO runs as flywheel (ie it is unlocked withinstability lt15 times 10minus13 from 1 to 100 s) It will still provide useable butless accurate references to the instruments including the reference to OFCCommon view ground-to-ground time- and frequency comparison and supplyof the remaining instruments remain possible at the USO provided accuracy

Mass volume and power consumption are 8 kg 7 l 12 W Controldata rateis 10 kBday 50 kBday (FCDP alone)

425 Microwave link (MWL)

The microwave link is a two-way triple-frequency microwave link (Fig 9) Itis designed to perform accurate frequency comparison as well as time-transfer

Ku-Band Up-linkPower Tx 5 WCarrier 13475 GHzPN-Code 100 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

S-Band Down-linkPower Tx 1 WCarrier 2248 MHzPN-Code 1 MChips1pps 1 time marker sData 5 kBits

Ku-Band Down-linkPower Tx 1 WCarrier 1470333 GHzPN-Code 100 MChips1pps 1 time marker sData 5 kBits

EGE SC

S

Ku

KaKa-Band Up-linkPower Tx 5 WCarrier 345 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

Ka-Band Down-LinkPower Tx 1 WCarrier 32 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 5 kBits

GS Antenna diameterEGE terminal 1 mACES terminal 06 m

Fig 9 EGE microwave two-way link concept and characteristics See text for a description Thelink will be able to perform frequency comparisons with up to ten ground stations simultaneously(only one active comparison is shown) pps pulse per second Tx transmitter SC spacecraft

Exp Astron

between on-board and on-ground clocks Two-way operation allows for com-pensation of the effects due to spacecraft location and its motion relative toground Its two bi-directional links operate in Ka-band (32345 GHz) andKu-band (147135 GHz) The high carrier frequency of the up- and down-links allows for a noticeable reduction of the ionospheric delay A third down-link channel is in S-band (22 GHz) and is used to determine the ionospheretotal electron content MWL uses a combination of carrier phase and codephase data to measure with high precision the absolute ionospheric delayas well as ionospheric delay variations during a ground contact The useof three frequency bands allows to resolve third order ionospheric termsKey modulation parameters are Ka band Tone 250 MHz Ku band PN100 MChips S-band 1 MChips

The EGE-MWL concept is a considerably improved version of the ACES-MWL Microwave carrier generation is directly from the 10 GHz microwaveoutput available from the MOLO and having exceptionally low close-in phasenoise without intermediate frequency dividers or other complex synthesiserstages commonly used in RF transponders The high frequency carriers arekey to achieve the desired frequency comparison accuracy

The Ka-band carrier is generated by tripling the FCDP-generated signalThe Ku-band is subsequently generated by division by two using a verystable regenerative divider Finally the assigned radio frequency is obtainedby adding a low frequency signal using an image reject mixer A mixedanalog-digital receiver design will enhance its performance by use of a flexibledigital signal processing and digital control loops The number of simultaneousground contacts is increased from four (as in the ACES mission) to ten or moreusing the digital concept while the number of physical receiver channels perband is reduced from four to two one for operation to ground and one for in-flight calibration The concept allows adaptation of receiver loop parametersto optimise for the different portions of a highly elliptic orbit while reducingsystematic uncertainties due to high Doppler and Doppler rate Real-timedata exchange to ground terminals is by additional data modulation on all RFsignals The spacecraft carries two different small antennae for perigee and forapogee with gains of 8 dBi 20 dBi beamwidths 100 (full earth coverage) 14respectively power is 1 W for each

The MWL end-to-end performance between the attached clocks includingspace- and ground terminals is predicted to be a relative Allan deviation of1 times 10minus16 at 1000 s (perigee) and 1 times 10minus17 at 10000 s (apogee) Thesevalues are based on conservative implementation losses and other instrumentallosses assumed as 6 dB at apogee and 10 dB at perigee including additionallosses due to atmosphere An attitude knowledge of 1 is required

The MWL link will also support precise orbitographyndashcomplementary tothe GNSS receiver Generation of ranging data is simultaneous to two-waytime transfer and is available under all weather conditions Orbit determina-tion accuracy is expected to be better than 1 cm well matching EGE clockrequirements The value is expected based on demonstrated results fromother missions (METEOR ERS-2 TOPEXPOSEIDON) Laser ranging to

Exp Astron

the CCR array is used a few times during the mission to calibrate the MWLranging

In case of non-availability of the ultrapure signal from MOLO an additional10 GHz output is available from the FCDP to support ground-to-ground clockcomparison in common-view mode even after the nominal mission periodwhen the neither the clocks nor MOLO may be available any longer Basedon the PRARE experimentrsquos 12 years of operation this can be assumed forthe FCDPMWL combination as well The potential extended operational pe-riod makes the proposed configuration particularly attractive to metrologicallaboratories that rely on availability of long-lasting data in support of thegeneration of UTC

The resources of MWL are as follows mass 18 kg volume 18 l power55 W TC 2 kBday TM 120 MBday

426 Auxiliary units

The star camera assembly on the EGE spacecraft will provide real-timeattitude data primarily for the orientation of the spacecraft (with accuracy of1 in real-time) and the proper reduction of the GNSSSLR data

A GNSS receiver is used for the precise orbit determination time andfrequency transfer coarse positioning (lt1 m) for real time use by the Attitudeand Orbit Control as well as the back-up for time tagging of all payloaddata For the time and frequency transfer the receiver is nominally using theonboard frequency reference The receiver assembly includes three specialantennas By the year 2015 more than 100 GNSS satellites are expected tobe available for precise orbit determination permitting orbit inaccuracy of 1ndash2 cm RMS for altitudes up to approx 10000 km At apogee a combinationof GNSS signals and MWL ranging will be used with expected inaccuracyof 10 cm

5 Basic spacecraft specifications

The mission foresees a single spacecraft with a mission duration limited toapprox 3 years for cost reasons Attitude control along three axes withless than 1 error on a timescale of 1000 s is required The control can beperformed with cold gas thrusters magnetotorquers or reaction wheels Theselection will be done on the basis of the vibrations induced by these systems(desired level arms lt 10minus6 g)

Cold gas thrusters or monopropellant thrusters will be required to maintainthe orbit parameters during the mission by small orbit correction manoeuvresand will also guarantee a proper tuning of the orbit parameters after the releasefrom the launcher The ground segment (telemetry control and data transfer)could be via non-ESA stations user-operated to save costs

Further resources are indicated in Table 2

Exp Astron

The satellite bus design drivers are nadir-pointing microwave antennasthermal stability for the payload instruments no strong magnetic sources lowmicrovibrations harsh radiation environment due to crossing of the radiationbelts once per orbit power demand by science payload (with the possibility toreduce power demand by stand-by of some components during specific missionphases eg eclipse) mass limit due to required high altitude with low costlaunch

The satellite bus design can be based on established platforms such asFLEXBUS or PRIMA used on several past missions Their design philosophydiffers one having body-mounted solar arrays the other wings The latter canprovide larger amounts of power The vibration level induced by the panelswhen they are fixed and the duration of solar array drive periods per orbitwill have to be considered in the selection The FLEXBUS concept was usedfor a mission already requiring a low level of residual accelerations and theplatform is optimised for this parameter

For most of the satellite components one can consider using compo-nents and electronic units from navigation or communication satellites whichtypically cruise at altitudes between 15000 and 25000 km altitude LISAPathfinder which will cross the radiation belts during spiralling to theLagrange point will face similar problems concerning radiation as EGE as wellas XMM which is in a Molniya-type orbit with similar radiation characteristics

The instruments will therefore be located in the satellite structure such asto realise ionizing radiation shielding by the spacecraft structure and non-sensitive equipment to the maximum extent possible the need for additionalradiation shielding (including trapped proton radiation) for especially sensitiveparts of the instruments must be determined by ground tests It should bementioned that an orbit with higher perigee and less radiation dose is also anoption as it leads to a tolerable reduction of the science goals

The required plusmn3C temperature stability at the interface to the payloadenclosure is feasible The thermal control will employ the usual means (anenclosure with high thermal inertia heaters and a suitable radiator) Thelocation of the radiator will need a special study given the non-standard orbit

Magnetic field present during the mission can affect the properties of theonboard clock Using SPENVIS the magnetic field has been calculated alongthe spacecraft orbit the field intensity presents maxima around 02 G anda minimum value of the order of 10minus3 G As discussed above this is not alimitation to the instrument performance

6 Discussion and conclusion

The accuracy of the measurements depends crucially on the stability andaccuracy of the clocks While the performance of ground clocks (even thoseeventually deployed in new ground stations dedicated to EGE) will likelyimprove significantly from the present level the space clock design will beldquofrozenrdquo approx 10 years before data acquisition Thus their performance

Exp Astron

level will be significantly below the future ground station clock performanceAs specification we have chosen a performance level that has been demon-strated with laboratory breadboards recently We see no fundamental reasonthat this laboratory performance cannot also be achieved in a space ion opticalclock The experience with the PHARAO cold atom clock has shown thatan engineering model of a laser-based cold atom clock has a performancesimilar to that of a laboratory prototype The perturbation of the ion clock fre-quency due to magnetic fields along the orbit can be modelled and suppressedwith shielding and the residual effects measured and compensated for Themoderate temperature stability (plusmn 3C) to be provided in the experimentalbay can be achieved with thermal shields Accurate temperature controlof particular units is done by active stabilization of the units themselves

One important aspect is radiation which may affect all optical and trapcomponents as well as the lasers and reference cavity The radiation envi-ronment on the EGE orbit is significantly harsher than the near-earth ISSorbit on which ACES will operate The radiation effects must therefore becharacterized comprehensively with appropriate tests and then the radiationshield for the EGE instruments designed

In summary the Einstein Gravity Explorer mission (EGE) is a satellitemission aimed at a precise measurement of the properties of space-time usingatomic clocks By testing one of the most fundamental predictions of EinsteinrsquosTheory of General Relativity the gravitational redshift and thereby searchingfor quantum effects in gravity it explores one of the most important frontiers infundamental physics The EGE mission reaches beyond fundamental physicsproviding an important spin-off in geophysics Furthermore the technologicalaspects of the mission are ground-breaking for future time-keeping and ref-erence frame systems in space and thus are also expected to be of wide useto science and technology in general The payload of the mission has beenconceived as a compromise between cost and performance needing to fit thecost envelope of an ESA medium-class mission1

Acknowledgement We thank S Cesare for helpful information

References

1 Will CM The confrontation between general relativity and experiment Living Rev Relativ9 3 (2006)

2 Ni W-T Empirical foundations of the relativistic gravity Int J Mod Phys D 14 901 (2005)doi101142S0218271805007139

1Finally we point out that a spacecraft with a significantly reduced and thus lower-cost payload (noclocks only MOLO FCDP MWL) on a circular orbit of appropriate height could be used for along-duration earth science mission dedicated specifically to the measurement of the geopotential

Exp Astron

GR General RelativityGRACE gravity recovery and climate experimentISS international space stationLPI Local Position InvarianceMWL microwave linkMOLO microwave-optical local oscillatorMampC monitoring and controlOFC optical frequency combPHARAO Projet drsquohorloge atomique par refroidissement drsquoatomes en

orbitePLL phase-locked loopPTB Physikalisch-Technische BundesanstaltPRARE Precise Range And Range-Rate EquipmentQCD quantum chromodynamicsRF radiofrequencyRx receiverSLR satellite laser rangingSME standard model extensionTC telecommandTM telemetryTampF time and frequencyULE ultralow-expansion glassUHV ultra-high vacuumUTC universal time coordinatedUSO ultrastable quartz oscillatorXPLC onboard computercontrol system

1 Introduction

Einsteinrsquos theory of General Relativity is a cornerstone of our current descrip-tion of the physical world It is used to describe the motion of bodies rangingin size from satellites to galaxy clusters the propagation of electromagneticwaves in the presence of massive bodies and the dynamics of the Universeas a whole In general the measurement of general relativistic effects is verychallenging due to their small size [1 2] Thus only few of its predictions havebeen tested with high accuracy for example the time delay of electromagneticwaves via the Cassini mission [3] and the existence of gravitational waves byobservation of binary star systems The accuracy of the determination of theseeffects is at the 20 ppm level

The situation is similar for one of the most fascinating effects predictedby General Relativity and other metric theories the slowing of clocks ina gravitational field (gravitational time dilation gravitational redshift) Thegravitational redshift was measured with 7 times 10minus5 relative inaccuracy in the1976 Gravity Probe-A experiment [4] by comparing a ground clock with asimilar clock on a rocket as the height changed The most performing clocks

Exp Astron

available at the time hydrogen masers were used for this experiment TheACES (Atomic Clock Ensemble in Space) mission planned to fly on the ISS in2014 seeks to improve the determination by a factor 25 by using a cold atomclock (PHARAO) and a hydrogen maser [5 6] Recent progress on cold atomclocks in the optical domain and in optical technology enable performing thisfundamental test with orders of magnitude better sensitivity

Although it has been very successful so far General Relativity as well asnumerous other alternative or more general theories proposed in the courseof the development of theoretical physics are classical theories As such theyare fundamentally incomplete because they do not include quantum effectsIn fact it has not yet been possible to develop a theory of gravity that includesquantum mechanics in a satisfactory way although enormous effort has beendevoted to this goal Such a theory would represent a crucial step towards aunified theory of all fundamental forces of nature Several approaches havebeen proposed and are currently under investigation (eg string theory) Afull understanding of gravity will require observations or experiments thatdetermine the relationship of gravity with the quantum world This topic isa prominent field of activity and includes the current studies of dark energy

The Einstein Gravity Explorer is a fundamental physics mission dedicatedto this question its primary task is to measure the gravitational frequency shiftwith unprecedented accuracy The EGE mission uses a satellite on a highlyelliptic orbit (see Fig 1) The ratio of the frequencies of two on-board clocksand the ratios between satellite and ground clocks are the main observables

Fig 1 General concept of the EGE mission Clocks on the satellite and on ground are intercom-pared as the satellite orbits the Earth on a highly elliptic orbit

Exp Astron

These can be interpreted toward tests of General Relativity (GR) and ofcompeting metric theories of gravity and as tests for the existence of newfields associated to matter (see Section 3) The outcome of the mission will beeither a confirmation of GR and metric theories within the accuracy providedby the instruments or the discovery of a deviation In the latter case themission would provide a first indication of the breakdown of current (classical)gravitational physics theories and could pave the way towards a unified theoryof all forces

2 Scientific objectives

21 Summary of science objectives

211 Fundamental physics science objectives (1 ppm = 1 part in 10 61 ppb = 1 part in 10 9)

Primary objectives

High-precision measurement of the earth gravitational frequency shift at25 ppb accuracy a factor 3000 improvementFirst precise measurement of the sun gravitational frequency shift at 1 ppmlevel a factor 2 times 104 improvementTest of space and time variability of fundamental constants with accuracy30 ppbFirst search for neutron scalar charge with accuracy 30 ppbTest of anomalous coupling of matter to the Standard Model quantumfields

Secondary objectives

Tests of Lorentz Invariance in the matter and photon sector (factor 20improvement)Contribution towards establishing a new definition of the unit of time

212 Spin-off to other fields (outside fundamental physics)

Establishment of a new approach to the determination of the geopotentialwith geoid height resolution of a few cmComparison of distant terrestrial clocks at the 10minus18 levelDemonstration of clock and link technology for future applications eg inprecision spacecraft navigationDemonstration on high performance real-time range determination withprecision 25 times (based on code phase) to 1000 times (based on carrier

Exp Astron

phase) higher compared to systems onboard previous missions (ERS-2TOPEX POSEIDON)Comparison of three different orbit determination systems Laser-Ranging μm-precision microwave ranging GPS-based orbit deter-minationAtmospheric signal propagation study precise determination of thirdorder ionospheric termMonitoring stability of the GalileoGPSGlonass time scales

22 Gravitational frequency shift measurement and search for new physics

The primary science goal of EGE is a measurement and test of the gravita-tional frequency shift in the weak-field limit as predicted by metric theoriesof gravitation These theories also predict that the shift is independent of thetype of clock used to measure it The comparison of the frequencies ν oftwo identical clocks 1 2 operating at different locations x1 x2 yields a fre-quency ratio

v2(xprime)v1

(xprime) sim= 1 + (U (x2) minus U (x1))

c2 (1)

Here νi(xprime) is the frequency of clock i located at xi as observed (measured) ata particular location xprime where the comparisons are performed The comparisonlocation can be arbitrary but will often be x1 or x2 or both U is thegravitational potential U(x) = minusGM|x| in case of a spherically symmetricbody of mass M In short time runs (or clocks tick) more slowly the closer thesystemclock is to a massive body Stationary clocks and observers and weakgravitational fields have been assumed in Eq 1

Through its measurement EGE searches for a possible violation of the grav-itational frequency shift A violation may be described phenomenologicallyby a dependence of one or more fundamental constants on the gravitationalpotential X = X(Uc2) where X is a generic dimensionless fundamentalconstant or a dimensionless combination of fundamental constants Such adependence is a violation of the principle of Local Position Invariance (LPI)which states that the outcome of local nongravitational experiments is inde-pendent of where and when in the Universe they are performed If a constantdepends on the gravitational potential then the results of those laboratory-type (ie local) experiments that depend on that constant will also depend onthe location namely on the location of the laboratory with respect to the massdistribution generating the potential and will violate LPI

LPI is one of the three cornerstones of the Einstein Equivalence Principle(EEP) the others being the Weak Equivalence Principle and Local LorentzInvariance It is a common belief that the EEP may not be absolutely valid -therefore spatial dependence and time dependence of fundamental constantshave become a very intense research topic in the last decade with experimentsundertaken worldwide to search for violations of various aspects of EEP For

Exp Astron

example extensive studies of atomic and molecular spectra of gas clouds in thedistant Universe are currently undertaken to search for a difference of theseparameters compared to todayrsquos values (see eg [7ndash11] and references therein)Also some differences in Big Bang nucleosynthesis data and calculations canbe naturally explained by a variation of fundamental constants [12 13]

The EGE mission is thus a probe of possible space-dependence of fun-damental constants The search may help to point the way toward a unifiedtheory of all forces reveal hypothetical extra dimensions in our Universe or theexistence of many sub-Universes with different physics chemistry and biologyIndeed theories unifying all forces applied to cosmology suggest space andtime dependence of the coupling constants [14ndash16] Another argument forspatial variation of fundamental constants comes from the anthropic principleThere must be a fine tuning of fundamental constants which allows humans andany life to appear Life in its present form is not consistent with fundamentalconstants whose values only slightly differ from the actual ones This finetuning can be naturally explained by the spatial variation of the fundamentalconstants we appeared in the area of the Universe where the values of thefundamental constants are consistent with our existence

In the Standard Model there are three main fundamental dimensionless pa-rameters determining the structure and energy levels of stable matter (atomsmolecules) the electromagnetic fine structure constant α and the two ratiosof electron mass me and light quark mass mq to the strong QCD energy scaleQCD Xeq = meqQCD The fundamental masses me and mq are proportionalto the vacuum Higgs field which determines the electroweak unification scaleTherefore the constants meqQCD can also be viewed as the ratio of the weakenergy scale to the strong energy scale

How can a space-time variation of the fundamental constants and a depen-dence on the gravitational potential occur Light scalar fields very naturallyappear in modern cosmological models affecting α and meqQCD Oneof these scalar fields is the famous ldquodark energyrdquo which causes acceleratedexpansion of the Universe Another hypothetical scalar field is the dilatonwhich appears in string theories together with the graviton Cosmologicalvariation of these scalar fields should occur because of drastic changes ofthe matter composition of the Universe During the Big Bang nucleosyn-thesis the Universe was dominated by radiation then by cold dark matterand now by dark energy Changes of the cosmic scalar field ϕ0(t) lead tothe variation of the fundamental constants X(ϕ0) Massive bodies (galaxiesstars planets) can also affect physical constants They have a large scalarcharge S = sp Z + se Z + sn N where Z is the number of protons and elec-trons and N is the number of neutrons sp se sn are the scalar charges ofthe proton electron and neutron respectively There is also a contributionof the nuclear binding energy (scalar charge of virtual mesons mediatingnuclear forces)

The scalar charge produces a Coulomb-like scalar field ϕs = SR where Ris the distance from the charge The total scalar field is ϕ = ϕ0 + ϕs therefore

Exp Astron

we may have a variation of the fundamental constants inversely proportionalto the distance from massive bodies

X (ϕ) = X (ϕ0 + ϕs) = X (ϕ0) + δX (R)

δX (R) = (dX

dϕ

)S

R

where a nonzero δX violates LPI and a small ϕs was assumedThe gravitational potential U is essentially proportional to the number of

baryons Z + N and inversely proportional to the distance R U = -GMR

Therefore the change of the fundamental constants near massive bodies canbe written

δX

X = K (X i) δ(Ui

c2

) (2)

where i refers to a particular composition of the mass M and where thecoefficients K(X i ) = minus(dXdϕ) c2SiGMi are not universal but are expectedto be different for eg the Sun and the Earth K(X Earth) = K(X Sun)Indeed the Sun consists mostly of hydrogen therefore the Sunrsquos scalar chargeis SSun Z (sp + se) The Earth contains heavier elements where the numberof neutrons exceeds the number of protons N 11 Z Therefore the scalarcharge of the Earth SEarth Z (11 sn + sp + se) and is sensitive to the neutronscalar charge (and also to the contribution of the nuclear binding) in contrastto the Sun

The most sensitive experiments suitable for a search of a space-time de-pendence in local or near-local experiments (such as satellite experiments) areclockndashclock comparisons In one type of terrestrial local experiment dissimilarclocks located nearby are compared for a duration spanning at least one yearDuring such periods the earth moves periodically closer and further awayfrom the sun so that the solar gravitational potential on Earth is modulatedThis enables a search for a dependence of the fundamental constants on thesunrsquos scalar charge [17 18] These experiments are able to set limits to K(XSun) for X = α meqQCD at the level of 10minus6 [12 13]

The proposed EGE satellite experiments are complementary to terrestrialexperiments allowing to set limits to K(X Earth) and combining with theresults on K(X Sun) from terrestrial comparisons to set limits on the depen-dence of the fundamental constants on the neutron scalar charge sn

In summary when the frequencies of two identical clocks (of nominalfrequency νa) that are located at different positions x1 x2 are comparedby measuring their frequency ratio at some position xrsquo the result may bewritten as

va2(xprime)

va1

(xprime) = 1 +

[1 +

sum

XAXa K (X i)

](Ui (x2) minus Ui (x1))

c2 (3)

The sensitivity factor AXa = X (d(ln νa)dX) is the sensitivity of the clockfrequency to a particular constant X and can be calculated using atomic and

Exp Astron

nuclear theory (see eg [12 13]) The measurement of this ratio which issensitive to K represents a test of General Relativityrsquos redshift

When the frequencies of two co-located but dissimilar clocks a b (ofnominal frequencies νa0 and νb 0) are compared in-situ at two locations x1 x2

(LPI test) the result is

[va1 (x1)

vb 1 (x1)

] [va2(x2)

vb 2 (x2)

] = 1 +[X

(AXa minus AXb

)K (X i)

](Ui (x2) minus Ui (x1))

c2 (4)

The result of the EGE mission may be given as a limits for (or nonzerovalues of) K(X Earth) for the three fundamental constants X = α meQCDand mqQCD

23 Measurement approaches for the gravitational frequency shift

The gravitational potential experienced by the ground and satellite clocksincludes the contributions from the earth and the sun (the contribution fromthe moon is an order smaller and may be neglected for the purposes of thediscussion but will be taken into account in data analysis)

Nonzero gravitational frequency shift signals (Eq 3) can be obtained inthree ways Method (1) below is the most sensitive and represents a strongreason to employ an optical clock which on short averaging times is ableto provide a higher frequency stability than a microwave clock Method (2)and (3) are less accurate but are complementary in terms of measurementapproach Finally method (3) addresses the solar gravitational frequencyshift while (1) and (2) measure the terrestrial effect The performance of theinstruments assumed in the following is shown in Fig 2

(1) Repeated measurements of satellite clock frequency variation betweenapogee and perigeeIn this measurement approach the difference (or ratio) between earthndashsatellite clock frequency ratio at perigee and at apogee is consideredThis difference is measured at each perigee and apogee passage andaveraged over the mission duration Systematic shifts (if not correlatedwith orbital motion) are expected to average out leading to a gain insensitivity of up to n12 where n = number of orbits This mode usesthe stability properties of the satellite clocks on the timescale fromperigee to apogee rather than their accuracy Assuming a specificationof le 3 times 10minus16 satellite optical clock fractional frequency instability forintegration times between 1000 and 25000 s n sim= 1000 comparisons(ie most orbits of the 3-year mission) and a perigee-apogee potentialdifference (U(apogee) minus U(perigee))c2 = Uc2 sim= 4 times 10minus10 allows a

Exp Astron

Fig 2 Performance of clocks and links on EGE Curve ldquooptical clocks EGErdquo is the specificationCurve ldquooptical clocks EGE (goal)rdquo refers to expected near-future performance of an opticalclock based on neutral atoms ldquoUpgraded PHARAOrdquo refers to the PHARAO microwave clockusing the high-quality microwave-optical local oscillator The microwave link (MWL) performanceexhibits a discontinuity since common-view contact time of ground stations with EGE is limited toseveral h The different slopes arise because on short time scales only white phase noise is relevantwhile on long time scales flicker noise also contributes The reproducibility is taken into account

measurement of the redshift at 25 ppb accuracy The ground clocks towhich the satellite clocks are compared to at perigee and at apogee willbe located in different ground stations The inaccuracy of these groundclocks must therefore be sufficiently smaller than 1 times 10minus17 in order notto contribute to the 25 ppb error Current state-of-the art inaccuracy is2 times 10minus17 and is likely to decrease further

(2) Absolute comparison between ground and satellite clocks with the satel-lite at apogeeHere the terrestrial gravitational potential difference is maximumU(apogee) minus U(perigee)c2 sim= 65 times 10minus10 Assuming a space optical clockinaccuracy of 1 times 10minus16 and a ground clock inaccuracy a factor 3lower (current state-of-the-art) allows a measurement of the redshift at150 ppb This measurement takes advantage of the repetitive nature ofthe orbit only to a small extent and relies on the accuracy of the opticalclocks in particular of the space clocks It is thus complementary to (1)albeit less accurateIn (1) and (2) the dominant contribution to the redshift is from Earthrsquosgravitational potential The contribution from the solar potential isstrongly reduced due to the fact that satellite and earth are an interactingsystem in free-fall toward the Sun [19] and can be modeled

(3) Comparisons between terrestrial clocksThe frequency comparison link to the EGE satellite allows terrestrialclock comparisons by a common-view method whereby the frequency

Exp Astron

signals from two distant ground clocks are sent to the satellite andcompared there The frequency difference between two identical groundclocks is determined if systematic instrumental errors are negligible bythe difference in gravitational potential at the two clock locations Thispotential difference includes a terrestrial and a solar contributionThe solar contribution varies according to the orientation of the clockpair with respect to the Sun with a 1 day-period For two clocks at asurface distance of sim6000 km and suitable orientation with respect tothe sun the normalized modulation amplitude is sim= 4 times 10minus13 This is thesame as the earth potential difference for 4 km height difference It ispossible that ground clocks will have reached an inaccuracy at the levelof 1 times 10minus18 at the time of the mission in 2015 Using the microwave linka comparison of ground clocks over sim3 h (typical common-view duration)with a resolution of 1 times 10minus17 can be performed limited by the instabilityof the link Repeating the clock comparison at each common-view win-dow (n = 1000 comparisons) will allow averaging and a measurementof the solar gravitational shift with fractional inaccuracy of 1 ppm Incomparison the best current results for the solar gravitational frequencyshift are at the few-percent level [20 21] A moderate averaging out ofground clock systematic shifts is assumed as well as modeling of the earthgravitational potential variations on the daily timescale at the appropriatelevel

(4) Comparisons of on-board clocksThe two on-board clocks are continuously compared these data areanalyzed for violation of LPI (Eq 4) Since the measurement is a nulltest the required accuracy of the gravitational potential at the satellitelocation is low and easily satisfied

24 Test of local position invariance and search for neutron scalar charge

The test of LPI is performed in conjunction with the measurement of thegravitational frequency shift

With approach 23 (1) the optical clock will be able to measure K (α Earth)with accuracy 30 ppb (spec)

Combining the results of the microwave and optical clocks will determineK(meQCD Earth) + εK(mqQCD Earth) with accuracy 85 ppb (spec)where ε asymp minus01

Simultaneously the on-board comparison 23 (4) yields the combination19 K(α Earth) +K(meQCD Earth) + εK(mqQCD Earth) with accuracy85 ppb (spec)

Note that these measurements are fully complementary to current andfuture terrestrial clock-clock comparisons which are sensitive to K(X Sun)and can be combined with the latter to determine or set limits to the neutronscalar charge

Exp Astron

25 Theoretical analysis of frequency comparison signals

The on-board clock-clock comparison signals can be analyzed straightfor-wardly in terms of a violation of LPI since signal propagation effects areabsent

The determination of the gravitational frequency dilation via ground-satellite clock comparisons is more complex (Eq 1 is strongly simplified) andrequires an accurate treatment of all relevant relativistic effects Gravitationin the solar system is described in the barycentric (BCRS) and geocentric(GCRS) non-rotating celestial reference systems by means of post-Newtoniansolutions of Einsteinrsquos equations for the metric tensor in harmonic gaugescodified in the conventions IAU2000 [22] In these conventions the relativisticstructure of Newton potential (order 1c2) and gravitomagnetic (order 1c3)

potentials are given For the Newton potential the multipolar expansion ofthe geopotential determined by gravity mapping is used The post-Newtonianeffects of the metric (described in terms of dimensionless parameters β and γ )

can be included Orbits of satellites are evaluated in the GCRS at the levelof a few cm by means of the Einstein-Infeld-Hoffmann equations at the order1c2 [23] using the relativistic Newton potential Instead the trajectories of theground clocks in the GCRS are evaluated from their positions fixed onthe Earth crust (ITRS International Terrestrial Reference System) by usingthe non-relativistic IERS2003 conventions [24 25]

For the propagation of light rays (here radio signals) between the satelliteand the ground stations one uses the null geodesics of the post-Newtoniansolution in the GCRS The timefrequency transfer properties have beentheoretically evaluated for an axisymmetric rotating body [26ndash29] at the order1c4 (ie with Newton and gravitomagnetic potentials developed in multipolarexpansions) These results have been developed for use in the ACES missionThe relative frequency dilation can be expressed as

ν

ν= 1 + 1

cL1 +

4sum

n = 2

1

cn(Ln + Gn)

where Ln describes special-relativistic Doppler effects and Gn the general-relativistic effects These quantities are time-dependent due to the orbitalmotion For concreteness we report the orders of the terms The kinematicaleffects are |L1c| lt 1 times 10minus5 |L2c2| lt 2 times 10minus10 |L3c3| lt 3 times 10minus15 |L4c4| lt

7 times 10minus20 The Newtonian contributions are G2 = G(M)2 + G(J2)

2 + G(J4)2 + G(J6)

2(M is the mass monopole Jn are the lowest multipoles) with estimates|G(M)

2 c2| lt 3 times 10minus11 |G(J2)2 c2| lt 2 times 10minus13 |G(J4)

2 c2| lt 3 times 10minus16 |G(J6)2 c2| lt

1 times 10minus16 For G3 = G(M)3 + G(J2)

3 the estimates are |G(M)3 c3| lt 2 times 10minus14

|G(J2)3 c3| lt 1 times 10minus18 Finally the G4 = G(M)

4 + G(S)4 contributions (G(S)

4 is thelowest gravitomagnetic effect) are of order |G(M)

4 c4| lt 1 times 10minus19 |G(S)4 c4| lt

1 times 10minus22 and thus not relevant for EGE

Exp Astron

The available theory is fairly complete for the data analysis in mode 23 (1)described below For modes 23(2 3) the theory will need to be generalizedto include the effect of the gravitational potential at the location of the groundclock(s) (including time-varying effects such as the tides) and of the solar andlunar potentials

The uncertainties of the various contributions above are minimized by ausing precise EGE orbit data obtained by satellite tracking and by usingcurrent and future accurate earth gravity information Required orbit positionknowledge is at 1 cm level near perigee and sim50 cm near apogee reachablealready today with the proposed orbitography approach described below Thevalidity of the special-relativistic Doppler shift has been and will continue to beindependently verified with increased accuracy by spectroscopy of relativisticatomic ion beams [30] Potential violations of this aspect of Lorentz Invari-ance are expected to be sufficiently bounded so as not to affect the signalinterpretation

26 Local Lorentz invariance tests

There has been an explosion of interest in Local Lorentz invariance in re-cent years [31 32] with numerous astronomical studies and high-precisionexperiments applied to search for violations of Lorentz invariance Signifi-cant developments have also occurred on the theoretical side and a theory(standard model extension SME [33]) has been worked out that provides aunified framework for describing and analyzing Lorentz Invariance violationsin a variety of systems Two aspects of Lorentz Invariance can be tested withEGE

261 Independence of the speed of light from the laboratory velocity

A possible dependence of the speed of light on the speed v of the laboratorycan be modeled according to the MansourindashSexl test theory [34] by

c (v) = c0(1 + Bv2

c2

)

Here v 377 kms is the velocity with respect to the cosmic microwavebackground the cosmologically preferred frame and B is a combinationof parameters describing deviations from the usual Lorentz transformationformulas B = 0 if Lorentz Invariance holds For a terrestrial experiment therotation of the Earth around its axis modulates v with a 300 ms amplitude Adependence c(v) can be searched for by measuring the frequency differencebetween a resonance frequency of a highly stable optical cavity (cavity fre-quencies are proportional to c) and an optical clock (Kennedy-Thorndike-typeexperiment [35]) and determining the modulation of the frequency differencecorrelated with the modulation of v The advantages of a space experimentare the high orbital velocity and strongly reduced cavity deformation thanksto microgravity [36 37] On the proposed elliptic orbit v varies between+4 and minus4 kms over approximately 1 h This variation is 13 times larger

Exp Astron

than for the case of a terrestrial experiment The shorter time scale is alsoadvantageous since the drift of the cavity is more predictable In the EGEmission the reference cavity of the clock laser is used The large numberof orbits improves the statistics An improvement of the accuracy of B by afactor 20 compared to the best current terrestrial results is realistic The dataanalysis can also be performed in the framework of the SME theory takinginto account complementary bounds obtained from terrestrial experimentsand astrophysical observations

262 Independence of Zeeman splitting frequency on the directionof the magnetic field

One class of tests of Lorentz Invariance consists in measuring the frequencysplitting between two levels of a quantum system induced by a static magneticfield as a function of orientation of the field direction with respect to thestars (HughesndashDrever-type experiments [31]) This class addresses the so-called matter sector and is complementary to combined-photonmatter-sectortests such as the above The most precise experiments are performed withatomic clocks eg masers or cold atom clocks [38] The EGE instruments allowsuch experiments since magnetic fields are applied continuously in the clockor repeatedly for calibration Compared to terrestrial experiments the highorbital velocity of EGE could lead to an improvement by a factor sim20

27 Application to geophysics

Using GPS coordinates of a point on the Earth can be obtained with aninaccuracy well below 1 cm in a well defined international terrestrial referencesystem The coordinates are purely geometrical and do not contain any gravityinformation A local gravitational potential UB is determined with respect toa reference gravitational potential U A by ldquolevellingrdquo the surface of the Earthie sequentially measuring gravity g and height differences every ca 50 m

UB minus U A = minusBint

A

g (r) middot d n

where n is the difference vector between subsequent locations The disadvan-tage of measuring height differences by the levelling method is a random walkeffect of accumulated errors

In the classical definition a geoid is defined as the particular equipotentialsurface nearest to mean sea level In these terms the geoid serves as areference surface for measuring the height and also to define a referencefor the gravitational potential Such a vertical reference frame historicallywas realized for a country or several countries by determining the mean ofsea level observations at tide gauge stations taken over long period of timeHowever modern satellite altimetry missions such as TOPEXPOSEIDON orENVISAT show that departure of the mean sea level from an equipotential

Exp Astron

surface may reach up to several meters on the global scale see eg [39]Therefore height systems based on different tide gauge stations may differin the realisation of the geoid and may differ with respect to the reference byseveral meters

The best gravity field models obtained from satellite data (eg missionGRACE) have reached a precision of about 1 cm over sim250 km half wave-length [40] Satellite data also reveals changes in time [41] However for typicalEarth topography with height variations of eg 1000 m over 30 km horizontaldistance one may expect variations in the geoid of about 80 cm Such ahigh-spatial-frequency signal in geoid variations cannot be detected by spacegravity missions and requires a combination of satellite and terrestrial gravitymeasurements like gravity anomalies deflections of vertical and GPSlevellingpoints The best combined global gravity field models are provided up to de-gree and order 360 in terms of spherical harmonic representation correspond-ing to a half-wavelength of about 55 km However in combining the satelliteand terrestrial gravity field measurements the problem of terrestrial data givenin different height systems between continents and different countries remainsand has to be tied to satellite measurements

The geodetic scientific community is currently establishing a Global Geo-detic Observing System (GGOS) [42] Its objectives are the early detectionof natural hazards the measurement of temporal changes of land ice andocean surfaces as well as the monitoring of mass transport processes in theEarth system Global change processes are small and therefore difficult toquantify Therefore the required precision relative to the Earthrsquos dimensionis 1 ppb GGOS will be established by the combination of geodetic spacetechniques (GPS Laser-Ranging very large baseline interferometry) andrealized by a very large number of terrestrial and space-borne observatoriesThe purely geometric terrestrial 3D coordinate system is in good shape andfully operational It has to be complemented however with a globally uniformheight system of similar precision The current precision level of regionalheight systems in terms of gravity potential differences is in the order of1 m2s2 (10 cm) with inconsistencies between these various systems up toseveral 10 m2s2 (several meters) The actual requirement in the context ofGGOS is 01 m2s2 (1 cm) with the need of a permanent ie dynamical controlThis requirement of high precision height control comes from the need tounderstand on a global scale processes such as sea level change global andcoastal dynamics of ocean circulation ice melting glacial isostatic adjustmentand land subsidence as well the interaction of these processes Only by meansof monitoring in terms of gravity potential changes at the above level ofprecision the change of ocean level can be understood as a global phenomenonand purely geometric height changes be complemented by information aboutthe associated density or mass changes

The geoid can also be defined in a relativistic way [43] as the surface whereaccurate clocks run with the same rate and where the surface is nearest tomean sea level The relation between the differences in the clock frequenciesand the gravitational potential is given in simplified form by Eq 1 At present

Exp Astron

there is no operational way to compare frequencies of the already availableoptical clocks on the global scale at the same level as their accuracy wouldallow With EGE and by then improved ground optical clocks it will becomepossible to obtain gravitational potential differences on a global scale by com-paring frequencies This would be a significant new dimension to gravity fielddetermination since such observables are given on the global scale and providein situ local gravity information at the same time EGE will allow clock-basedgravitational potential mapping with the same payload instruments used forthe fundamental physics experiments The on-board clocks are actually notrequired for this purpose MOLO MWL and FCDP suffice Thus EGE willallow establishing a global reference frame for the Earth gravitational fieldwith accuracy in the order of a few cm in terms of the geoid heights Thisassumes that by the time EGE is operational (mobile) ground clocks withfractional inaccuracy at the level of 10minus18 are available the measurementsare limited by MWL noise and thus require a long integration times to matchthe expected ground clock performance Such a reference frame will serve asreference for all gravity field modeling and clocks will be used to define theinternational atomic time scale and the Earth gravitational potential at thesame time

In summary with optical clocks geodesy will undergo a second revolutionafter geometry now being measured by clock-based GNSS systems also physi-cal heights and potential will be measured by (mobile) optical clocks using thegravitational redshift effect

28 Application to frequency standards and terrestrial fundamentalphysics studies

As has been described repeating ground clock comparisons and integratingover several months will provide global ground clock frequency comparisonsat the level of 1 times 10minus18 more than two orders of magnitude below the currentGPS and two-way satellite time and frequency transfer methods In the timeand frequency community the availability of such a global high performancemicrowave link will accelerate the process leading to a redefinition of the SIsecond based on clocks operating in the optical domain

The global character of the ground clock comparisons made possible byEGE can also contribute to the search for a time-variation of the fundamentalconstants

The comparisons do not require availability of the on-board clocks The riskof this type of measurements is therefore reduced Moreover the lifetime ofmany subunits required for this measurement (MWL and FCDP see below)can exceed 10 years (from previous experience with similar devices) andthus offers a motivation to increase the mission duration beyond the nominal3 years

Further applications to metrology are listed under the secondary goals inSection 21 above

Exp Astron

3 Mission profile

31 Orbit

The orbit must satisfy the following requirements

Primary 1 Large gravitational potential difference between apogee andperigee

2 sufficient contact time with at least two ground stations atperigee

Secondary Simultaneous visibility of satellite from distant (inter-continental) ground stations for several hours

To satisfy the primary objective the orbit must be highly elliptic The perigeeshould be low A very low perigee means a short contact time which decreasesthe accuracy of the frequency measurement A compromise is a perigeealtitude of sim2500 km For a sufficiently high apogee this yields a U of 60of the theoretical maximum between the earth surface and infinity

The baseline scenario is an orbit of approx 12 h period with the followingproperties

Apogee Altitude 37856 km (minus 10 deg latitude one over Atlantic andPacific resp)

Perigee Altitude 2500 km (+ 10 deg latitude one over Guatemala andMalaysia resp)

Inclination 634 argument of perigee 170RAAN 25 True Anomaly 220Apogee drift rate 70 kmday perigee drift rate sim 14 kmday

True anomaly and right ascension of the ascending node (RAAN) wereoptimized to minimize orbital precession Figure 3 shows the ground track andthe ground coverage at one of the perigees Common-view contact time forstations in Europe is about 6 h while between a station in Europe and in theUSA it is about 35 h Frequency comparison windows between a given groundstation and EGE repeat every 24 h

A final selection of the orbit requires an in-depth study of the costs of thepropellant and maneuvers as well as of the setting up of new ground stations

32 Ground segment

The mission requires two separate ground segments the science groundsegment and the standard satellite control segment which supports telemetryand telecommand (TMTC)

Exp Astron

Fig 3 Baseline EGE orbit One (out of two) apogee and one (out of two) perigee ground pointsare shown The ground coverage at one of the perigees is shown with circles at elevation anglesteps of 10 starting from 10

321 Science ground segment

As described above one part of the mission is to perform frequency com-parisons between ground-to-space and ground-to-ground (common-view andnon-common view) The use of a two-way microwave link (MWL) systemhas the advantage that a precise orbit or the precise relative velocity of thesatellite must not be known [4] The orbit determination requirements fora two-way system are significantly relaxed compared to the requirements tomodel the relativistic effects on the clocks The frequency comparison resultsare available both on ground and on the satellite

For ground-to-satellite clock comparisons five ground stations arefavourable two capable of viewing perigee three for viewing apogee Thisallows for the classic three-clock configuration where the performance of eachclock can be obtained individually The perigee stations will be observing thesatellite for sim25 min every perigee passage for the duration of the missionThe apogee stations will observe for sim3 h every apogee passage Of coursea larger number of stations as well as longer observation time are usefulfor robustness redundancy and orbit determination The stations could beinstalled on a temporary basis for the duration of the mission

The stations must have access to optical clocks of performance at least ashigh as todayrsquos state-of-the-art ground clocks but preferably higher Theywill include an optical frequency comb by means of which the optical clockfrequency is transformed to a radio frequency for comparison with the EGEclocks High-accuracy fiber-optic link technology already existing today allowsdistances of sim100 km between the ground station and the correspondingoptical clock allowing a certain freedom in the choice of the clock locationWith further increase of the fiber-optic link range currently being investigated

Exp Astron

the area over which the ground station can be positioned relative to the opticalclock will increase This option is of importance for the implementation of ageophysics measurement campaign for which a large number of clock locationsis of interest

The above ground stations will be complemented by some 15 additionalground terminals provided by the world-wide laboratories equipped with highperformance primary clocks at the time of the mission These will participate inorbit determination to 1 cm accuracy along the full orbit which can be achievedwith non-optical atomic clocks The EGE MWL is fully compatible to plannedACES ground terminals which might reduce the number of required stations

With a larger number of transportable optical clock ground stations de-ployed and moved across the continents a geophysics campaign could beimplemented to map the gravitational potential

Ground station equipment MWL uses dedicated small transportable termi-nals which can be easily installed at any location of interest Each terminalconsists of a steerable dish antenna of 1 m diameter with pointing accuracy of01 using a transmitted power of 5 W per band The high directivity antenna(beam aperture Ka-band 07 Ku-band 14 S-band 20) avoids multi-pathfrom nearby objects Antenna noise temperatures are Ka-band 450 K Ku-band 340 K Operation is at elevations above 10 and measurements areperformed above 20 For ranging purposes terminal location is determined byGPS surveying Each ground station computes its own ionospheric correctionsbased on a triple-band receiver The terminal has a built-in delay monitorwhich allows calibrated ranging during several months once its bias has beendetermined by laser ranging Its M and C interface is via local area networkto either a userrsquos computer or directly to the science network control centreby remote MampC The stationrsquos operation is fully automatic and unattendedSchedule and orbital data are received via LAN once per day Signal acquisi-tion is fully autonomous Data transmission is after each pass The terminal hasan uplink telecommand capability of 1 kBits and a downlink data capacity ofup to 10 kBitss For the local user there is a visual interface providing quick-look data from real-time comparison to the spacecraft clocks to verify theirgood health The final full performance product is calculated at the sciencedata centre

Local metrological data (temperature pressure humidity) are recorded tocorrect for tropospheric delay of the ranging data The main advantage ofMWL terminals compared to laser based systems is their all-weather capabil-ity although microwaves suffer higher propagation errors in the ionosphereand troposphere

The EGE ground terminal will profit from the ERS-2 mission PRARE(precise range and range rate equipment) [45] experiments whose terminalshad similar design and have been operated under extreme climatic conditionsincl Antarctica and high temperature regions from existing designs fromthe space segment (mission ACES) and no further special developments aredeemed necessary Cost per EGE master station is modest

Exp Astron

Satellite tracking via laser ranging Laser ranging is alternative to MWL rang-ing and also serves for absolute calibration of the latter For this purpose anon-board corner cube reflector (CCR) is provided Ranging will be done withthe International Laser Ranging Service Ideally the satellite will be rangedat every perigee passage and every apogee passage simultaneously with thefrequency comparison procedures Todayrsquos ranging precision is sufficient tosatisfy the science goals For apogee observations in the northern hemispherea large number of laser ranging stations is available The number of stationscapable of ranging at or near perigee in the southern hemisphere is six andthus is sufficient even considering unfavourable weather conditions at someof these The MWL ground stations are able to range even under thoseconditions

322 Control ground segment

Average data transfer rate is estimated at 300 MBday (approx 40 kBs onaverage) Data will be stored on board (2 GB capacity required) and will betransferred to a receiving station on the Earth once a day

XPLC

incl Data StorageIF to TMTC Link

HK amp TMTC

to ground

Opt Clock

AOM

MOLO

PHARAO

FCDP

10 GHz

100 MHz

250 MHz

GNSS Rx

MWLTampF Link

uplink Data

UTC(k)

TampF Space vs Ground

TM from GS

Tim

e S

cale

Orbit GNSS Time UTC

TampF Space vs Ground

MampC Time Scale

MampC

MampC

MampC

MampC

Mamp

C

PHARAO vs Opt Clock

PHARAO vs Opt Clock

10 GHz amp Epoch

10 MHz

amp Epoch

Epoch

10 MHz

amp Epoch

Opt Link

Fig 4 Schematic of the payload showing subunits and their signals The results of the frequencycomparison between the two on-board clocks (optical clock and microwave clock PHARAO) areobtained from a combination of the frequency values applied to the frequency shifters AOM (seeFigs 7 and 8) and those output from the frequency control and distribution package (FCDP)as indicated by the arrows ldquoPHARAO vs Opt Clockrdquo The satellite clock to ground clockcomparison results are output from the MWL XPLC is the on-board computer and is involvedin generation of signals for instrument control The optical link shown dashed is a possible add-onproviding an alternative to the MWL GS ground stations MampC monitoring and control HKhousekeeping TampF time and frequency

Exp Astron

4 Main payload instruments

41 Overview

The satellite payload consists of a pair of cold-atom clocks a microwave-optical local oscillator (MOLO) an on-board frequency distribution unit(FCDP) a satellite-to-ground clock comparison unit (MWL) and the onboardcomputercontrol system (XPLC) see Fig 4 Auxiliary units are the posi-tion determination GPSGalileo receiver (GNSS Rx) attitude monitor (startracker not shown) and the corner cube reflector array (CCR not shown)

The science goals require clocks of very high frequency stability on thetimescales of half the orbital period (6 h) as well as perigee passage (sim1000 s)see Fig 2 Only clocks based on cold atoms are suited for the science goalsTable 1 reports values for a subset of atomic clocks that have been investigatednot including clocks that require cryogenic operation or are too complex foruse in an medium-class mission The choice of a clock pair rather than a singleone fulfils several requirements it addresses more science goals improves thevalidity of some science results and it enables calibration accuracy evaluationand redundancy The validity of the science results is enhanced since theclocks differ not only in the employed atoms but also in terms of havingdifferent sensitivity to perturbations by electric and magnetic fields electronicserrors and misalignments A comparison of the results arising from each clockallows for an understanding of measurement errors Both the clocksrsquo and thefrequency comparison unitrsquos performances will be determined after launchand regularly verified during the course of the mission The only reliable

Table 1 Properties of different clock candidates for EGE

Clock Clock Fractional σ = Fractional Sensitivity Figure of Sensitivityatom frequency inaccuracy instability to α Aα merit |Aα| to meq

at 1000 s σ [1016] QCD

Sr+ 445 THz 3 times 10minus17 (p) 5 times 10minus17 (p) 042 08 (p) ndashYb+ (1) 688 THz 3 times 10minus16 (d) 3 times 10minus16 (d) 088 029 (d) ndash

3 times 10minus17 (p) 1 (p)Yb+ (2) 642 THz lt 1 times 10minus17 (p) 1 times 10minus17 (p) minus53 53 (p) ndashSr 435 THz 1 times 10minus16 (d) 1 times 10minus16 (d) 009 009 (d) ndash

lt 1 times 10minus17 (p) 09 (p)Yb 519 THz 7 times 10minus16 (d) 1 times 10minus16 (d) 031 03 (d) ndash

lt 1 times 10minus17 (p) 1 times 10minus17 (p) 31 (p)Cs 91 GHz 2 times 10minus16 (d) 1 times 10minus15 (d) 275 027 (d) 1Yb+ 126 GHz 1 times 10minus15 (p) 1 times 10minus15 (p) 34 034 (p) 1

A relevant figure of merit is the ratio of |AX | and fractional frequency instability σ on the perigeevisibility timescale The Cs clock is the only high performance clock with adequate sensitivity tothe particle masses me mq The indicated Cs clock performance was achieved using a microwavelocal oscillator based on a cryogenic oscillator On EGE the latter is to be replaced by an opticallocal oscillator-based device (MOLO) with similar short-term performance but without the needto operate at cryogenic temperature The clock instability at 10000 s is typically similar to theindicated inaccuracy(p) projected (d) demonstrated (1) quadropole (2) octupole

Exp Astron

way to achieve this is by having at least two clocks on board The on-boardclock-clock comparison provides performance evaluation independent of theground-satellite comparison unit Once the satellite clock performances areestablished the link performance can be established as well

The clocks are similar in basic structure but differ in the detailed implemen-tation They consist of

a source of atomsa preparation subunit which cools the atoms to near-standstill (millikelvintemperature or below)a trapping subunit that confines the atom(s) to a tiny volumean interrogation subunita clock control electronics package

The clocks share a common central unit the optical-microwave local oscil-lator which serves as a flywheel oscillator with outstanding stability on shorttimescales (01ndash10 s) for both clocks It consists of a laser stabilized to an ultra-low loss optical resonator made of ultra-low-expansion glass A microwavesignal is derived from the laser by phase-locking a frequency comb to the laserwave and using a multiple of the repetition rate

While the subunits of any optical clock have a similar technological basis(optics electrooptics acoustooptics thermal control vacuum systems) theydiffer in their specifications Depending on the choice of the atom species de-livery confinement and interrogation methods clocks of different perfor-mance result A number of laboratory optical clock demonstrators have beendeveloped worldwide This ensures availability of the know-how for the indus-trial implementation of the flight models

The first clock is a microwave clock based on the interrogation of slowlymoving ensembles of cold Cs atoms It is essentially based on the PHARAOclock [5 6] developed to engineering model level by CNES in the frame-work of the ESA ISS mission ACES In EGE this clock is upgraded usingthe microwave-optical local oscillator This improves significantly the clockperformance The specification is a relative instability (Allan deviation) of3 times 10minus14 (τ s)minus12 ie at the integration time τ = 1000 s it drops to 1 times 10minus15and to 3 times 10minus16 at 10000 s While this specification is at least a factor 3 lessstringent than for the optical clocks the significantly lower cost of obtaining aflight model and the sensitivity of its frequency to the particle masses makes ita suitable choice

The optical clock baseline instrument is a single-ion optical clock Amongvarious suitable ion species (Hg+ Al+ Sr+ Yb+) [44ndash47] the Yb+ ion hasbeen selected Its breadboard performance is established at the instability levelof 3 times 10minus16 over 1000 s it exhibits a large sensitivity coefficient Aα hassmall sensitivity to magnetic fields and exhibits significant potential for furtherimprovements (use of the octupole transition with sim6 times larger Aα andlower overall instability [48] see below) A trapped Yb+ ion is also a veryrobust system it has been demonstrated that a single Yb+ ion can remaintrapped uninterruptedly for many months Finally Yb+ is the only ion with

Exp Astron

which optical clocks demonstrators have been operated in not a single but twomajor metrology laboratories

Another attractive optical clock is based on ensembles of cold neutral Sror Yb atoms trapped in an optical lattice produced by laser beams This clocktype has a lower instability than a single-ion clock thanks to the large numberof atoms (sim105) used Its development is in progress in several laboratoriesin the world (in part supported by space agencies) and has to date reached6 times 10minus17 instability at 5000 s and 1 times 10minus16 uncertainty [49 50] with potentialfor even higher performance However it is more complex and costly thanthe Yb+ ion clock since the cold atom preparation and trapping subunitsinclude more components require higher laser powers and the respectiveoperating procedures are more complex It is therefore not considered for thepresent mission designed to be developed in the very near future with limitedresources but is an interesting candidate for missions without such constraintsIts estimated physical parameters are indicated in Table 2 for comparison

42 Main instruments

421 Microwave clock

The PHARAO cold atom clock is one of the two atomic clocks of the spacemission ACES managed by ESA It is proposed here to include a secondversion of this instrument as its costperformance ratio is very attractive forthe EGE scientific objectives Its concept is very similar to ground basedatomic fountains but with the major difference of zero-g operation Atomsslowly launched in free flight cross two microwave fields tuned to the transitionbetween the two hyperfine levels of the cesium ground state The interrogationmethod based on two separate oscillating fields (Ramsey scheme) allowsthe detection of an atomic line whose width is inversely proportional to thetransit time between the two microwave cavities The resonant microwavefield at 9192631770 GHz (SI definition of the second) is synthesized startingfrom an ultrastable quartz oscillator (USO) in the FCDP phaselocked to themicrowave output of the microwave-optical local oscillator and stabilized tothe clock line using the error signal generated by the cesium resonator In thisway the intrinsic qualities of the cesium hyperfine transition both in terms ofaccuracy and frequency stability are transferred to the USO In a microgravityenvironment the velocity of atoms along the ballistic trajectories is constantand can be changed continuously over almost two orders of magnitude (5ndash500 cms) Differently from atomic fountain clocks presently operated onground very long interaction times (up to few seconds) will be possible whilekeeping reasonable the size of the instrument

The instrument consists of 4 main elements a laser source which providesthe light required to cool and detect the cesium atoms (Fig 5 left) a frequencychain which generates the 92 GHz microwave signal a cesium vacuum tubewhere the interaction between the microwave field and the atoms occurs

Exp Astron

Table 2 Resources and technology readiness level (TRL) of the EGE spacecraft

Unit Volume Mass Power TRL Comment[l] [kg] [W]

Ion clock 100 70 60Atom source 4Atom preparation 6Atom trap 4Atom interrogation 6Clock control 6

Microwave clock 200 91 110 8 TRL by end 2008Local oscillator

Stabilized laser 18 15 23 6Frequency comb 20 16 38 6

Microwave link 18 18 55 8FCDP 7 8 12 8GPS receiver 2 5 7 9CCR 2 2 0 9Payload sum 367 225 305Thermal control 14 17 30 8OBDH 20 24 38 9Star tracker 1 1 8 9Radiation shield 10 40 0 9Support structure 170 204 0 9Power incl solar arr 50 58 10 9Propulsion 21 25 20 9AOCS 36 43 82 9TMTC 8 10 15 9Satellite bus sum 330 422 203 Flexbus conceptPayload plus bus incl 20 margin 8364 7764 6096Propellant 40Launch adapter 100SC wet masspower incl Margins 9164Lattice clock 135 105 152 Alternative clock

Atom source 4Atom preparation 6Atom trap 4Atom interrogation 6Clock control 6

Also included are the resources for an optical lattice clock based on neutral Strontium atoms

(Fig 5 right) and a computer The engineering model of PHARAO is fullyassembled at CNES Toulouse and undergoes functional and performancetests until the end of 2008 The operating temperature range is 10ndash33 andnon-operating temperatures are minus45 +60 The cesium tube consists in aUHV chamber pumped by getters and a 3 ls ion pump a cesium reservoira microwave cavity coils to provide a uniform magnetic field and three layersof magnetic shields Vacuum windows with fiber-optics collimators enable thetransmission of the cooling and detection light onto the cesium atoms

In autonomous mode PHARAO uses the USO as interrogation oscillatorand will provide a clock signal with fractional frequency stability below1 times 10minus13 (τs) minus12 and inaccuracy near 1 times 10minus16 When using the microwave-

Exp Astron

Fig 5 Left The PHARAO laser source engineering model with cover removed Dimensionsare 530 times 350 times 150 mm3 and the mass is 200 kg Ten polarization maintaining optical fibers(yellow) guide the laser beams to the cesium tube All diode lasers are mounted on Peltier coolersfor temperature regulation within 2 mK (courtesy EADS SODERN) Right The PHARAOengineering model cesium tube without the two external magnetic shields The volume is 990 times336 times 444 mm3 and the total mass is 44 kg

optical local oscillator PHARAO will provide a reduced frequency instabilityof 3 times 10minus14 (τs) minus12 This corresponds to 1 times 10minus15 at 1000 s and 3 times 10minus16

at 104 s

422 Ion optical clock

The 171Yb+ ion clock shown in Figs 6 and 7 provides an atomic frequencyreference based on a single 171Yb+ ion confined within an RF electric end-captrap (similar to a Paul trap) within an ultra high vacuum (UHV) chamber andlaser cooled to sim1 mK close to the Doppler cooling limit As a result the ionrsquosfirst order Doppler motion is completely removed and the ion experienceslittle collisional perturbation from its environment Under these conditionsthe narrow linewidth 2S12 (F = 0 mF = 0) minus 2 D32 (F = 2 mF = 0) quadrupole

Fig 6 Left Simplified level scheme of 171Yb+ showing relevant cooling (blue) auxiliary (red)quadrupole clock (green) and octupole clock (dark blue) transitions δνnat is the natural linewidthof the clock transition Middle close-up of Paul-type RF trap (PTB Braunschweig) the ion istrapped in the hole (diameter 14 mm) Right Complete ion trap with vacuum flange height 10 cm(NPL Teddington)

Exp Astron

935 nm D32-state repumper laser

638 nm F72-state repumper laser

739 nm extended cavity diode laser

AOM

PMT

Ion trap UHV chamber

vacuum pumps

Fibre-free space interface

Mu-metal shielding

EC

T

B

Magnetic field

Temperature

Ion fluorescence Processor

Clock laser frequency step and control

4355 nm

Frequengy doubling II

3695 nm cooling wave

Low-drift cavity

Low-drift cavity 171Yb+ 435 nm ion clock unit

876 nm clock laser wave from MOLO

Frequengy doubling

Fig 7 Simplified schematic of the single-ion optical clock unit AOM acousto-optic frequencyshifter PMT photomultiplier tube EC endcap T temperature stabilization system B magneticfield

clock transition at 4355 nm can be interrogated by spectrally narrow and stableclock laser light A diagram of the simplified energy level scheme for 171Yb+showing the relevant cooling auxiliary and clock transitions is given in Fig 6All these wavelengths are provided by diode lasers operated in fundamentalor frequency doubled mode The clock laser light is obtained from the 871 nmwave provided by the MOLO unit (see Section 423) after frequency-doublingin single pass by a KNbO3 nonlinear-optical crystal to 4355 nm The clocktransition spectral profile is observed by stepping the clock laser frequency bymeans of an acousto-optic frequency shifter (AOM in Fig 7) and recordingthe statistics of the quantum jumps in cooling laser induced fluorescence asa function of this frequency Under clock operation the AOM repeatedlysteps back and forward between the half-intensity points on the transitionprofile monitoring the signal imbalance between these points Any detectedimbalance is servo-corrected to zero by feedback to the AOM

In detail the optical clock architecture will comprise (Fig 7)

bull an RF end-cap trap for ionising and confining a single 171Yb+ ion withinan ultra-high vacuum chamber pumped by a small ion pump and non-evaporable getter pump The trap is driven by means of an ac voltage ofa few hundred volts at a drive frequency of sim10 MHz A small oven withseveral mg of 171Yb isotope is heated to provide a low flux of Yb atomsfrom which single ions can be ionised by electron bombardment from ahot wire filament within the trap potential well Low voltage dc is applied

Exp Astron

to additional electrodes to position the ion precisely at trap potentialcentre in order to minimise the ionrsquos micromotion at the drive frequencyThis needs automatic periodic monitoring and correction especially in theperiod after re-loading an ion into the trap External magnetic field coilsin three orthogonal axes allow the nulling of external fields and settingof a fixed field of sim1 μT The trap and coils are surrounded by mu-metalshielding to minimise external field changes during the orbit A level oftemperature control of 1 K of the region surrounding the trap is requiredfor maintaining the blackbody-induced frequency shift uncertainty withtemperature at the 10minus17 level

bull a laser platform to provide frequency doubling of an extended cavity739 nm diode laser for Doppler cooling of the ion on the 2S12(F = 1) minus2 P12 (F = 0) dipole transition at 3695 nm Light from an amplified diodelaser device frequency-doubled in a periodically poled LiTaO3 crystal insingle-pass will be used Typical laser powers used for driving the coolingtransition below saturation are about 2 μW for a beam waist in the trapof about 50 μm By modulation of the injection current of the 739 nmdiode laser at 147 GHz a sideband is generated that excites the 2S12 (F =0) rarr2 P12 (F = 1) ldquorepumperrdquo transition in order to avoid optical pump-ing between the ground hyperfine states

bull The platform also houses the auxiliary lasers at 935 and 638 nm The935 nm diode provides repumping of the ion from the 2 D32(F = 1)metastable level after occasional branching decays to this level during thecooling sequence The 638 nm diode allows fast recovery of the ion fromthe very-long-lived 2F72 metastable state after very occasional collisionaldecay to that state Currently extended cavity lasers are forseen butdistributed-feedback lasers may be available in the near future

bull a high NA lens imaging system and photomultiplier detection system torecord the statistics of 369 nm fluorescence quantum jumps as a functionof clock laser frequency step

bull a fibre system to deliver the various cooling auxiliary and clock light fromsource to trap making use of achromatic doublets where necessary at thefibre-free space interface for launching into the trap

bull a monitoring and control processor which provides the primary coolingand clock laser pulse magnetic field and detection sequencing to observeand lock to the ion clock transition frequency The processor also monitorsfrequency and amplitude data necessary to determine normal laser andion operational conditions and initiate resetting and recovery algorithmswhere necessary and laser unit failure The actual magnetic field presenton the atomic volume is determined and actively stabilized so that aninaccuracy at the level of 1 times 10minus17 can be achieved

bull a redundancy level of at least two units for cooling clock and repumperlasers plus similar for frequency doubling crystals All redundancy unitsfor each wavelength will be fibre multiplexed as standard allowing redun-dant unit activation on determination of prior unit failure

Exp Astron

For an optimized quantum-projection-noise-limited performance of thesingle-ion clock Ramsey interrogation with a cycle time equal to the lifetimeof the metastable D32 level of 50 ms will be performed leading to an Allandeviation of about 27 times 10minus15 (τ s)minus12 This is sim45 times 10minus17 at 1 h satisfyingrequired stabilities for the science objectives Current laboratory results are astability of 1 times 10minus15 at 30 s averaging down to 3 times 10minus16 at 1000 s (PTBGermany) Also comparisons between two independent traps showed agree-ment at the level of several parts in 10minus16 over a series of eight measurementruns

The 171Yb+ ion clock offers a second option namely the 2S12ndash2 F72 octu-pole transition at 476 nm (see Table 1 and Fig 6) [48] It has an extremelylong upper state lifetime low electric quadrupole and second order Zeemansystematic shifts and therefore is a candidate for an even higher performanceclock if a clock laser system with lower instability than the above is providedApart from the clock laser it shares the laser system of the quadrupoleclock Thus with moderate additional resources the Yb+ ion apparatus couldprovide two optical clocks in a single package and specifically perform directfrequency comparisons between the quadrupole and the octupole transitionwith a large sensitivity Aα

The ion clock will operate under microprocessor control providing thecontrol algorithms for the integrated pulse sequencing for cooling repumpingclock interrogation magnetic field switching cooling fluorescence monitoringThe controller will also provide error flags for critical processes and systemreset and correction where necessary This will include ion re-loading andmicromotion reduction algorithms Automatic monitoring of laser power andspectral quality with signal re-optimisation or laser failure determination isrequired Redundant laser units will be pre-aligned and multiplexed into thefibre delivery to the trap so that redundant units can be readily activated in thecase of diode laser failure

423 Microwave-optical local oscillator (MOLO)

This oscillator provides two ultrastable frequency outputs one in the opticalregion the other in the microwave region the two frequencies being coher-ently related The MOLO is composed of three subunits the clock laser thereference cavity and the frequency comb see Fig 8

The clock laser subunit delivers the light for the excitation of the opticalclock transition It is generated by an extended-cavity diode laser at a wave-length of 871 nm Most of the output wave of this laser is sent to the ionclock The laser subunit containing two lasers for redundancy will require3 kg 3 l 15 W The frequency instability of the clock laser must be superiorto the minimum clock instability 27 times 10minus15 (τ s)minus12 for τ up to the timeconstant (sim10 s) of the servo system that locks the laser to the ionrsquos resonancesignal In order to achieve this the laser frequency is stabilized to the reference

Exp Astron

871 nm clock laser

EO

Detector and servo

AOM Femtosecond

frequency comb

To optical clock 10 GHz To FCDP

Microwave-optical local oscillator

ULE cavity

Fig 8 Microwave-optical local oscillator (MOLO) consisting of clock laser reference ULE cavityand femtosecond frequency comb (OFC) Radiofrequency inputs and outputs are not shown EOelectro-optic modulator ULE ultra-low expansion AOM frequency shifter

cavity where the PoundndashDreverndashHall method is used About 10 μW of thelaser output are used for this

The system will exhibit sim04 Hz (sim1 times 10minus15 relative) linewidth and afrequency drift lt01 Hzs The cavity consists of two high-reflectivity mirrorsoptically contacted to a cylinder of diameter 70 mm length 150 mm made ofULE (a glass exhibiting ultra low thermal expansion coefficient) with a cavityfinesse of about 300000 Residual vibrations of the satellite (arms lt 1 times 10minus6 g)should not be a limiting factor to the desired linewidth for safety marginan optimized cavity support can reduce the sensitivity to accelerations Foracoustic and thermal isolation the cavity resides inside an aluminium vacuumchamber equipped with a small (3 ls) ion-getter pump Two stages of polishedaluminium shields are implemented around the cavity which are activelytemperature stabilized so that the cavity is precisely kept at the zero crossingtemperature of the ULE thermal expansion coefficient Heat application andremoval is by thermoelectric elements between the shield and the vacuumchamber Additionally the entire vacuum chamber is temperature-stabilized

During transportation and launch the cavity will be rigidly fixed insidethe vacuum chamber by pressing on it from the sides by piezo-mechanicalactuators allowing accelerations of several g After bringing the satellite onthe orbit the actuators will be released

The optical setup for frequency stabilization will be a classic Pound-Drever-Hall scheme in reflection using an electro-optic modulator (EO) Couplingof the clock laser radiation to the cavity is via single-mode polarization main-taining optical fiber The fiber in- and out-couplers are mounted on miniaturepiezo-motor driven multiaxis translation stages Microprocessor control ofthe translation stages performs a laser-to-fiber incoupling stabilization andlaser-to-cavity mode-matching optimization (positioning inaccuracy lt10 μmalignment inaccuracy lt10 μrad)

The frequency stability of the clock laser will be limited by thermal 1f noisein the mirrors with an fractional Allan deviation slightly below 1 times 10minus15 from1 s to a few 10 s Resources including control electronics are 15 l 12 kg 8 W

Exp Astron

A femtosecond fiber laser based optical frequency comb generator (OFCFig 8) will be used for the clock frequency comparisons and to obtain aspectrally ultrapure microwave local oscillator output for the operation ofPHARAO for the microwave link and for the GNSS receiver The fem-tosecond laser is based on erbium-doped fiber pumped by fiber-coupled laserdiodes It emits 100 fs pulses at a repetition rate of 250 MHz with an averagepower of 100 mW The emission spectrum is centered around 1560 nm Spec-tral broadening to more than one octave is obtained via nonlinear interactionin a microstructure fibre so that the range from about 1050 to 2100 nm iscovered

Second harmonic generation of a portion of the broadened spectrum allowsobtaining the required light for the measurement of the clock laser frequencyat 871 nm The OFC will be referenced to the clock laser frequency (afterintroduction a controllable shift via the frequency shifter AOM) by makingthe carrier envelope offset frequency vanish so that the repetition rate fr isa high-order subharmonic of the optical clock reference frequency f1 = m1 frThe link between the optical and the microwave domain is then simply thedistribution of the repetition rate frequency or more conveniently a high(order 40) harmonic of the comb repetition rate 10 GHz

It has been demonstrated by several groups that frequency comparison andfrequency division with an OFC can be performed to very high accuracy andstability beyond the limitations imposed by the optical clocks themselves Forexample a fiber laser OFC with an instability (Allan deviation) of 6 times 10minus17

in 1 s has been demonstrated [51] OFCs based on fiber technology possessgood long-term reliability and high power efficiency The projection for a spacequalified OFC based on current industrial terrestrial technology is 16 kg mass20 l volume and 38 W power requirement

424 Frequency comparison and distribution package (FCDP)

This package distributes the ultra-stable 10 GHz reference frequency outputof the MOLO to the MWL performs frequency comparisons between MOLOand PHARAO and generates a variety of different reference signals for theindividual instruments including the 250 MHz reference clock to the OFCAll signals are coherent to the MOLO In the FCDP a USO is locked to theMOLO signal by a digital PLL A divider chain generates timing pulses (1 pps1000 pps etc) and keeps a continuous local time scale From the digital phasecomparison data the phase noise density spectrum between the USO and theMOLO is available by telemetry (TM) to assess the overall frequency purityin-flight

As the FCDP distributes the full performance reference from the MOLOdirectly to the MWL there is minimum impact of the FCDP on overall clockcomparison performance The USO-phase lock loop uses a similar designas for ACES-FCDP The digital PLL with linear phase detector allows theUSO to be slightly offset from its reference which is selectable by groundcommand This results in an accurate on-board time scale despite any offset

Exp Astron

of the MOLO reference frequency which can be synchronised to UTC Thefrequency synthesis chain re-uses modules developed for ACES-FCDP andtheir frequency inaccuracy Expected relative uncertainty (Allan deviation)of the frequency comparison between two signals provided to the FCDP is47 times 10minus17 at 1000 s 95 times 10minus18 at 10000 s

During non-availability of the MOLO ie during system start-up andcontingency conditions the USO runs as flywheel (ie it is unlocked withinstability lt15 times 10minus13 from 1 to 100 s) It will still provide useable butless accurate references to the instruments including the reference to OFCCommon view ground-to-ground time- and frequency comparison and supplyof the remaining instruments remain possible at the USO provided accuracy

Mass volume and power consumption are 8 kg 7 l 12 W Controldata rateis 10 kBday 50 kBday (FCDP alone)

425 Microwave link (MWL)

The microwave link is a two-way triple-frequency microwave link (Fig 9) Itis designed to perform accurate frequency comparison as well as time-transfer

Ku-Band Up-linkPower Tx 5 WCarrier 13475 GHzPN-Code 100 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

S-Band Down-linkPower Tx 1 WCarrier 2248 MHzPN-Code 1 MChips1pps 1 time marker sData 5 kBits

Ku-Band Down-linkPower Tx 1 WCarrier 1470333 GHzPN-Code 100 MChips1pps 1 time marker sData 5 kBits

EGE SC

S

Ku

KaKa-Band Up-linkPower Tx 5 WCarrier 345 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

Ka-Band Down-LinkPower Tx 1 WCarrier 32 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 5 kBits

GS Antenna diameterEGE terminal 1 mACES terminal 06 m

Fig 9 EGE microwave two-way link concept and characteristics See text for a description Thelink will be able to perform frequency comparisons with up to ten ground stations simultaneously(only one active comparison is shown) pps pulse per second Tx transmitter SC spacecraft

Exp Astron

between on-board and on-ground clocks Two-way operation allows for com-pensation of the effects due to spacecraft location and its motion relative toground Its two bi-directional links operate in Ka-band (32345 GHz) andKu-band (147135 GHz) The high carrier frequency of the up- and down-links allows for a noticeable reduction of the ionospheric delay A third down-link channel is in S-band (22 GHz) and is used to determine the ionospheretotal electron content MWL uses a combination of carrier phase and codephase data to measure with high precision the absolute ionospheric delayas well as ionospheric delay variations during a ground contact The useof three frequency bands allows to resolve third order ionospheric termsKey modulation parameters are Ka band Tone 250 MHz Ku band PN100 MChips S-band 1 MChips

The EGE-MWL concept is a considerably improved version of the ACES-MWL Microwave carrier generation is directly from the 10 GHz microwaveoutput available from the MOLO and having exceptionally low close-in phasenoise without intermediate frequency dividers or other complex synthesiserstages commonly used in RF transponders The high frequency carriers arekey to achieve the desired frequency comparison accuracy

The Ka-band carrier is generated by tripling the FCDP-generated signalThe Ku-band is subsequently generated by division by two using a verystable regenerative divider Finally the assigned radio frequency is obtainedby adding a low frequency signal using an image reject mixer A mixedanalog-digital receiver design will enhance its performance by use of a flexibledigital signal processing and digital control loops The number of simultaneousground contacts is increased from four (as in the ACES mission) to ten or moreusing the digital concept while the number of physical receiver channels perband is reduced from four to two one for operation to ground and one for in-flight calibration The concept allows adaptation of receiver loop parametersto optimise for the different portions of a highly elliptic orbit while reducingsystematic uncertainties due to high Doppler and Doppler rate Real-timedata exchange to ground terminals is by additional data modulation on all RFsignals The spacecraft carries two different small antennae for perigee and forapogee with gains of 8 dBi 20 dBi beamwidths 100 (full earth coverage) 14respectively power is 1 W for each

The MWL end-to-end performance between the attached clocks includingspace- and ground terminals is predicted to be a relative Allan deviation of1 times 10minus16 at 1000 s (perigee) and 1 times 10minus17 at 10000 s (apogee) Thesevalues are based on conservative implementation losses and other instrumentallosses assumed as 6 dB at apogee and 10 dB at perigee including additionallosses due to atmosphere An attitude knowledge of 1 is required

The MWL link will also support precise orbitographyndashcomplementary tothe GNSS receiver Generation of ranging data is simultaneous to two-waytime transfer and is available under all weather conditions Orbit determina-tion accuracy is expected to be better than 1 cm well matching EGE clockrequirements The value is expected based on demonstrated results fromother missions (METEOR ERS-2 TOPEXPOSEIDON) Laser ranging to

Exp Astron

the CCR array is used a few times during the mission to calibrate the MWLranging

In case of non-availability of the ultrapure signal from MOLO an additional10 GHz output is available from the FCDP to support ground-to-ground clockcomparison in common-view mode even after the nominal mission periodwhen the neither the clocks nor MOLO may be available any longer Basedon the PRARE experimentrsquos 12 years of operation this can be assumed forthe FCDPMWL combination as well The potential extended operational pe-riod makes the proposed configuration particularly attractive to metrologicallaboratories that rely on availability of long-lasting data in support of thegeneration of UTC

The resources of MWL are as follows mass 18 kg volume 18 l power55 W TC 2 kBday TM 120 MBday

426 Auxiliary units

The star camera assembly on the EGE spacecraft will provide real-timeattitude data primarily for the orientation of the spacecraft (with accuracy of1 in real-time) and the proper reduction of the GNSSSLR data

A GNSS receiver is used for the precise orbit determination time andfrequency transfer coarse positioning (lt1 m) for real time use by the Attitudeand Orbit Control as well as the back-up for time tagging of all payloaddata For the time and frequency transfer the receiver is nominally using theonboard frequency reference The receiver assembly includes three specialantennas By the year 2015 more than 100 GNSS satellites are expected tobe available for precise orbit determination permitting orbit inaccuracy of 1ndash2 cm RMS for altitudes up to approx 10000 km At apogee a combinationof GNSS signals and MWL ranging will be used with expected inaccuracyof 10 cm

5 Basic spacecraft specifications

The mission foresees a single spacecraft with a mission duration limited toapprox 3 years for cost reasons Attitude control along three axes withless than 1 error on a timescale of 1000 s is required The control can beperformed with cold gas thrusters magnetotorquers or reaction wheels Theselection will be done on the basis of the vibrations induced by these systems(desired level arms lt 10minus6 g)

Cold gas thrusters or monopropellant thrusters will be required to maintainthe orbit parameters during the mission by small orbit correction manoeuvresand will also guarantee a proper tuning of the orbit parameters after the releasefrom the launcher The ground segment (telemetry control and data transfer)could be via non-ESA stations user-operated to save costs

Further resources are indicated in Table 2

Exp Astron

The satellite bus design drivers are nadir-pointing microwave antennasthermal stability for the payload instruments no strong magnetic sources lowmicrovibrations harsh radiation environment due to crossing of the radiationbelts once per orbit power demand by science payload (with the possibility toreduce power demand by stand-by of some components during specific missionphases eg eclipse) mass limit due to required high altitude with low costlaunch

The satellite bus design can be based on established platforms such asFLEXBUS or PRIMA used on several past missions Their design philosophydiffers one having body-mounted solar arrays the other wings The latter canprovide larger amounts of power The vibration level induced by the panelswhen they are fixed and the duration of solar array drive periods per orbitwill have to be considered in the selection The FLEXBUS concept was usedfor a mission already requiring a low level of residual accelerations and theplatform is optimised for this parameter

For most of the satellite components one can consider using compo-nents and electronic units from navigation or communication satellites whichtypically cruise at altitudes between 15000 and 25000 km altitude LISAPathfinder which will cross the radiation belts during spiralling to theLagrange point will face similar problems concerning radiation as EGE as wellas XMM which is in a Molniya-type orbit with similar radiation characteristics

The instruments will therefore be located in the satellite structure such asto realise ionizing radiation shielding by the spacecraft structure and non-sensitive equipment to the maximum extent possible the need for additionalradiation shielding (including trapped proton radiation) for especially sensitiveparts of the instruments must be determined by ground tests It should bementioned that an orbit with higher perigee and less radiation dose is also anoption as it leads to a tolerable reduction of the science goals

The required plusmn3C temperature stability at the interface to the payloadenclosure is feasible The thermal control will employ the usual means (anenclosure with high thermal inertia heaters and a suitable radiator) Thelocation of the radiator will need a special study given the non-standard orbit

Magnetic field present during the mission can affect the properties of theonboard clock Using SPENVIS the magnetic field has been calculated alongthe spacecraft orbit the field intensity presents maxima around 02 G anda minimum value of the order of 10minus3 G As discussed above this is not alimitation to the instrument performance

6 Discussion and conclusion

The accuracy of the measurements depends crucially on the stability andaccuracy of the clocks While the performance of ground clocks (even thoseeventually deployed in new ground stations dedicated to EGE) will likelyimprove significantly from the present level the space clock design will beldquofrozenrdquo approx 10 years before data acquisition Thus their performance

Exp Astron

level will be significantly below the future ground station clock performanceAs specification we have chosen a performance level that has been demon-strated with laboratory breadboards recently We see no fundamental reasonthat this laboratory performance cannot also be achieved in a space ion opticalclock The experience with the PHARAO cold atom clock has shown thatan engineering model of a laser-based cold atom clock has a performancesimilar to that of a laboratory prototype The perturbation of the ion clock fre-quency due to magnetic fields along the orbit can be modelled and suppressedwith shielding and the residual effects measured and compensated for Themoderate temperature stability (plusmn 3C) to be provided in the experimentalbay can be achieved with thermal shields Accurate temperature controlof particular units is done by active stabilization of the units themselves

One important aspect is radiation which may affect all optical and trapcomponents as well as the lasers and reference cavity The radiation envi-ronment on the EGE orbit is significantly harsher than the near-earth ISSorbit on which ACES will operate The radiation effects must therefore becharacterized comprehensively with appropriate tests and then the radiationshield for the EGE instruments designed

In summary the Einstein Gravity Explorer mission (EGE) is a satellitemission aimed at a precise measurement of the properties of space-time usingatomic clocks By testing one of the most fundamental predictions of EinsteinrsquosTheory of General Relativity the gravitational redshift and thereby searchingfor quantum effects in gravity it explores one of the most important frontiers infundamental physics The EGE mission reaches beyond fundamental physicsproviding an important spin-off in geophysics Furthermore the technologicalaspects of the mission are ground-breaking for future time-keeping and ref-erence frame systems in space and thus are also expected to be of wide useto science and technology in general The payload of the mission has beenconceived as a compromise between cost and performance needing to fit thecost envelope of an ESA medium-class mission1

Acknowledgement We thank S Cesare for helpful information

References

1 Will CM The confrontation between general relativity and experiment Living Rev Relativ9 3 (2006)

2 Ni W-T Empirical foundations of the relativistic gravity Int J Mod Phys D 14 901 (2005)doi101142S0218271805007139

1Finally we point out that a spacecraft with a significantly reduced and thus lower-cost payload (noclocks only MOLO FCDP MWL) on a circular orbit of appropriate height could be used for along-duration earth science mission dedicated specifically to the measurement of the geopotential

Exp Astron

available at the time hydrogen masers were used for this experiment TheACES (Atomic Clock Ensemble in Space) mission planned to fly on the ISS in2014 seeks to improve the determination by a factor 25 by using a cold atomclock (PHARAO) and a hydrogen maser [5 6] Recent progress on cold atomclocks in the optical domain and in optical technology enable performing thisfundamental test with orders of magnitude better sensitivity

Although it has been very successful so far General Relativity as well asnumerous other alternative or more general theories proposed in the courseof the development of theoretical physics are classical theories As such theyare fundamentally incomplete because they do not include quantum effectsIn fact it has not yet been possible to develop a theory of gravity that includesquantum mechanics in a satisfactory way although enormous effort has beendevoted to this goal Such a theory would represent a crucial step towards aunified theory of all fundamental forces of nature Several approaches havebeen proposed and are currently under investigation (eg string theory) Afull understanding of gravity will require observations or experiments thatdetermine the relationship of gravity with the quantum world This topic isa prominent field of activity and includes the current studies of dark energy

The Einstein Gravity Explorer is a fundamental physics mission dedicatedto this question its primary task is to measure the gravitational frequency shiftwith unprecedented accuracy The EGE mission uses a satellite on a highlyelliptic orbit (see Fig 1) The ratio of the frequencies of two on-board clocksand the ratios between satellite and ground clocks are the main observables

Fig 1 General concept of the EGE mission Clocks on the satellite and on ground are intercom-pared as the satellite orbits the Earth on a highly elliptic orbit

Exp Astron

These can be interpreted toward tests of General Relativity (GR) and ofcompeting metric theories of gravity and as tests for the existence of newfields associated to matter (see Section 3) The outcome of the mission will beeither a confirmation of GR and metric theories within the accuracy providedby the instruments or the discovery of a deviation In the latter case themission would provide a first indication of the breakdown of current (classical)gravitational physics theories and could pave the way towards a unified theoryof all forces

2 Scientific objectives

21 Summary of science objectives

211 Fundamental physics science objectives (1 ppm = 1 part in 10 61 ppb = 1 part in 10 9)

Primary objectives

High-precision measurement of the earth gravitational frequency shift at25 ppb accuracy a factor 3000 improvementFirst precise measurement of the sun gravitational frequency shift at 1 ppmlevel a factor 2 times 104 improvementTest of space and time variability of fundamental constants with accuracy30 ppbFirst search for neutron scalar charge with accuracy 30 ppbTest of anomalous coupling of matter to the Standard Model quantumfields

Secondary objectives

Tests of Lorentz Invariance in the matter and photon sector (factor 20improvement)Contribution towards establishing a new definition of the unit of time

212 Spin-off to other fields (outside fundamental physics)

Establishment of a new approach to the determination of the geopotentialwith geoid height resolution of a few cmComparison of distant terrestrial clocks at the 10minus18 levelDemonstration of clock and link technology for future applications eg inprecision spacecraft navigationDemonstration on high performance real-time range determination withprecision 25 times (based on code phase) to 1000 times (based on carrier

Exp Astron

phase) higher compared to systems onboard previous missions (ERS-2TOPEX POSEIDON)Comparison of three different orbit determination systems Laser-Ranging μm-precision microwave ranging GPS-based orbit deter-minationAtmospheric signal propagation study precise determination of thirdorder ionospheric termMonitoring stability of the GalileoGPSGlonass time scales

22 Gravitational frequency shift measurement and search for new physics

The primary science goal of EGE is a measurement and test of the gravita-tional frequency shift in the weak-field limit as predicted by metric theoriesof gravitation These theories also predict that the shift is independent of thetype of clock used to measure it The comparison of the frequencies ν oftwo identical clocks 1 2 operating at different locations x1 x2 yields a fre-quency ratio

v2(xprime)v1

(xprime) sim= 1 + (U (x2) minus U (x1))

c2 (1)

Here νi(xprime) is the frequency of clock i located at xi as observed (measured) ata particular location xprime where the comparisons are performed The comparisonlocation can be arbitrary but will often be x1 or x2 or both U is thegravitational potential U(x) = minusGM|x| in case of a spherically symmetricbody of mass M In short time runs (or clocks tick) more slowly the closer thesystemclock is to a massive body Stationary clocks and observers and weakgravitational fields have been assumed in Eq 1

Through its measurement EGE searches for a possible violation of the grav-itational frequency shift A violation may be described phenomenologicallyby a dependence of one or more fundamental constants on the gravitationalpotential X = X(Uc2) where X is a generic dimensionless fundamentalconstant or a dimensionless combination of fundamental constants Such adependence is a violation of the principle of Local Position Invariance (LPI)which states that the outcome of local nongravitational experiments is inde-pendent of where and when in the Universe they are performed If a constantdepends on the gravitational potential then the results of those laboratory-type (ie local) experiments that depend on that constant will also depend onthe location namely on the location of the laboratory with respect to the massdistribution generating the potential and will violate LPI

LPI is one of the three cornerstones of the Einstein Equivalence Principle(EEP) the others being the Weak Equivalence Principle and Local LorentzInvariance It is a common belief that the EEP may not be absolutely valid -therefore spatial dependence and time dependence of fundamental constantshave become a very intense research topic in the last decade with experimentsundertaken worldwide to search for violations of various aspects of EEP For

Exp Astron

example extensive studies of atomic and molecular spectra of gas clouds in thedistant Universe are currently undertaken to search for a difference of theseparameters compared to todayrsquos values (see eg [7ndash11] and references therein)Also some differences in Big Bang nucleosynthesis data and calculations canbe naturally explained by a variation of fundamental constants [12 13]

The EGE mission is thus a probe of possible space-dependence of fun-damental constants The search may help to point the way toward a unifiedtheory of all forces reveal hypothetical extra dimensions in our Universe or theexistence of many sub-Universes with different physics chemistry and biologyIndeed theories unifying all forces applied to cosmology suggest space andtime dependence of the coupling constants [14ndash16] Another argument forspatial variation of fundamental constants comes from the anthropic principleThere must be a fine tuning of fundamental constants which allows humans andany life to appear Life in its present form is not consistent with fundamentalconstants whose values only slightly differ from the actual ones This finetuning can be naturally explained by the spatial variation of the fundamentalconstants we appeared in the area of the Universe where the values of thefundamental constants are consistent with our existence

In the Standard Model there are three main fundamental dimensionless pa-rameters determining the structure and energy levels of stable matter (atomsmolecules) the electromagnetic fine structure constant α and the two ratiosof electron mass me and light quark mass mq to the strong QCD energy scaleQCD Xeq = meqQCD The fundamental masses me and mq are proportionalto the vacuum Higgs field which determines the electroweak unification scaleTherefore the constants meqQCD can also be viewed as the ratio of the weakenergy scale to the strong energy scale

How can a space-time variation of the fundamental constants and a depen-dence on the gravitational potential occur Light scalar fields very naturallyappear in modern cosmological models affecting α and meqQCD Oneof these scalar fields is the famous ldquodark energyrdquo which causes acceleratedexpansion of the Universe Another hypothetical scalar field is the dilatonwhich appears in string theories together with the graviton Cosmologicalvariation of these scalar fields should occur because of drastic changes ofthe matter composition of the Universe During the Big Bang nucleosyn-thesis the Universe was dominated by radiation then by cold dark matterand now by dark energy Changes of the cosmic scalar field ϕ0(t) lead tothe variation of the fundamental constants X(ϕ0) Massive bodies (galaxiesstars planets) can also affect physical constants They have a large scalarcharge S = sp Z + se Z + sn N where Z is the number of protons and elec-trons and N is the number of neutrons sp se sn are the scalar charges ofthe proton electron and neutron respectively There is also a contributionof the nuclear binding energy (scalar charge of virtual mesons mediatingnuclear forces)

The scalar charge produces a Coulomb-like scalar field ϕs = SR where Ris the distance from the charge The total scalar field is ϕ = ϕ0 + ϕs therefore

Exp Astron

we may have a variation of the fundamental constants inversely proportionalto the distance from massive bodies

X (ϕ) = X (ϕ0 + ϕs) = X (ϕ0) + δX (R)

δX (R) = (dX

dϕ

)S

R

where a nonzero δX violates LPI and a small ϕs was assumedThe gravitational potential U is essentially proportional to the number of

baryons Z + N and inversely proportional to the distance R U = -GMR

Therefore the change of the fundamental constants near massive bodies canbe written

δX

X = K (X i) δ(Ui

c2

) (2)

where i refers to a particular composition of the mass M and where thecoefficients K(X i ) = minus(dXdϕ) c2SiGMi are not universal but are expectedto be different for eg the Sun and the Earth K(X Earth) = K(X Sun)Indeed the Sun consists mostly of hydrogen therefore the Sunrsquos scalar chargeis SSun Z (sp + se) The Earth contains heavier elements where the numberof neutrons exceeds the number of protons N 11 Z Therefore the scalarcharge of the Earth SEarth Z (11 sn + sp + se) and is sensitive to the neutronscalar charge (and also to the contribution of the nuclear binding) in contrastto the Sun

The most sensitive experiments suitable for a search of a space-time de-pendence in local or near-local experiments (such as satellite experiments) areclockndashclock comparisons In one type of terrestrial local experiment dissimilarclocks located nearby are compared for a duration spanning at least one yearDuring such periods the earth moves periodically closer and further awayfrom the sun so that the solar gravitational potential on Earth is modulatedThis enables a search for a dependence of the fundamental constants on thesunrsquos scalar charge [17 18] These experiments are able to set limits to K(XSun) for X = α meqQCD at the level of 10minus6 [12 13]

The proposed EGE satellite experiments are complementary to terrestrialexperiments allowing to set limits to K(X Earth) and combining with theresults on K(X Sun) from terrestrial comparisons to set limits on the depen-dence of the fundamental constants on the neutron scalar charge sn

In summary when the frequencies of two identical clocks (of nominalfrequency νa) that are located at different positions x1 x2 are comparedby measuring their frequency ratio at some position xrsquo the result may bewritten as

va2(xprime)

va1

(xprime) = 1 +

[1 +

sum

XAXa K (X i)

](Ui (x2) minus Ui (x1))

c2 (3)

The sensitivity factor AXa = X (d(ln νa)dX) is the sensitivity of the clockfrequency to a particular constant X and can be calculated using atomic and

Exp Astron

nuclear theory (see eg [12 13]) The measurement of this ratio which issensitive to K represents a test of General Relativityrsquos redshift

When the frequencies of two co-located but dissimilar clocks a b (ofnominal frequencies νa0 and νb 0) are compared in-situ at two locations x1 x2

(LPI test) the result is

[va1 (x1)

vb 1 (x1)

] [va2(x2)

vb 2 (x2)

] = 1 +[X

(AXa minus AXb

)K (X i)

](Ui (x2) minus Ui (x1))

c2 (4)

The result of the EGE mission may be given as a limits for (or nonzerovalues of) K(X Earth) for the three fundamental constants X = α meQCDand mqQCD

23 Measurement approaches for the gravitational frequency shift

The gravitational potential experienced by the ground and satellite clocksincludes the contributions from the earth and the sun (the contribution fromthe moon is an order smaller and may be neglected for the purposes of thediscussion but will be taken into account in data analysis)

Nonzero gravitational frequency shift signals (Eq 3) can be obtained inthree ways Method (1) below is the most sensitive and represents a strongreason to employ an optical clock which on short averaging times is ableto provide a higher frequency stability than a microwave clock Method (2)and (3) are less accurate but are complementary in terms of measurementapproach Finally method (3) addresses the solar gravitational frequencyshift while (1) and (2) measure the terrestrial effect The performance of theinstruments assumed in the following is shown in Fig 2

(1) Repeated measurements of satellite clock frequency variation betweenapogee and perigeeIn this measurement approach the difference (or ratio) between earthndashsatellite clock frequency ratio at perigee and at apogee is consideredThis difference is measured at each perigee and apogee passage andaveraged over the mission duration Systematic shifts (if not correlatedwith orbital motion) are expected to average out leading to a gain insensitivity of up to n12 where n = number of orbits This mode usesthe stability properties of the satellite clocks on the timescale fromperigee to apogee rather than their accuracy Assuming a specificationof le 3 times 10minus16 satellite optical clock fractional frequency instability forintegration times between 1000 and 25000 s n sim= 1000 comparisons(ie most orbits of the 3-year mission) and a perigee-apogee potentialdifference (U(apogee) minus U(perigee))c2 = Uc2 sim= 4 times 10minus10 allows a

Exp Astron

Fig 2 Performance of clocks and links on EGE Curve ldquooptical clocks EGErdquo is the specificationCurve ldquooptical clocks EGE (goal)rdquo refers to expected near-future performance of an opticalclock based on neutral atoms ldquoUpgraded PHARAOrdquo refers to the PHARAO microwave clockusing the high-quality microwave-optical local oscillator The microwave link (MWL) performanceexhibits a discontinuity since common-view contact time of ground stations with EGE is limited toseveral h The different slopes arise because on short time scales only white phase noise is relevantwhile on long time scales flicker noise also contributes The reproducibility is taken into account

measurement of the redshift at 25 ppb accuracy The ground clocks towhich the satellite clocks are compared to at perigee and at apogee willbe located in different ground stations The inaccuracy of these groundclocks must therefore be sufficiently smaller than 1 times 10minus17 in order notto contribute to the 25 ppb error Current state-of-the art inaccuracy is2 times 10minus17 and is likely to decrease further

(2) Absolute comparison between ground and satellite clocks with the satel-lite at apogeeHere the terrestrial gravitational potential difference is maximumU(apogee) minus U(perigee)c2 sim= 65 times 10minus10 Assuming a space optical clockinaccuracy of 1 times 10minus16 and a ground clock inaccuracy a factor 3lower (current state-of-the-art) allows a measurement of the redshift at150 ppb This measurement takes advantage of the repetitive nature ofthe orbit only to a small extent and relies on the accuracy of the opticalclocks in particular of the space clocks It is thus complementary to (1)albeit less accurateIn (1) and (2) the dominant contribution to the redshift is from Earthrsquosgravitational potential The contribution from the solar potential isstrongly reduced due to the fact that satellite and earth are an interactingsystem in free-fall toward the Sun [19] and can be modeled

(3) Comparisons between terrestrial clocksThe frequency comparison link to the EGE satellite allows terrestrialclock comparisons by a common-view method whereby the frequency

Exp Astron

signals from two distant ground clocks are sent to the satellite andcompared there The frequency difference between two identical groundclocks is determined if systematic instrumental errors are negligible bythe difference in gravitational potential at the two clock locations Thispotential difference includes a terrestrial and a solar contributionThe solar contribution varies according to the orientation of the clockpair with respect to the Sun with a 1 day-period For two clocks at asurface distance of sim6000 km and suitable orientation with respect tothe sun the normalized modulation amplitude is sim= 4 times 10minus13 This is thesame as the earth potential difference for 4 km height difference It ispossible that ground clocks will have reached an inaccuracy at the levelof 1 times 10minus18 at the time of the mission in 2015 Using the microwave linka comparison of ground clocks over sim3 h (typical common-view duration)with a resolution of 1 times 10minus17 can be performed limited by the instabilityof the link Repeating the clock comparison at each common-view win-dow (n = 1000 comparisons) will allow averaging and a measurementof the solar gravitational shift with fractional inaccuracy of 1 ppm Incomparison the best current results for the solar gravitational frequencyshift are at the few-percent level [20 21] A moderate averaging out ofground clock systematic shifts is assumed as well as modeling of the earthgravitational potential variations on the daily timescale at the appropriatelevel

(4) Comparisons of on-board clocksThe two on-board clocks are continuously compared these data areanalyzed for violation of LPI (Eq 4) Since the measurement is a nulltest the required accuracy of the gravitational potential at the satellitelocation is low and easily satisfied

24 Test of local position invariance and search for neutron scalar charge

The test of LPI is performed in conjunction with the measurement of thegravitational frequency shift

With approach 23 (1) the optical clock will be able to measure K (α Earth)with accuracy 30 ppb (spec)

Combining the results of the microwave and optical clocks will determineK(meQCD Earth) + εK(mqQCD Earth) with accuracy 85 ppb (spec)where ε asymp minus01

Simultaneously the on-board comparison 23 (4) yields the combination19 K(α Earth) +K(meQCD Earth) + εK(mqQCD Earth) with accuracy85 ppb (spec)

Note that these measurements are fully complementary to current andfuture terrestrial clock-clock comparisons which are sensitive to K(X Sun)and can be combined with the latter to determine or set limits to the neutronscalar charge

Exp Astron

25 Theoretical analysis of frequency comparison signals

The on-board clock-clock comparison signals can be analyzed straightfor-wardly in terms of a violation of LPI since signal propagation effects areabsent

The determination of the gravitational frequency dilation via ground-satellite clock comparisons is more complex (Eq 1 is strongly simplified) andrequires an accurate treatment of all relevant relativistic effects Gravitationin the solar system is described in the barycentric (BCRS) and geocentric(GCRS) non-rotating celestial reference systems by means of post-Newtoniansolutions of Einsteinrsquos equations for the metric tensor in harmonic gaugescodified in the conventions IAU2000 [22] In these conventions the relativisticstructure of Newton potential (order 1c2) and gravitomagnetic (order 1c3)

potentials are given For the Newton potential the multipolar expansion ofthe geopotential determined by gravity mapping is used The post-Newtonianeffects of the metric (described in terms of dimensionless parameters β and γ )

can be included Orbits of satellites are evaluated in the GCRS at the levelof a few cm by means of the Einstein-Infeld-Hoffmann equations at the order1c2 [23] using the relativistic Newton potential Instead the trajectories of theground clocks in the GCRS are evaluated from their positions fixed onthe Earth crust (ITRS International Terrestrial Reference System) by usingthe non-relativistic IERS2003 conventions [24 25]

For the propagation of light rays (here radio signals) between the satelliteand the ground stations one uses the null geodesics of the post-Newtoniansolution in the GCRS The timefrequency transfer properties have beentheoretically evaluated for an axisymmetric rotating body [26ndash29] at the order1c4 (ie with Newton and gravitomagnetic potentials developed in multipolarexpansions) These results have been developed for use in the ACES missionThe relative frequency dilation can be expressed as

ν

ν= 1 + 1

cL1 +

4sum

n = 2

1

cn(Ln + Gn)

where Ln describes special-relativistic Doppler effects and Gn the general-relativistic effects These quantities are time-dependent due to the orbitalmotion For concreteness we report the orders of the terms The kinematicaleffects are |L1c| lt 1 times 10minus5 |L2c2| lt 2 times 10minus10 |L3c3| lt 3 times 10minus15 |L4c4| lt

7 times 10minus20 The Newtonian contributions are G2 = G(M)2 + G(J2)

2 + G(J4)2 + G(J6)

2(M is the mass monopole Jn are the lowest multipoles) with estimates|G(M)

2 c2| lt 3 times 10minus11 |G(J2)2 c2| lt 2 times 10minus13 |G(J4)

2 c2| lt 3 times 10minus16 |G(J6)2 c2| lt

1 times 10minus16 For G3 = G(M)3 + G(J2)

3 the estimates are |G(M)3 c3| lt 2 times 10minus14

|G(J2)3 c3| lt 1 times 10minus18 Finally the G4 = G(M)

4 + G(S)4 contributions (G(S)

4 is thelowest gravitomagnetic effect) are of order |G(M)

4 c4| lt 1 times 10minus19 |G(S)4 c4| lt

1 times 10minus22 and thus not relevant for EGE

Exp Astron

The available theory is fairly complete for the data analysis in mode 23 (1)described below For modes 23(2 3) the theory will need to be generalizedto include the effect of the gravitational potential at the location of the groundclock(s) (including time-varying effects such as the tides) and of the solar andlunar potentials

The uncertainties of the various contributions above are minimized by ausing precise EGE orbit data obtained by satellite tracking and by usingcurrent and future accurate earth gravity information Required orbit positionknowledge is at 1 cm level near perigee and sim50 cm near apogee reachablealready today with the proposed orbitography approach described below Thevalidity of the special-relativistic Doppler shift has been and will continue to beindependently verified with increased accuracy by spectroscopy of relativisticatomic ion beams [30] Potential violations of this aspect of Lorentz Invari-ance are expected to be sufficiently bounded so as not to affect the signalinterpretation

26 Local Lorentz invariance tests

There has been an explosion of interest in Local Lorentz invariance in re-cent years [31 32] with numerous astronomical studies and high-precisionexperiments applied to search for violations of Lorentz invariance Signifi-cant developments have also occurred on the theoretical side and a theory(standard model extension SME [33]) has been worked out that provides aunified framework for describing and analyzing Lorentz Invariance violationsin a variety of systems Two aspects of Lorentz Invariance can be tested withEGE

261 Independence of the speed of light from the laboratory velocity

A possible dependence of the speed of light on the speed v of the laboratorycan be modeled according to the MansourindashSexl test theory [34] by

c (v) = c0(1 + Bv2

c2

)

Here v 377 kms is the velocity with respect to the cosmic microwavebackground the cosmologically preferred frame and B is a combinationof parameters describing deviations from the usual Lorentz transformationformulas B = 0 if Lorentz Invariance holds For a terrestrial experiment therotation of the Earth around its axis modulates v with a 300 ms amplitude Adependence c(v) can be searched for by measuring the frequency differencebetween a resonance frequency of a highly stable optical cavity (cavity fre-quencies are proportional to c) and an optical clock (Kennedy-Thorndike-typeexperiment [35]) and determining the modulation of the frequency differencecorrelated with the modulation of v The advantages of a space experimentare the high orbital velocity and strongly reduced cavity deformation thanksto microgravity [36 37] On the proposed elliptic orbit v varies between+4 and minus4 kms over approximately 1 h This variation is 13 times larger

Exp Astron

than for the case of a terrestrial experiment The shorter time scale is alsoadvantageous since the drift of the cavity is more predictable In the EGEmission the reference cavity of the clock laser is used The large numberof orbits improves the statistics An improvement of the accuracy of B by afactor 20 compared to the best current terrestrial results is realistic The dataanalysis can also be performed in the framework of the SME theory takinginto account complementary bounds obtained from terrestrial experimentsand astrophysical observations

262 Independence of Zeeman splitting frequency on the directionof the magnetic field

One class of tests of Lorentz Invariance consists in measuring the frequencysplitting between two levels of a quantum system induced by a static magneticfield as a function of orientation of the field direction with respect to thestars (HughesndashDrever-type experiments [31]) This class addresses the so-called matter sector and is complementary to combined-photonmatter-sectortests such as the above The most precise experiments are performed withatomic clocks eg masers or cold atom clocks [38] The EGE instruments allowsuch experiments since magnetic fields are applied continuously in the clockor repeatedly for calibration Compared to terrestrial experiments the highorbital velocity of EGE could lead to an improvement by a factor sim20

27 Application to geophysics

Using GPS coordinates of a point on the Earth can be obtained with aninaccuracy well below 1 cm in a well defined international terrestrial referencesystem The coordinates are purely geometrical and do not contain any gravityinformation A local gravitational potential UB is determined with respect toa reference gravitational potential U A by ldquolevellingrdquo the surface of the Earthie sequentially measuring gravity g and height differences every ca 50 m

UB minus U A = minusBint

A

g (r) middot d n

where n is the difference vector between subsequent locations The disadvan-tage of measuring height differences by the levelling method is a random walkeffect of accumulated errors

In the classical definition a geoid is defined as the particular equipotentialsurface nearest to mean sea level In these terms the geoid serves as areference surface for measuring the height and also to define a referencefor the gravitational potential Such a vertical reference frame historicallywas realized for a country or several countries by determining the mean ofsea level observations at tide gauge stations taken over long period of timeHowever modern satellite altimetry missions such as TOPEXPOSEIDON orENVISAT show that departure of the mean sea level from an equipotential

Exp Astron

surface may reach up to several meters on the global scale see eg [39]Therefore height systems based on different tide gauge stations may differin the realisation of the geoid and may differ with respect to the reference byseveral meters

The best gravity field models obtained from satellite data (eg missionGRACE) have reached a precision of about 1 cm over sim250 km half wave-length [40] Satellite data also reveals changes in time [41] However for typicalEarth topography with height variations of eg 1000 m over 30 km horizontaldistance one may expect variations in the geoid of about 80 cm Such ahigh-spatial-frequency signal in geoid variations cannot be detected by spacegravity missions and requires a combination of satellite and terrestrial gravitymeasurements like gravity anomalies deflections of vertical and GPSlevellingpoints The best combined global gravity field models are provided up to de-gree and order 360 in terms of spherical harmonic representation correspond-ing to a half-wavelength of about 55 km However in combining the satelliteand terrestrial gravity field measurements the problem of terrestrial data givenin different height systems between continents and different countries remainsand has to be tied to satellite measurements

The geodetic scientific community is currently establishing a Global Geo-detic Observing System (GGOS) [42] Its objectives are the early detectionof natural hazards the measurement of temporal changes of land ice andocean surfaces as well as the monitoring of mass transport processes in theEarth system Global change processes are small and therefore difficult toquantify Therefore the required precision relative to the Earthrsquos dimensionis 1 ppb GGOS will be established by the combination of geodetic spacetechniques (GPS Laser-Ranging very large baseline interferometry) andrealized by a very large number of terrestrial and space-borne observatoriesThe purely geometric terrestrial 3D coordinate system is in good shape andfully operational It has to be complemented however with a globally uniformheight system of similar precision The current precision level of regionalheight systems in terms of gravity potential differences is in the order of1 m2s2 (10 cm) with inconsistencies between these various systems up toseveral 10 m2s2 (several meters) The actual requirement in the context ofGGOS is 01 m2s2 (1 cm) with the need of a permanent ie dynamical controlThis requirement of high precision height control comes from the need tounderstand on a global scale processes such as sea level change global andcoastal dynamics of ocean circulation ice melting glacial isostatic adjustmentand land subsidence as well the interaction of these processes Only by meansof monitoring in terms of gravity potential changes at the above level ofprecision the change of ocean level can be understood as a global phenomenonand purely geometric height changes be complemented by information aboutthe associated density or mass changes

The geoid can also be defined in a relativistic way [43] as the surface whereaccurate clocks run with the same rate and where the surface is nearest tomean sea level The relation between the differences in the clock frequenciesand the gravitational potential is given in simplified form by Eq 1 At present

Exp Astron

there is no operational way to compare frequencies of the already availableoptical clocks on the global scale at the same level as their accuracy wouldallow With EGE and by then improved ground optical clocks it will becomepossible to obtain gravitational potential differences on a global scale by com-paring frequencies This would be a significant new dimension to gravity fielddetermination since such observables are given on the global scale and providein situ local gravity information at the same time EGE will allow clock-basedgravitational potential mapping with the same payload instruments used forthe fundamental physics experiments The on-board clocks are actually notrequired for this purpose MOLO MWL and FCDP suffice Thus EGE willallow establishing a global reference frame for the Earth gravitational fieldwith accuracy in the order of a few cm in terms of the geoid heights Thisassumes that by the time EGE is operational (mobile) ground clocks withfractional inaccuracy at the level of 10minus18 are available the measurementsare limited by MWL noise and thus require a long integration times to matchthe expected ground clock performance Such a reference frame will serve asreference for all gravity field modeling and clocks will be used to define theinternational atomic time scale and the Earth gravitational potential at thesame time

In summary with optical clocks geodesy will undergo a second revolutionafter geometry now being measured by clock-based GNSS systems also physi-cal heights and potential will be measured by (mobile) optical clocks using thegravitational redshift effect

28 Application to frequency standards and terrestrial fundamentalphysics studies

As has been described repeating ground clock comparisons and integratingover several months will provide global ground clock frequency comparisonsat the level of 1 times 10minus18 more than two orders of magnitude below the currentGPS and two-way satellite time and frequency transfer methods In the timeand frequency community the availability of such a global high performancemicrowave link will accelerate the process leading to a redefinition of the SIsecond based on clocks operating in the optical domain

The global character of the ground clock comparisons made possible byEGE can also contribute to the search for a time-variation of the fundamentalconstants

The comparisons do not require availability of the on-board clocks The riskof this type of measurements is therefore reduced Moreover the lifetime ofmany subunits required for this measurement (MWL and FCDP see below)can exceed 10 years (from previous experience with similar devices) andthus offers a motivation to increase the mission duration beyond the nominal3 years

Further applications to metrology are listed under the secondary goals inSection 21 above

Exp Astron

3 Mission profile

31 Orbit

The orbit must satisfy the following requirements

Primary 1 Large gravitational potential difference between apogee andperigee

2 sufficient contact time with at least two ground stations atperigee

Secondary Simultaneous visibility of satellite from distant (inter-continental) ground stations for several hours

To satisfy the primary objective the orbit must be highly elliptic The perigeeshould be low A very low perigee means a short contact time which decreasesthe accuracy of the frequency measurement A compromise is a perigeealtitude of sim2500 km For a sufficiently high apogee this yields a U of 60of the theoretical maximum between the earth surface and infinity

The baseline scenario is an orbit of approx 12 h period with the followingproperties

Apogee Altitude 37856 km (minus 10 deg latitude one over Atlantic andPacific resp)

Perigee Altitude 2500 km (+ 10 deg latitude one over Guatemala andMalaysia resp)

Inclination 634 argument of perigee 170RAAN 25 True Anomaly 220Apogee drift rate 70 kmday perigee drift rate sim 14 kmday

True anomaly and right ascension of the ascending node (RAAN) wereoptimized to minimize orbital precession Figure 3 shows the ground track andthe ground coverage at one of the perigees Common-view contact time forstations in Europe is about 6 h while between a station in Europe and in theUSA it is about 35 h Frequency comparison windows between a given groundstation and EGE repeat every 24 h

A final selection of the orbit requires an in-depth study of the costs of thepropellant and maneuvers as well as of the setting up of new ground stations

32 Ground segment

The mission requires two separate ground segments the science groundsegment and the standard satellite control segment which supports telemetryand telecommand (TMTC)

Exp Astron

Fig 3 Baseline EGE orbit One (out of two) apogee and one (out of two) perigee ground pointsare shown The ground coverage at one of the perigees is shown with circles at elevation anglesteps of 10 starting from 10

321 Science ground segment

As described above one part of the mission is to perform frequency com-parisons between ground-to-space and ground-to-ground (common-view andnon-common view) The use of a two-way microwave link (MWL) systemhas the advantage that a precise orbit or the precise relative velocity of thesatellite must not be known [4] The orbit determination requirements fora two-way system are significantly relaxed compared to the requirements tomodel the relativistic effects on the clocks The frequency comparison resultsare available both on ground and on the satellite

For ground-to-satellite clock comparisons five ground stations arefavourable two capable of viewing perigee three for viewing apogee Thisallows for the classic three-clock configuration where the performance of eachclock can be obtained individually The perigee stations will be observing thesatellite for sim25 min every perigee passage for the duration of the missionThe apogee stations will observe for sim3 h every apogee passage Of coursea larger number of stations as well as longer observation time are usefulfor robustness redundancy and orbit determination The stations could beinstalled on a temporary basis for the duration of the mission

The stations must have access to optical clocks of performance at least ashigh as todayrsquos state-of-the-art ground clocks but preferably higher Theywill include an optical frequency comb by means of which the optical clockfrequency is transformed to a radio frequency for comparison with the EGEclocks High-accuracy fiber-optic link technology already existing today allowsdistances of sim100 km between the ground station and the correspondingoptical clock allowing a certain freedom in the choice of the clock locationWith further increase of the fiber-optic link range currently being investigated

Exp Astron

the area over which the ground station can be positioned relative to the opticalclock will increase This option is of importance for the implementation of ageophysics measurement campaign for which a large number of clock locationsis of interest

The above ground stations will be complemented by some 15 additionalground terminals provided by the world-wide laboratories equipped with highperformance primary clocks at the time of the mission These will participate inorbit determination to 1 cm accuracy along the full orbit which can be achievedwith non-optical atomic clocks The EGE MWL is fully compatible to plannedACES ground terminals which might reduce the number of required stations

With a larger number of transportable optical clock ground stations de-ployed and moved across the continents a geophysics campaign could beimplemented to map the gravitational potential

Ground station equipment MWL uses dedicated small transportable termi-nals which can be easily installed at any location of interest Each terminalconsists of a steerable dish antenna of 1 m diameter with pointing accuracy of01 using a transmitted power of 5 W per band The high directivity antenna(beam aperture Ka-band 07 Ku-band 14 S-band 20) avoids multi-pathfrom nearby objects Antenna noise temperatures are Ka-band 450 K Ku-band 340 K Operation is at elevations above 10 and measurements areperformed above 20 For ranging purposes terminal location is determined byGPS surveying Each ground station computes its own ionospheric correctionsbased on a triple-band receiver The terminal has a built-in delay monitorwhich allows calibrated ranging during several months once its bias has beendetermined by laser ranging Its M and C interface is via local area networkto either a userrsquos computer or directly to the science network control centreby remote MampC The stationrsquos operation is fully automatic and unattendedSchedule and orbital data are received via LAN once per day Signal acquisi-tion is fully autonomous Data transmission is after each pass The terminal hasan uplink telecommand capability of 1 kBits and a downlink data capacity ofup to 10 kBitss For the local user there is a visual interface providing quick-look data from real-time comparison to the spacecraft clocks to verify theirgood health The final full performance product is calculated at the sciencedata centre

Local metrological data (temperature pressure humidity) are recorded tocorrect for tropospheric delay of the ranging data The main advantage ofMWL terminals compared to laser based systems is their all-weather capabil-ity although microwaves suffer higher propagation errors in the ionosphereand troposphere

The EGE ground terminal will profit from the ERS-2 mission PRARE(precise range and range rate equipment) [45] experiments whose terminalshad similar design and have been operated under extreme climatic conditionsincl Antarctica and high temperature regions from existing designs fromthe space segment (mission ACES) and no further special developments aredeemed necessary Cost per EGE master station is modest

Exp Astron

Satellite tracking via laser ranging Laser ranging is alternative to MWL rang-ing and also serves for absolute calibration of the latter For this purpose anon-board corner cube reflector (CCR) is provided Ranging will be done withthe International Laser Ranging Service Ideally the satellite will be rangedat every perigee passage and every apogee passage simultaneously with thefrequency comparison procedures Todayrsquos ranging precision is sufficient tosatisfy the science goals For apogee observations in the northern hemispherea large number of laser ranging stations is available The number of stationscapable of ranging at or near perigee in the southern hemisphere is six andthus is sufficient even considering unfavourable weather conditions at someof these The MWL ground stations are able to range even under thoseconditions

322 Control ground segment

Average data transfer rate is estimated at 300 MBday (approx 40 kBs onaverage) Data will be stored on board (2 GB capacity required) and will betransferred to a receiving station on the Earth once a day

XPLC

incl Data StorageIF to TMTC Link

HK amp TMTC

to ground

Opt Clock

AOM

MOLO

PHARAO

FCDP

10 GHz

100 MHz

250 MHz

GNSS Rx

MWLTampF Link

uplink Data

UTC(k)

TampF Space vs Ground

TM from GS

Tim

e S

cale

Orbit GNSS Time UTC

TampF Space vs Ground

MampC Time Scale

MampC

MampC

MampC

MampC

Mamp

C

PHARAO vs Opt Clock

PHARAO vs Opt Clock

10 GHz amp Epoch

10 MHz

amp Epoch

Epoch

10 MHz

amp Epoch

Opt Link

Fig 4 Schematic of the payload showing subunits and their signals The results of the frequencycomparison between the two on-board clocks (optical clock and microwave clock PHARAO) areobtained from a combination of the frequency values applied to the frequency shifters AOM (seeFigs 7 and 8) and those output from the frequency control and distribution package (FCDP)as indicated by the arrows ldquoPHARAO vs Opt Clockrdquo The satellite clock to ground clockcomparison results are output from the MWL XPLC is the on-board computer and is involvedin generation of signals for instrument control The optical link shown dashed is a possible add-onproviding an alternative to the MWL GS ground stations MampC monitoring and control HKhousekeeping TampF time and frequency

Exp Astron

4 Main payload instruments

41 Overview

The satellite payload consists of a pair of cold-atom clocks a microwave-optical local oscillator (MOLO) an on-board frequency distribution unit(FCDP) a satellite-to-ground clock comparison unit (MWL) and the onboardcomputercontrol system (XPLC) see Fig 4 Auxiliary units are the posi-tion determination GPSGalileo receiver (GNSS Rx) attitude monitor (startracker not shown) and the corner cube reflector array (CCR not shown)

The science goals require clocks of very high frequency stability on thetimescales of half the orbital period (6 h) as well as perigee passage (sim1000 s)see Fig 2 Only clocks based on cold atoms are suited for the science goalsTable 1 reports values for a subset of atomic clocks that have been investigatednot including clocks that require cryogenic operation or are too complex foruse in an medium-class mission The choice of a clock pair rather than a singleone fulfils several requirements it addresses more science goals improves thevalidity of some science results and it enables calibration accuracy evaluationand redundancy The validity of the science results is enhanced since theclocks differ not only in the employed atoms but also in terms of havingdifferent sensitivity to perturbations by electric and magnetic fields electronicserrors and misalignments A comparison of the results arising from each clockallows for an understanding of measurement errors Both the clocksrsquo and thefrequency comparison unitrsquos performances will be determined after launchand regularly verified during the course of the mission The only reliable

Table 1 Properties of different clock candidates for EGE

Clock Clock Fractional σ = Fractional Sensitivity Figure of Sensitivityatom frequency inaccuracy instability to α Aα merit |Aα| to meq

at 1000 s σ [1016] QCD

Sr+ 445 THz 3 times 10minus17 (p) 5 times 10minus17 (p) 042 08 (p) ndashYb+ (1) 688 THz 3 times 10minus16 (d) 3 times 10minus16 (d) 088 029 (d) ndash

3 times 10minus17 (p) 1 (p)Yb+ (2) 642 THz lt 1 times 10minus17 (p) 1 times 10minus17 (p) minus53 53 (p) ndashSr 435 THz 1 times 10minus16 (d) 1 times 10minus16 (d) 009 009 (d) ndash

lt 1 times 10minus17 (p) 09 (p)Yb 519 THz 7 times 10minus16 (d) 1 times 10minus16 (d) 031 03 (d) ndash

lt 1 times 10minus17 (p) 1 times 10minus17 (p) 31 (p)Cs 91 GHz 2 times 10minus16 (d) 1 times 10minus15 (d) 275 027 (d) 1Yb+ 126 GHz 1 times 10minus15 (p) 1 times 10minus15 (p) 34 034 (p) 1

A relevant figure of merit is the ratio of |AX | and fractional frequency instability σ on the perigeevisibility timescale The Cs clock is the only high performance clock with adequate sensitivity tothe particle masses me mq The indicated Cs clock performance was achieved using a microwavelocal oscillator based on a cryogenic oscillator On EGE the latter is to be replaced by an opticallocal oscillator-based device (MOLO) with similar short-term performance but without the needto operate at cryogenic temperature The clock instability at 10000 s is typically similar to theindicated inaccuracy(p) projected (d) demonstrated (1) quadropole (2) octupole

Exp Astron

way to achieve this is by having at least two clocks on board The on-boardclock-clock comparison provides performance evaluation independent of theground-satellite comparison unit Once the satellite clock performances areestablished the link performance can be established as well

The clocks are similar in basic structure but differ in the detailed implemen-tation They consist of

a source of atomsa preparation subunit which cools the atoms to near-standstill (millikelvintemperature or below)a trapping subunit that confines the atom(s) to a tiny volumean interrogation subunita clock control electronics package

The clocks share a common central unit the optical-microwave local oscil-lator which serves as a flywheel oscillator with outstanding stability on shorttimescales (01ndash10 s) for both clocks It consists of a laser stabilized to an ultra-low loss optical resonator made of ultra-low-expansion glass A microwavesignal is derived from the laser by phase-locking a frequency comb to the laserwave and using a multiple of the repetition rate

While the subunits of any optical clock have a similar technological basis(optics electrooptics acoustooptics thermal control vacuum systems) theydiffer in their specifications Depending on the choice of the atom species de-livery confinement and interrogation methods clocks of different perfor-mance result A number of laboratory optical clock demonstrators have beendeveloped worldwide This ensures availability of the know-how for the indus-trial implementation of the flight models

The first clock is a microwave clock based on the interrogation of slowlymoving ensembles of cold Cs atoms It is essentially based on the PHARAOclock [5 6] developed to engineering model level by CNES in the frame-work of the ESA ISS mission ACES In EGE this clock is upgraded usingthe microwave-optical local oscillator This improves significantly the clockperformance The specification is a relative instability (Allan deviation) of3 times 10minus14 (τ s)minus12 ie at the integration time τ = 1000 s it drops to 1 times 10minus15and to 3 times 10minus16 at 10000 s While this specification is at least a factor 3 lessstringent than for the optical clocks the significantly lower cost of obtaining aflight model and the sensitivity of its frequency to the particle masses makes ita suitable choice

The optical clock baseline instrument is a single-ion optical clock Amongvarious suitable ion species (Hg+ Al+ Sr+ Yb+) [44ndash47] the Yb+ ion hasbeen selected Its breadboard performance is established at the instability levelof 3 times 10minus16 over 1000 s it exhibits a large sensitivity coefficient Aα hassmall sensitivity to magnetic fields and exhibits significant potential for furtherimprovements (use of the octupole transition with sim6 times larger Aα andlower overall instability [48] see below) A trapped Yb+ ion is also a veryrobust system it has been demonstrated that a single Yb+ ion can remaintrapped uninterruptedly for many months Finally Yb+ is the only ion with

Exp Astron

which optical clocks demonstrators have been operated in not a single but twomajor metrology laboratories

Another attractive optical clock is based on ensembles of cold neutral Sror Yb atoms trapped in an optical lattice produced by laser beams This clocktype has a lower instability than a single-ion clock thanks to the large numberof atoms (sim105) used Its development is in progress in several laboratoriesin the world (in part supported by space agencies) and has to date reached6 times 10minus17 instability at 5000 s and 1 times 10minus16 uncertainty [49 50] with potentialfor even higher performance However it is more complex and costly thanthe Yb+ ion clock since the cold atom preparation and trapping subunitsinclude more components require higher laser powers and the respectiveoperating procedures are more complex It is therefore not considered for thepresent mission designed to be developed in the very near future with limitedresources but is an interesting candidate for missions without such constraintsIts estimated physical parameters are indicated in Table 2 for comparison

42 Main instruments

421 Microwave clock

The PHARAO cold atom clock is one of the two atomic clocks of the spacemission ACES managed by ESA It is proposed here to include a secondversion of this instrument as its costperformance ratio is very attractive forthe EGE scientific objectives Its concept is very similar to ground basedatomic fountains but with the major difference of zero-g operation Atomsslowly launched in free flight cross two microwave fields tuned to the transitionbetween the two hyperfine levels of the cesium ground state The interrogationmethod based on two separate oscillating fields (Ramsey scheme) allowsthe detection of an atomic line whose width is inversely proportional to thetransit time between the two microwave cavities The resonant microwavefield at 9192631770 GHz (SI definition of the second) is synthesized startingfrom an ultrastable quartz oscillator (USO) in the FCDP phaselocked to themicrowave output of the microwave-optical local oscillator and stabilized tothe clock line using the error signal generated by the cesium resonator In thisway the intrinsic qualities of the cesium hyperfine transition both in terms ofaccuracy and frequency stability are transferred to the USO In a microgravityenvironment the velocity of atoms along the ballistic trajectories is constantand can be changed continuously over almost two orders of magnitude (5ndash500 cms) Differently from atomic fountain clocks presently operated onground very long interaction times (up to few seconds) will be possible whilekeeping reasonable the size of the instrument

The instrument consists of 4 main elements a laser source which providesthe light required to cool and detect the cesium atoms (Fig 5 left) a frequencychain which generates the 92 GHz microwave signal a cesium vacuum tubewhere the interaction between the microwave field and the atoms occurs

Exp Astron

Table 2 Resources and technology readiness level (TRL) of the EGE spacecraft

Unit Volume Mass Power TRL Comment[l] [kg] [W]

Ion clock 100 70 60Atom source 4Atom preparation 6Atom trap 4Atom interrogation 6Clock control 6

Microwave clock 200 91 110 8 TRL by end 2008Local oscillator

Stabilized laser 18 15 23 6Frequency comb 20 16 38 6

Microwave link 18 18 55 8FCDP 7 8 12 8GPS receiver 2 5 7 9CCR 2 2 0 9Payload sum 367 225 305Thermal control 14 17 30 8OBDH 20 24 38 9Star tracker 1 1 8 9Radiation shield 10 40 0 9Support structure 170 204 0 9Power incl solar arr 50 58 10 9Propulsion 21 25 20 9AOCS 36 43 82 9TMTC 8 10 15 9Satellite bus sum 330 422 203 Flexbus conceptPayload plus bus incl 20 margin 8364 7764 6096Propellant 40Launch adapter 100SC wet masspower incl Margins 9164Lattice clock 135 105 152 Alternative clock

Atom source 4Atom preparation 6Atom trap 4Atom interrogation 6Clock control 6

Also included are the resources for an optical lattice clock based on neutral Strontium atoms

(Fig 5 right) and a computer The engineering model of PHARAO is fullyassembled at CNES Toulouse and undergoes functional and performancetests until the end of 2008 The operating temperature range is 10ndash33 andnon-operating temperatures are minus45 +60 The cesium tube consists in aUHV chamber pumped by getters and a 3 ls ion pump a cesium reservoira microwave cavity coils to provide a uniform magnetic field and three layersof magnetic shields Vacuum windows with fiber-optics collimators enable thetransmission of the cooling and detection light onto the cesium atoms

In autonomous mode PHARAO uses the USO as interrogation oscillatorand will provide a clock signal with fractional frequency stability below1 times 10minus13 (τs) minus12 and inaccuracy near 1 times 10minus16 When using the microwave-

Exp Astron

Fig 5 Left The PHARAO laser source engineering model with cover removed Dimensionsare 530 times 350 times 150 mm3 and the mass is 200 kg Ten polarization maintaining optical fibers(yellow) guide the laser beams to the cesium tube All diode lasers are mounted on Peltier coolersfor temperature regulation within 2 mK (courtesy EADS SODERN) Right The PHARAOengineering model cesium tube without the two external magnetic shields The volume is 990 times336 times 444 mm3 and the total mass is 44 kg

optical local oscillator PHARAO will provide a reduced frequency instabilityof 3 times 10minus14 (τs) minus12 This corresponds to 1 times 10minus15 at 1000 s and 3 times 10minus16

at 104 s

422 Ion optical clock

The 171Yb+ ion clock shown in Figs 6 and 7 provides an atomic frequencyreference based on a single 171Yb+ ion confined within an RF electric end-captrap (similar to a Paul trap) within an ultra high vacuum (UHV) chamber andlaser cooled to sim1 mK close to the Doppler cooling limit As a result the ionrsquosfirst order Doppler motion is completely removed and the ion experienceslittle collisional perturbation from its environment Under these conditionsthe narrow linewidth 2S12 (F = 0 mF = 0) minus 2 D32 (F = 2 mF = 0) quadrupole

Fig 6 Left Simplified level scheme of 171Yb+ showing relevant cooling (blue) auxiliary (red)quadrupole clock (green) and octupole clock (dark blue) transitions δνnat is the natural linewidthof the clock transition Middle close-up of Paul-type RF trap (PTB Braunschweig) the ion istrapped in the hole (diameter 14 mm) Right Complete ion trap with vacuum flange height 10 cm(NPL Teddington)

Exp Astron

935 nm D32-state repumper laser

638 nm F72-state repumper laser

739 nm extended cavity diode laser

AOM

PMT

Ion trap UHV chamber

vacuum pumps

Fibre-free space interface

Mu-metal shielding

EC

T

B

Magnetic field

Temperature

Ion fluorescence Processor

Clock laser frequency step and control

4355 nm

Frequengy doubling II

3695 nm cooling wave

Low-drift cavity

Low-drift cavity 171Yb+ 435 nm ion clock unit

876 nm clock laser wave from MOLO

Frequengy doubling

Fig 7 Simplified schematic of the single-ion optical clock unit AOM acousto-optic frequencyshifter PMT photomultiplier tube EC endcap T temperature stabilization system B magneticfield

clock transition at 4355 nm can be interrogated by spectrally narrow and stableclock laser light A diagram of the simplified energy level scheme for 171Yb+showing the relevant cooling auxiliary and clock transitions is given in Fig 6All these wavelengths are provided by diode lasers operated in fundamentalor frequency doubled mode The clock laser light is obtained from the 871 nmwave provided by the MOLO unit (see Section 423) after frequency-doublingin single pass by a KNbO3 nonlinear-optical crystal to 4355 nm The clocktransition spectral profile is observed by stepping the clock laser frequency bymeans of an acousto-optic frequency shifter (AOM in Fig 7) and recordingthe statistics of the quantum jumps in cooling laser induced fluorescence asa function of this frequency Under clock operation the AOM repeatedlysteps back and forward between the half-intensity points on the transitionprofile monitoring the signal imbalance between these points Any detectedimbalance is servo-corrected to zero by feedback to the AOM

In detail the optical clock architecture will comprise (Fig 7)

bull an RF end-cap trap for ionising and confining a single 171Yb+ ion withinan ultra-high vacuum chamber pumped by a small ion pump and non-evaporable getter pump The trap is driven by means of an ac voltage ofa few hundred volts at a drive frequency of sim10 MHz A small oven withseveral mg of 171Yb isotope is heated to provide a low flux of Yb atomsfrom which single ions can be ionised by electron bombardment from ahot wire filament within the trap potential well Low voltage dc is applied

Exp Astron

to additional electrodes to position the ion precisely at trap potentialcentre in order to minimise the ionrsquos micromotion at the drive frequencyThis needs automatic periodic monitoring and correction especially in theperiod after re-loading an ion into the trap External magnetic field coilsin three orthogonal axes allow the nulling of external fields and settingof a fixed field of sim1 μT The trap and coils are surrounded by mu-metalshielding to minimise external field changes during the orbit A level oftemperature control of 1 K of the region surrounding the trap is requiredfor maintaining the blackbody-induced frequency shift uncertainty withtemperature at the 10minus17 level

bull a laser platform to provide frequency doubling of an extended cavity739 nm diode laser for Doppler cooling of the ion on the 2S12(F = 1) minus2 P12 (F = 0) dipole transition at 3695 nm Light from an amplified diodelaser device frequency-doubled in a periodically poled LiTaO3 crystal insingle-pass will be used Typical laser powers used for driving the coolingtransition below saturation are about 2 μW for a beam waist in the trapof about 50 μm By modulation of the injection current of the 739 nmdiode laser at 147 GHz a sideband is generated that excites the 2S12 (F =0) rarr2 P12 (F = 1) ldquorepumperrdquo transition in order to avoid optical pump-ing between the ground hyperfine states

bull The platform also houses the auxiliary lasers at 935 and 638 nm The935 nm diode provides repumping of the ion from the 2 D32(F = 1)metastable level after occasional branching decays to this level during thecooling sequence The 638 nm diode allows fast recovery of the ion fromthe very-long-lived 2F72 metastable state after very occasional collisionaldecay to that state Currently extended cavity lasers are forseen butdistributed-feedback lasers may be available in the near future

bull a high NA lens imaging system and photomultiplier detection system torecord the statistics of 369 nm fluorescence quantum jumps as a functionof clock laser frequency step

bull a fibre system to deliver the various cooling auxiliary and clock light fromsource to trap making use of achromatic doublets where necessary at thefibre-free space interface for launching into the trap

bull a monitoring and control processor which provides the primary coolingand clock laser pulse magnetic field and detection sequencing to observeand lock to the ion clock transition frequency The processor also monitorsfrequency and amplitude data necessary to determine normal laser andion operational conditions and initiate resetting and recovery algorithmswhere necessary and laser unit failure The actual magnetic field presenton the atomic volume is determined and actively stabilized so that aninaccuracy at the level of 1 times 10minus17 can be achieved

bull a redundancy level of at least two units for cooling clock and repumperlasers plus similar for frequency doubling crystals All redundancy unitsfor each wavelength will be fibre multiplexed as standard allowing redun-dant unit activation on determination of prior unit failure

Exp Astron

For an optimized quantum-projection-noise-limited performance of thesingle-ion clock Ramsey interrogation with a cycle time equal to the lifetimeof the metastable D32 level of 50 ms will be performed leading to an Allandeviation of about 27 times 10minus15 (τ s)minus12 This is sim45 times 10minus17 at 1 h satisfyingrequired stabilities for the science objectives Current laboratory results are astability of 1 times 10minus15 at 30 s averaging down to 3 times 10minus16 at 1000 s (PTBGermany) Also comparisons between two independent traps showed agree-ment at the level of several parts in 10minus16 over a series of eight measurementruns

The 171Yb+ ion clock offers a second option namely the 2S12ndash2 F72 octu-pole transition at 476 nm (see Table 1 and Fig 6) [48] It has an extremelylong upper state lifetime low electric quadrupole and second order Zeemansystematic shifts and therefore is a candidate for an even higher performanceclock if a clock laser system with lower instability than the above is providedApart from the clock laser it shares the laser system of the quadrupoleclock Thus with moderate additional resources the Yb+ ion apparatus couldprovide two optical clocks in a single package and specifically perform directfrequency comparisons between the quadrupole and the octupole transitionwith a large sensitivity Aα

The ion clock will operate under microprocessor control providing thecontrol algorithms for the integrated pulse sequencing for cooling repumpingclock interrogation magnetic field switching cooling fluorescence monitoringThe controller will also provide error flags for critical processes and systemreset and correction where necessary This will include ion re-loading andmicromotion reduction algorithms Automatic monitoring of laser power andspectral quality with signal re-optimisation or laser failure determination isrequired Redundant laser units will be pre-aligned and multiplexed into thefibre delivery to the trap so that redundant units can be readily activated in thecase of diode laser failure

423 Microwave-optical local oscillator (MOLO)

This oscillator provides two ultrastable frequency outputs one in the opticalregion the other in the microwave region the two frequencies being coher-ently related The MOLO is composed of three subunits the clock laser thereference cavity and the frequency comb see Fig 8

The clock laser subunit delivers the light for the excitation of the opticalclock transition It is generated by an extended-cavity diode laser at a wave-length of 871 nm Most of the output wave of this laser is sent to the ionclock The laser subunit containing two lasers for redundancy will require3 kg 3 l 15 W The frequency instability of the clock laser must be superiorto the minimum clock instability 27 times 10minus15 (τ s)minus12 for τ up to the timeconstant (sim10 s) of the servo system that locks the laser to the ionrsquos resonancesignal In order to achieve this the laser frequency is stabilized to the reference

Exp Astron

871 nm clock laser

EO

Detector and servo

AOM Femtosecond

frequency comb

To optical clock 10 GHz To FCDP

Microwave-optical local oscillator

ULE cavity

Fig 8 Microwave-optical local oscillator (MOLO) consisting of clock laser reference ULE cavityand femtosecond frequency comb (OFC) Radiofrequency inputs and outputs are not shown EOelectro-optic modulator ULE ultra-low expansion AOM frequency shifter

cavity where the PoundndashDreverndashHall method is used About 10 μW of thelaser output are used for this

The system will exhibit sim04 Hz (sim1 times 10minus15 relative) linewidth and afrequency drift lt01 Hzs The cavity consists of two high-reflectivity mirrorsoptically contacted to a cylinder of diameter 70 mm length 150 mm made ofULE (a glass exhibiting ultra low thermal expansion coefficient) with a cavityfinesse of about 300000 Residual vibrations of the satellite (arms lt 1 times 10minus6 g)should not be a limiting factor to the desired linewidth for safety marginan optimized cavity support can reduce the sensitivity to accelerations Foracoustic and thermal isolation the cavity resides inside an aluminium vacuumchamber equipped with a small (3 ls) ion-getter pump Two stages of polishedaluminium shields are implemented around the cavity which are activelytemperature stabilized so that the cavity is precisely kept at the zero crossingtemperature of the ULE thermal expansion coefficient Heat application andremoval is by thermoelectric elements between the shield and the vacuumchamber Additionally the entire vacuum chamber is temperature-stabilized

During transportation and launch the cavity will be rigidly fixed insidethe vacuum chamber by pressing on it from the sides by piezo-mechanicalactuators allowing accelerations of several g After bringing the satellite onthe orbit the actuators will be released

The optical setup for frequency stabilization will be a classic Pound-Drever-Hall scheme in reflection using an electro-optic modulator (EO) Couplingof the clock laser radiation to the cavity is via single-mode polarization main-taining optical fiber The fiber in- and out-couplers are mounted on miniaturepiezo-motor driven multiaxis translation stages Microprocessor control ofthe translation stages performs a laser-to-fiber incoupling stabilization andlaser-to-cavity mode-matching optimization (positioning inaccuracy lt10 μmalignment inaccuracy lt10 μrad)

The frequency stability of the clock laser will be limited by thermal 1f noisein the mirrors with an fractional Allan deviation slightly below 1 times 10minus15 from1 s to a few 10 s Resources including control electronics are 15 l 12 kg 8 W

Exp Astron

A femtosecond fiber laser based optical frequency comb generator (OFCFig 8) will be used for the clock frequency comparisons and to obtain aspectrally ultrapure microwave local oscillator output for the operation ofPHARAO for the microwave link and for the GNSS receiver The fem-tosecond laser is based on erbium-doped fiber pumped by fiber-coupled laserdiodes It emits 100 fs pulses at a repetition rate of 250 MHz with an averagepower of 100 mW The emission spectrum is centered around 1560 nm Spec-tral broadening to more than one octave is obtained via nonlinear interactionin a microstructure fibre so that the range from about 1050 to 2100 nm iscovered

Second harmonic generation of a portion of the broadened spectrum allowsobtaining the required light for the measurement of the clock laser frequencyat 871 nm The OFC will be referenced to the clock laser frequency (afterintroduction a controllable shift via the frequency shifter AOM) by makingthe carrier envelope offset frequency vanish so that the repetition rate fr isa high-order subharmonic of the optical clock reference frequency f1 = m1 frThe link between the optical and the microwave domain is then simply thedistribution of the repetition rate frequency or more conveniently a high(order 40) harmonic of the comb repetition rate 10 GHz

It has been demonstrated by several groups that frequency comparison andfrequency division with an OFC can be performed to very high accuracy andstability beyond the limitations imposed by the optical clocks themselves Forexample a fiber laser OFC with an instability (Allan deviation) of 6 times 10minus17

in 1 s has been demonstrated [51] OFCs based on fiber technology possessgood long-term reliability and high power efficiency The projection for a spacequalified OFC based on current industrial terrestrial technology is 16 kg mass20 l volume and 38 W power requirement

424 Frequency comparison and distribution package (FCDP)

This package distributes the ultra-stable 10 GHz reference frequency outputof the MOLO to the MWL performs frequency comparisons between MOLOand PHARAO and generates a variety of different reference signals for theindividual instruments including the 250 MHz reference clock to the OFCAll signals are coherent to the MOLO In the FCDP a USO is locked to theMOLO signal by a digital PLL A divider chain generates timing pulses (1 pps1000 pps etc) and keeps a continuous local time scale From the digital phasecomparison data the phase noise density spectrum between the USO and theMOLO is available by telemetry (TM) to assess the overall frequency purityin-flight

As the FCDP distributes the full performance reference from the MOLOdirectly to the MWL there is minimum impact of the FCDP on overall clockcomparison performance The USO-phase lock loop uses a similar designas for ACES-FCDP The digital PLL with linear phase detector allows theUSO to be slightly offset from its reference which is selectable by groundcommand This results in an accurate on-board time scale despite any offset

Exp Astron

of the MOLO reference frequency which can be synchronised to UTC Thefrequency synthesis chain re-uses modules developed for ACES-FCDP andtheir frequency inaccuracy Expected relative uncertainty (Allan deviation)of the frequency comparison between two signals provided to the FCDP is47 times 10minus17 at 1000 s 95 times 10minus18 at 10000 s

During non-availability of the MOLO ie during system start-up andcontingency conditions the USO runs as flywheel (ie it is unlocked withinstability lt15 times 10minus13 from 1 to 100 s) It will still provide useable butless accurate references to the instruments including the reference to OFCCommon view ground-to-ground time- and frequency comparison and supplyof the remaining instruments remain possible at the USO provided accuracy

Mass volume and power consumption are 8 kg 7 l 12 W Controldata rateis 10 kBday 50 kBday (FCDP alone)

425 Microwave link (MWL)

The microwave link is a two-way triple-frequency microwave link (Fig 9) Itis designed to perform accurate frequency comparison as well as time-transfer

Ku-Band Up-linkPower Tx 5 WCarrier 13475 GHzPN-Code 100 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

S-Band Down-linkPower Tx 1 WCarrier 2248 MHzPN-Code 1 MChips1pps 1 time marker sData 5 kBits

Ku-Band Down-linkPower Tx 1 WCarrier 1470333 GHzPN-Code 100 MChips1pps 1 time marker sData 5 kBits

EGE SC

S

Ku

KaKa-Band Up-linkPower Tx 5 WCarrier 345 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

Ka-Band Down-LinkPower Tx 1 WCarrier 32 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 5 kBits

GS Antenna diameterEGE terminal 1 mACES terminal 06 m

Fig 9 EGE microwave two-way link concept and characteristics See text for a description Thelink will be able to perform frequency comparisons with up to ten ground stations simultaneously(only one active comparison is shown) pps pulse per second Tx transmitter SC spacecraft

Exp Astron

between on-board and on-ground clocks Two-way operation allows for com-pensation of the effects due to spacecraft location and its motion relative toground Its two bi-directional links operate in Ka-band (32345 GHz) andKu-band (147135 GHz) The high carrier frequency of the up- and down-links allows for a noticeable reduction of the ionospheric delay A third down-link channel is in S-band (22 GHz) and is used to determine the ionospheretotal electron content MWL uses a combination of carrier phase and codephase data to measure with high precision the absolute ionospheric delayas well as ionospheric delay variations during a ground contact The useof three frequency bands allows to resolve third order ionospheric termsKey modulation parameters are Ka band Tone 250 MHz Ku band PN100 MChips S-band 1 MChips

The EGE-MWL concept is a considerably improved version of the ACES-MWL Microwave carrier generation is directly from the 10 GHz microwaveoutput available from the MOLO and having exceptionally low close-in phasenoise without intermediate frequency dividers or other complex synthesiserstages commonly used in RF transponders The high frequency carriers arekey to achieve the desired frequency comparison accuracy

The Ka-band carrier is generated by tripling the FCDP-generated signalThe Ku-band is subsequently generated by division by two using a verystable regenerative divider Finally the assigned radio frequency is obtainedby adding a low frequency signal using an image reject mixer A mixedanalog-digital receiver design will enhance its performance by use of a flexibledigital signal processing and digital control loops The number of simultaneousground contacts is increased from four (as in the ACES mission) to ten or moreusing the digital concept while the number of physical receiver channels perband is reduced from four to two one for operation to ground and one for in-flight calibration The concept allows adaptation of receiver loop parametersto optimise for the different portions of a highly elliptic orbit while reducingsystematic uncertainties due to high Doppler and Doppler rate Real-timedata exchange to ground terminals is by additional data modulation on all RFsignals The spacecraft carries two different small antennae for perigee and forapogee with gains of 8 dBi 20 dBi beamwidths 100 (full earth coverage) 14respectively power is 1 W for each

The MWL end-to-end performance between the attached clocks includingspace- and ground terminals is predicted to be a relative Allan deviation of1 times 10minus16 at 1000 s (perigee) and 1 times 10minus17 at 10000 s (apogee) Thesevalues are based on conservative implementation losses and other instrumentallosses assumed as 6 dB at apogee and 10 dB at perigee including additionallosses due to atmosphere An attitude knowledge of 1 is required

The MWL link will also support precise orbitographyndashcomplementary tothe GNSS receiver Generation of ranging data is simultaneous to two-waytime transfer and is available under all weather conditions Orbit determina-tion accuracy is expected to be better than 1 cm well matching EGE clockrequirements The value is expected based on demonstrated results fromother missions (METEOR ERS-2 TOPEXPOSEIDON) Laser ranging to

Exp Astron

the CCR array is used a few times during the mission to calibrate the MWLranging

In case of non-availability of the ultrapure signal from MOLO an additional10 GHz output is available from the FCDP to support ground-to-ground clockcomparison in common-view mode even after the nominal mission periodwhen the neither the clocks nor MOLO may be available any longer Basedon the PRARE experimentrsquos 12 years of operation this can be assumed forthe FCDPMWL combination as well The potential extended operational pe-riod makes the proposed configuration particularly attractive to metrologicallaboratories that rely on availability of long-lasting data in support of thegeneration of UTC

The resources of MWL are as follows mass 18 kg volume 18 l power55 W TC 2 kBday TM 120 MBday

426 Auxiliary units

The star camera assembly on the EGE spacecraft will provide real-timeattitude data primarily for the orientation of the spacecraft (with accuracy of1 in real-time) and the proper reduction of the GNSSSLR data

A GNSS receiver is used for the precise orbit determination time andfrequency transfer coarse positioning (lt1 m) for real time use by the Attitudeand Orbit Control as well as the back-up for time tagging of all payloaddata For the time and frequency transfer the receiver is nominally using theonboard frequency reference The receiver assembly includes three specialantennas By the year 2015 more than 100 GNSS satellites are expected tobe available for precise orbit determination permitting orbit inaccuracy of 1ndash2 cm RMS for altitudes up to approx 10000 km At apogee a combinationof GNSS signals and MWL ranging will be used with expected inaccuracyof 10 cm

5 Basic spacecraft specifications

The mission foresees a single spacecraft with a mission duration limited toapprox 3 years for cost reasons Attitude control along three axes withless than 1 error on a timescale of 1000 s is required The control can beperformed with cold gas thrusters magnetotorquers or reaction wheels Theselection will be done on the basis of the vibrations induced by these systems(desired level arms lt 10minus6 g)

Cold gas thrusters or monopropellant thrusters will be required to maintainthe orbit parameters during the mission by small orbit correction manoeuvresand will also guarantee a proper tuning of the orbit parameters after the releasefrom the launcher The ground segment (telemetry control and data transfer)could be via non-ESA stations user-operated to save costs

Further resources are indicated in Table 2

Exp Astron

The satellite bus design drivers are nadir-pointing microwave antennasthermal stability for the payload instruments no strong magnetic sources lowmicrovibrations harsh radiation environment due to crossing of the radiationbelts once per orbit power demand by science payload (with the possibility toreduce power demand by stand-by of some components during specific missionphases eg eclipse) mass limit due to required high altitude with low costlaunch

The satellite bus design can be based on established platforms such asFLEXBUS or PRIMA used on several past missions Their design philosophydiffers one having body-mounted solar arrays the other wings The latter canprovide larger amounts of power The vibration level induced by the panelswhen they are fixed and the duration of solar array drive periods per orbitwill have to be considered in the selection The FLEXBUS concept was usedfor a mission already requiring a low level of residual accelerations and theplatform is optimised for this parameter

For most of the satellite components one can consider using compo-nents and electronic units from navigation or communication satellites whichtypically cruise at altitudes between 15000 and 25000 km altitude LISAPathfinder which will cross the radiation belts during spiralling to theLagrange point will face similar problems concerning radiation as EGE as wellas XMM which is in a Molniya-type orbit with similar radiation characteristics

The instruments will therefore be located in the satellite structure such asto realise ionizing radiation shielding by the spacecraft structure and non-sensitive equipment to the maximum extent possible the need for additionalradiation shielding (including trapped proton radiation) for especially sensitiveparts of the instruments must be determined by ground tests It should bementioned that an orbit with higher perigee and less radiation dose is also anoption as it leads to a tolerable reduction of the science goals

The required plusmn3C temperature stability at the interface to the payloadenclosure is feasible The thermal control will employ the usual means (anenclosure with high thermal inertia heaters and a suitable radiator) Thelocation of the radiator will need a special study given the non-standard orbit

Magnetic field present during the mission can affect the properties of theonboard clock Using SPENVIS the magnetic field has been calculated alongthe spacecraft orbit the field intensity presents maxima around 02 G anda minimum value of the order of 10minus3 G As discussed above this is not alimitation to the instrument performance

6 Discussion and conclusion

The accuracy of the measurements depends crucially on the stability andaccuracy of the clocks While the performance of ground clocks (even thoseeventually deployed in new ground stations dedicated to EGE) will likelyimprove significantly from the present level the space clock design will beldquofrozenrdquo approx 10 years before data acquisition Thus their performance

Exp Astron

level will be significantly below the future ground station clock performanceAs specification we have chosen a performance level that has been demon-strated with laboratory breadboards recently We see no fundamental reasonthat this laboratory performance cannot also be achieved in a space ion opticalclock The experience with the PHARAO cold atom clock has shown thatan engineering model of a laser-based cold atom clock has a performancesimilar to that of a laboratory prototype The perturbation of the ion clock fre-quency due to magnetic fields along the orbit can be modelled and suppressedwith shielding and the residual effects measured and compensated for Themoderate temperature stability (plusmn 3C) to be provided in the experimentalbay can be achieved with thermal shields Accurate temperature controlof particular units is done by active stabilization of the units themselves

One important aspect is radiation which may affect all optical and trapcomponents as well as the lasers and reference cavity The radiation envi-ronment on the EGE orbit is significantly harsher than the near-earth ISSorbit on which ACES will operate The radiation effects must therefore becharacterized comprehensively with appropriate tests and then the radiationshield for the EGE instruments designed

In summary the Einstein Gravity Explorer mission (EGE) is a satellitemission aimed at a precise measurement of the properties of space-time usingatomic clocks By testing one of the most fundamental predictions of EinsteinrsquosTheory of General Relativity the gravitational redshift and thereby searchingfor quantum effects in gravity it explores one of the most important frontiers infundamental physics The EGE mission reaches beyond fundamental physicsproviding an important spin-off in geophysics Furthermore the technologicalaspects of the mission are ground-breaking for future time-keeping and ref-erence frame systems in space and thus are also expected to be of wide useto science and technology in general The payload of the mission has beenconceived as a compromise between cost and performance needing to fit thecost envelope of an ESA medium-class mission1

Acknowledgement We thank S Cesare for helpful information

References

1 Will CM The confrontation between general relativity and experiment Living Rev Relativ9 3 (2006)

2 Ni W-T Empirical foundations of the relativistic gravity Int J Mod Phys D 14 901 (2005)doi101142S0218271805007139

1Finally we point out that a spacecraft with a significantly reduced and thus lower-cost payload (noclocks only MOLO FCDP MWL) on a circular orbit of appropriate height could be used for along-duration earth science mission dedicated specifically to the measurement of the geopotential

Exp Astron

These can be interpreted toward tests of General Relativity (GR) and ofcompeting metric theories of gravity and as tests for the existence of newfields associated to matter (see Section 3) The outcome of the mission will beeither a confirmation of GR and metric theories within the accuracy providedby the instruments or the discovery of a deviation In the latter case themission would provide a first indication of the breakdown of current (classical)gravitational physics theories and could pave the way towards a unified theoryof all forces

2 Scientific objectives

21 Summary of science objectives

211 Fundamental physics science objectives (1 ppm = 1 part in 10 61 ppb = 1 part in 10 9)

Primary objectives

High-precision measurement of the earth gravitational frequency shift at25 ppb accuracy a factor 3000 improvementFirst precise measurement of the sun gravitational frequency shift at 1 ppmlevel a factor 2 times 104 improvementTest of space and time variability of fundamental constants with accuracy30 ppbFirst search for neutron scalar charge with accuracy 30 ppbTest of anomalous coupling of matter to the Standard Model quantumfields

Secondary objectives

Tests of Lorentz Invariance in the matter and photon sector (factor 20improvement)Contribution towards establishing a new definition of the unit of time

212 Spin-off to other fields (outside fundamental physics)

Establishment of a new approach to the determination of the geopotentialwith geoid height resolution of a few cmComparison of distant terrestrial clocks at the 10minus18 levelDemonstration of clock and link technology for future applications eg inprecision spacecraft navigationDemonstration on high performance real-time range determination withprecision 25 times (based on code phase) to 1000 times (based on carrier

Exp Astron

phase) higher compared to systems onboard previous missions (ERS-2TOPEX POSEIDON)Comparison of three different orbit determination systems Laser-Ranging μm-precision microwave ranging GPS-based orbit deter-minationAtmospheric signal propagation study precise determination of thirdorder ionospheric termMonitoring stability of the GalileoGPSGlonass time scales

22 Gravitational frequency shift measurement and search for new physics

The primary science goal of EGE is a measurement and test of the gravita-tional frequency shift in the weak-field limit as predicted by metric theoriesof gravitation These theories also predict that the shift is independent of thetype of clock used to measure it The comparison of the frequencies ν oftwo identical clocks 1 2 operating at different locations x1 x2 yields a fre-quency ratio

v2(xprime)v1

(xprime) sim= 1 + (U (x2) minus U (x1))

c2 (1)

Here νi(xprime) is the frequency of clock i located at xi as observed (measured) ata particular location xprime where the comparisons are performed The comparisonlocation can be arbitrary but will often be x1 or x2 or both U is thegravitational potential U(x) = minusGM|x| in case of a spherically symmetricbody of mass M In short time runs (or clocks tick) more slowly the closer thesystemclock is to a massive body Stationary clocks and observers and weakgravitational fields have been assumed in Eq 1

Through its measurement EGE searches for a possible violation of the grav-itational frequency shift A violation may be described phenomenologicallyby a dependence of one or more fundamental constants on the gravitationalpotential X = X(Uc2) where X is a generic dimensionless fundamentalconstant or a dimensionless combination of fundamental constants Such adependence is a violation of the principle of Local Position Invariance (LPI)which states that the outcome of local nongravitational experiments is inde-pendent of where and when in the Universe they are performed If a constantdepends on the gravitational potential then the results of those laboratory-type (ie local) experiments that depend on that constant will also depend onthe location namely on the location of the laboratory with respect to the massdistribution generating the potential and will violate LPI

LPI is one of the three cornerstones of the Einstein Equivalence Principle(EEP) the others being the Weak Equivalence Principle and Local LorentzInvariance It is a common belief that the EEP may not be absolutely valid -therefore spatial dependence and time dependence of fundamental constantshave become a very intense research topic in the last decade with experimentsundertaken worldwide to search for violations of various aspects of EEP For

Exp Astron

example extensive studies of atomic and molecular spectra of gas clouds in thedistant Universe are currently undertaken to search for a difference of theseparameters compared to todayrsquos values (see eg [7ndash11] and references therein)Also some differences in Big Bang nucleosynthesis data and calculations canbe naturally explained by a variation of fundamental constants [12 13]

The EGE mission is thus a probe of possible space-dependence of fun-damental constants The search may help to point the way toward a unifiedtheory of all forces reveal hypothetical extra dimensions in our Universe or theexistence of many sub-Universes with different physics chemistry and biologyIndeed theories unifying all forces applied to cosmology suggest space andtime dependence of the coupling constants [14ndash16] Another argument forspatial variation of fundamental constants comes from the anthropic principleThere must be a fine tuning of fundamental constants which allows humans andany life to appear Life in its present form is not consistent with fundamentalconstants whose values only slightly differ from the actual ones This finetuning can be naturally explained by the spatial variation of the fundamentalconstants we appeared in the area of the Universe where the values of thefundamental constants are consistent with our existence

In the Standard Model there are three main fundamental dimensionless pa-rameters determining the structure and energy levels of stable matter (atomsmolecules) the electromagnetic fine structure constant α and the two ratiosof electron mass me and light quark mass mq to the strong QCD energy scaleQCD Xeq = meqQCD The fundamental masses me and mq are proportionalto the vacuum Higgs field which determines the electroweak unification scaleTherefore the constants meqQCD can also be viewed as the ratio of the weakenergy scale to the strong energy scale

How can a space-time variation of the fundamental constants and a depen-dence on the gravitational potential occur Light scalar fields very naturallyappear in modern cosmological models affecting α and meqQCD Oneof these scalar fields is the famous ldquodark energyrdquo which causes acceleratedexpansion of the Universe Another hypothetical scalar field is the dilatonwhich appears in string theories together with the graviton Cosmologicalvariation of these scalar fields should occur because of drastic changes ofthe matter composition of the Universe During the Big Bang nucleosyn-thesis the Universe was dominated by radiation then by cold dark matterand now by dark energy Changes of the cosmic scalar field ϕ0(t) lead tothe variation of the fundamental constants X(ϕ0) Massive bodies (galaxiesstars planets) can also affect physical constants They have a large scalarcharge S = sp Z + se Z + sn N where Z is the number of protons and elec-trons and N is the number of neutrons sp se sn are the scalar charges ofthe proton electron and neutron respectively There is also a contributionof the nuclear binding energy (scalar charge of virtual mesons mediatingnuclear forces)

The scalar charge produces a Coulomb-like scalar field ϕs = SR where Ris the distance from the charge The total scalar field is ϕ = ϕ0 + ϕs therefore

Exp Astron

we may have a variation of the fundamental constants inversely proportionalto the distance from massive bodies

X (ϕ) = X (ϕ0 + ϕs) = X (ϕ0) + δX (R)

δX (R) = (dX

dϕ

)S

R

where a nonzero δX violates LPI and a small ϕs was assumedThe gravitational potential U is essentially proportional to the number of

baryons Z + N and inversely proportional to the distance R U = -GMR

Therefore the change of the fundamental constants near massive bodies canbe written

δX

X = K (X i) δ(Ui

c2

) (2)

where i refers to a particular composition of the mass M and where thecoefficients K(X i ) = minus(dXdϕ) c2SiGMi are not universal but are expectedto be different for eg the Sun and the Earth K(X Earth) = K(X Sun)Indeed the Sun consists mostly of hydrogen therefore the Sunrsquos scalar chargeis SSun Z (sp + se) The Earth contains heavier elements where the numberof neutrons exceeds the number of protons N 11 Z Therefore the scalarcharge of the Earth SEarth Z (11 sn + sp + se) and is sensitive to the neutronscalar charge (and also to the contribution of the nuclear binding) in contrastto the Sun

The most sensitive experiments suitable for a search of a space-time de-pendence in local or near-local experiments (such as satellite experiments) areclockndashclock comparisons In one type of terrestrial local experiment dissimilarclocks located nearby are compared for a duration spanning at least one yearDuring such periods the earth moves periodically closer and further awayfrom the sun so that the solar gravitational potential on Earth is modulatedThis enables a search for a dependence of the fundamental constants on thesunrsquos scalar charge [17 18] These experiments are able to set limits to K(XSun) for X = α meqQCD at the level of 10minus6 [12 13]

The proposed EGE satellite experiments are complementary to terrestrialexperiments allowing to set limits to K(X Earth) and combining with theresults on K(X Sun) from terrestrial comparisons to set limits on the depen-dence of the fundamental constants on the neutron scalar charge sn

In summary when the frequencies of two identical clocks (of nominalfrequency νa) that are located at different positions x1 x2 are comparedby measuring their frequency ratio at some position xrsquo the result may bewritten as

va2(xprime)

va1

(xprime) = 1 +

[1 +

sum

XAXa K (X i)

](Ui (x2) minus Ui (x1))

c2 (3)

The sensitivity factor AXa = X (d(ln νa)dX) is the sensitivity of the clockfrequency to a particular constant X and can be calculated using atomic and

Exp Astron

nuclear theory (see eg [12 13]) The measurement of this ratio which issensitive to K represents a test of General Relativityrsquos redshift

When the frequencies of two co-located but dissimilar clocks a b (ofnominal frequencies νa0 and νb 0) are compared in-situ at two locations x1 x2

(LPI test) the result is

[va1 (x1)

vb 1 (x1)

] [va2(x2)

vb 2 (x2)

] = 1 +[X

(AXa minus AXb

)K (X i)

](Ui (x2) minus Ui (x1))

c2 (4)

The result of the EGE mission may be given as a limits for (or nonzerovalues of) K(X Earth) for the three fundamental constants X = α meQCDand mqQCD

23 Measurement approaches for the gravitational frequency shift

The gravitational potential experienced by the ground and satellite clocksincludes the contributions from the earth and the sun (the contribution fromthe moon is an order smaller and may be neglected for the purposes of thediscussion but will be taken into account in data analysis)

Nonzero gravitational frequency shift signals (Eq 3) can be obtained inthree ways Method (1) below is the most sensitive and represents a strongreason to employ an optical clock which on short averaging times is ableto provide a higher frequency stability than a microwave clock Method (2)and (3) are less accurate but are complementary in terms of measurementapproach Finally method (3) addresses the solar gravitational frequencyshift while (1) and (2) measure the terrestrial effect The performance of theinstruments assumed in the following is shown in Fig 2

(1) Repeated measurements of satellite clock frequency variation betweenapogee and perigeeIn this measurement approach the difference (or ratio) between earthndashsatellite clock frequency ratio at perigee and at apogee is consideredThis difference is measured at each perigee and apogee passage andaveraged over the mission duration Systematic shifts (if not correlatedwith orbital motion) are expected to average out leading to a gain insensitivity of up to n12 where n = number of orbits This mode usesthe stability properties of the satellite clocks on the timescale fromperigee to apogee rather than their accuracy Assuming a specificationof le 3 times 10minus16 satellite optical clock fractional frequency instability forintegration times between 1000 and 25000 s n sim= 1000 comparisons(ie most orbits of the 3-year mission) and a perigee-apogee potentialdifference (U(apogee) minus U(perigee))c2 = Uc2 sim= 4 times 10minus10 allows a

Exp Astron

Fig 2 Performance of clocks and links on EGE Curve ldquooptical clocks EGErdquo is the specificationCurve ldquooptical clocks EGE (goal)rdquo refers to expected near-future performance of an opticalclock based on neutral atoms ldquoUpgraded PHARAOrdquo refers to the PHARAO microwave clockusing the high-quality microwave-optical local oscillator The microwave link (MWL) performanceexhibits a discontinuity since common-view contact time of ground stations with EGE is limited toseveral h The different slopes arise because on short time scales only white phase noise is relevantwhile on long time scales flicker noise also contributes The reproducibility is taken into account

measurement of the redshift at 25 ppb accuracy The ground clocks towhich the satellite clocks are compared to at perigee and at apogee willbe located in different ground stations The inaccuracy of these groundclocks must therefore be sufficiently smaller than 1 times 10minus17 in order notto contribute to the 25 ppb error Current state-of-the art inaccuracy is2 times 10minus17 and is likely to decrease further

(2) Absolute comparison between ground and satellite clocks with the satel-lite at apogeeHere the terrestrial gravitational potential difference is maximumU(apogee) minus U(perigee)c2 sim= 65 times 10minus10 Assuming a space optical clockinaccuracy of 1 times 10minus16 and a ground clock inaccuracy a factor 3lower (current state-of-the-art) allows a measurement of the redshift at150 ppb This measurement takes advantage of the repetitive nature ofthe orbit only to a small extent and relies on the accuracy of the opticalclocks in particular of the space clocks It is thus complementary to (1)albeit less accurateIn (1) and (2) the dominant contribution to the redshift is from Earthrsquosgravitational potential The contribution from the solar potential isstrongly reduced due to the fact that satellite and earth are an interactingsystem in free-fall toward the Sun [19] and can be modeled

(3) Comparisons between terrestrial clocksThe frequency comparison link to the EGE satellite allows terrestrialclock comparisons by a common-view method whereby the frequency

Exp Astron

signals from two distant ground clocks are sent to the satellite andcompared there The frequency difference between two identical groundclocks is determined if systematic instrumental errors are negligible bythe difference in gravitational potential at the two clock locations Thispotential difference includes a terrestrial and a solar contributionThe solar contribution varies according to the orientation of the clockpair with respect to the Sun with a 1 day-period For two clocks at asurface distance of sim6000 km and suitable orientation with respect tothe sun the normalized modulation amplitude is sim= 4 times 10minus13 This is thesame as the earth potential difference for 4 km height difference It ispossible that ground clocks will have reached an inaccuracy at the levelof 1 times 10minus18 at the time of the mission in 2015 Using the microwave linka comparison of ground clocks over sim3 h (typical common-view duration)with a resolution of 1 times 10minus17 can be performed limited by the instabilityof the link Repeating the clock comparison at each common-view win-dow (n = 1000 comparisons) will allow averaging and a measurementof the solar gravitational shift with fractional inaccuracy of 1 ppm Incomparison the best current results for the solar gravitational frequencyshift are at the few-percent level [20 21] A moderate averaging out ofground clock systematic shifts is assumed as well as modeling of the earthgravitational potential variations on the daily timescale at the appropriatelevel

(4) Comparisons of on-board clocksThe two on-board clocks are continuously compared these data areanalyzed for violation of LPI (Eq 4) Since the measurement is a nulltest the required accuracy of the gravitational potential at the satellitelocation is low and easily satisfied

24 Test of local position invariance and search for neutron scalar charge

The test of LPI is performed in conjunction with the measurement of thegravitational frequency shift

With approach 23 (1) the optical clock will be able to measure K (α Earth)with accuracy 30 ppb (spec)

Combining the results of the microwave and optical clocks will determineK(meQCD Earth) + εK(mqQCD Earth) with accuracy 85 ppb (spec)where ε asymp minus01

Simultaneously the on-board comparison 23 (4) yields the combination19 K(α Earth) +K(meQCD Earth) + εK(mqQCD Earth) with accuracy85 ppb (spec)

Note that these measurements are fully complementary to current andfuture terrestrial clock-clock comparisons which are sensitive to K(X Sun)and can be combined with the latter to determine or set limits to the neutronscalar charge

Exp Astron

25 Theoretical analysis of frequency comparison signals

The on-board clock-clock comparison signals can be analyzed straightfor-wardly in terms of a violation of LPI since signal propagation effects areabsent

The determination of the gravitational frequency dilation via ground-satellite clock comparisons is more complex (Eq 1 is strongly simplified) andrequires an accurate treatment of all relevant relativistic effects Gravitationin the solar system is described in the barycentric (BCRS) and geocentric(GCRS) non-rotating celestial reference systems by means of post-Newtoniansolutions of Einsteinrsquos equations for the metric tensor in harmonic gaugescodified in the conventions IAU2000 [22] In these conventions the relativisticstructure of Newton potential (order 1c2) and gravitomagnetic (order 1c3)

potentials are given For the Newton potential the multipolar expansion ofthe geopotential determined by gravity mapping is used The post-Newtonianeffects of the metric (described in terms of dimensionless parameters β and γ )

can be included Orbits of satellites are evaluated in the GCRS at the levelof a few cm by means of the Einstein-Infeld-Hoffmann equations at the order1c2 [23] using the relativistic Newton potential Instead the trajectories of theground clocks in the GCRS are evaluated from their positions fixed onthe Earth crust (ITRS International Terrestrial Reference System) by usingthe non-relativistic IERS2003 conventions [24 25]

For the propagation of light rays (here radio signals) between the satelliteand the ground stations one uses the null geodesics of the post-Newtoniansolution in the GCRS The timefrequency transfer properties have beentheoretically evaluated for an axisymmetric rotating body [26ndash29] at the order1c4 (ie with Newton and gravitomagnetic potentials developed in multipolarexpansions) These results have been developed for use in the ACES missionThe relative frequency dilation can be expressed as

ν

ν= 1 + 1

cL1 +

4sum

n = 2

1

cn(Ln + Gn)

where Ln describes special-relativistic Doppler effects and Gn the general-relativistic effects These quantities are time-dependent due to the orbitalmotion For concreteness we report the orders of the terms The kinematicaleffects are |L1c| lt 1 times 10minus5 |L2c2| lt 2 times 10minus10 |L3c3| lt 3 times 10minus15 |L4c4| lt

7 times 10minus20 The Newtonian contributions are G2 = G(M)2 + G(J2)

2 + G(J4)2 + G(J6)

2(M is the mass monopole Jn are the lowest multipoles) with estimates|G(M)

2 c2| lt 3 times 10minus11 |G(J2)2 c2| lt 2 times 10minus13 |G(J4)

2 c2| lt 3 times 10minus16 |G(J6)2 c2| lt

1 times 10minus16 For G3 = G(M)3 + G(J2)

3 the estimates are |G(M)3 c3| lt 2 times 10minus14

|G(J2)3 c3| lt 1 times 10minus18 Finally the G4 = G(M)

4 + G(S)4 contributions (G(S)

4 is thelowest gravitomagnetic effect) are of order |G(M)

4 c4| lt 1 times 10minus19 |G(S)4 c4| lt

1 times 10minus22 and thus not relevant for EGE

Exp Astron

The available theory is fairly complete for the data analysis in mode 23 (1)described below For modes 23(2 3) the theory will need to be generalizedto include the effect of the gravitational potential at the location of the groundclock(s) (including time-varying effects such as the tides) and of the solar andlunar potentials

The uncertainties of the various contributions above are minimized by ausing precise EGE orbit data obtained by satellite tracking and by usingcurrent and future accurate earth gravity information Required orbit positionknowledge is at 1 cm level near perigee and sim50 cm near apogee reachablealready today with the proposed orbitography approach described below Thevalidity of the special-relativistic Doppler shift has been and will continue to beindependently verified with increased accuracy by spectroscopy of relativisticatomic ion beams [30] Potential violations of this aspect of Lorentz Invari-ance are expected to be sufficiently bounded so as not to affect the signalinterpretation

26 Local Lorentz invariance tests

There has been an explosion of interest in Local Lorentz invariance in re-cent years [31 32] with numerous astronomical studies and high-precisionexperiments applied to search for violations of Lorentz invariance Signifi-cant developments have also occurred on the theoretical side and a theory(standard model extension SME [33]) has been worked out that provides aunified framework for describing and analyzing Lorentz Invariance violationsin a variety of systems Two aspects of Lorentz Invariance can be tested withEGE

261 Independence of the speed of light from the laboratory velocity

A possible dependence of the speed of light on the speed v of the laboratorycan be modeled according to the MansourindashSexl test theory [34] by

c (v) = c0(1 + Bv2

c2

)

Here v 377 kms is the velocity with respect to the cosmic microwavebackground the cosmologically preferred frame and B is a combinationof parameters describing deviations from the usual Lorentz transformationformulas B = 0 if Lorentz Invariance holds For a terrestrial experiment therotation of the Earth around its axis modulates v with a 300 ms amplitude Adependence c(v) can be searched for by measuring the frequency differencebetween a resonance frequency of a highly stable optical cavity (cavity fre-quencies are proportional to c) and an optical clock (Kennedy-Thorndike-typeexperiment [35]) and determining the modulation of the frequency differencecorrelated with the modulation of v The advantages of a space experimentare the high orbital velocity and strongly reduced cavity deformation thanksto microgravity [36 37] On the proposed elliptic orbit v varies between+4 and minus4 kms over approximately 1 h This variation is 13 times larger

Exp Astron

than for the case of a terrestrial experiment The shorter time scale is alsoadvantageous since the drift of the cavity is more predictable In the EGEmission the reference cavity of the clock laser is used The large numberof orbits improves the statistics An improvement of the accuracy of B by afactor 20 compared to the best current terrestrial results is realistic The dataanalysis can also be performed in the framework of the SME theory takinginto account complementary bounds obtained from terrestrial experimentsand astrophysical observations

262 Independence of Zeeman splitting frequency on the directionof the magnetic field

One class of tests of Lorentz Invariance consists in measuring the frequencysplitting between two levels of a quantum system induced by a static magneticfield as a function of orientation of the field direction with respect to thestars (HughesndashDrever-type experiments [31]) This class addresses the so-called matter sector and is complementary to combined-photonmatter-sectortests such as the above The most precise experiments are performed withatomic clocks eg masers or cold atom clocks [38] The EGE instruments allowsuch experiments since magnetic fields are applied continuously in the clockor repeatedly for calibration Compared to terrestrial experiments the highorbital velocity of EGE could lead to an improvement by a factor sim20

27 Application to geophysics

Using GPS coordinates of a point on the Earth can be obtained with aninaccuracy well below 1 cm in a well defined international terrestrial referencesystem The coordinates are purely geometrical and do not contain any gravityinformation A local gravitational potential UB is determined with respect toa reference gravitational potential U A by ldquolevellingrdquo the surface of the Earthie sequentially measuring gravity g and height differences every ca 50 m

UB minus U A = minusBint

A

g (r) middot d n

where n is the difference vector between subsequent locations The disadvan-tage of measuring height differences by the levelling method is a random walkeffect of accumulated errors

In the classical definition a geoid is defined as the particular equipotentialsurface nearest to mean sea level In these terms the geoid serves as areference surface for measuring the height and also to define a referencefor the gravitational potential Such a vertical reference frame historicallywas realized for a country or several countries by determining the mean ofsea level observations at tide gauge stations taken over long period of timeHowever modern satellite altimetry missions such as TOPEXPOSEIDON orENVISAT show that departure of the mean sea level from an equipotential

Exp Astron

surface may reach up to several meters on the global scale see eg [39]Therefore height systems based on different tide gauge stations may differin the realisation of the geoid and may differ with respect to the reference byseveral meters

The best gravity field models obtained from satellite data (eg missionGRACE) have reached a precision of about 1 cm over sim250 km half wave-length [40] Satellite data also reveals changes in time [41] However for typicalEarth topography with height variations of eg 1000 m over 30 km horizontaldistance one may expect variations in the geoid of about 80 cm Such ahigh-spatial-frequency signal in geoid variations cannot be detected by spacegravity missions and requires a combination of satellite and terrestrial gravitymeasurements like gravity anomalies deflections of vertical and GPSlevellingpoints The best combined global gravity field models are provided up to de-gree and order 360 in terms of spherical harmonic representation correspond-ing to a half-wavelength of about 55 km However in combining the satelliteand terrestrial gravity field measurements the problem of terrestrial data givenin different height systems between continents and different countries remainsand has to be tied to satellite measurements

The geodetic scientific community is currently establishing a Global Geo-detic Observing System (GGOS) [42] Its objectives are the early detectionof natural hazards the measurement of temporal changes of land ice andocean surfaces as well as the monitoring of mass transport processes in theEarth system Global change processes are small and therefore difficult toquantify Therefore the required precision relative to the Earthrsquos dimensionis 1 ppb GGOS will be established by the combination of geodetic spacetechniques (GPS Laser-Ranging very large baseline interferometry) andrealized by a very large number of terrestrial and space-borne observatoriesThe purely geometric terrestrial 3D coordinate system is in good shape andfully operational It has to be complemented however with a globally uniformheight system of similar precision The current precision level of regionalheight systems in terms of gravity potential differences is in the order of1 m2s2 (10 cm) with inconsistencies between these various systems up toseveral 10 m2s2 (several meters) The actual requirement in the context ofGGOS is 01 m2s2 (1 cm) with the need of a permanent ie dynamical controlThis requirement of high precision height control comes from the need tounderstand on a global scale processes such as sea level change global andcoastal dynamics of ocean circulation ice melting glacial isostatic adjustmentand land subsidence as well the interaction of these processes Only by meansof monitoring in terms of gravity potential changes at the above level ofprecision the change of ocean level can be understood as a global phenomenonand purely geometric height changes be complemented by information aboutthe associated density or mass changes

The geoid can also be defined in a relativistic way [43] as the surface whereaccurate clocks run with the same rate and where the surface is nearest tomean sea level The relation between the differences in the clock frequenciesand the gravitational potential is given in simplified form by Eq 1 At present

Exp Astron

there is no operational way to compare frequencies of the already availableoptical clocks on the global scale at the same level as their accuracy wouldallow With EGE and by then improved ground optical clocks it will becomepossible to obtain gravitational potential differences on a global scale by com-paring frequencies This would be a significant new dimension to gravity fielddetermination since such observables are given on the global scale and providein situ local gravity information at the same time EGE will allow clock-basedgravitational potential mapping with the same payload instruments used forthe fundamental physics experiments The on-board clocks are actually notrequired for this purpose MOLO MWL and FCDP suffice Thus EGE willallow establishing a global reference frame for the Earth gravitational fieldwith accuracy in the order of a few cm in terms of the geoid heights Thisassumes that by the time EGE is operational (mobile) ground clocks withfractional inaccuracy at the level of 10minus18 are available the measurementsare limited by MWL noise and thus require a long integration times to matchthe expected ground clock performance Such a reference frame will serve asreference for all gravity field modeling and clocks will be used to define theinternational atomic time scale and the Earth gravitational potential at thesame time

In summary with optical clocks geodesy will undergo a second revolutionafter geometry now being measured by clock-based GNSS systems also physi-cal heights and potential will be measured by (mobile) optical clocks using thegravitational redshift effect

28 Application to frequency standards and terrestrial fundamentalphysics studies

As has been described repeating ground clock comparisons and integratingover several months will provide global ground clock frequency comparisonsat the level of 1 times 10minus18 more than two orders of magnitude below the currentGPS and two-way satellite time and frequency transfer methods In the timeand frequency community the availability of such a global high performancemicrowave link will accelerate the process leading to a redefinition of the SIsecond based on clocks operating in the optical domain

The global character of the ground clock comparisons made possible byEGE can also contribute to the search for a time-variation of the fundamentalconstants

The comparisons do not require availability of the on-board clocks The riskof this type of measurements is therefore reduced Moreover the lifetime ofmany subunits required for this measurement (MWL and FCDP see below)can exceed 10 years (from previous experience with similar devices) andthus offers a motivation to increase the mission duration beyond the nominal3 years

Further applications to metrology are listed under the secondary goals inSection 21 above

Exp Astron

3 Mission profile

31 Orbit

The orbit must satisfy the following requirements

Primary 1 Large gravitational potential difference between apogee andperigee

2 sufficient contact time with at least two ground stations atperigee

Secondary Simultaneous visibility of satellite from distant (inter-continental) ground stations for several hours

To satisfy the primary objective the orbit must be highly elliptic The perigeeshould be low A very low perigee means a short contact time which decreasesthe accuracy of the frequency measurement A compromise is a perigeealtitude of sim2500 km For a sufficiently high apogee this yields a U of 60of the theoretical maximum between the earth surface and infinity

The baseline scenario is an orbit of approx 12 h period with the followingproperties

Apogee Altitude 37856 km (minus 10 deg latitude one over Atlantic andPacific resp)

Perigee Altitude 2500 km (+ 10 deg latitude one over Guatemala andMalaysia resp)

Inclination 634 argument of perigee 170RAAN 25 True Anomaly 220Apogee drift rate 70 kmday perigee drift rate sim 14 kmday

True anomaly and right ascension of the ascending node (RAAN) wereoptimized to minimize orbital precession Figure 3 shows the ground track andthe ground coverage at one of the perigees Common-view contact time forstations in Europe is about 6 h while between a station in Europe and in theUSA it is about 35 h Frequency comparison windows between a given groundstation and EGE repeat every 24 h

A final selection of the orbit requires an in-depth study of the costs of thepropellant and maneuvers as well as of the setting up of new ground stations

32 Ground segment

The mission requires two separate ground segments the science groundsegment and the standard satellite control segment which supports telemetryand telecommand (TMTC)

Exp Astron

Fig 3 Baseline EGE orbit One (out of two) apogee and one (out of two) perigee ground pointsare shown The ground coverage at one of the perigees is shown with circles at elevation anglesteps of 10 starting from 10

321 Science ground segment

As described above one part of the mission is to perform frequency com-parisons between ground-to-space and ground-to-ground (common-view andnon-common view) The use of a two-way microwave link (MWL) systemhas the advantage that a precise orbit or the precise relative velocity of thesatellite must not be known [4] The orbit determination requirements fora two-way system are significantly relaxed compared to the requirements tomodel the relativistic effects on the clocks The frequency comparison resultsare available both on ground and on the satellite

For ground-to-satellite clock comparisons five ground stations arefavourable two capable of viewing perigee three for viewing apogee Thisallows for the classic three-clock configuration where the performance of eachclock can be obtained individually The perigee stations will be observing thesatellite for sim25 min every perigee passage for the duration of the missionThe apogee stations will observe for sim3 h every apogee passage Of coursea larger number of stations as well as longer observation time are usefulfor robustness redundancy and orbit determination The stations could beinstalled on a temporary basis for the duration of the mission

The stations must have access to optical clocks of performance at least ashigh as todayrsquos state-of-the-art ground clocks but preferably higher Theywill include an optical frequency comb by means of which the optical clockfrequency is transformed to a radio frequency for comparison with the EGEclocks High-accuracy fiber-optic link technology already existing today allowsdistances of sim100 km between the ground station and the correspondingoptical clock allowing a certain freedom in the choice of the clock locationWith further increase of the fiber-optic link range currently being investigated

Exp Astron

the area over which the ground station can be positioned relative to the opticalclock will increase This option is of importance for the implementation of ageophysics measurement campaign for which a large number of clock locationsis of interest

The above ground stations will be complemented by some 15 additionalground terminals provided by the world-wide laboratories equipped with highperformance primary clocks at the time of the mission These will participate inorbit determination to 1 cm accuracy along the full orbit which can be achievedwith non-optical atomic clocks The EGE MWL is fully compatible to plannedACES ground terminals which might reduce the number of required stations

With a larger number of transportable optical clock ground stations de-ployed and moved across the continents a geophysics campaign could beimplemented to map the gravitational potential

Ground station equipment MWL uses dedicated small transportable termi-nals which can be easily installed at any location of interest Each terminalconsists of a steerable dish antenna of 1 m diameter with pointing accuracy of01 using a transmitted power of 5 W per band The high directivity antenna(beam aperture Ka-band 07 Ku-band 14 S-band 20) avoids multi-pathfrom nearby objects Antenna noise temperatures are Ka-band 450 K Ku-band 340 K Operation is at elevations above 10 and measurements areperformed above 20 For ranging purposes terminal location is determined byGPS surveying Each ground station computes its own ionospheric correctionsbased on a triple-band receiver The terminal has a built-in delay monitorwhich allows calibrated ranging during several months once its bias has beendetermined by laser ranging Its M and C interface is via local area networkto either a userrsquos computer or directly to the science network control centreby remote MampC The stationrsquos operation is fully automatic and unattendedSchedule and orbital data are received via LAN once per day Signal acquisi-tion is fully autonomous Data transmission is after each pass The terminal hasan uplink telecommand capability of 1 kBits and a downlink data capacity ofup to 10 kBitss For the local user there is a visual interface providing quick-look data from real-time comparison to the spacecraft clocks to verify theirgood health The final full performance product is calculated at the sciencedata centre

Local metrological data (temperature pressure humidity) are recorded tocorrect for tropospheric delay of the ranging data The main advantage ofMWL terminals compared to laser based systems is their all-weather capabil-ity although microwaves suffer higher propagation errors in the ionosphereand troposphere

The EGE ground terminal will profit from the ERS-2 mission PRARE(precise range and range rate equipment) [45] experiments whose terminalshad similar design and have been operated under extreme climatic conditionsincl Antarctica and high temperature regions from existing designs fromthe space segment (mission ACES) and no further special developments aredeemed necessary Cost per EGE master station is modest

Exp Astron

Satellite tracking via laser ranging Laser ranging is alternative to MWL rang-ing and also serves for absolute calibration of the latter For this purpose anon-board corner cube reflector (CCR) is provided Ranging will be done withthe International Laser Ranging Service Ideally the satellite will be rangedat every perigee passage and every apogee passage simultaneously with thefrequency comparison procedures Todayrsquos ranging precision is sufficient tosatisfy the science goals For apogee observations in the northern hemispherea large number of laser ranging stations is available The number of stationscapable of ranging at or near perigee in the southern hemisphere is six andthus is sufficient even considering unfavourable weather conditions at someof these The MWL ground stations are able to range even under thoseconditions

322 Control ground segment

Average data transfer rate is estimated at 300 MBday (approx 40 kBs onaverage) Data will be stored on board (2 GB capacity required) and will betransferred to a receiving station on the Earth once a day

XPLC

incl Data StorageIF to TMTC Link

HK amp TMTC

to ground

Opt Clock

AOM

MOLO

PHARAO

FCDP

10 GHz

100 MHz

250 MHz

GNSS Rx

MWLTampF Link

uplink Data

UTC(k)

TampF Space vs Ground

TM from GS

Tim

e S

cale

Orbit GNSS Time UTC

TampF Space vs Ground

MampC Time Scale

MampC

MampC

MampC

MampC

Mamp

C

PHARAO vs Opt Clock

PHARAO vs Opt Clock

10 GHz amp Epoch

10 MHz

amp Epoch

Epoch

10 MHz

amp Epoch

Opt Link

Fig 4 Schematic of the payload showing subunits and their signals The results of the frequencycomparison between the two on-board clocks (optical clock and microwave clock PHARAO) areobtained from a combination of the frequency values applied to the frequency shifters AOM (seeFigs 7 and 8) and those output from the frequency control and distribution package (FCDP)as indicated by the arrows ldquoPHARAO vs Opt Clockrdquo The satellite clock to ground clockcomparison results are output from the MWL XPLC is the on-board computer and is involvedin generation of signals for instrument control The optical link shown dashed is a possible add-onproviding an alternative to the MWL GS ground stations MampC monitoring and control HKhousekeeping TampF time and frequency

Exp Astron

4 Main payload instruments

41 Overview

The satellite payload consists of a pair of cold-atom clocks a microwave-optical local oscillator (MOLO) an on-board frequency distribution unit(FCDP) a satellite-to-ground clock comparison unit (MWL) and the onboardcomputercontrol system (XPLC) see Fig 4 Auxiliary units are the posi-tion determination GPSGalileo receiver (GNSS Rx) attitude monitor (startracker not shown) and the corner cube reflector array (CCR not shown)

The science goals require clocks of very high frequency stability on thetimescales of half the orbital period (6 h) as well as perigee passage (sim1000 s)see Fig 2 Only clocks based on cold atoms are suited for the science goalsTable 1 reports values for a subset of atomic clocks that have been investigatednot including clocks that require cryogenic operation or are too complex foruse in an medium-class mission The choice of a clock pair rather than a singleone fulfils several requirements it addresses more science goals improves thevalidity of some science results and it enables calibration accuracy evaluationand redundancy The validity of the science results is enhanced since theclocks differ not only in the employed atoms but also in terms of havingdifferent sensitivity to perturbations by electric and magnetic fields electronicserrors and misalignments A comparison of the results arising from each clockallows for an understanding of measurement errors Both the clocksrsquo and thefrequency comparison unitrsquos performances will be determined after launchand regularly verified during the course of the mission The only reliable

Table 1 Properties of different clock candidates for EGE

Clock Clock Fractional σ = Fractional Sensitivity Figure of Sensitivityatom frequency inaccuracy instability to α Aα merit |Aα| to meq

at 1000 s σ [1016] QCD

Sr+ 445 THz 3 times 10minus17 (p) 5 times 10minus17 (p) 042 08 (p) ndashYb+ (1) 688 THz 3 times 10minus16 (d) 3 times 10minus16 (d) 088 029 (d) ndash

3 times 10minus17 (p) 1 (p)Yb+ (2) 642 THz lt 1 times 10minus17 (p) 1 times 10minus17 (p) minus53 53 (p) ndashSr 435 THz 1 times 10minus16 (d) 1 times 10minus16 (d) 009 009 (d) ndash

lt 1 times 10minus17 (p) 09 (p)Yb 519 THz 7 times 10minus16 (d) 1 times 10minus16 (d) 031 03 (d) ndash

lt 1 times 10minus17 (p) 1 times 10minus17 (p) 31 (p)Cs 91 GHz 2 times 10minus16 (d) 1 times 10minus15 (d) 275 027 (d) 1Yb+ 126 GHz 1 times 10minus15 (p) 1 times 10minus15 (p) 34 034 (p) 1

A relevant figure of merit is the ratio of |AX | and fractional frequency instability σ on the perigeevisibility timescale The Cs clock is the only high performance clock with adequate sensitivity tothe particle masses me mq The indicated Cs clock performance was achieved using a microwavelocal oscillator based on a cryogenic oscillator On EGE the latter is to be replaced by an opticallocal oscillator-based device (MOLO) with similar short-term performance but without the needto operate at cryogenic temperature The clock instability at 10000 s is typically similar to theindicated inaccuracy(p) projected (d) demonstrated (1) quadropole (2) octupole

Exp Astron

way to achieve this is by having at least two clocks on board The on-boardclock-clock comparison provides performance evaluation independent of theground-satellite comparison unit Once the satellite clock performances areestablished the link performance can be established as well

The clocks are similar in basic structure but differ in the detailed implemen-tation They consist of

a source of atomsa preparation subunit which cools the atoms to near-standstill (millikelvintemperature or below)a trapping subunit that confines the atom(s) to a tiny volumean interrogation subunita clock control electronics package

The clocks share a common central unit the optical-microwave local oscil-lator which serves as a flywheel oscillator with outstanding stability on shorttimescales (01ndash10 s) for both clocks It consists of a laser stabilized to an ultra-low loss optical resonator made of ultra-low-expansion glass A microwavesignal is derived from the laser by phase-locking a frequency comb to the laserwave and using a multiple of the repetition rate

While the subunits of any optical clock have a similar technological basis(optics electrooptics acoustooptics thermal control vacuum systems) theydiffer in their specifications Depending on the choice of the atom species de-livery confinement and interrogation methods clocks of different perfor-mance result A number of laboratory optical clock demonstrators have beendeveloped worldwide This ensures availability of the know-how for the indus-trial implementation of the flight models

The first clock is a microwave clock based on the interrogation of slowlymoving ensembles of cold Cs atoms It is essentially based on the PHARAOclock [5 6] developed to engineering model level by CNES in the frame-work of the ESA ISS mission ACES In EGE this clock is upgraded usingthe microwave-optical local oscillator This improves significantly the clockperformance The specification is a relative instability (Allan deviation) of3 times 10minus14 (τ s)minus12 ie at the integration time τ = 1000 s it drops to 1 times 10minus15and to 3 times 10minus16 at 10000 s While this specification is at least a factor 3 lessstringent than for the optical clocks the significantly lower cost of obtaining aflight model and the sensitivity of its frequency to the particle masses makes ita suitable choice

The optical clock baseline instrument is a single-ion optical clock Amongvarious suitable ion species (Hg+ Al+ Sr+ Yb+) [44ndash47] the Yb+ ion hasbeen selected Its breadboard performance is established at the instability levelof 3 times 10minus16 over 1000 s it exhibits a large sensitivity coefficient Aα hassmall sensitivity to magnetic fields and exhibits significant potential for furtherimprovements (use of the octupole transition with sim6 times larger Aα andlower overall instability [48] see below) A trapped Yb+ ion is also a veryrobust system it has been demonstrated that a single Yb+ ion can remaintrapped uninterruptedly for many months Finally Yb+ is the only ion with

Exp Astron

which optical clocks demonstrators have been operated in not a single but twomajor metrology laboratories

Another attractive optical clock is based on ensembles of cold neutral Sror Yb atoms trapped in an optical lattice produced by laser beams This clocktype has a lower instability than a single-ion clock thanks to the large numberof atoms (sim105) used Its development is in progress in several laboratoriesin the world (in part supported by space agencies) and has to date reached6 times 10minus17 instability at 5000 s and 1 times 10minus16 uncertainty [49 50] with potentialfor even higher performance However it is more complex and costly thanthe Yb+ ion clock since the cold atom preparation and trapping subunitsinclude more components require higher laser powers and the respectiveoperating procedures are more complex It is therefore not considered for thepresent mission designed to be developed in the very near future with limitedresources but is an interesting candidate for missions without such constraintsIts estimated physical parameters are indicated in Table 2 for comparison

42 Main instruments

421 Microwave clock

The PHARAO cold atom clock is one of the two atomic clocks of the spacemission ACES managed by ESA It is proposed here to include a secondversion of this instrument as its costperformance ratio is very attractive forthe EGE scientific objectives Its concept is very similar to ground basedatomic fountains but with the major difference of zero-g operation Atomsslowly launched in free flight cross two microwave fields tuned to the transitionbetween the two hyperfine levels of the cesium ground state The interrogationmethod based on two separate oscillating fields (Ramsey scheme) allowsthe detection of an atomic line whose width is inversely proportional to thetransit time between the two microwave cavities The resonant microwavefield at 9192631770 GHz (SI definition of the second) is synthesized startingfrom an ultrastable quartz oscillator (USO) in the FCDP phaselocked to themicrowave output of the microwave-optical local oscillator and stabilized tothe clock line using the error signal generated by the cesium resonator In thisway the intrinsic qualities of the cesium hyperfine transition both in terms ofaccuracy and frequency stability are transferred to the USO In a microgravityenvironment the velocity of atoms along the ballistic trajectories is constantand can be changed continuously over almost two orders of magnitude (5ndash500 cms) Differently from atomic fountain clocks presently operated onground very long interaction times (up to few seconds) will be possible whilekeeping reasonable the size of the instrument

The instrument consists of 4 main elements a laser source which providesthe light required to cool and detect the cesium atoms (Fig 5 left) a frequencychain which generates the 92 GHz microwave signal a cesium vacuum tubewhere the interaction between the microwave field and the atoms occurs

Exp Astron

Table 2 Resources and technology readiness level (TRL) of the EGE spacecraft

Unit Volume Mass Power TRL Comment[l] [kg] [W]

Ion clock 100 70 60Atom source 4Atom preparation 6Atom trap 4Atom interrogation 6Clock control 6

Microwave clock 200 91 110 8 TRL by end 2008Local oscillator

Stabilized laser 18 15 23 6Frequency comb 20 16 38 6

Microwave link 18 18 55 8FCDP 7 8 12 8GPS receiver 2 5 7 9CCR 2 2 0 9Payload sum 367 225 305Thermal control 14 17 30 8OBDH 20 24 38 9Star tracker 1 1 8 9Radiation shield 10 40 0 9Support structure 170 204 0 9Power incl solar arr 50 58 10 9Propulsion 21 25 20 9AOCS 36 43 82 9TMTC 8 10 15 9Satellite bus sum 330 422 203 Flexbus conceptPayload plus bus incl 20 margin 8364 7764 6096Propellant 40Launch adapter 100SC wet masspower incl Margins 9164Lattice clock 135 105 152 Alternative clock

Atom source 4Atom preparation 6Atom trap 4Atom interrogation 6Clock control 6

Also included are the resources for an optical lattice clock based on neutral Strontium atoms

(Fig 5 right) and a computer The engineering model of PHARAO is fullyassembled at CNES Toulouse and undergoes functional and performancetests until the end of 2008 The operating temperature range is 10ndash33 andnon-operating temperatures are minus45 +60 The cesium tube consists in aUHV chamber pumped by getters and a 3 ls ion pump a cesium reservoira microwave cavity coils to provide a uniform magnetic field and three layersof magnetic shields Vacuum windows with fiber-optics collimators enable thetransmission of the cooling and detection light onto the cesium atoms

In autonomous mode PHARAO uses the USO as interrogation oscillatorand will provide a clock signal with fractional frequency stability below1 times 10minus13 (τs) minus12 and inaccuracy near 1 times 10minus16 When using the microwave-

Exp Astron

Fig 5 Left The PHARAO laser source engineering model with cover removed Dimensionsare 530 times 350 times 150 mm3 and the mass is 200 kg Ten polarization maintaining optical fibers(yellow) guide the laser beams to the cesium tube All diode lasers are mounted on Peltier coolersfor temperature regulation within 2 mK (courtesy EADS SODERN) Right The PHARAOengineering model cesium tube without the two external magnetic shields The volume is 990 times336 times 444 mm3 and the total mass is 44 kg

optical local oscillator PHARAO will provide a reduced frequency instabilityof 3 times 10minus14 (τs) minus12 This corresponds to 1 times 10minus15 at 1000 s and 3 times 10minus16

at 104 s

422 Ion optical clock

The 171Yb+ ion clock shown in Figs 6 and 7 provides an atomic frequencyreference based on a single 171Yb+ ion confined within an RF electric end-captrap (similar to a Paul trap) within an ultra high vacuum (UHV) chamber andlaser cooled to sim1 mK close to the Doppler cooling limit As a result the ionrsquosfirst order Doppler motion is completely removed and the ion experienceslittle collisional perturbation from its environment Under these conditionsthe narrow linewidth 2S12 (F = 0 mF = 0) minus 2 D32 (F = 2 mF = 0) quadrupole

Fig 6 Left Simplified level scheme of 171Yb+ showing relevant cooling (blue) auxiliary (red)quadrupole clock (green) and octupole clock (dark blue) transitions δνnat is the natural linewidthof the clock transition Middle close-up of Paul-type RF trap (PTB Braunschweig) the ion istrapped in the hole (diameter 14 mm) Right Complete ion trap with vacuum flange height 10 cm(NPL Teddington)

Exp Astron

935 nm D32-state repumper laser

638 nm F72-state repumper laser

739 nm extended cavity diode laser

AOM

PMT

Ion trap UHV chamber

vacuum pumps

Fibre-free space interface

Mu-metal shielding

EC

T

B

Magnetic field

Temperature

Ion fluorescence Processor

Clock laser frequency step and control

4355 nm

Frequengy doubling II

3695 nm cooling wave

Low-drift cavity

Low-drift cavity 171Yb+ 435 nm ion clock unit

876 nm clock laser wave from MOLO

Frequengy doubling

Fig 7 Simplified schematic of the single-ion optical clock unit AOM acousto-optic frequencyshifter PMT photomultiplier tube EC endcap T temperature stabilization system B magneticfield

clock transition at 4355 nm can be interrogated by spectrally narrow and stableclock laser light A diagram of the simplified energy level scheme for 171Yb+showing the relevant cooling auxiliary and clock transitions is given in Fig 6All these wavelengths are provided by diode lasers operated in fundamentalor frequency doubled mode The clock laser light is obtained from the 871 nmwave provided by the MOLO unit (see Section 423) after frequency-doublingin single pass by a KNbO3 nonlinear-optical crystal to 4355 nm The clocktransition spectral profile is observed by stepping the clock laser frequency bymeans of an acousto-optic frequency shifter (AOM in Fig 7) and recordingthe statistics of the quantum jumps in cooling laser induced fluorescence asa function of this frequency Under clock operation the AOM repeatedlysteps back and forward between the half-intensity points on the transitionprofile monitoring the signal imbalance between these points Any detectedimbalance is servo-corrected to zero by feedback to the AOM

In detail the optical clock architecture will comprise (Fig 7)

bull an RF end-cap trap for ionising and confining a single 171Yb+ ion withinan ultra-high vacuum chamber pumped by a small ion pump and non-evaporable getter pump The trap is driven by means of an ac voltage ofa few hundred volts at a drive frequency of sim10 MHz A small oven withseveral mg of 171Yb isotope is heated to provide a low flux of Yb atomsfrom which single ions can be ionised by electron bombardment from ahot wire filament within the trap potential well Low voltage dc is applied

Exp Astron

to additional electrodes to position the ion precisely at trap potentialcentre in order to minimise the ionrsquos micromotion at the drive frequencyThis needs automatic periodic monitoring and correction especially in theperiod after re-loading an ion into the trap External magnetic field coilsin three orthogonal axes allow the nulling of external fields and settingof a fixed field of sim1 μT The trap and coils are surrounded by mu-metalshielding to minimise external field changes during the orbit A level oftemperature control of 1 K of the region surrounding the trap is requiredfor maintaining the blackbody-induced frequency shift uncertainty withtemperature at the 10minus17 level

bull a laser platform to provide frequency doubling of an extended cavity739 nm diode laser for Doppler cooling of the ion on the 2S12(F = 1) minus2 P12 (F = 0) dipole transition at 3695 nm Light from an amplified diodelaser device frequency-doubled in a periodically poled LiTaO3 crystal insingle-pass will be used Typical laser powers used for driving the coolingtransition below saturation are about 2 μW for a beam waist in the trapof about 50 μm By modulation of the injection current of the 739 nmdiode laser at 147 GHz a sideband is generated that excites the 2S12 (F =0) rarr2 P12 (F = 1) ldquorepumperrdquo transition in order to avoid optical pump-ing between the ground hyperfine states

bull The platform also houses the auxiliary lasers at 935 and 638 nm The935 nm diode provides repumping of the ion from the 2 D32(F = 1)metastable level after occasional branching decays to this level during thecooling sequence The 638 nm diode allows fast recovery of the ion fromthe very-long-lived 2F72 metastable state after very occasional collisionaldecay to that state Currently extended cavity lasers are forseen butdistributed-feedback lasers may be available in the near future

bull a high NA lens imaging system and photomultiplier detection system torecord the statistics of 369 nm fluorescence quantum jumps as a functionof clock laser frequency step

bull a fibre system to deliver the various cooling auxiliary and clock light fromsource to trap making use of achromatic doublets where necessary at thefibre-free space interface for launching into the trap

bull a monitoring and control processor which provides the primary coolingand clock laser pulse magnetic field and detection sequencing to observeand lock to the ion clock transition frequency The processor also monitorsfrequency and amplitude data necessary to determine normal laser andion operational conditions and initiate resetting and recovery algorithmswhere necessary and laser unit failure The actual magnetic field presenton the atomic volume is determined and actively stabilized so that aninaccuracy at the level of 1 times 10minus17 can be achieved

bull a redundancy level of at least two units for cooling clock and repumperlasers plus similar for frequency doubling crystals All redundancy unitsfor each wavelength will be fibre multiplexed as standard allowing redun-dant unit activation on determination of prior unit failure

Exp Astron

For an optimized quantum-projection-noise-limited performance of thesingle-ion clock Ramsey interrogation with a cycle time equal to the lifetimeof the metastable D32 level of 50 ms will be performed leading to an Allandeviation of about 27 times 10minus15 (τ s)minus12 This is sim45 times 10minus17 at 1 h satisfyingrequired stabilities for the science objectives Current laboratory results are astability of 1 times 10minus15 at 30 s averaging down to 3 times 10minus16 at 1000 s (PTBGermany) Also comparisons between two independent traps showed agree-ment at the level of several parts in 10minus16 over a series of eight measurementruns

The 171Yb+ ion clock offers a second option namely the 2S12ndash2 F72 octu-pole transition at 476 nm (see Table 1 and Fig 6) [48] It has an extremelylong upper state lifetime low electric quadrupole and second order Zeemansystematic shifts and therefore is a candidate for an even higher performanceclock if a clock laser system with lower instability than the above is providedApart from the clock laser it shares the laser system of the quadrupoleclock Thus with moderate additional resources the Yb+ ion apparatus couldprovide two optical clocks in a single package and specifically perform directfrequency comparisons between the quadrupole and the octupole transitionwith a large sensitivity Aα

The ion clock will operate under microprocessor control providing thecontrol algorithms for the integrated pulse sequencing for cooling repumpingclock interrogation magnetic field switching cooling fluorescence monitoringThe controller will also provide error flags for critical processes and systemreset and correction where necessary This will include ion re-loading andmicromotion reduction algorithms Automatic monitoring of laser power andspectral quality with signal re-optimisation or laser failure determination isrequired Redundant laser units will be pre-aligned and multiplexed into thefibre delivery to the trap so that redundant units can be readily activated in thecase of diode laser failure

423 Microwave-optical local oscillator (MOLO)

This oscillator provides two ultrastable frequency outputs one in the opticalregion the other in the microwave region the two frequencies being coher-ently related The MOLO is composed of three subunits the clock laser thereference cavity and the frequency comb see Fig 8

The clock laser subunit delivers the light for the excitation of the opticalclock transition It is generated by an extended-cavity diode laser at a wave-length of 871 nm Most of the output wave of this laser is sent to the ionclock The laser subunit containing two lasers for redundancy will require3 kg 3 l 15 W The frequency instability of the clock laser must be superiorto the minimum clock instability 27 times 10minus15 (τ s)minus12 for τ up to the timeconstant (sim10 s) of the servo system that locks the laser to the ionrsquos resonancesignal In order to achieve this the laser frequency is stabilized to the reference

Exp Astron

871 nm clock laser

EO

Detector and servo

AOM Femtosecond

frequency comb

To optical clock 10 GHz To FCDP

Microwave-optical local oscillator

ULE cavity

Fig 8 Microwave-optical local oscillator (MOLO) consisting of clock laser reference ULE cavityand femtosecond frequency comb (OFC) Radiofrequency inputs and outputs are not shown EOelectro-optic modulator ULE ultra-low expansion AOM frequency shifter

cavity where the PoundndashDreverndashHall method is used About 10 μW of thelaser output are used for this

The system will exhibit sim04 Hz (sim1 times 10minus15 relative) linewidth and afrequency drift lt01 Hzs The cavity consists of two high-reflectivity mirrorsoptically contacted to a cylinder of diameter 70 mm length 150 mm made ofULE (a glass exhibiting ultra low thermal expansion coefficient) with a cavityfinesse of about 300000 Residual vibrations of the satellite (arms lt 1 times 10minus6 g)should not be a limiting factor to the desired linewidth for safety marginan optimized cavity support can reduce the sensitivity to accelerations Foracoustic and thermal isolation the cavity resides inside an aluminium vacuumchamber equipped with a small (3 ls) ion-getter pump Two stages of polishedaluminium shields are implemented around the cavity which are activelytemperature stabilized so that the cavity is precisely kept at the zero crossingtemperature of the ULE thermal expansion coefficient Heat application andremoval is by thermoelectric elements between the shield and the vacuumchamber Additionally the entire vacuum chamber is temperature-stabilized

During transportation and launch the cavity will be rigidly fixed insidethe vacuum chamber by pressing on it from the sides by piezo-mechanicalactuators allowing accelerations of several g After bringing the satellite onthe orbit the actuators will be released

The optical setup for frequency stabilization will be a classic Pound-Drever-Hall scheme in reflection using an electro-optic modulator (EO) Couplingof the clock laser radiation to the cavity is via single-mode polarization main-taining optical fiber The fiber in- and out-couplers are mounted on miniaturepiezo-motor driven multiaxis translation stages Microprocessor control ofthe translation stages performs a laser-to-fiber incoupling stabilization andlaser-to-cavity mode-matching optimization (positioning inaccuracy lt10 μmalignment inaccuracy lt10 μrad)

The frequency stability of the clock laser will be limited by thermal 1f noisein the mirrors with an fractional Allan deviation slightly below 1 times 10minus15 from1 s to a few 10 s Resources including control electronics are 15 l 12 kg 8 W

Exp Astron

A femtosecond fiber laser based optical frequency comb generator (OFCFig 8) will be used for the clock frequency comparisons and to obtain aspectrally ultrapure microwave local oscillator output for the operation ofPHARAO for the microwave link and for the GNSS receiver The fem-tosecond laser is based on erbium-doped fiber pumped by fiber-coupled laserdiodes It emits 100 fs pulses at a repetition rate of 250 MHz with an averagepower of 100 mW The emission spectrum is centered around 1560 nm Spec-tral broadening to more than one octave is obtained via nonlinear interactionin a microstructure fibre so that the range from about 1050 to 2100 nm iscovered

Second harmonic generation of a portion of the broadened spectrum allowsobtaining the required light for the measurement of the clock laser frequencyat 871 nm The OFC will be referenced to the clock laser frequency (afterintroduction a controllable shift via the frequency shifter AOM) by makingthe carrier envelope offset frequency vanish so that the repetition rate fr isa high-order subharmonic of the optical clock reference frequency f1 = m1 frThe link between the optical and the microwave domain is then simply thedistribution of the repetition rate frequency or more conveniently a high(order 40) harmonic of the comb repetition rate 10 GHz

It has been demonstrated by several groups that frequency comparison andfrequency division with an OFC can be performed to very high accuracy andstability beyond the limitations imposed by the optical clocks themselves Forexample a fiber laser OFC with an instability (Allan deviation) of 6 times 10minus17

in 1 s has been demonstrated [51] OFCs based on fiber technology possessgood long-term reliability and high power efficiency The projection for a spacequalified OFC based on current industrial terrestrial technology is 16 kg mass20 l volume and 38 W power requirement

424 Frequency comparison and distribution package (FCDP)

This package distributes the ultra-stable 10 GHz reference frequency outputof the MOLO to the MWL performs frequency comparisons between MOLOand PHARAO and generates a variety of different reference signals for theindividual instruments including the 250 MHz reference clock to the OFCAll signals are coherent to the MOLO In the FCDP a USO is locked to theMOLO signal by a digital PLL A divider chain generates timing pulses (1 pps1000 pps etc) and keeps a continuous local time scale From the digital phasecomparison data the phase noise density spectrum between the USO and theMOLO is available by telemetry (TM) to assess the overall frequency purityin-flight

As the FCDP distributes the full performance reference from the MOLOdirectly to the MWL there is minimum impact of the FCDP on overall clockcomparison performance The USO-phase lock loop uses a similar designas for ACES-FCDP The digital PLL with linear phase detector allows theUSO to be slightly offset from its reference which is selectable by groundcommand This results in an accurate on-board time scale despite any offset

Exp Astron

of the MOLO reference frequency which can be synchronised to UTC Thefrequency synthesis chain re-uses modules developed for ACES-FCDP andtheir frequency inaccuracy Expected relative uncertainty (Allan deviation)of the frequency comparison between two signals provided to the FCDP is47 times 10minus17 at 1000 s 95 times 10minus18 at 10000 s

During non-availability of the MOLO ie during system start-up andcontingency conditions the USO runs as flywheel (ie it is unlocked withinstability lt15 times 10minus13 from 1 to 100 s) It will still provide useable butless accurate references to the instruments including the reference to OFCCommon view ground-to-ground time- and frequency comparison and supplyof the remaining instruments remain possible at the USO provided accuracy

Mass volume and power consumption are 8 kg 7 l 12 W Controldata rateis 10 kBday 50 kBday (FCDP alone)

425 Microwave link (MWL)

The microwave link is a two-way triple-frequency microwave link (Fig 9) Itis designed to perform accurate frequency comparison as well as time-transfer

Ku-Band Up-linkPower Tx 5 WCarrier 13475 GHzPN-Code 100 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

S-Band Down-linkPower Tx 1 WCarrier 2248 MHzPN-Code 1 MChips1pps 1 time marker sData 5 kBits

Ku-Band Down-linkPower Tx 1 WCarrier 1470333 GHzPN-Code 100 MChips1pps 1 time marker sData 5 kBits

EGE SC

S

Ku

KaKa-Band Up-linkPower Tx 5 WCarrier 345 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

Ka-Band Down-LinkPower Tx 1 WCarrier 32 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 5 kBits

GS Antenna diameterEGE terminal 1 mACES terminal 06 m

Fig 9 EGE microwave two-way link concept and characteristics See text for a description Thelink will be able to perform frequency comparisons with up to ten ground stations simultaneously(only one active comparison is shown) pps pulse per second Tx transmitter SC spacecraft

Exp Astron

between on-board and on-ground clocks Two-way operation allows for com-pensation of the effects due to spacecraft location and its motion relative toground Its two bi-directional links operate in Ka-band (32345 GHz) andKu-band (147135 GHz) The high carrier frequency of the up- and down-links allows for a noticeable reduction of the ionospheric delay A third down-link channel is in S-band (22 GHz) and is used to determine the ionospheretotal electron content MWL uses a combination of carrier phase and codephase data to measure with high precision the absolute ionospheric delayas well as ionospheric delay variations during a ground contact The useof three frequency bands allows to resolve third order ionospheric termsKey modulation parameters are Ka band Tone 250 MHz Ku band PN100 MChips S-band 1 MChips

The EGE-MWL concept is a considerably improved version of the ACES-MWL Microwave carrier generation is directly from the 10 GHz microwaveoutput available from the MOLO and having exceptionally low close-in phasenoise without intermediate frequency dividers or other complex synthesiserstages commonly used in RF transponders The high frequency carriers arekey to achieve the desired frequency comparison accuracy

The Ka-band carrier is generated by tripling the FCDP-generated signalThe Ku-band is subsequently generated by division by two using a verystable regenerative divider Finally the assigned radio frequency is obtainedby adding a low frequency signal using an image reject mixer A mixedanalog-digital receiver design will enhance its performance by use of a flexibledigital signal processing and digital control loops The number of simultaneousground contacts is increased from four (as in the ACES mission) to ten or moreusing the digital concept while the number of physical receiver channels perband is reduced from four to two one for operation to ground and one for in-flight calibration The concept allows adaptation of receiver loop parametersto optimise for the different portions of a highly elliptic orbit while reducingsystematic uncertainties due to high Doppler and Doppler rate Real-timedata exchange to ground terminals is by additional data modulation on all RFsignals The spacecraft carries two different small antennae for perigee and forapogee with gains of 8 dBi 20 dBi beamwidths 100 (full earth coverage) 14respectively power is 1 W for each

The MWL end-to-end performance between the attached clocks includingspace- and ground terminals is predicted to be a relative Allan deviation of1 times 10minus16 at 1000 s (perigee) and 1 times 10minus17 at 10000 s (apogee) Thesevalues are based on conservative implementation losses and other instrumentallosses assumed as 6 dB at apogee and 10 dB at perigee including additionallosses due to atmosphere An attitude knowledge of 1 is required

The MWL link will also support precise orbitographyndashcomplementary tothe GNSS receiver Generation of ranging data is simultaneous to two-waytime transfer and is available under all weather conditions Orbit determina-tion accuracy is expected to be better than 1 cm well matching EGE clockrequirements The value is expected based on demonstrated results fromother missions (METEOR ERS-2 TOPEXPOSEIDON) Laser ranging to

Exp Astron

the CCR array is used a few times during the mission to calibrate the MWLranging

In case of non-availability of the ultrapure signal from MOLO an additional10 GHz output is available from the FCDP to support ground-to-ground clockcomparison in common-view mode even after the nominal mission periodwhen the neither the clocks nor MOLO may be available any longer Basedon the PRARE experimentrsquos 12 years of operation this can be assumed forthe FCDPMWL combination as well The potential extended operational pe-riod makes the proposed configuration particularly attractive to metrologicallaboratories that rely on availability of long-lasting data in support of thegeneration of UTC

The resources of MWL are as follows mass 18 kg volume 18 l power55 W TC 2 kBday TM 120 MBday

426 Auxiliary units

The star camera assembly on the EGE spacecraft will provide real-timeattitude data primarily for the orientation of the spacecraft (with accuracy of1 in real-time) and the proper reduction of the GNSSSLR data

A GNSS receiver is used for the precise orbit determination time andfrequency transfer coarse positioning (lt1 m) for real time use by the Attitudeand Orbit Control as well as the back-up for time tagging of all payloaddata For the time and frequency transfer the receiver is nominally using theonboard frequency reference The receiver assembly includes three specialantennas By the year 2015 more than 100 GNSS satellites are expected tobe available for precise orbit determination permitting orbit inaccuracy of 1ndash2 cm RMS for altitudes up to approx 10000 km At apogee a combinationof GNSS signals and MWL ranging will be used with expected inaccuracyof 10 cm

5 Basic spacecraft specifications

The mission foresees a single spacecraft with a mission duration limited toapprox 3 years for cost reasons Attitude control along three axes withless than 1 error on a timescale of 1000 s is required The control can beperformed with cold gas thrusters magnetotorquers or reaction wheels Theselection will be done on the basis of the vibrations induced by these systems(desired level arms lt 10minus6 g)

Cold gas thrusters or monopropellant thrusters will be required to maintainthe orbit parameters during the mission by small orbit correction manoeuvresand will also guarantee a proper tuning of the orbit parameters after the releasefrom the launcher The ground segment (telemetry control and data transfer)could be via non-ESA stations user-operated to save costs

Further resources are indicated in Table 2

Exp Astron

The satellite bus design drivers are nadir-pointing microwave antennasthermal stability for the payload instruments no strong magnetic sources lowmicrovibrations harsh radiation environment due to crossing of the radiationbelts once per orbit power demand by science payload (with the possibility toreduce power demand by stand-by of some components during specific missionphases eg eclipse) mass limit due to required high altitude with low costlaunch

The satellite bus design can be based on established platforms such asFLEXBUS or PRIMA used on several past missions Their design philosophydiffers one having body-mounted solar arrays the other wings The latter canprovide larger amounts of power The vibration level induced by the panelswhen they are fixed and the duration of solar array drive periods per orbitwill have to be considered in the selection The FLEXBUS concept was usedfor a mission already requiring a low level of residual accelerations and theplatform is optimised for this parameter

For most of the satellite components one can consider using compo-nents and electronic units from navigation or communication satellites whichtypically cruise at altitudes between 15000 and 25000 km altitude LISAPathfinder which will cross the radiation belts during spiralling to theLagrange point will face similar problems concerning radiation as EGE as wellas XMM which is in a Molniya-type orbit with similar radiation characteristics

The instruments will therefore be located in the satellite structure such asto realise ionizing radiation shielding by the spacecraft structure and non-sensitive equipment to the maximum extent possible the need for additionalradiation shielding (including trapped proton radiation) for especially sensitiveparts of the instruments must be determined by ground tests It should bementioned that an orbit with higher perigee and less radiation dose is also anoption as it leads to a tolerable reduction of the science goals

The required plusmn3C temperature stability at the interface to the payloadenclosure is feasible The thermal control will employ the usual means (anenclosure with high thermal inertia heaters and a suitable radiator) Thelocation of the radiator will need a special study given the non-standard orbit

Magnetic field present during the mission can affect the properties of theonboard clock Using SPENVIS the magnetic field has been calculated alongthe spacecraft orbit the field intensity presents maxima around 02 G anda minimum value of the order of 10minus3 G As discussed above this is not alimitation to the instrument performance

6 Discussion and conclusion

The accuracy of the measurements depends crucially on the stability andaccuracy of the clocks While the performance of ground clocks (even thoseeventually deployed in new ground stations dedicated to EGE) will likelyimprove significantly from the present level the space clock design will beldquofrozenrdquo approx 10 years before data acquisition Thus their performance

Exp Astron

level will be significantly below the future ground station clock performanceAs specification we have chosen a performance level that has been demon-strated with laboratory breadboards recently We see no fundamental reasonthat this laboratory performance cannot also be achieved in a space ion opticalclock The experience with the PHARAO cold atom clock has shown thatan engineering model of a laser-based cold atom clock has a performancesimilar to that of a laboratory prototype The perturbation of the ion clock fre-quency due to magnetic fields along the orbit can be modelled and suppressedwith shielding and the residual effects measured and compensated for Themoderate temperature stability (plusmn 3C) to be provided in the experimentalbay can be achieved with thermal shields Accurate temperature controlof particular units is done by active stabilization of the units themselves

One important aspect is radiation which may affect all optical and trapcomponents as well as the lasers and reference cavity The radiation envi-ronment on the EGE orbit is significantly harsher than the near-earth ISSorbit on which ACES will operate The radiation effects must therefore becharacterized comprehensively with appropriate tests and then the radiationshield for the EGE instruments designed

In summary the Einstein Gravity Explorer mission (EGE) is a satellitemission aimed at a precise measurement of the properties of space-time usingatomic clocks By testing one of the most fundamental predictions of EinsteinrsquosTheory of General Relativity the gravitational redshift and thereby searchingfor quantum effects in gravity it explores one of the most important frontiers infundamental physics The EGE mission reaches beyond fundamental physicsproviding an important spin-off in geophysics Furthermore the technologicalaspects of the mission are ground-breaking for future time-keeping and ref-erence frame systems in space and thus are also expected to be of wide useto science and technology in general The payload of the mission has beenconceived as a compromise between cost and performance needing to fit thecost envelope of an ESA medium-class mission1

Acknowledgement We thank S Cesare for helpful information

References

1 Will CM The confrontation between general relativity and experiment Living Rev Relativ9 3 (2006)

2 Ni W-T Empirical foundations of the relativistic gravity Int J Mod Phys D 14 901 (2005)doi101142S0218271805007139

1Finally we point out that a spacecraft with a significantly reduced and thus lower-cost payload (noclocks only MOLO FCDP MWL) on a circular orbit of appropriate height could be used for along-duration earth science mission dedicated specifically to the measurement of the geopotential

Exp Astron

phase) higher compared to systems onboard previous missions (ERS-2TOPEX POSEIDON)Comparison of three different orbit determination systems Laser-Ranging μm-precision microwave ranging GPS-based orbit deter-minationAtmospheric signal propagation study precise determination of thirdorder ionospheric termMonitoring stability of the GalileoGPSGlonass time scales

22 Gravitational frequency shift measurement and search for new physics

The primary science goal of EGE is a measurement and test of the gravita-tional frequency shift in the weak-field limit as predicted by metric theoriesof gravitation These theories also predict that the shift is independent of thetype of clock used to measure it The comparison of the frequencies ν oftwo identical clocks 1 2 operating at different locations x1 x2 yields a fre-quency ratio

v2(xprime)v1

(xprime) sim= 1 + (U (x2) minus U (x1))

c2 (1)

Here νi(xprime) is the frequency of clock i located at xi as observed (measured) ata particular location xprime where the comparisons are performed The comparisonlocation can be arbitrary but will often be x1 or x2 or both U is thegravitational potential U(x) = minusGM|x| in case of a spherically symmetricbody of mass M In short time runs (or clocks tick) more slowly the closer thesystemclock is to a massive body Stationary clocks and observers and weakgravitational fields have been assumed in Eq 1

Through its measurement EGE searches for a possible violation of the grav-itational frequency shift A violation may be described phenomenologicallyby a dependence of one or more fundamental constants on the gravitationalpotential X = X(Uc2) where X is a generic dimensionless fundamentalconstant or a dimensionless combination of fundamental constants Such adependence is a violation of the principle of Local Position Invariance (LPI)which states that the outcome of local nongravitational experiments is inde-pendent of where and when in the Universe they are performed If a constantdepends on the gravitational potential then the results of those laboratory-type (ie local) experiments that depend on that constant will also depend onthe location namely on the location of the laboratory with respect to the massdistribution generating the potential and will violate LPI

LPI is one of the three cornerstones of the Einstein Equivalence Principle(EEP) the others being the Weak Equivalence Principle and Local LorentzInvariance It is a common belief that the EEP may not be absolutely valid -therefore spatial dependence and time dependence of fundamental constantshave become a very intense research topic in the last decade with experimentsundertaken worldwide to search for violations of various aspects of EEP For

Exp Astron

example extensive studies of atomic and molecular spectra of gas clouds in thedistant Universe are currently undertaken to search for a difference of theseparameters compared to todayrsquos values (see eg [7ndash11] and references therein)Also some differences in Big Bang nucleosynthesis data and calculations canbe naturally explained by a variation of fundamental constants [12 13]

The EGE mission is thus a probe of possible space-dependence of fun-damental constants The search may help to point the way toward a unifiedtheory of all forces reveal hypothetical extra dimensions in our Universe or theexistence of many sub-Universes with different physics chemistry and biologyIndeed theories unifying all forces applied to cosmology suggest space andtime dependence of the coupling constants [14ndash16] Another argument forspatial variation of fundamental constants comes from the anthropic principleThere must be a fine tuning of fundamental constants which allows humans andany life to appear Life in its present form is not consistent with fundamentalconstants whose values only slightly differ from the actual ones This finetuning can be naturally explained by the spatial variation of the fundamentalconstants we appeared in the area of the Universe where the values of thefundamental constants are consistent with our existence

In the Standard Model there are three main fundamental dimensionless pa-rameters determining the structure and energy levels of stable matter (atomsmolecules) the electromagnetic fine structure constant α and the two ratiosof electron mass me and light quark mass mq to the strong QCD energy scaleQCD Xeq = meqQCD The fundamental masses me and mq are proportionalto the vacuum Higgs field which determines the electroweak unification scaleTherefore the constants meqQCD can also be viewed as the ratio of the weakenergy scale to the strong energy scale

How can a space-time variation of the fundamental constants and a depen-dence on the gravitational potential occur Light scalar fields very naturallyappear in modern cosmological models affecting α and meqQCD Oneof these scalar fields is the famous ldquodark energyrdquo which causes acceleratedexpansion of the Universe Another hypothetical scalar field is the dilatonwhich appears in string theories together with the graviton Cosmologicalvariation of these scalar fields should occur because of drastic changes ofthe matter composition of the Universe During the Big Bang nucleosyn-thesis the Universe was dominated by radiation then by cold dark matterand now by dark energy Changes of the cosmic scalar field ϕ0(t) lead tothe variation of the fundamental constants X(ϕ0) Massive bodies (galaxiesstars planets) can also affect physical constants They have a large scalarcharge S = sp Z + se Z + sn N where Z is the number of protons and elec-trons and N is the number of neutrons sp se sn are the scalar charges ofthe proton electron and neutron respectively There is also a contributionof the nuclear binding energy (scalar charge of virtual mesons mediatingnuclear forces)

The scalar charge produces a Coulomb-like scalar field ϕs = SR where Ris the distance from the charge The total scalar field is ϕ = ϕ0 + ϕs therefore

Exp Astron

we may have a variation of the fundamental constants inversely proportionalto the distance from massive bodies

X (ϕ) = X (ϕ0 + ϕs) = X (ϕ0) + δX (R)

δX (R) = (dX

dϕ

)S

R

where a nonzero δX violates LPI and a small ϕs was assumedThe gravitational potential U is essentially proportional to the number of

baryons Z + N and inversely proportional to the distance R U = -GMR

Therefore the change of the fundamental constants near massive bodies canbe written

δX

X = K (X i) δ(Ui

c2

) (2)

where i refers to a particular composition of the mass M and where thecoefficients K(X i ) = minus(dXdϕ) c2SiGMi are not universal but are expectedto be different for eg the Sun and the Earth K(X Earth) = K(X Sun)Indeed the Sun consists mostly of hydrogen therefore the Sunrsquos scalar chargeis SSun Z (sp + se) The Earth contains heavier elements where the numberof neutrons exceeds the number of protons N 11 Z Therefore the scalarcharge of the Earth SEarth Z (11 sn + sp + se) and is sensitive to the neutronscalar charge (and also to the contribution of the nuclear binding) in contrastto the Sun

The most sensitive experiments suitable for a search of a space-time de-pendence in local or near-local experiments (such as satellite experiments) areclockndashclock comparisons In one type of terrestrial local experiment dissimilarclocks located nearby are compared for a duration spanning at least one yearDuring such periods the earth moves periodically closer and further awayfrom the sun so that the solar gravitational potential on Earth is modulatedThis enables a search for a dependence of the fundamental constants on thesunrsquos scalar charge [17 18] These experiments are able to set limits to K(XSun) for X = α meqQCD at the level of 10minus6 [12 13]

The proposed EGE satellite experiments are complementary to terrestrialexperiments allowing to set limits to K(X Earth) and combining with theresults on K(X Sun) from terrestrial comparisons to set limits on the depen-dence of the fundamental constants on the neutron scalar charge sn

In summary when the frequencies of two identical clocks (of nominalfrequency νa) that are located at different positions x1 x2 are comparedby measuring their frequency ratio at some position xrsquo the result may bewritten as

va2(xprime)

va1

(xprime) = 1 +

[1 +

sum

XAXa K (X i)

](Ui (x2) minus Ui (x1))

c2 (3)

The sensitivity factor AXa = X (d(ln νa)dX) is the sensitivity of the clockfrequency to a particular constant X and can be calculated using atomic and

Exp Astron

nuclear theory (see eg [12 13]) The measurement of this ratio which issensitive to K represents a test of General Relativityrsquos redshift

When the frequencies of two co-located but dissimilar clocks a b (ofnominal frequencies νa0 and νb 0) are compared in-situ at two locations x1 x2

(LPI test) the result is

[va1 (x1)

vb 1 (x1)

] [va2(x2)

vb 2 (x2)

] = 1 +[X

(AXa minus AXb

)K (X i)

](Ui (x2) minus Ui (x1))

c2 (4)

The result of the EGE mission may be given as a limits for (or nonzerovalues of) K(X Earth) for the three fundamental constants X = α meQCDand mqQCD

23 Measurement approaches for the gravitational frequency shift

The gravitational potential experienced by the ground and satellite clocksincludes the contributions from the earth and the sun (the contribution fromthe moon is an order smaller and may be neglected for the purposes of thediscussion but will be taken into account in data analysis)

Nonzero gravitational frequency shift signals (Eq 3) can be obtained inthree ways Method (1) below is the most sensitive and represents a strongreason to employ an optical clock which on short averaging times is ableto provide a higher frequency stability than a microwave clock Method (2)and (3) are less accurate but are complementary in terms of measurementapproach Finally method (3) addresses the solar gravitational frequencyshift while (1) and (2) measure the terrestrial effect The performance of theinstruments assumed in the following is shown in Fig 2

(1) Repeated measurements of satellite clock frequency variation betweenapogee and perigeeIn this measurement approach the difference (or ratio) between earthndashsatellite clock frequency ratio at perigee and at apogee is consideredThis difference is measured at each perigee and apogee passage andaveraged over the mission duration Systematic shifts (if not correlatedwith orbital motion) are expected to average out leading to a gain insensitivity of up to n12 where n = number of orbits This mode usesthe stability properties of the satellite clocks on the timescale fromperigee to apogee rather than their accuracy Assuming a specificationof le 3 times 10minus16 satellite optical clock fractional frequency instability forintegration times between 1000 and 25000 s n sim= 1000 comparisons(ie most orbits of the 3-year mission) and a perigee-apogee potentialdifference (U(apogee) minus U(perigee))c2 = Uc2 sim= 4 times 10minus10 allows a

Exp Astron

Fig 2 Performance of clocks and links on EGE Curve ldquooptical clocks EGErdquo is the specificationCurve ldquooptical clocks EGE (goal)rdquo refers to expected near-future performance of an opticalclock based on neutral atoms ldquoUpgraded PHARAOrdquo refers to the PHARAO microwave clockusing the high-quality microwave-optical local oscillator The microwave link (MWL) performanceexhibits a discontinuity since common-view contact time of ground stations with EGE is limited toseveral h The different slopes arise because on short time scales only white phase noise is relevantwhile on long time scales flicker noise also contributes The reproducibility is taken into account

measurement of the redshift at 25 ppb accuracy The ground clocks towhich the satellite clocks are compared to at perigee and at apogee willbe located in different ground stations The inaccuracy of these groundclocks must therefore be sufficiently smaller than 1 times 10minus17 in order notto contribute to the 25 ppb error Current state-of-the art inaccuracy is2 times 10minus17 and is likely to decrease further

(2) Absolute comparison between ground and satellite clocks with the satel-lite at apogeeHere the terrestrial gravitational potential difference is maximumU(apogee) minus U(perigee)c2 sim= 65 times 10minus10 Assuming a space optical clockinaccuracy of 1 times 10minus16 and a ground clock inaccuracy a factor 3lower (current state-of-the-art) allows a measurement of the redshift at150 ppb This measurement takes advantage of the repetitive nature ofthe orbit only to a small extent and relies on the accuracy of the opticalclocks in particular of the space clocks It is thus complementary to (1)albeit less accurateIn (1) and (2) the dominant contribution to the redshift is from Earthrsquosgravitational potential The contribution from the solar potential isstrongly reduced due to the fact that satellite and earth are an interactingsystem in free-fall toward the Sun [19] and can be modeled

(3) Comparisons between terrestrial clocksThe frequency comparison link to the EGE satellite allows terrestrialclock comparisons by a common-view method whereby the frequency

Exp Astron

signals from two distant ground clocks are sent to the satellite andcompared there The frequency difference between two identical groundclocks is determined if systematic instrumental errors are negligible bythe difference in gravitational potential at the two clock locations Thispotential difference includes a terrestrial and a solar contributionThe solar contribution varies according to the orientation of the clockpair with respect to the Sun with a 1 day-period For two clocks at asurface distance of sim6000 km and suitable orientation with respect tothe sun the normalized modulation amplitude is sim= 4 times 10minus13 This is thesame as the earth potential difference for 4 km height difference It ispossible that ground clocks will have reached an inaccuracy at the levelof 1 times 10minus18 at the time of the mission in 2015 Using the microwave linka comparison of ground clocks over sim3 h (typical common-view duration)with a resolution of 1 times 10minus17 can be performed limited by the instabilityof the link Repeating the clock comparison at each common-view win-dow (n = 1000 comparisons) will allow averaging and a measurementof the solar gravitational shift with fractional inaccuracy of 1 ppm Incomparison the best current results for the solar gravitational frequencyshift are at the few-percent level [20 21] A moderate averaging out ofground clock systematic shifts is assumed as well as modeling of the earthgravitational potential variations on the daily timescale at the appropriatelevel

(4) Comparisons of on-board clocksThe two on-board clocks are continuously compared these data areanalyzed for violation of LPI (Eq 4) Since the measurement is a nulltest the required accuracy of the gravitational potential at the satellitelocation is low and easily satisfied

24 Test of local position invariance and search for neutron scalar charge

The test of LPI is performed in conjunction with the measurement of thegravitational frequency shift

With approach 23 (1) the optical clock will be able to measure K (α Earth)with accuracy 30 ppb (spec)

Combining the results of the microwave and optical clocks will determineK(meQCD Earth) + εK(mqQCD Earth) with accuracy 85 ppb (spec)where ε asymp minus01

Simultaneously the on-board comparison 23 (4) yields the combination19 K(α Earth) +K(meQCD Earth) + εK(mqQCD Earth) with accuracy85 ppb (spec)

Note that these measurements are fully complementary to current andfuture terrestrial clock-clock comparisons which are sensitive to K(X Sun)and can be combined with the latter to determine or set limits to the neutronscalar charge

Exp Astron

25 Theoretical analysis of frequency comparison signals

The on-board clock-clock comparison signals can be analyzed straightfor-wardly in terms of a violation of LPI since signal propagation effects areabsent

The determination of the gravitational frequency dilation via ground-satellite clock comparisons is more complex (Eq 1 is strongly simplified) andrequires an accurate treatment of all relevant relativistic effects Gravitationin the solar system is described in the barycentric (BCRS) and geocentric(GCRS) non-rotating celestial reference systems by means of post-Newtoniansolutions of Einsteinrsquos equations for the metric tensor in harmonic gaugescodified in the conventions IAU2000 [22] In these conventions the relativisticstructure of Newton potential (order 1c2) and gravitomagnetic (order 1c3)

potentials are given For the Newton potential the multipolar expansion ofthe geopotential determined by gravity mapping is used The post-Newtonianeffects of the metric (described in terms of dimensionless parameters β and γ )

can be included Orbits of satellites are evaluated in the GCRS at the levelof a few cm by means of the Einstein-Infeld-Hoffmann equations at the order1c2 [23] using the relativistic Newton potential Instead the trajectories of theground clocks in the GCRS are evaluated from their positions fixed onthe Earth crust (ITRS International Terrestrial Reference System) by usingthe non-relativistic IERS2003 conventions [24 25]

For the propagation of light rays (here radio signals) between the satelliteand the ground stations one uses the null geodesics of the post-Newtoniansolution in the GCRS The timefrequency transfer properties have beentheoretically evaluated for an axisymmetric rotating body [26ndash29] at the order1c4 (ie with Newton and gravitomagnetic potentials developed in multipolarexpansions) These results have been developed for use in the ACES missionThe relative frequency dilation can be expressed as

ν

ν= 1 + 1

cL1 +

4sum

n = 2

1

cn(Ln + Gn)

where Ln describes special-relativistic Doppler effects and Gn the general-relativistic effects These quantities are time-dependent due to the orbitalmotion For concreteness we report the orders of the terms The kinematicaleffects are |L1c| lt 1 times 10minus5 |L2c2| lt 2 times 10minus10 |L3c3| lt 3 times 10minus15 |L4c4| lt

7 times 10minus20 The Newtonian contributions are G2 = G(M)2 + G(J2)

2 + G(J4)2 + G(J6)

2(M is the mass monopole Jn are the lowest multipoles) with estimates|G(M)

2 c2| lt 3 times 10minus11 |G(J2)2 c2| lt 2 times 10minus13 |G(J4)

2 c2| lt 3 times 10minus16 |G(J6)2 c2| lt

1 times 10minus16 For G3 = G(M)3 + G(J2)

3 the estimates are |G(M)3 c3| lt 2 times 10minus14

|G(J2)3 c3| lt 1 times 10minus18 Finally the G4 = G(M)

4 + G(S)4 contributions (G(S)

4 is thelowest gravitomagnetic effect) are of order |G(M)

4 c4| lt 1 times 10minus19 |G(S)4 c4| lt

1 times 10minus22 and thus not relevant for EGE

Exp Astron

The available theory is fairly complete for the data analysis in mode 23 (1)described below For modes 23(2 3) the theory will need to be generalizedto include the effect of the gravitational potential at the location of the groundclock(s) (including time-varying effects such as the tides) and of the solar andlunar potentials

The uncertainties of the various contributions above are minimized by ausing precise EGE orbit data obtained by satellite tracking and by usingcurrent and future accurate earth gravity information Required orbit positionknowledge is at 1 cm level near perigee and sim50 cm near apogee reachablealready today with the proposed orbitography approach described below Thevalidity of the special-relativistic Doppler shift has been and will continue to beindependently verified with increased accuracy by spectroscopy of relativisticatomic ion beams [30] Potential violations of this aspect of Lorentz Invari-ance are expected to be sufficiently bounded so as not to affect the signalinterpretation

26 Local Lorentz invariance tests

There has been an explosion of interest in Local Lorentz invariance in re-cent years [31 32] with numerous astronomical studies and high-precisionexperiments applied to search for violations of Lorentz invariance Signifi-cant developments have also occurred on the theoretical side and a theory(standard model extension SME [33]) has been worked out that provides aunified framework for describing and analyzing Lorentz Invariance violationsin a variety of systems Two aspects of Lorentz Invariance can be tested withEGE

261 Independence of the speed of light from the laboratory velocity

A possible dependence of the speed of light on the speed v of the laboratorycan be modeled according to the MansourindashSexl test theory [34] by

c (v) = c0(1 + Bv2

c2

)

Here v 377 kms is the velocity with respect to the cosmic microwavebackground the cosmologically preferred frame and B is a combinationof parameters describing deviations from the usual Lorentz transformationformulas B = 0 if Lorentz Invariance holds For a terrestrial experiment therotation of the Earth around its axis modulates v with a 300 ms amplitude Adependence c(v) can be searched for by measuring the frequency differencebetween a resonance frequency of a highly stable optical cavity (cavity fre-quencies are proportional to c) and an optical clock (Kennedy-Thorndike-typeexperiment [35]) and determining the modulation of the frequency differencecorrelated with the modulation of v The advantages of a space experimentare the high orbital velocity and strongly reduced cavity deformation thanksto microgravity [36 37] On the proposed elliptic orbit v varies between+4 and minus4 kms over approximately 1 h This variation is 13 times larger

Exp Astron

than for the case of a terrestrial experiment The shorter time scale is alsoadvantageous since the drift of the cavity is more predictable In the EGEmission the reference cavity of the clock laser is used The large numberof orbits improves the statistics An improvement of the accuracy of B by afactor 20 compared to the best current terrestrial results is realistic The dataanalysis can also be performed in the framework of the SME theory takinginto account complementary bounds obtained from terrestrial experimentsand astrophysical observations

262 Independence of Zeeman splitting frequency on the directionof the magnetic field

One class of tests of Lorentz Invariance consists in measuring the frequencysplitting between two levels of a quantum system induced by a static magneticfield as a function of orientation of the field direction with respect to thestars (HughesndashDrever-type experiments [31]) This class addresses the so-called matter sector and is complementary to combined-photonmatter-sectortests such as the above The most precise experiments are performed withatomic clocks eg masers or cold atom clocks [38] The EGE instruments allowsuch experiments since magnetic fields are applied continuously in the clockor repeatedly for calibration Compared to terrestrial experiments the highorbital velocity of EGE could lead to an improvement by a factor sim20

27 Application to geophysics

Using GPS coordinates of a point on the Earth can be obtained with aninaccuracy well below 1 cm in a well defined international terrestrial referencesystem The coordinates are purely geometrical and do not contain any gravityinformation A local gravitational potential UB is determined with respect toa reference gravitational potential U A by ldquolevellingrdquo the surface of the Earthie sequentially measuring gravity g and height differences every ca 50 m

UB minus U A = minusBint

A

g (r) middot d n

where n is the difference vector between subsequent locations The disadvan-tage of measuring height differences by the levelling method is a random walkeffect of accumulated errors

In the classical definition a geoid is defined as the particular equipotentialsurface nearest to mean sea level In these terms the geoid serves as areference surface for measuring the height and also to define a referencefor the gravitational potential Such a vertical reference frame historicallywas realized for a country or several countries by determining the mean ofsea level observations at tide gauge stations taken over long period of timeHowever modern satellite altimetry missions such as TOPEXPOSEIDON orENVISAT show that departure of the mean sea level from an equipotential

Exp Astron

surface may reach up to several meters on the global scale see eg [39]Therefore height systems based on different tide gauge stations may differin the realisation of the geoid and may differ with respect to the reference byseveral meters

The best gravity field models obtained from satellite data (eg missionGRACE) have reached a precision of about 1 cm over sim250 km half wave-length [40] Satellite data also reveals changes in time [41] However for typicalEarth topography with height variations of eg 1000 m over 30 km horizontaldistance one may expect variations in the geoid of about 80 cm Such ahigh-spatial-frequency signal in geoid variations cannot be detected by spacegravity missions and requires a combination of satellite and terrestrial gravitymeasurements like gravity anomalies deflections of vertical and GPSlevellingpoints The best combined global gravity field models are provided up to de-gree and order 360 in terms of spherical harmonic representation correspond-ing to a half-wavelength of about 55 km However in combining the satelliteand terrestrial gravity field measurements the problem of terrestrial data givenin different height systems between continents and different countries remainsand has to be tied to satellite measurements

The geodetic scientific community is currently establishing a Global Geo-detic Observing System (GGOS) [42] Its objectives are the early detectionof natural hazards the measurement of temporal changes of land ice andocean surfaces as well as the monitoring of mass transport processes in theEarth system Global change processes are small and therefore difficult toquantify Therefore the required precision relative to the Earthrsquos dimensionis 1 ppb GGOS will be established by the combination of geodetic spacetechniques (GPS Laser-Ranging very large baseline interferometry) andrealized by a very large number of terrestrial and space-borne observatoriesThe purely geometric terrestrial 3D coordinate system is in good shape andfully operational It has to be complemented however with a globally uniformheight system of similar precision The current precision level of regionalheight systems in terms of gravity potential differences is in the order of1 m2s2 (10 cm) with inconsistencies between these various systems up toseveral 10 m2s2 (several meters) The actual requirement in the context ofGGOS is 01 m2s2 (1 cm) with the need of a permanent ie dynamical controlThis requirement of high precision height control comes from the need tounderstand on a global scale processes such as sea level change global andcoastal dynamics of ocean circulation ice melting glacial isostatic adjustmentand land subsidence as well the interaction of these processes Only by meansof monitoring in terms of gravity potential changes at the above level ofprecision the change of ocean level can be understood as a global phenomenonand purely geometric height changes be complemented by information aboutthe associated density or mass changes

The geoid can also be defined in a relativistic way [43] as the surface whereaccurate clocks run with the same rate and where the surface is nearest tomean sea level The relation between the differences in the clock frequenciesand the gravitational potential is given in simplified form by Eq 1 At present

Exp Astron

there is no operational way to compare frequencies of the already availableoptical clocks on the global scale at the same level as their accuracy wouldallow With EGE and by then improved ground optical clocks it will becomepossible to obtain gravitational potential differences on a global scale by com-paring frequencies This would be a significant new dimension to gravity fielddetermination since such observables are given on the global scale and providein situ local gravity information at the same time EGE will allow clock-basedgravitational potential mapping with the same payload instruments used forthe fundamental physics experiments The on-board clocks are actually notrequired for this purpose MOLO MWL and FCDP suffice Thus EGE willallow establishing a global reference frame for the Earth gravitational fieldwith accuracy in the order of a few cm in terms of the geoid heights Thisassumes that by the time EGE is operational (mobile) ground clocks withfractional inaccuracy at the level of 10minus18 are available the measurementsare limited by MWL noise and thus require a long integration times to matchthe expected ground clock performance Such a reference frame will serve asreference for all gravity field modeling and clocks will be used to define theinternational atomic time scale and the Earth gravitational potential at thesame time

In summary with optical clocks geodesy will undergo a second revolutionafter geometry now being measured by clock-based GNSS systems also physi-cal heights and potential will be measured by (mobile) optical clocks using thegravitational redshift effect

28 Application to frequency standards and terrestrial fundamentalphysics studies

As has been described repeating ground clock comparisons and integratingover several months will provide global ground clock frequency comparisonsat the level of 1 times 10minus18 more than two orders of magnitude below the currentGPS and two-way satellite time and frequency transfer methods In the timeand frequency community the availability of such a global high performancemicrowave link will accelerate the process leading to a redefinition of the SIsecond based on clocks operating in the optical domain

The global character of the ground clock comparisons made possible byEGE can also contribute to the search for a time-variation of the fundamentalconstants

The comparisons do not require availability of the on-board clocks The riskof this type of measurements is therefore reduced Moreover the lifetime ofmany subunits required for this measurement (MWL and FCDP see below)can exceed 10 years (from previous experience with similar devices) andthus offers a motivation to increase the mission duration beyond the nominal3 years

Further applications to metrology are listed under the secondary goals inSection 21 above

Exp Astron

3 Mission profile

31 Orbit

The orbit must satisfy the following requirements

Primary 1 Large gravitational potential difference between apogee andperigee

2 sufficient contact time with at least two ground stations atperigee

Secondary Simultaneous visibility of satellite from distant (inter-continental) ground stations for several hours

To satisfy the primary objective the orbit must be highly elliptic The perigeeshould be low A very low perigee means a short contact time which decreasesthe accuracy of the frequency measurement A compromise is a perigeealtitude of sim2500 km For a sufficiently high apogee this yields a U of 60of the theoretical maximum between the earth surface and infinity

The baseline scenario is an orbit of approx 12 h period with the followingproperties

Apogee Altitude 37856 km (minus 10 deg latitude one over Atlantic andPacific resp)

Perigee Altitude 2500 km (+ 10 deg latitude one over Guatemala andMalaysia resp)

Inclination 634 argument of perigee 170RAAN 25 True Anomaly 220Apogee drift rate 70 kmday perigee drift rate sim 14 kmday

True anomaly and right ascension of the ascending node (RAAN) wereoptimized to minimize orbital precession Figure 3 shows the ground track andthe ground coverage at one of the perigees Common-view contact time forstations in Europe is about 6 h while between a station in Europe and in theUSA it is about 35 h Frequency comparison windows between a given groundstation and EGE repeat every 24 h

A final selection of the orbit requires an in-depth study of the costs of thepropellant and maneuvers as well as of the setting up of new ground stations

32 Ground segment

The mission requires two separate ground segments the science groundsegment and the standard satellite control segment which supports telemetryand telecommand (TMTC)

Exp Astron

Fig 3 Baseline EGE orbit One (out of two) apogee and one (out of two) perigee ground pointsare shown The ground coverage at one of the perigees is shown with circles at elevation anglesteps of 10 starting from 10

321 Science ground segment

As described above one part of the mission is to perform frequency com-parisons between ground-to-space and ground-to-ground (common-view andnon-common view) The use of a two-way microwave link (MWL) systemhas the advantage that a precise orbit or the precise relative velocity of thesatellite must not be known [4] The orbit determination requirements fora two-way system are significantly relaxed compared to the requirements tomodel the relativistic effects on the clocks The frequency comparison resultsare available both on ground and on the satellite

For ground-to-satellite clock comparisons five ground stations arefavourable two capable of viewing perigee three for viewing apogee Thisallows for the classic three-clock configuration where the performance of eachclock can be obtained individually The perigee stations will be observing thesatellite for sim25 min every perigee passage for the duration of the missionThe apogee stations will observe for sim3 h every apogee passage Of coursea larger number of stations as well as longer observation time are usefulfor robustness redundancy and orbit determination The stations could beinstalled on a temporary basis for the duration of the mission

The stations must have access to optical clocks of performance at least ashigh as todayrsquos state-of-the-art ground clocks but preferably higher Theywill include an optical frequency comb by means of which the optical clockfrequency is transformed to a radio frequency for comparison with the EGEclocks High-accuracy fiber-optic link technology already existing today allowsdistances of sim100 km between the ground station and the correspondingoptical clock allowing a certain freedom in the choice of the clock locationWith further increase of the fiber-optic link range currently being investigated

Exp Astron

the area over which the ground station can be positioned relative to the opticalclock will increase This option is of importance for the implementation of ageophysics measurement campaign for which a large number of clock locationsis of interest

The above ground stations will be complemented by some 15 additionalground terminals provided by the world-wide laboratories equipped with highperformance primary clocks at the time of the mission These will participate inorbit determination to 1 cm accuracy along the full orbit which can be achievedwith non-optical atomic clocks The EGE MWL is fully compatible to plannedACES ground terminals which might reduce the number of required stations

With a larger number of transportable optical clock ground stations de-ployed and moved across the continents a geophysics campaign could beimplemented to map the gravitational potential

Ground station equipment MWL uses dedicated small transportable termi-nals which can be easily installed at any location of interest Each terminalconsists of a steerable dish antenna of 1 m diameter with pointing accuracy of01 using a transmitted power of 5 W per band The high directivity antenna(beam aperture Ka-band 07 Ku-band 14 S-band 20) avoids multi-pathfrom nearby objects Antenna noise temperatures are Ka-band 450 K Ku-band 340 K Operation is at elevations above 10 and measurements areperformed above 20 For ranging purposes terminal location is determined byGPS surveying Each ground station computes its own ionospheric correctionsbased on a triple-band receiver The terminal has a built-in delay monitorwhich allows calibrated ranging during several months once its bias has beendetermined by laser ranging Its M and C interface is via local area networkto either a userrsquos computer or directly to the science network control centreby remote MampC The stationrsquos operation is fully automatic and unattendedSchedule and orbital data are received via LAN once per day Signal acquisi-tion is fully autonomous Data transmission is after each pass The terminal hasan uplink telecommand capability of 1 kBits and a downlink data capacity ofup to 10 kBitss For the local user there is a visual interface providing quick-look data from real-time comparison to the spacecraft clocks to verify theirgood health The final full performance product is calculated at the sciencedata centre

Local metrological data (temperature pressure humidity) are recorded tocorrect for tropospheric delay of the ranging data The main advantage ofMWL terminals compared to laser based systems is their all-weather capabil-ity although microwaves suffer higher propagation errors in the ionosphereand troposphere

The EGE ground terminal will profit from the ERS-2 mission PRARE(precise range and range rate equipment) [45] experiments whose terminalshad similar design and have been operated under extreme climatic conditionsincl Antarctica and high temperature regions from existing designs fromthe space segment (mission ACES) and no further special developments aredeemed necessary Cost per EGE master station is modest

Exp Astron

Satellite tracking via laser ranging Laser ranging is alternative to MWL rang-ing and also serves for absolute calibration of the latter For this purpose anon-board corner cube reflector (CCR) is provided Ranging will be done withthe International Laser Ranging Service Ideally the satellite will be rangedat every perigee passage and every apogee passage simultaneously with thefrequency comparison procedures Todayrsquos ranging precision is sufficient tosatisfy the science goals For apogee observations in the northern hemispherea large number of laser ranging stations is available The number of stationscapable of ranging at or near perigee in the southern hemisphere is six andthus is sufficient even considering unfavourable weather conditions at someof these The MWL ground stations are able to range even under thoseconditions

322 Control ground segment

Average data transfer rate is estimated at 300 MBday (approx 40 kBs onaverage) Data will be stored on board (2 GB capacity required) and will betransferred to a receiving station on the Earth once a day

XPLC

incl Data StorageIF to TMTC Link

HK amp TMTC

to ground

Opt Clock

AOM

MOLO

PHARAO

FCDP

10 GHz

100 MHz

250 MHz

GNSS Rx

MWLTampF Link

uplink Data

UTC(k)

TampF Space vs Ground

TM from GS

Tim

e S

cale

Orbit GNSS Time UTC

TampF Space vs Ground

MampC Time Scale

MampC

MampC

MampC

MampC

Mamp

C

PHARAO vs Opt Clock

PHARAO vs Opt Clock

10 GHz amp Epoch

10 MHz

amp Epoch

Epoch

10 MHz

amp Epoch

Opt Link

Fig 4 Schematic of the payload showing subunits and their signals The results of the frequencycomparison between the two on-board clocks (optical clock and microwave clock PHARAO) areobtained from a combination of the frequency values applied to the frequency shifters AOM (seeFigs 7 and 8) and those output from the frequency control and distribution package (FCDP)as indicated by the arrows ldquoPHARAO vs Opt Clockrdquo The satellite clock to ground clockcomparison results are output from the MWL XPLC is the on-board computer and is involvedin generation of signals for instrument control The optical link shown dashed is a possible add-onproviding an alternative to the MWL GS ground stations MampC monitoring and control HKhousekeeping TampF time and frequency

Exp Astron

4 Main payload instruments

41 Overview

The satellite payload consists of a pair of cold-atom clocks a microwave-optical local oscillator (MOLO) an on-board frequency distribution unit(FCDP) a satellite-to-ground clock comparison unit (MWL) and the onboardcomputercontrol system (XPLC) see Fig 4 Auxiliary units are the posi-tion determination GPSGalileo receiver (GNSS Rx) attitude monitor (startracker not shown) and the corner cube reflector array (CCR not shown)

The science goals require clocks of very high frequency stability on thetimescales of half the orbital period (6 h) as well as perigee passage (sim1000 s)see Fig 2 Only clocks based on cold atoms are suited for the science goalsTable 1 reports values for a subset of atomic clocks that have been investigatednot including clocks that require cryogenic operation or are too complex foruse in an medium-class mission The choice of a clock pair rather than a singleone fulfils several requirements it addresses more science goals improves thevalidity of some science results and it enables calibration accuracy evaluationand redundancy The validity of the science results is enhanced since theclocks differ not only in the employed atoms but also in terms of havingdifferent sensitivity to perturbations by electric and magnetic fields electronicserrors and misalignments A comparison of the results arising from each clockallows for an understanding of measurement errors Both the clocksrsquo and thefrequency comparison unitrsquos performances will be determined after launchand regularly verified during the course of the mission The only reliable

Table 1 Properties of different clock candidates for EGE

Clock Clock Fractional σ = Fractional Sensitivity Figure of Sensitivityatom frequency inaccuracy instability to α Aα merit |Aα| to meq

at 1000 s σ [1016] QCD

Sr+ 445 THz 3 times 10minus17 (p) 5 times 10minus17 (p) 042 08 (p) ndashYb+ (1) 688 THz 3 times 10minus16 (d) 3 times 10minus16 (d) 088 029 (d) ndash

3 times 10minus17 (p) 1 (p)Yb+ (2) 642 THz lt 1 times 10minus17 (p) 1 times 10minus17 (p) minus53 53 (p) ndashSr 435 THz 1 times 10minus16 (d) 1 times 10minus16 (d) 009 009 (d) ndash

lt 1 times 10minus17 (p) 09 (p)Yb 519 THz 7 times 10minus16 (d) 1 times 10minus16 (d) 031 03 (d) ndash

lt 1 times 10minus17 (p) 1 times 10minus17 (p) 31 (p)Cs 91 GHz 2 times 10minus16 (d) 1 times 10minus15 (d) 275 027 (d) 1Yb+ 126 GHz 1 times 10minus15 (p) 1 times 10minus15 (p) 34 034 (p) 1

A relevant figure of merit is the ratio of |AX | and fractional frequency instability σ on the perigeevisibility timescale The Cs clock is the only high performance clock with adequate sensitivity tothe particle masses me mq The indicated Cs clock performance was achieved using a microwavelocal oscillator based on a cryogenic oscillator On EGE the latter is to be replaced by an opticallocal oscillator-based device (MOLO) with similar short-term performance but without the needto operate at cryogenic temperature The clock instability at 10000 s is typically similar to theindicated inaccuracy(p) projected (d) demonstrated (1) quadropole (2) octupole

Exp Astron

way to achieve this is by having at least two clocks on board The on-boardclock-clock comparison provides performance evaluation independent of theground-satellite comparison unit Once the satellite clock performances areestablished the link performance can be established as well

The clocks are similar in basic structure but differ in the detailed implemen-tation They consist of

a source of atomsa preparation subunit which cools the atoms to near-standstill (millikelvintemperature or below)a trapping subunit that confines the atom(s) to a tiny volumean interrogation subunita clock control electronics package

The clocks share a common central unit the optical-microwave local oscil-lator which serves as a flywheel oscillator with outstanding stability on shorttimescales (01ndash10 s) for both clocks It consists of a laser stabilized to an ultra-low loss optical resonator made of ultra-low-expansion glass A microwavesignal is derived from the laser by phase-locking a frequency comb to the laserwave and using a multiple of the repetition rate

While the subunits of any optical clock have a similar technological basis(optics electrooptics acoustooptics thermal control vacuum systems) theydiffer in their specifications Depending on the choice of the atom species de-livery confinement and interrogation methods clocks of different perfor-mance result A number of laboratory optical clock demonstrators have beendeveloped worldwide This ensures availability of the know-how for the indus-trial implementation of the flight models

The first clock is a microwave clock based on the interrogation of slowlymoving ensembles of cold Cs atoms It is essentially based on the PHARAOclock [5 6] developed to engineering model level by CNES in the frame-work of the ESA ISS mission ACES In EGE this clock is upgraded usingthe microwave-optical local oscillator This improves significantly the clockperformance The specification is a relative instability (Allan deviation) of3 times 10minus14 (τ s)minus12 ie at the integration time τ = 1000 s it drops to 1 times 10minus15and to 3 times 10minus16 at 10000 s While this specification is at least a factor 3 lessstringent than for the optical clocks the significantly lower cost of obtaining aflight model and the sensitivity of its frequency to the particle masses makes ita suitable choice

The optical clock baseline instrument is a single-ion optical clock Amongvarious suitable ion species (Hg+ Al+ Sr+ Yb+) [44ndash47] the Yb+ ion hasbeen selected Its breadboard performance is established at the instability levelof 3 times 10minus16 over 1000 s it exhibits a large sensitivity coefficient Aα hassmall sensitivity to magnetic fields and exhibits significant potential for furtherimprovements (use of the octupole transition with sim6 times larger Aα andlower overall instability [48] see below) A trapped Yb+ ion is also a veryrobust system it has been demonstrated that a single Yb+ ion can remaintrapped uninterruptedly for many months Finally Yb+ is the only ion with

Exp Astron

which optical clocks demonstrators have been operated in not a single but twomajor metrology laboratories

Another attractive optical clock is based on ensembles of cold neutral Sror Yb atoms trapped in an optical lattice produced by laser beams This clocktype has a lower instability than a single-ion clock thanks to the large numberof atoms (sim105) used Its development is in progress in several laboratoriesin the world (in part supported by space agencies) and has to date reached6 times 10minus17 instability at 5000 s and 1 times 10minus16 uncertainty [49 50] with potentialfor even higher performance However it is more complex and costly thanthe Yb+ ion clock since the cold atom preparation and trapping subunitsinclude more components require higher laser powers and the respectiveoperating procedures are more complex It is therefore not considered for thepresent mission designed to be developed in the very near future with limitedresources but is an interesting candidate for missions without such constraintsIts estimated physical parameters are indicated in Table 2 for comparison

42 Main instruments

421 Microwave clock

The PHARAO cold atom clock is one of the two atomic clocks of the spacemission ACES managed by ESA It is proposed here to include a secondversion of this instrument as its costperformance ratio is very attractive forthe EGE scientific objectives Its concept is very similar to ground basedatomic fountains but with the major difference of zero-g operation Atomsslowly launched in free flight cross two microwave fields tuned to the transitionbetween the two hyperfine levels of the cesium ground state The interrogationmethod based on two separate oscillating fields (Ramsey scheme) allowsthe detection of an atomic line whose width is inversely proportional to thetransit time between the two microwave cavities The resonant microwavefield at 9192631770 GHz (SI definition of the second) is synthesized startingfrom an ultrastable quartz oscillator (USO) in the FCDP phaselocked to themicrowave output of the microwave-optical local oscillator and stabilized tothe clock line using the error signal generated by the cesium resonator In thisway the intrinsic qualities of the cesium hyperfine transition both in terms ofaccuracy and frequency stability are transferred to the USO In a microgravityenvironment the velocity of atoms along the ballistic trajectories is constantand can be changed continuously over almost two orders of magnitude (5ndash500 cms) Differently from atomic fountain clocks presently operated onground very long interaction times (up to few seconds) will be possible whilekeeping reasonable the size of the instrument

The instrument consists of 4 main elements a laser source which providesthe light required to cool and detect the cesium atoms (Fig 5 left) a frequencychain which generates the 92 GHz microwave signal a cesium vacuum tubewhere the interaction between the microwave field and the atoms occurs

Exp Astron

Table 2 Resources and technology readiness level (TRL) of the EGE spacecraft

Unit Volume Mass Power TRL Comment[l] [kg] [W]

Ion clock 100 70 60Atom source 4Atom preparation 6Atom trap 4Atom interrogation 6Clock control 6

Microwave clock 200 91 110 8 TRL by end 2008Local oscillator

Stabilized laser 18 15 23 6Frequency comb 20 16 38 6

Microwave link 18 18 55 8FCDP 7 8 12 8GPS receiver 2 5 7 9CCR 2 2 0 9Payload sum 367 225 305Thermal control 14 17 30 8OBDH 20 24 38 9Star tracker 1 1 8 9Radiation shield 10 40 0 9Support structure 170 204 0 9Power incl solar arr 50 58 10 9Propulsion 21 25 20 9AOCS 36 43 82 9TMTC 8 10 15 9Satellite bus sum 330 422 203 Flexbus conceptPayload plus bus incl 20 margin 8364 7764 6096Propellant 40Launch adapter 100SC wet masspower incl Margins 9164Lattice clock 135 105 152 Alternative clock

Atom source 4Atom preparation 6Atom trap 4Atom interrogation 6Clock control 6

Also included are the resources for an optical lattice clock based on neutral Strontium atoms

(Fig 5 right) and a computer The engineering model of PHARAO is fullyassembled at CNES Toulouse and undergoes functional and performancetests until the end of 2008 The operating temperature range is 10ndash33 andnon-operating temperatures are minus45 +60 The cesium tube consists in aUHV chamber pumped by getters and a 3 ls ion pump a cesium reservoira microwave cavity coils to provide a uniform magnetic field and three layersof magnetic shields Vacuum windows with fiber-optics collimators enable thetransmission of the cooling and detection light onto the cesium atoms

In autonomous mode PHARAO uses the USO as interrogation oscillatorand will provide a clock signal with fractional frequency stability below1 times 10minus13 (τs) minus12 and inaccuracy near 1 times 10minus16 When using the microwave-

Exp Astron

Fig 5 Left The PHARAO laser source engineering model with cover removed Dimensionsare 530 times 350 times 150 mm3 and the mass is 200 kg Ten polarization maintaining optical fibers(yellow) guide the laser beams to the cesium tube All diode lasers are mounted on Peltier coolersfor temperature regulation within 2 mK (courtesy EADS SODERN) Right The PHARAOengineering model cesium tube without the two external magnetic shields The volume is 990 times336 times 444 mm3 and the total mass is 44 kg

optical local oscillator PHARAO will provide a reduced frequency instabilityof 3 times 10minus14 (τs) minus12 This corresponds to 1 times 10minus15 at 1000 s and 3 times 10minus16

at 104 s

422 Ion optical clock

The 171Yb+ ion clock shown in Figs 6 and 7 provides an atomic frequencyreference based on a single 171Yb+ ion confined within an RF electric end-captrap (similar to a Paul trap) within an ultra high vacuum (UHV) chamber andlaser cooled to sim1 mK close to the Doppler cooling limit As a result the ionrsquosfirst order Doppler motion is completely removed and the ion experienceslittle collisional perturbation from its environment Under these conditionsthe narrow linewidth 2S12 (F = 0 mF = 0) minus 2 D32 (F = 2 mF = 0) quadrupole

Fig 6 Left Simplified level scheme of 171Yb+ showing relevant cooling (blue) auxiliary (red)quadrupole clock (green) and octupole clock (dark blue) transitions δνnat is the natural linewidthof the clock transition Middle close-up of Paul-type RF trap (PTB Braunschweig) the ion istrapped in the hole (diameter 14 mm) Right Complete ion trap with vacuum flange height 10 cm(NPL Teddington)

Exp Astron

935 nm D32-state repumper laser

638 nm F72-state repumper laser

739 nm extended cavity diode laser

AOM

PMT

Ion trap UHV chamber

vacuum pumps

Fibre-free space interface

Mu-metal shielding

EC

T

B

Magnetic field

Temperature

Ion fluorescence Processor

Clock laser frequency step and control

4355 nm

Frequengy doubling II

3695 nm cooling wave

Low-drift cavity

Low-drift cavity 171Yb+ 435 nm ion clock unit

876 nm clock laser wave from MOLO

Frequengy doubling

Fig 7 Simplified schematic of the single-ion optical clock unit AOM acousto-optic frequencyshifter PMT photomultiplier tube EC endcap T temperature stabilization system B magneticfield

clock transition at 4355 nm can be interrogated by spectrally narrow and stableclock laser light A diagram of the simplified energy level scheme for 171Yb+showing the relevant cooling auxiliary and clock transitions is given in Fig 6All these wavelengths are provided by diode lasers operated in fundamentalor frequency doubled mode The clock laser light is obtained from the 871 nmwave provided by the MOLO unit (see Section 423) after frequency-doublingin single pass by a KNbO3 nonlinear-optical crystal to 4355 nm The clocktransition spectral profile is observed by stepping the clock laser frequency bymeans of an acousto-optic frequency shifter (AOM in Fig 7) and recordingthe statistics of the quantum jumps in cooling laser induced fluorescence asa function of this frequency Under clock operation the AOM repeatedlysteps back and forward between the half-intensity points on the transitionprofile monitoring the signal imbalance between these points Any detectedimbalance is servo-corrected to zero by feedback to the AOM

In detail the optical clock architecture will comprise (Fig 7)

bull an RF end-cap trap for ionising and confining a single 171Yb+ ion withinan ultra-high vacuum chamber pumped by a small ion pump and non-evaporable getter pump The trap is driven by means of an ac voltage ofa few hundred volts at a drive frequency of sim10 MHz A small oven withseveral mg of 171Yb isotope is heated to provide a low flux of Yb atomsfrom which single ions can be ionised by electron bombardment from ahot wire filament within the trap potential well Low voltage dc is applied

Exp Astron

to additional electrodes to position the ion precisely at trap potentialcentre in order to minimise the ionrsquos micromotion at the drive frequencyThis needs automatic periodic monitoring and correction especially in theperiod after re-loading an ion into the trap External magnetic field coilsin three orthogonal axes allow the nulling of external fields and settingof a fixed field of sim1 μT The trap and coils are surrounded by mu-metalshielding to minimise external field changes during the orbit A level oftemperature control of 1 K of the region surrounding the trap is requiredfor maintaining the blackbody-induced frequency shift uncertainty withtemperature at the 10minus17 level

bull a laser platform to provide frequency doubling of an extended cavity739 nm diode laser for Doppler cooling of the ion on the 2S12(F = 1) minus2 P12 (F = 0) dipole transition at 3695 nm Light from an amplified diodelaser device frequency-doubled in a periodically poled LiTaO3 crystal insingle-pass will be used Typical laser powers used for driving the coolingtransition below saturation are about 2 μW for a beam waist in the trapof about 50 μm By modulation of the injection current of the 739 nmdiode laser at 147 GHz a sideband is generated that excites the 2S12 (F =0) rarr2 P12 (F = 1) ldquorepumperrdquo transition in order to avoid optical pump-ing between the ground hyperfine states

bull The platform also houses the auxiliary lasers at 935 and 638 nm The935 nm diode provides repumping of the ion from the 2 D32(F = 1)metastable level after occasional branching decays to this level during thecooling sequence The 638 nm diode allows fast recovery of the ion fromthe very-long-lived 2F72 metastable state after very occasional collisionaldecay to that state Currently extended cavity lasers are forseen butdistributed-feedback lasers may be available in the near future

bull a high NA lens imaging system and photomultiplier detection system torecord the statistics of 369 nm fluorescence quantum jumps as a functionof clock laser frequency step

bull a fibre system to deliver the various cooling auxiliary and clock light fromsource to trap making use of achromatic doublets where necessary at thefibre-free space interface for launching into the trap

bull a monitoring and control processor which provides the primary coolingand clock laser pulse magnetic field and detection sequencing to observeand lock to the ion clock transition frequency The processor also monitorsfrequency and amplitude data necessary to determine normal laser andion operational conditions and initiate resetting and recovery algorithmswhere necessary and laser unit failure The actual magnetic field presenton the atomic volume is determined and actively stabilized so that aninaccuracy at the level of 1 times 10minus17 can be achieved

bull a redundancy level of at least two units for cooling clock and repumperlasers plus similar for frequency doubling crystals All redundancy unitsfor each wavelength will be fibre multiplexed as standard allowing redun-dant unit activation on determination of prior unit failure

Exp Astron

For an optimized quantum-projection-noise-limited performance of thesingle-ion clock Ramsey interrogation with a cycle time equal to the lifetimeof the metastable D32 level of 50 ms will be performed leading to an Allandeviation of about 27 times 10minus15 (τ s)minus12 This is sim45 times 10minus17 at 1 h satisfyingrequired stabilities for the science objectives Current laboratory results are astability of 1 times 10minus15 at 30 s averaging down to 3 times 10minus16 at 1000 s (PTBGermany) Also comparisons between two independent traps showed agree-ment at the level of several parts in 10minus16 over a series of eight measurementruns

The 171Yb+ ion clock offers a second option namely the 2S12ndash2 F72 octu-pole transition at 476 nm (see Table 1 and Fig 6) [48] It has an extremelylong upper state lifetime low electric quadrupole and second order Zeemansystematic shifts and therefore is a candidate for an even higher performanceclock if a clock laser system with lower instability than the above is providedApart from the clock laser it shares the laser system of the quadrupoleclock Thus with moderate additional resources the Yb+ ion apparatus couldprovide two optical clocks in a single package and specifically perform directfrequency comparisons between the quadrupole and the octupole transitionwith a large sensitivity Aα

The ion clock will operate under microprocessor control providing thecontrol algorithms for the integrated pulse sequencing for cooling repumpingclock interrogation magnetic field switching cooling fluorescence monitoringThe controller will also provide error flags for critical processes and systemreset and correction where necessary This will include ion re-loading andmicromotion reduction algorithms Automatic monitoring of laser power andspectral quality with signal re-optimisation or laser failure determination isrequired Redundant laser units will be pre-aligned and multiplexed into thefibre delivery to the trap so that redundant units can be readily activated in thecase of diode laser failure

423 Microwave-optical local oscillator (MOLO)

This oscillator provides two ultrastable frequency outputs one in the opticalregion the other in the microwave region the two frequencies being coher-ently related The MOLO is composed of three subunits the clock laser thereference cavity and the frequency comb see Fig 8

The clock laser subunit delivers the light for the excitation of the opticalclock transition It is generated by an extended-cavity diode laser at a wave-length of 871 nm Most of the output wave of this laser is sent to the ionclock The laser subunit containing two lasers for redundancy will require3 kg 3 l 15 W The frequency instability of the clock laser must be superiorto the minimum clock instability 27 times 10minus15 (τ s)minus12 for τ up to the timeconstant (sim10 s) of the servo system that locks the laser to the ionrsquos resonancesignal In order to achieve this the laser frequency is stabilized to the reference

Exp Astron

871 nm clock laser

EO

Detector and servo

AOM Femtosecond

frequency comb

To optical clock 10 GHz To FCDP

Microwave-optical local oscillator

ULE cavity

Fig 8 Microwave-optical local oscillator (MOLO) consisting of clock laser reference ULE cavityand femtosecond frequency comb (OFC) Radiofrequency inputs and outputs are not shown EOelectro-optic modulator ULE ultra-low expansion AOM frequency shifter

cavity where the PoundndashDreverndashHall method is used About 10 μW of thelaser output are used for this

The system will exhibit sim04 Hz (sim1 times 10minus15 relative) linewidth and afrequency drift lt01 Hzs The cavity consists of two high-reflectivity mirrorsoptically contacted to a cylinder of diameter 70 mm length 150 mm made ofULE (a glass exhibiting ultra low thermal expansion coefficient) with a cavityfinesse of about 300000 Residual vibrations of the satellite (arms lt 1 times 10minus6 g)should not be a limiting factor to the desired linewidth for safety marginan optimized cavity support can reduce the sensitivity to accelerations Foracoustic and thermal isolation the cavity resides inside an aluminium vacuumchamber equipped with a small (3 ls) ion-getter pump Two stages of polishedaluminium shields are implemented around the cavity which are activelytemperature stabilized so that the cavity is precisely kept at the zero crossingtemperature of the ULE thermal expansion coefficient Heat application andremoval is by thermoelectric elements between the shield and the vacuumchamber Additionally the entire vacuum chamber is temperature-stabilized

During transportation and launch the cavity will be rigidly fixed insidethe vacuum chamber by pressing on it from the sides by piezo-mechanicalactuators allowing accelerations of several g After bringing the satellite onthe orbit the actuators will be released

The optical setup for frequency stabilization will be a classic Pound-Drever-Hall scheme in reflection using an electro-optic modulator (EO) Couplingof the clock laser radiation to the cavity is via single-mode polarization main-taining optical fiber The fiber in- and out-couplers are mounted on miniaturepiezo-motor driven multiaxis translation stages Microprocessor control ofthe translation stages performs a laser-to-fiber incoupling stabilization andlaser-to-cavity mode-matching optimization (positioning inaccuracy lt10 μmalignment inaccuracy lt10 μrad)

The frequency stability of the clock laser will be limited by thermal 1f noisein the mirrors with an fractional Allan deviation slightly below 1 times 10minus15 from1 s to a few 10 s Resources including control electronics are 15 l 12 kg 8 W

Exp Astron

A femtosecond fiber laser based optical frequency comb generator (OFCFig 8) will be used for the clock frequency comparisons and to obtain aspectrally ultrapure microwave local oscillator output for the operation ofPHARAO for the microwave link and for the GNSS receiver The fem-tosecond laser is based on erbium-doped fiber pumped by fiber-coupled laserdiodes It emits 100 fs pulses at a repetition rate of 250 MHz with an averagepower of 100 mW The emission spectrum is centered around 1560 nm Spec-tral broadening to more than one octave is obtained via nonlinear interactionin a microstructure fibre so that the range from about 1050 to 2100 nm iscovered

Second harmonic generation of a portion of the broadened spectrum allowsobtaining the required light for the measurement of the clock laser frequencyat 871 nm The OFC will be referenced to the clock laser frequency (afterintroduction a controllable shift via the frequency shifter AOM) by makingthe carrier envelope offset frequency vanish so that the repetition rate fr isa high-order subharmonic of the optical clock reference frequency f1 = m1 frThe link between the optical and the microwave domain is then simply thedistribution of the repetition rate frequency or more conveniently a high(order 40) harmonic of the comb repetition rate 10 GHz

It has been demonstrated by several groups that frequency comparison andfrequency division with an OFC can be performed to very high accuracy andstability beyond the limitations imposed by the optical clocks themselves Forexample a fiber laser OFC with an instability (Allan deviation) of 6 times 10minus17

in 1 s has been demonstrated [51] OFCs based on fiber technology possessgood long-term reliability and high power efficiency The projection for a spacequalified OFC based on current industrial terrestrial technology is 16 kg mass20 l volume and 38 W power requirement

424 Frequency comparison and distribution package (FCDP)

This package distributes the ultra-stable 10 GHz reference frequency outputof the MOLO to the MWL performs frequency comparisons between MOLOand PHARAO and generates a variety of different reference signals for theindividual instruments including the 250 MHz reference clock to the OFCAll signals are coherent to the MOLO In the FCDP a USO is locked to theMOLO signal by a digital PLL A divider chain generates timing pulses (1 pps1000 pps etc) and keeps a continuous local time scale From the digital phasecomparison data the phase noise density spectrum between the USO and theMOLO is available by telemetry (TM) to assess the overall frequency purityin-flight

As the FCDP distributes the full performance reference from the MOLOdirectly to the MWL there is minimum impact of the FCDP on overall clockcomparison performance The USO-phase lock loop uses a similar designas for ACES-FCDP The digital PLL with linear phase detector allows theUSO to be slightly offset from its reference which is selectable by groundcommand This results in an accurate on-board time scale despite any offset

Exp Astron

of the MOLO reference frequency which can be synchronised to UTC Thefrequency synthesis chain re-uses modules developed for ACES-FCDP andtheir frequency inaccuracy Expected relative uncertainty (Allan deviation)of the frequency comparison between two signals provided to the FCDP is47 times 10minus17 at 1000 s 95 times 10minus18 at 10000 s

During non-availability of the MOLO ie during system start-up andcontingency conditions the USO runs as flywheel (ie it is unlocked withinstability lt15 times 10minus13 from 1 to 100 s) It will still provide useable butless accurate references to the instruments including the reference to OFCCommon view ground-to-ground time- and frequency comparison and supplyof the remaining instruments remain possible at the USO provided accuracy

Mass volume and power consumption are 8 kg 7 l 12 W Controldata rateis 10 kBday 50 kBday (FCDP alone)

425 Microwave link (MWL)

The microwave link is a two-way triple-frequency microwave link (Fig 9) Itis designed to perform accurate frequency comparison as well as time-transfer

Ku-Band Up-linkPower Tx 5 WCarrier 13475 GHzPN-Code 100 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

S-Band Down-linkPower Tx 1 WCarrier 2248 MHzPN-Code 1 MChips1pps 1 time marker sData 5 kBits

Ku-Band Down-linkPower Tx 1 WCarrier 1470333 GHzPN-Code 100 MChips1pps 1 time marker sData 5 kBits

EGE SC

S

Ku

KaKa-Band Up-linkPower Tx 5 WCarrier 345 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

Ka-Band Down-LinkPower Tx 1 WCarrier 32 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 5 kBits

GS Antenna diameterEGE terminal 1 mACES terminal 06 m

Fig 9 EGE microwave two-way link concept and characteristics See text for a description Thelink will be able to perform frequency comparisons with up to ten ground stations simultaneously(only one active comparison is shown) pps pulse per second Tx transmitter SC spacecraft

Exp Astron

between on-board and on-ground clocks Two-way operation allows for com-pensation of the effects due to spacecraft location and its motion relative toground Its two bi-directional links operate in Ka-band (32345 GHz) andKu-band (147135 GHz) The high carrier frequency of the up- and down-links allows for a noticeable reduction of the ionospheric delay A third down-link channel is in S-band (22 GHz) and is used to determine the ionospheretotal electron content MWL uses a combination of carrier phase and codephase data to measure with high precision the absolute ionospheric delayas well as ionospheric delay variations during a ground contact The useof three frequency bands allows to resolve third order ionospheric termsKey modulation parameters are Ka band Tone 250 MHz Ku band PN100 MChips S-band 1 MChips

The EGE-MWL concept is a considerably improved version of the ACES-MWL Microwave carrier generation is directly from the 10 GHz microwaveoutput available from the MOLO and having exceptionally low close-in phasenoise without intermediate frequency dividers or other complex synthesiserstages commonly used in RF transponders The high frequency carriers arekey to achieve the desired frequency comparison accuracy

The Ka-band carrier is generated by tripling the FCDP-generated signalThe Ku-band is subsequently generated by division by two using a verystable regenerative divider Finally the assigned radio frequency is obtainedby adding a low frequency signal using an image reject mixer A mixedanalog-digital receiver design will enhance its performance by use of a flexibledigital signal processing and digital control loops The number of simultaneousground contacts is increased from four (as in the ACES mission) to ten or moreusing the digital concept while the number of physical receiver channels perband is reduced from four to two one for operation to ground and one for in-flight calibration The concept allows adaptation of receiver loop parametersto optimise for the different portions of a highly elliptic orbit while reducingsystematic uncertainties due to high Doppler and Doppler rate Real-timedata exchange to ground terminals is by additional data modulation on all RFsignals The spacecraft carries two different small antennae for perigee and forapogee with gains of 8 dBi 20 dBi beamwidths 100 (full earth coverage) 14respectively power is 1 W for each

The MWL end-to-end performance between the attached clocks includingspace- and ground terminals is predicted to be a relative Allan deviation of1 times 10minus16 at 1000 s (perigee) and 1 times 10minus17 at 10000 s (apogee) Thesevalues are based on conservative implementation losses and other instrumentallosses assumed as 6 dB at apogee and 10 dB at perigee including additionallosses due to atmosphere An attitude knowledge of 1 is required

The MWL link will also support precise orbitographyndashcomplementary tothe GNSS receiver Generation of ranging data is simultaneous to two-waytime transfer and is available under all weather conditions Orbit determina-tion accuracy is expected to be better than 1 cm well matching EGE clockrequirements The value is expected based on demonstrated results fromother missions (METEOR ERS-2 TOPEXPOSEIDON) Laser ranging to

Exp Astron

the CCR array is used a few times during the mission to calibrate the MWLranging

In case of non-availability of the ultrapure signal from MOLO an additional10 GHz output is available from the FCDP to support ground-to-ground clockcomparison in common-view mode even after the nominal mission periodwhen the neither the clocks nor MOLO may be available any longer Basedon the PRARE experimentrsquos 12 years of operation this can be assumed forthe FCDPMWL combination as well The potential extended operational pe-riod makes the proposed configuration particularly attractive to metrologicallaboratories that rely on availability of long-lasting data in support of thegeneration of UTC

The resources of MWL are as follows mass 18 kg volume 18 l power55 W TC 2 kBday TM 120 MBday

426 Auxiliary units

The star camera assembly on the EGE spacecraft will provide real-timeattitude data primarily for the orientation of the spacecraft (with accuracy of1 in real-time) and the proper reduction of the GNSSSLR data

A GNSS receiver is used for the precise orbit determination time andfrequency transfer coarse positioning (lt1 m) for real time use by the Attitudeand Orbit Control as well as the back-up for time tagging of all payloaddata For the time and frequency transfer the receiver is nominally using theonboard frequency reference The receiver assembly includes three specialantennas By the year 2015 more than 100 GNSS satellites are expected tobe available for precise orbit determination permitting orbit inaccuracy of 1ndash2 cm RMS for altitudes up to approx 10000 km At apogee a combinationof GNSS signals and MWL ranging will be used with expected inaccuracyof 10 cm

5 Basic spacecraft specifications

The mission foresees a single spacecraft with a mission duration limited toapprox 3 years for cost reasons Attitude control along three axes withless than 1 error on a timescale of 1000 s is required The control can beperformed with cold gas thrusters magnetotorquers or reaction wheels Theselection will be done on the basis of the vibrations induced by these systems(desired level arms lt 10minus6 g)

Cold gas thrusters or monopropellant thrusters will be required to maintainthe orbit parameters during the mission by small orbit correction manoeuvresand will also guarantee a proper tuning of the orbit parameters after the releasefrom the launcher The ground segment (telemetry control and data transfer)could be via non-ESA stations user-operated to save costs

Further resources are indicated in Table 2

Exp Astron

The satellite bus design drivers are nadir-pointing microwave antennasthermal stability for the payload instruments no strong magnetic sources lowmicrovibrations harsh radiation environment due to crossing of the radiationbelts once per orbit power demand by science payload (with the possibility toreduce power demand by stand-by of some components during specific missionphases eg eclipse) mass limit due to required high altitude with low costlaunch

The satellite bus design can be based on established platforms such asFLEXBUS or PRIMA used on several past missions Their design philosophydiffers one having body-mounted solar arrays the other wings The latter canprovide larger amounts of power The vibration level induced by the panelswhen they are fixed and the duration of solar array drive periods per orbitwill have to be considered in the selection The FLEXBUS concept was usedfor a mission already requiring a low level of residual accelerations and theplatform is optimised for this parameter

For most of the satellite components one can consider using compo-nents and electronic units from navigation or communication satellites whichtypically cruise at altitudes between 15000 and 25000 km altitude LISAPathfinder which will cross the radiation belts during spiralling to theLagrange point will face similar problems concerning radiation as EGE as wellas XMM which is in a Molniya-type orbit with similar radiation characteristics

The instruments will therefore be located in the satellite structure such asto realise ionizing radiation shielding by the spacecraft structure and non-sensitive equipment to the maximum extent possible the need for additionalradiation shielding (including trapped proton radiation) for especially sensitiveparts of the instruments must be determined by ground tests It should bementioned that an orbit with higher perigee and less radiation dose is also anoption as it leads to a tolerable reduction of the science goals

The required plusmn3C temperature stability at the interface to the payloadenclosure is feasible The thermal control will employ the usual means (anenclosure with high thermal inertia heaters and a suitable radiator) Thelocation of the radiator will need a special study given the non-standard orbit

Magnetic field present during the mission can affect the properties of theonboard clock Using SPENVIS the magnetic field has been calculated alongthe spacecraft orbit the field intensity presents maxima around 02 G anda minimum value of the order of 10minus3 G As discussed above this is not alimitation to the instrument performance

6 Discussion and conclusion

The accuracy of the measurements depends crucially on the stability andaccuracy of the clocks While the performance of ground clocks (even thoseeventually deployed in new ground stations dedicated to EGE) will likelyimprove significantly from the present level the space clock design will beldquofrozenrdquo approx 10 years before data acquisition Thus their performance

Exp Astron

level will be significantly below the future ground station clock performanceAs specification we have chosen a performance level that has been demon-strated with laboratory breadboards recently We see no fundamental reasonthat this laboratory performance cannot also be achieved in a space ion opticalclock The experience with the PHARAO cold atom clock has shown thatan engineering model of a laser-based cold atom clock has a performancesimilar to that of a laboratory prototype The perturbation of the ion clock fre-quency due to magnetic fields along the orbit can be modelled and suppressedwith shielding and the residual effects measured and compensated for Themoderate temperature stability (plusmn 3C) to be provided in the experimentalbay can be achieved with thermal shields Accurate temperature controlof particular units is done by active stabilization of the units themselves

One important aspect is radiation which may affect all optical and trapcomponents as well as the lasers and reference cavity The radiation envi-ronment on the EGE orbit is significantly harsher than the near-earth ISSorbit on which ACES will operate The radiation effects must therefore becharacterized comprehensively with appropriate tests and then the radiationshield for the EGE instruments designed

In summary the Einstein Gravity Explorer mission (EGE) is a satellitemission aimed at a precise measurement of the properties of space-time usingatomic clocks By testing one of the most fundamental predictions of EinsteinrsquosTheory of General Relativity the gravitational redshift and thereby searchingfor quantum effects in gravity it explores one of the most important frontiers infundamental physics The EGE mission reaches beyond fundamental physicsproviding an important spin-off in geophysics Furthermore the technologicalaspects of the mission are ground-breaking for future time-keeping and ref-erence frame systems in space and thus are also expected to be of wide useto science and technology in general The payload of the mission has beenconceived as a compromise between cost and performance needing to fit thecost envelope of an ESA medium-class mission1

Acknowledgement We thank S Cesare for helpful information

References

1 Will CM The confrontation between general relativity and experiment Living Rev Relativ9 3 (2006)

2 Ni W-T Empirical foundations of the relativistic gravity Int J Mod Phys D 14 901 (2005)doi101142S0218271805007139

1Finally we point out that a spacecraft with a significantly reduced and thus lower-cost payload (noclocks only MOLO FCDP MWL) on a circular orbit of appropriate height could be used for along-duration earth science mission dedicated specifically to the measurement of the geopotential

Exp Astron

example extensive studies of atomic and molecular spectra of gas clouds in thedistant Universe are currently undertaken to search for a difference of theseparameters compared to todayrsquos values (see eg [7ndash11] and references therein)Also some differences in Big Bang nucleosynthesis data and calculations canbe naturally explained by a variation of fundamental constants [12 13]

The EGE mission is thus a probe of possible space-dependence of fun-damental constants The search may help to point the way toward a unifiedtheory of all forces reveal hypothetical extra dimensions in our Universe or theexistence of many sub-Universes with different physics chemistry and biologyIndeed theories unifying all forces applied to cosmology suggest space andtime dependence of the coupling constants [14ndash16] Another argument forspatial variation of fundamental constants comes from the anthropic principleThere must be a fine tuning of fundamental constants which allows humans andany life to appear Life in its present form is not consistent with fundamentalconstants whose values only slightly differ from the actual ones This finetuning can be naturally explained by the spatial variation of the fundamentalconstants we appeared in the area of the Universe where the values of thefundamental constants are consistent with our existence

In the Standard Model there are three main fundamental dimensionless pa-rameters determining the structure and energy levels of stable matter (atomsmolecules) the electromagnetic fine structure constant α and the two ratiosof electron mass me and light quark mass mq to the strong QCD energy scaleQCD Xeq = meqQCD The fundamental masses me and mq are proportionalto the vacuum Higgs field which determines the electroweak unification scaleTherefore the constants meqQCD can also be viewed as the ratio of the weakenergy scale to the strong energy scale

How can a space-time variation of the fundamental constants and a depen-dence on the gravitational potential occur Light scalar fields very naturallyappear in modern cosmological models affecting α and meqQCD Oneof these scalar fields is the famous ldquodark energyrdquo which causes acceleratedexpansion of the Universe Another hypothetical scalar field is the dilatonwhich appears in string theories together with the graviton Cosmologicalvariation of these scalar fields should occur because of drastic changes ofthe matter composition of the Universe During the Big Bang nucleosyn-thesis the Universe was dominated by radiation then by cold dark matterand now by dark energy Changes of the cosmic scalar field ϕ0(t) lead tothe variation of the fundamental constants X(ϕ0) Massive bodies (galaxiesstars planets) can also affect physical constants They have a large scalarcharge S = sp Z + se Z + sn N where Z is the number of protons and elec-trons and N is the number of neutrons sp se sn are the scalar charges ofthe proton electron and neutron respectively There is also a contributionof the nuclear binding energy (scalar charge of virtual mesons mediatingnuclear forces)

The scalar charge produces a Coulomb-like scalar field ϕs = SR where Ris the distance from the charge The total scalar field is ϕ = ϕ0 + ϕs therefore

Exp Astron

we may have a variation of the fundamental constants inversely proportionalto the distance from massive bodies

X (ϕ) = X (ϕ0 + ϕs) = X (ϕ0) + δX (R)

δX (R) = (dX

dϕ

)S

R

where a nonzero δX violates LPI and a small ϕs was assumedThe gravitational potential U is essentially proportional to the number of

baryons Z + N and inversely proportional to the distance R U = -GMR

Therefore the change of the fundamental constants near massive bodies canbe written

δX

X = K (X i) δ(Ui

c2

) (2)

where i refers to a particular composition of the mass M and where thecoefficients K(X i ) = minus(dXdϕ) c2SiGMi are not universal but are expectedto be different for eg the Sun and the Earth K(X Earth) = K(X Sun)Indeed the Sun consists mostly of hydrogen therefore the Sunrsquos scalar chargeis SSun Z (sp + se) The Earth contains heavier elements where the numberof neutrons exceeds the number of protons N 11 Z Therefore the scalarcharge of the Earth SEarth Z (11 sn + sp + se) and is sensitive to the neutronscalar charge (and also to the contribution of the nuclear binding) in contrastto the Sun

The most sensitive experiments suitable for a search of a space-time de-pendence in local or near-local experiments (such as satellite experiments) areclockndashclock comparisons In one type of terrestrial local experiment dissimilarclocks located nearby are compared for a duration spanning at least one yearDuring such periods the earth moves periodically closer and further awayfrom the sun so that the solar gravitational potential on Earth is modulatedThis enables a search for a dependence of the fundamental constants on thesunrsquos scalar charge [17 18] These experiments are able to set limits to K(XSun) for X = α meqQCD at the level of 10minus6 [12 13]

The proposed EGE satellite experiments are complementary to terrestrialexperiments allowing to set limits to K(X Earth) and combining with theresults on K(X Sun) from terrestrial comparisons to set limits on the depen-dence of the fundamental constants on the neutron scalar charge sn

In summary when the frequencies of two identical clocks (of nominalfrequency νa) that are located at different positions x1 x2 are comparedby measuring their frequency ratio at some position xrsquo the result may bewritten as

va2(xprime)

va1

(xprime) = 1 +

[1 +

sum

XAXa K (X i)

](Ui (x2) minus Ui (x1))

c2 (3)

The sensitivity factor AXa = X (d(ln νa)dX) is the sensitivity of the clockfrequency to a particular constant X and can be calculated using atomic and

Exp Astron

nuclear theory (see eg [12 13]) The measurement of this ratio which issensitive to K represents a test of General Relativityrsquos redshift

When the frequencies of two co-located but dissimilar clocks a b (ofnominal frequencies νa0 and νb 0) are compared in-situ at two locations x1 x2

(LPI test) the result is

[va1 (x1)

vb 1 (x1)

] [va2(x2)

vb 2 (x2)

] = 1 +[X

(AXa minus AXb

)K (X i)

](Ui (x2) minus Ui (x1))

c2 (4)

The result of the EGE mission may be given as a limits for (or nonzerovalues of) K(X Earth) for the three fundamental constants X = α meQCDand mqQCD

23 Measurement approaches for the gravitational frequency shift

The gravitational potential experienced by the ground and satellite clocksincludes the contributions from the earth and the sun (the contribution fromthe moon is an order smaller and may be neglected for the purposes of thediscussion but will be taken into account in data analysis)

Nonzero gravitational frequency shift signals (Eq 3) can be obtained inthree ways Method (1) below is the most sensitive and represents a strongreason to employ an optical clock which on short averaging times is ableto provide a higher frequency stability than a microwave clock Method (2)and (3) are less accurate but are complementary in terms of measurementapproach Finally method (3) addresses the solar gravitational frequencyshift while (1) and (2) measure the terrestrial effect The performance of theinstruments assumed in the following is shown in Fig 2

(1) Repeated measurements of satellite clock frequency variation betweenapogee and perigeeIn this measurement approach the difference (or ratio) between earthndashsatellite clock frequency ratio at perigee and at apogee is consideredThis difference is measured at each perigee and apogee passage andaveraged over the mission duration Systematic shifts (if not correlatedwith orbital motion) are expected to average out leading to a gain insensitivity of up to n12 where n = number of orbits This mode usesthe stability properties of the satellite clocks on the timescale fromperigee to apogee rather than their accuracy Assuming a specificationof le 3 times 10minus16 satellite optical clock fractional frequency instability forintegration times between 1000 and 25000 s n sim= 1000 comparisons(ie most orbits of the 3-year mission) and a perigee-apogee potentialdifference (U(apogee) minus U(perigee))c2 = Uc2 sim= 4 times 10minus10 allows a

Exp Astron

Fig 2 Performance of clocks and links on EGE Curve ldquooptical clocks EGErdquo is the specificationCurve ldquooptical clocks EGE (goal)rdquo refers to expected near-future performance of an opticalclock based on neutral atoms ldquoUpgraded PHARAOrdquo refers to the PHARAO microwave clockusing the high-quality microwave-optical local oscillator The microwave link (MWL) performanceexhibits a discontinuity since common-view contact time of ground stations with EGE is limited toseveral h The different slopes arise because on short time scales only white phase noise is relevantwhile on long time scales flicker noise also contributes The reproducibility is taken into account

measurement of the redshift at 25 ppb accuracy The ground clocks towhich the satellite clocks are compared to at perigee and at apogee willbe located in different ground stations The inaccuracy of these groundclocks must therefore be sufficiently smaller than 1 times 10minus17 in order notto contribute to the 25 ppb error Current state-of-the art inaccuracy is2 times 10minus17 and is likely to decrease further

(2) Absolute comparison between ground and satellite clocks with the satel-lite at apogeeHere the terrestrial gravitational potential difference is maximumU(apogee) minus U(perigee)c2 sim= 65 times 10minus10 Assuming a space optical clockinaccuracy of 1 times 10minus16 and a ground clock inaccuracy a factor 3lower (current state-of-the-art) allows a measurement of the redshift at150 ppb This measurement takes advantage of the repetitive nature ofthe orbit only to a small extent and relies on the accuracy of the opticalclocks in particular of the space clocks It is thus complementary to (1)albeit less accurateIn (1) and (2) the dominant contribution to the redshift is from Earthrsquosgravitational potential The contribution from the solar potential isstrongly reduced due to the fact that satellite and earth are an interactingsystem in free-fall toward the Sun [19] and can be modeled

(3) Comparisons between terrestrial clocksThe frequency comparison link to the EGE satellite allows terrestrialclock comparisons by a common-view method whereby the frequency

Exp Astron

signals from two distant ground clocks are sent to the satellite andcompared there The frequency difference between two identical groundclocks is determined if systematic instrumental errors are negligible bythe difference in gravitational potential at the two clock locations Thispotential difference includes a terrestrial and a solar contributionThe solar contribution varies according to the orientation of the clockpair with respect to the Sun with a 1 day-period For two clocks at asurface distance of sim6000 km and suitable orientation with respect tothe sun the normalized modulation amplitude is sim= 4 times 10minus13 This is thesame as the earth potential difference for 4 km height difference It ispossible that ground clocks will have reached an inaccuracy at the levelof 1 times 10minus18 at the time of the mission in 2015 Using the microwave linka comparison of ground clocks over sim3 h (typical common-view duration)with a resolution of 1 times 10minus17 can be performed limited by the instabilityof the link Repeating the clock comparison at each common-view win-dow (n = 1000 comparisons) will allow averaging and a measurementof the solar gravitational shift with fractional inaccuracy of 1 ppm Incomparison the best current results for the solar gravitational frequencyshift are at the few-percent level [20 21] A moderate averaging out ofground clock systematic shifts is assumed as well as modeling of the earthgravitational potential variations on the daily timescale at the appropriatelevel

(4) Comparisons of on-board clocksThe two on-board clocks are continuously compared these data areanalyzed for violation of LPI (Eq 4) Since the measurement is a nulltest the required accuracy of the gravitational potential at the satellitelocation is low and easily satisfied

24 Test of local position invariance and search for neutron scalar charge

The test of LPI is performed in conjunction with the measurement of thegravitational frequency shift

With approach 23 (1) the optical clock will be able to measure K (α Earth)with accuracy 30 ppb (spec)

Combining the results of the microwave and optical clocks will determineK(meQCD Earth) + εK(mqQCD Earth) with accuracy 85 ppb (spec)where ε asymp minus01

Simultaneously the on-board comparison 23 (4) yields the combination19 K(α Earth) +K(meQCD Earth) + εK(mqQCD Earth) with accuracy85 ppb (spec)

Note that these measurements are fully complementary to current andfuture terrestrial clock-clock comparisons which are sensitive to K(X Sun)and can be combined with the latter to determine or set limits to the neutronscalar charge

Exp Astron

25 Theoretical analysis of frequency comparison signals

The on-board clock-clock comparison signals can be analyzed straightfor-wardly in terms of a violation of LPI since signal propagation effects areabsent

The determination of the gravitational frequency dilation via ground-satellite clock comparisons is more complex (Eq 1 is strongly simplified) andrequires an accurate treatment of all relevant relativistic effects Gravitationin the solar system is described in the barycentric (BCRS) and geocentric(GCRS) non-rotating celestial reference systems by means of post-Newtoniansolutions of Einsteinrsquos equations for the metric tensor in harmonic gaugescodified in the conventions IAU2000 [22] In these conventions the relativisticstructure of Newton potential (order 1c2) and gravitomagnetic (order 1c3)

potentials are given For the Newton potential the multipolar expansion ofthe geopotential determined by gravity mapping is used The post-Newtonianeffects of the metric (described in terms of dimensionless parameters β and γ )

can be included Orbits of satellites are evaluated in the GCRS at the levelof a few cm by means of the Einstein-Infeld-Hoffmann equations at the order1c2 [23] using the relativistic Newton potential Instead the trajectories of theground clocks in the GCRS are evaluated from their positions fixed onthe Earth crust (ITRS International Terrestrial Reference System) by usingthe non-relativistic IERS2003 conventions [24 25]

For the propagation of light rays (here radio signals) between the satelliteand the ground stations one uses the null geodesics of the post-Newtoniansolution in the GCRS The timefrequency transfer properties have beentheoretically evaluated for an axisymmetric rotating body [26ndash29] at the order1c4 (ie with Newton and gravitomagnetic potentials developed in multipolarexpansions) These results have been developed for use in the ACES missionThe relative frequency dilation can be expressed as

ν

ν= 1 + 1

cL1 +

4sum

n = 2

1

cn(Ln + Gn)

where Ln describes special-relativistic Doppler effects and Gn the general-relativistic effects These quantities are time-dependent due to the orbitalmotion For concreteness we report the orders of the terms The kinematicaleffects are |L1c| lt 1 times 10minus5 |L2c2| lt 2 times 10minus10 |L3c3| lt 3 times 10minus15 |L4c4| lt

7 times 10minus20 The Newtonian contributions are G2 = G(M)2 + G(J2)

2 + G(J4)2 + G(J6)

2(M is the mass monopole Jn are the lowest multipoles) with estimates|G(M)

2 c2| lt 3 times 10minus11 |G(J2)2 c2| lt 2 times 10minus13 |G(J4)

2 c2| lt 3 times 10minus16 |G(J6)2 c2| lt

1 times 10minus16 For G3 = G(M)3 + G(J2)

3 the estimates are |G(M)3 c3| lt 2 times 10minus14

|G(J2)3 c3| lt 1 times 10minus18 Finally the G4 = G(M)

4 + G(S)4 contributions (G(S)

4 is thelowest gravitomagnetic effect) are of order |G(M)

4 c4| lt 1 times 10minus19 |G(S)4 c4| lt

1 times 10minus22 and thus not relevant for EGE

Exp Astron

The available theory is fairly complete for the data analysis in mode 23 (1)described below For modes 23(2 3) the theory will need to be generalizedto include the effect of the gravitational potential at the location of the groundclock(s) (including time-varying effects such as the tides) and of the solar andlunar potentials

The uncertainties of the various contributions above are minimized by ausing precise EGE orbit data obtained by satellite tracking and by usingcurrent and future accurate earth gravity information Required orbit positionknowledge is at 1 cm level near perigee and sim50 cm near apogee reachablealready today with the proposed orbitography approach described below Thevalidity of the special-relativistic Doppler shift has been and will continue to beindependently verified with increased accuracy by spectroscopy of relativisticatomic ion beams [30] Potential violations of this aspect of Lorentz Invari-ance are expected to be sufficiently bounded so as not to affect the signalinterpretation

26 Local Lorentz invariance tests

There has been an explosion of interest in Local Lorentz invariance in re-cent years [31 32] with numerous astronomical studies and high-precisionexperiments applied to search for violations of Lorentz invariance Signifi-cant developments have also occurred on the theoretical side and a theory(standard model extension SME [33]) has been worked out that provides aunified framework for describing and analyzing Lorentz Invariance violationsin a variety of systems Two aspects of Lorentz Invariance can be tested withEGE

261 Independence of the speed of light from the laboratory velocity

A possible dependence of the speed of light on the speed v of the laboratorycan be modeled according to the MansourindashSexl test theory [34] by

c (v) = c0(1 + Bv2

c2

)

Here v 377 kms is the velocity with respect to the cosmic microwavebackground the cosmologically preferred frame and B is a combinationof parameters describing deviations from the usual Lorentz transformationformulas B = 0 if Lorentz Invariance holds For a terrestrial experiment therotation of the Earth around its axis modulates v with a 300 ms amplitude Adependence c(v) can be searched for by measuring the frequency differencebetween a resonance frequency of a highly stable optical cavity (cavity fre-quencies are proportional to c) and an optical clock (Kennedy-Thorndike-typeexperiment [35]) and determining the modulation of the frequency differencecorrelated with the modulation of v The advantages of a space experimentare the high orbital velocity and strongly reduced cavity deformation thanksto microgravity [36 37] On the proposed elliptic orbit v varies between+4 and minus4 kms over approximately 1 h This variation is 13 times larger

Exp Astron

than for the case of a terrestrial experiment The shorter time scale is alsoadvantageous since the drift of the cavity is more predictable In the EGEmission the reference cavity of the clock laser is used The large numberof orbits improves the statistics An improvement of the accuracy of B by afactor 20 compared to the best current terrestrial results is realistic The dataanalysis can also be performed in the framework of the SME theory takinginto account complementary bounds obtained from terrestrial experimentsand astrophysical observations

262 Independence of Zeeman splitting frequency on the directionof the magnetic field

One class of tests of Lorentz Invariance consists in measuring the frequencysplitting between two levels of a quantum system induced by a static magneticfield as a function of orientation of the field direction with respect to thestars (HughesndashDrever-type experiments [31]) This class addresses the so-called matter sector and is complementary to combined-photonmatter-sectortests such as the above The most precise experiments are performed withatomic clocks eg masers or cold atom clocks [38] The EGE instruments allowsuch experiments since magnetic fields are applied continuously in the clockor repeatedly for calibration Compared to terrestrial experiments the highorbital velocity of EGE could lead to an improvement by a factor sim20

27 Application to geophysics

Using GPS coordinates of a point on the Earth can be obtained with aninaccuracy well below 1 cm in a well defined international terrestrial referencesystem The coordinates are purely geometrical and do not contain any gravityinformation A local gravitational potential UB is determined with respect toa reference gravitational potential U A by ldquolevellingrdquo the surface of the Earthie sequentially measuring gravity g and height differences every ca 50 m

UB minus U A = minusBint

A

g (r) middot d n

where n is the difference vector between subsequent locations The disadvan-tage of measuring height differences by the levelling method is a random walkeffect of accumulated errors

In the classical definition a geoid is defined as the particular equipotentialsurface nearest to mean sea level In these terms the geoid serves as areference surface for measuring the height and also to define a referencefor the gravitational potential Such a vertical reference frame historicallywas realized for a country or several countries by determining the mean ofsea level observations at tide gauge stations taken over long period of timeHowever modern satellite altimetry missions such as TOPEXPOSEIDON orENVISAT show that departure of the mean sea level from an equipotential

Exp Astron

surface may reach up to several meters on the global scale see eg [39]Therefore height systems based on different tide gauge stations may differin the realisation of the geoid and may differ with respect to the reference byseveral meters

The best gravity field models obtained from satellite data (eg missionGRACE) have reached a precision of about 1 cm over sim250 km half wave-length [40] Satellite data also reveals changes in time [41] However for typicalEarth topography with height variations of eg 1000 m over 30 km horizontaldistance one may expect variations in the geoid of about 80 cm Such ahigh-spatial-frequency signal in geoid variations cannot be detected by spacegravity missions and requires a combination of satellite and terrestrial gravitymeasurements like gravity anomalies deflections of vertical and GPSlevellingpoints The best combined global gravity field models are provided up to de-gree and order 360 in terms of spherical harmonic representation correspond-ing to a half-wavelength of about 55 km However in combining the satelliteand terrestrial gravity field measurements the problem of terrestrial data givenin different height systems between continents and different countries remainsand has to be tied to satellite measurements

The geodetic scientific community is currently establishing a Global Geo-detic Observing System (GGOS) [42] Its objectives are the early detectionof natural hazards the measurement of temporal changes of land ice andocean surfaces as well as the monitoring of mass transport processes in theEarth system Global change processes are small and therefore difficult toquantify Therefore the required precision relative to the Earthrsquos dimensionis 1 ppb GGOS will be established by the combination of geodetic spacetechniques (GPS Laser-Ranging very large baseline interferometry) andrealized by a very large number of terrestrial and space-borne observatoriesThe purely geometric terrestrial 3D coordinate system is in good shape andfully operational It has to be complemented however with a globally uniformheight system of similar precision The current precision level of regionalheight systems in terms of gravity potential differences is in the order of1 m2s2 (10 cm) with inconsistencies between these various systems up toseveral 10 m2s2 (several meters) The actual requirement in the context ofGGOS is 01 m2s2 (1 cm) with the need of a permanent ie dynamical controlThis requirement of high precision height control comes from the need tounderstand on a global scale processes such as sea level change global andcoastal dynamics of ocean circulation ice melting glacial isostatic adjustmentand land subsidence as well the interaction of these processes Only by meansof monitoring in terms of gravity potential changes at the above level ofprecision the change of ocean level can be understood as a global phenomenonand purely geometric height changes be complemented by information aboutthe associated density or mass changes

The geoid can also be defined in a relativistic way [43] as the surface whereaccurate clocks run with the same rate and where the surface is nearest tomean sea level The relation between the differences in the clock frequenciesand the gravitational potential is given in simplified form by Eq 1 At present

Exp Astron

there is no operational way to compare frequencies of the already availableoptical clocks on the global scale at the same level as their accuracy wouldallow With EGE and by then improved ground optical clocks it will becomepossible to obtain gravitational potential differences on a global scale by com-paring frequencies This would be a significant new dimension to gravity fielddetermination since such observables are given on the global scale and providein situ local gravity information at the same time EGE will allow clock-basedgravitational potential mapping with the same payload instruments used forthe fundamental physics experiments The on-board clocks are actually notrequired for this purpose MOLO MWL and FCDP suffice Thus EGE willallow establishing a global reference frame for the Earth gravitational fieldwith accuracy in the order of a few cm in terms of the geoid heights Thisassumes that by the time EGE is operational (mobile) ground clocks withfractional inaccuracy at the level of 10minus18 are available the measurementsare limited by MWL noise and thus require a long integration times to matchthe expected ground clock performance Such a reference frame will serve asreference for all gravity field modeling and clocks will be used to define theinternational atomic time scale and the Earth gravitational potential at thesame time

In summary with optical clocks geodesy will undergo a second revolutionafter geometry now being measured by clock-based GNSS systems also physi-cal heights and potential will be measured by (mobile) optical clocks using thegravitational redshift effect

28 Application to frequency standards and terrestrial fundamentalphysics studies

As has been described repeating ground clock comparisons and integratingover several months will provide global ground clock frequency comparisonsat the level of 1 times 10minus18 more than two orders of magnitude below the currentGPS and two-way satellite time and frequency transfer methods In the timeand frequency community the availability of such a global high performancemicrowave link will accelerate the process leading to a redefinition of the SIsecond based on clocks operating in the optical domain

The global character of the ground clock comparisons made possible byEGE can also contribute to the search for a time-variation of the fundamentalconstants

The comparisons do not require availability of the on-board clocks The riskof this type of measurements is therefore reduced Moreover the lifetime ofmany subunits required for this measurement (MWL and FCDP see below)can exceed 10 years (from previous experience with similar devices) andthus offers a motivation to increase the mission duration beyond the nominal3 years

Further applications to metrology are listed under the secondary goals inSection 21 above

Exp Astron

3 Mission profile

31 Orbit

The orbit must satisfy the following requirements

Primary 1 Large gravitational potential difference between apogee andperigee

2 sufficient contact time with at least two ground stations atperigee

Secondary Simultaneous visibility of satellite from distant (inter-continental) ground stations for several hours

To satisfy the primary objective the orbit must be highly elliptic The perigeeshould be low A very low perigee means a short contact time which decreasesthe accuracy of the frequency measurement A compromise is a perigeealtitude of sim2500 km For a sufficiently high apogee this yields a U of 60of the theoretical maximum between the earth surface and infinity

The baseline scenario is an orbit of approx 12 h period with the followingproperties

Apogee Altitude 37856 km (minus 10 deg latitude one over Atlantic andPacific resp)

Perigee Altitude 2500 km (+ 10 deg latitude one over Guatemala andMalaysia resp)

Inclination 634 argument of perigee 170RAAN 25 True Anomaly 220Apogee drift rate 70 kmday perigee drift rate sim 14 kmday

True anomaly and right ascension of the ascending node (RAAN) wereoptimized to minimize orbital precession Figure 3 shows the ground track andthe ground coverage at one of the perigees Common-view contact time forstations in Europe is about 6 h while between a station in Europe and in theUSA it is about 35 h Frequency comparison windows between a given groundstation and EGE repeat every 24 h

A final selection of the orbit requires an in-depth study of the costs of thepropellant and maneuvers as well as of the setting up of new ground stations

32 Ground segment

The mission requires two separate ground segments the science groundsegment and the standard satellite control segment which supports telemetryand telecommand (TMTC)

Exp Astron

Fig 3 Baseline EGE orbit One (out of two) apogee and one (out of two) perigee ground pointsare shown The ground coverage at one of the perigees is shown with circles at elevation anglesteps of 10 starting from 10

321 Science ground segment

As described above one part of the mission is to perform frequency com-parisons between ground-to-space and ground-to-ground (common-view andnon-common view) The use of a two-way microwave link (MWL) systemhas the advantage that a precise orbit or the precise relative velocity of thesatellite must not be known [4] The orbit determination requirements fora two-way system are significantly relaxed compared to the requirements tomodel the relativistic effects on the clocks The frequency comparison resultsare available both on ground and on the satellite

For ground-to-satellite clock comparisons five ground stations arefavourable two capable of viewing perigee three for viewing apogee Thisallows for the classic three-clock configuration where the performance of eachclock can be obtained individually The perigee stations will be observing thesatellite for sim25 min every perigee passage for the duration of the missionThe apogee stations will observe for sim3 h every apogee passage Of coursea larger number of stations as well as longer observation time are usefulfor robustness redundancy and orbit determination The stations could beinstalled on a temporary basis for the duration of the mission

The stations must have access to optical clocks of performance at least ashigh as todayrsquos state-of-the-art ground clocks but preferably higher Theywill include an optical frequency comb by means of which the optical clockfrequency is transformed to a radio frequency for comparison with the EGEclocks High-accuracy fiber-optic link technology already existing today allowsdistances of sim100 km between the ground station and the correspondingoptical clock allowing a certain freedom in the choice of the clock locationWith further increase of the fiber-optic link range currently being investigated

Exp Astron

the area over which the ground station can be positioned relative to the opticalclock will increase This option is of importance for the implementation of ageophysics measurement campaign for which a large number of clock locationsis of interest

The above ground stations will be complemented by some 15 additionalground terminals provided by the world-wide laboratories equipped with highperformance primary clocks at the time of the mission These will participate inorbit determination to 1 cm accuracy along the full orbit which can be achievedwith non-optical atomic clocks The EGE MWL is fully compatible to plannedACES ground terminals which might reduce the number of required stations

With a larger number of transportable optical clock ground stations de-ployed and moved across the continents a geophysics campaign could beimplemented to map the gravitational potential

Ground station equipment MWL uses dedicated small transportable termi-nals which can be easily installed at any location of interest Each terminalconsists of a steerable dish antenna of 1 m diameter with pointing accuracy of01 using a transmitted power of 5 W per band The high directivity antenna(beam aperture Ka-band 07 Ku-band 14 S-band 20) avoids multi-pathfrom nearby objects Antenna noise temperatures are Ka-band 450 K Ku-band 340 K Operation is at elevations above 10 and measurements areperformed above 20 For ranging purposes terminal location is determined byGPS surveying Each ground station computes its own ionospheric correctionsbased on a triple-band receiver The terminal has a built-in delay monitorwhich allows calibrated ranging during several months once its bias has beendetermined by laser ranging Its M and C interface is via local area networkto either a userrsquos computer or directly to the science network control centreby remote MampC The stationrsquos operation is fully automatic and unattendedSchedule and orbital data are received via LAN once per day Signal acquisi-tion is fully autonomous Data transmission is after each pass The terminal hasan uplink telecommand capability of 1 kBits and a downlink data capacity ofup to 10 kBitss For the local user there is a visual interface providing quick-look data from real-time comparison to the spacecraft clocks to verify theirgood health The final full performance product is calculated at the sciencedata centre

Local metrological data (temperature pressure humidity) are recorded tocorrect for tropospheric delay of the ranging data The main advantage ofMWL terminals compared to laser based systems is their all-weather capabil-ity although microwaves suffer higher propagation errors in the ionosphereand troposphere

The EGE ground terminal will profit from the ERS-2 mission PRARE(precise range and range rate equipment) [45] experiments whose terminalshad similar design and have been operated under extreme climatic conditionsincl Antarctica and high temperature regions from existing designs fromthe space segment (mission ACES) and no further special developments aredeemed necessary Cost per EGE master station is modest

Exp Astron

Satellite tracking via laser ranging Laser ranging is alternative to MWL rang-ing and also serves for absolute calibration of the latter For this purpose anon-board corner cube reflector (CCR) is provided Ranging will be done withthe International Laser Ranging Service Ideally the satellite will be rangedat every perigee passage and every apogee passage simultaneously with thefrequency comparison procedures Todayrsquos ranging precision is sufficient tosatisfy the science goals For apogee observations in the northern hemispherea large number of laser ranging stations is available The number of stationscapable of ranging at or near perigee in the southern hemisphere is six andthus is sufficient even considering unfavourable weather conditions at someof these The MWL ground stations are able to range even under thoseconditions

322 Control ground segment

Average data transfer rate is estimated at 300 MBday (approx 40 kBs onaverage) Data will be stored on board (2 GB capacity required) and will betransferred to a receiving station on the Earth once a day

XPLC

incl Data StorageIF to TMTC Link

HK amp TMTC

to ground

Opt Clock

AOM

MOLO

PHARAO

FCDP

10 GHz

100 MHz

250 MHz

GNSS Rx

MWLTampF Link

uplink Data

UTC(k)

TampF Space vs Ground

TM from GS

Tim

e S

cale

Orbit GNSS Time UTC

TampF Space vs Ground

MampC Time Scale

MampC

MampC

MampC

MampC

Mamp

C

PHARAO vs Opt Clock

PHARAO vs Opt Clock

10 GHz amp Epoch

10 MHz

amp Epoch

Epoch

10 MHz

amp Epoch

Opt Link

Fig 4 Schematic of the payload showing subunits and their signals The results of the frequencycomparison between the two on-board clocks (optical clock and microwave clock PHARAO) areobtained from a combination of the frequency values applied to the frequency shifters AOM (seeFigs 7 and 8) and those output from the frequency control and distribution package (FCDP)as indicated by the arrows ldquoPHARAO vs Opt Clockrdquo The satellite clock to ground clockcomparison results are output from the MWL XPLC is the on-board computer and is involvedin generation of signals for instrument control The optical link shown dashed is a possible add-onproviding an alternative to the MWL GS ground stations MampC monitoring and control HKhousekeeping TampF time and frequency

Exp Astron

4 Main payload instruments

41 Overview

The satellite payload consists of a pair of cold-atom clocks a microwave-optical local oscillator (MOLO) an on-board frequency distribution unit(FCDP) a satellite-to-ground clock comparison unit (MWL) and the onboardcomputercontrol system (XPLC) see Fig 4 Auxiliary units are the posi-tion determination GPSGalileo receiver (GNSS Rx) attitude monitor (startracker not shown) and the corner cube reflector array (CCR not shown)

The science goals require clocks of very high frequency stability on thetimescales of half the orbital period (6 h) as well as perigee passage (sim1000 s)see Fig 2 Only clocks based on cold atoms are suited for the science goalsTable 1 reports values for a subset of atomic clocks that have been investigatednot including clocks that require cryogenic operation or are too complex foruse in an medium-class mission The choice of a clock pair rather than a singleone fulfils several requirements it addresses more science goals improves thevalidity of some science results and it enables calibration accuracy evaluationand redundancy The validity of the science results is enhanced since theclocks differ not only in the employed atoms but also in terms of havingdifferent sensitivity to perturbations by electric and magnetic fields electronicserrors and misalignments A comparison of the results arising from each clockallows for an understanding of measurement errors Both the clocksrsquo and thefrequency comparison unitrsquos performances will be determined after launchand regularly verified during the course of the mission The only reliable

Table 1 Properties of different clock candidates for EGE

Clock Clock Fractional σ = Fractional Sensitivity Figure of Sensitivityatom frequency inaccuracy instability to α Aα merit |Aα| to meq

at 1000 s σ [1016] QCD

Sr+ 445 THz 3 times 10minus17 (p) 5 times 10minus17 (p) 042 08 (p) ndashYb+ (1) 688 THz 3 times 10minus16 (d) 3 times 10minus16 (d) 088 029 (d) ndash

3 times 10minus17 (p) 1 (p)Yb+ (2) 642 THz lt 1 times 10minus17 (p) 1 times 10minus17 (p) minus53 53 (p) ndashSr 435 THz 1 times 10minus16 (d) 1 times 10minus16 (d) 009 009 (d) ndash

lt 1 times 10minus17 (p) 09 (p)Yb 519 THz 7 times 10minus16 (d) 1 times 10minus16 (d) 031 03 (d) ndash

lt 1 times 10minus17 (p) 1 times 10minus17 (p) 31 (p)Cs 91 GHz 2 times 10minus16 (d) 1 times 10minus15 (d) 275 027 (d) 1Yb+ 126 GHz 1 times 10minus15 (p) 1 times 10minus15 (p) 34 034 (p) 1

A relevant figure of merit is the ratio of |AX | and fractional frequency instability σ on the perigeevisibility timescale The Cs clock is the only high performance clock with adequate sensitivity tothe particle masses me mq The indicated Cs clock performance was achieved using a microwavelocal oscillator based on a cryogenic oscillator On EGE the latter is to be replaced by an opticallocal oscillator-based device (MOLO) with similar short-term performance but without the needto operate at cryogenic temperature The clock instability at 10000 s is typically similar to theindicated inaccuracy(p) projected (d) demonstrated (1) quadropole (2) octupole

Exp Astron

way to achieve this is by having at least two clocks on board The on-boardclock-clock comparison provides performance evaluation independent of theground-satellite comparison unit Once the satellite clock performances areestablished the link performance can be established as well

The clocks are similar in basic structure but differ in the detailed implemen-tation They consist of

a source of atomsa preparation subunit which cools the atoms to near-standstill (millikelvintemperature or below)a trapping subunit that confines the atom(s) to a tiny volumean interrogation subunita clock control electronics package

The clocks share a common central unit the optical-microwave local oscil-lator which serves as a flywheel oscillator with outstanding stability on shorttimescales (01ndash10 s) for both clocks It consists of a laser stabilized to an ultra-low loss optical resonator made of ultra-low-expansion glass A microwavesignal is derived from the laser by phase-locking a frequency comb to the laserwave and using a multiple of the repetition rate

While the subunits of any optical clock have a similar technological basis(optics electrooptics acoustooptics thermal control vacuum systems) theydiffer in their specifications Depending on the choice of the atom species de-livery confinement and interrogation methods clocks of different perfor-mance result A number of laboratory optical clock demonstrators have beendeveloped worldwide This ensures availability of the know-how for the indus-trial implementation of the flight models

The first clock is a microwave clock based on the interrogation of slowlymoving ensembles of cold Cs atoms It is essentially based on the PHARAOclock [5 6] developed to engineering model level by CNES in the frame-work of the ESA ISS mission ACES In EGE this clock is upgraded usingthe microwave-optical local oscillator This improves significantly the clockperformance The specification is a relative instability (Allan deviation) of3 times 10minus14 (τ s)minus12 ie at the integration time τ = 1000 s it drops to 1 times 10minus15and to 3 times 10minus16 at 10000 s While this specification is at least a factor 3 lessstringent than for the optical clocks the significantly lower cost of obtaining aflight model and the sensitivity of its frequency to the particle masses makes ita suitable choice

The optical clock baseline instrument is a single-ion optical clock Amongvarious suitable ion species (Hg+ Al+ Sr+ Yb+) [44ndash47] the Yb+ ion hasbeen selected Its breadboard performance is established at the instability levelof 3 times 10minus16 over 1000 s it exhibits a large sensitivity coefficient Aα hassmall sensitivity to magnetic fields and exhibits significant potential for furtherimprovements (use of the octupole transition with sim6 times larger Aα andlower overall instability [48] see below) A trapped Yb+ ion is also a veryrobust system it has been demonstrated that a single Yb+ ion can remaintrapped uninterruptedly for many months Finally Yb+ is the only ion with

Exp Astron

which optical clocks demonstrators have been operated in not a single but twomajor metrology laboratories

Another attractive optical clock is based on ensembles of cold neutral Sror Yb atoms trapped in an optical lattice produced by laser beams This clocktype has a lower instability than a single-ion clock thanks to the large numberof atoms (sim105) used Its development is in progress in several laboratoriesin the world (in part supported by space agencies) and has to date reached6 times 10minus17 instability at 5000 s and 1 times 10minus16 uncertainty [49 50] with potentialfor even higher performance However it is more complex and costly thanthe Yb+ ion clock since the cold atom preparation and trapping subunitsinclude more components require higher laser powers and the respectiveoperating procedures are more complex It is therefore not considered for thepresent mission designed to be developed in the very near future with limitedresources but is an interesting candidate for missions without such constraintsIts estimated physical parameters are indicated in Table 2 for comparison

42 Main instruments

421 Microwave clock

The PHARAO cold atom clock is one of the two atomic clocks of the spacemission ACES managed by ESA It is proposed here to include a secondversion of this instrument as its costperformance ratio is very attractive forthe EGE scientific objectives Its concept is very similar to ground basedatomic fountains but with the major difference of zero-g operation Atomsslowly launched in free flight cross two microwave fields tuned to the transitionbetween the two hyperfine levels of the cesium ground state The interrogationmethod based on two separate oscillating fields (Ramsey scheme) allowsthe detection of an atomic line whose width is inversely proportional to thetransit time between the two microwave cavities The resonant microwavefield at 9192631770 GHz (SI definition of the second) is synthesized startingfrom an ultrastable quartz oscillator (USO) in the FCDP phaselocked to themicrowave output of the microwave-optical local oscillator and stabilized tothe clock line using the error signal generated by the cesium resonator In thisway the intrinsic qualities of the cesium hyperfine transition both in terms ofaccuracy and frequency stability are transferred to the USO In a microgravityenvironment the velocity of atoms along the ballistic trajectories is constantand can be changed continuously over almost two orders of magnitude (5ndash500 cms) Differently from atomic fountain clocks presently operated onground very long interaction times (up to few seconds) will be possible whilekeeping reasonable the size of the instrument

The instrument consists of 4 main elements a laser source which providesthe light required to cool and detect the cesium atoms (Fig 5 left) a frequencychain which generates the 92 GHz microwave signal a cesium vacuum tubewhere the interaction between the microwave field and the atoms occurs

Exp Astron

Table 2 Resources and technology readiness level (TRL) of the EGE spacecraft

Unit Volume Mass Power TRL Comment[l] [kg] [W]

Ion clock 100 70 60Atom source 4Atom preparation 6Atom trap 4Atom interrogation 6Clock control 6

Microwave clock 200 91 110 8 TRL by end 2008Local oscillator

Stabilized laser 18 15 23 6Frequency comb 20 16 38 6

Microwave link 18 18 55 8FCDP 7 8 12 8GPS receiver 2 5 7 9CCR 2 2 0 9Payload sum 367 225 305Thermal control 14 17 30 8OBDH 20 24 38 9Star tracker 1 1 8 9Radiation shield 10 40 0 9Support structure 170 204 0 9Power incl solar arr 50 58 10 9Propulsion 21 25 20 9AOCS 36 43 82 9TMTC 8 10 15 9Satellite bus sum 330 422 203 Flexbus conceptPayload plus bus incl 20 margin 8364 7764 6096Propellant 40Launch adapter 100SC wet masspower incl Margins 9164Lattice clock 135 105 152 Alternative clock

Atom source 4Atom preparation 6Atom trap 4Atom interrogation 6Clock control 6

Also included are the resources for an optical lattice clock based on neutral Strontium atoms

(Fig 5 right) and a computer The engineering model of PHARAO is fullyassembled at CNES Toulouse and undergoes functional and performancetests until the end of 2008 The operating temperature range is 10ndash33 andnon-operating temperatures are minus45 +60 The cesium tube consists in aUHV chamber pumped by getters and a 3 ls ion pump a cesium reservoira microwave cavity coils to provide a uniform magnetic field and three layersof magnetic shields Vacuum windows with fiber-optics collimators enable thetransmission of the cooling and detection light onto the cesium atoms

In autonomous mode PHARAO uses the USO as interrogation oscillatorand will provide a clock signal with fractional frequency stability below1 times 10minus13 (τs) minus12 and inaccuracy near 1 times 10minus16 When using the microwave-

Exp Astron

Fig 5 Left The PHARAO laser source engineering model with cover removed Dimensionsare 530 times 350 times 150 mm3 and the mass is 200 kg Ten polarization maintaining optical fibers(yellow) guide the laser beams to the cesium tube All diode lasers are mounted on Peltier coolersfor temperature regulation within 2 mK (courtesy EADS SODERN) Right The PHARAOengineering model cesium tube without the two external magnetic shields The volume is 990 times336 times 444 mm3 and the total mass is 44 kg

optical local oscillator PHARAO will provide a reduced frequency instabilityof 3 times 10minus14 (τs) minus12 This corresponds to 1 times 10minus15 at 1000 s and 3 times 10minus16

at 104 s

422 Ion optical clock

The 171Yb+ ion clock shown in Figs 6 and 7 provides an atomic frequencyreference based on a single 171Yb+ ion confined within an RF electric end-captrap (similar to a Paul trap) within an ultra high vacuum (UHV) chamber andlaser cooled to sim1 mK close to the Doppler cooling limit As a result the ionrsquosfirst order Doppler motion is completely removed and the ion experienceslittle collisional perturbation from its environment Under these conditionsthe narrow linewidth 2S12 (F = 0 mF = 0) minus 2 D32 (F = 2 mF = 0) quadrupole

Fig 6 Left Simplified level scheme of 171Yb+ showing relevant cooling (blue) auxiliary (red)quadrupole clock (green) and octupole clock (dark blue) transitions δνnat is the natural linewidthof the clock transition Middle close-up of Paul-type RF trap (PTB Braunschweig) the ion istrapped in the hole (diameter 14 mm) Right Complete ion trap with vacuum flange height 10 cm(NPL Teddington)

Exp Astron

935 nm D32-state repumper laser

638 nm F72-state repumper laser

739 nm extended cavity diode laser

AOM

PMT

Ion trap UHV chamber

vacuum pumps

Fibre-free space interface

Mu-metal shielding

EC

T

B

Magnetic field

Temperature

Ion fluorescence Processor

Clock laser frequency step and control

4355 nm

Frequengy doubling II

3695 nm cooling wave

Low-drift cavity

Low-drift cavity 171Yb+ 435 nm ion clock unit

876 nm clock laser wave from MOLO

Frequengy doubling

Fig 7 Simplified schematic of the single-ion optical clock unit AOM acousto-optic frequencyshifter PMT photomultiplier tube EC endcap T temperature stabilization system B magneticfield

clock transition at 4355 nm can be interrogated by spectrally narrow and stableclock laser light A diagram of the simplified energy level scheme for 171Yb+showing the relevant cooling auxiliary and clock transitions is given in Fig 6All these wavelengths are provided by diode lasers operated in fundamentalor frequency doubled mode The clock laser light is obtained from the 871 nmwave provided by the MOLO unit (see Section 423) after frequency-doublingin single pass by a KNbO3 nonlinear-optical crystal to 4355 nm The clocktransition spectral profile is observed by stepping the clock laser frequency bymeans of an acousto-optic frequency shifter (AOM in Fig 7) and recordingthe statistics of the quantum jumps in cooling laser induced fluorescence asa function of this frequency Under clock operation the AOM repeatedlysteps back and forward between the half-intensity points on the transitionprofile monitoring the signal imbalance between these points Any detectedimbalance is servo-corrected to zero by feedback to the AOM

In detail the optical clock architecture will comprise (Fig 7)

bull an RF end-cap trap for ionising and confining a single 171Yb+ ion withinan ultra-high vacuum chamber pumped by a small ion pump and non-evaporable getter pump The trap is driven by means of an ac voltage ofa few hundred volts at a drive frequency of sim10 MHz A small oven withseveral mg of 171Yb isotope is heated to provide a low flux of Yb atomsfrom which single ions can be ionised by electron bombardment from ahot wire filament within the trap potential well Low voltage dc is applied

Exp Astron

to additional electrodes to position the ion precisely at trap potentialcentre in order to minimise the ionrsquos micromotion at the drive frequencyThis needs automatic periodic monitoring and correction especially in theperiod after re-loading an ion into the trap External magnetic field coilsin three orthogonal axes allow the nulling of external fields and settingof a fixed field of sim1 μT The trap and coils are surrounded by mu-metalshielding to minimise external field changes during the orbit A level oftemperature control of 1 K of the region surrounding the trap is requiredfor maintaining the blackbody-induced frequency shift uncertainty withtemperature at the 10minus17 level

bull a laser platform to provide frequency doubling of an extended cavity739 nm diode laser for Doppler cooling of the ion on the 2S12(F = 1) minus2 P12 (F = 0) dipole transition at 3695 nm Light from an amplified diodelaser device frequency-doubled in a periodically poled LiTaO3 crystal insingle-pass will be used Typical laser powers used for driving the coolingtransition below saturation are about 2 μW for a beam waist in the trapof about 50 μm By modulation of the injection current of the 739 nmdiode laser at 147 GHz a sideband is generated that excites the 2S12 (F =0) rarr2 P12 (F = 1) ldquorepumperrdquo transition in order to avoid optical pump-ing between the ground hyperfine states

bull The platform also houses the auxiliary lasers at 935 and 638 nm The935 nm diode provides repumping of the ion from the 2 D32(F = 1)metastable level after occasional branching decays to this level during thecooling sequence The 638 nm diode allows fast recovery of the ion fromthe very-long-lived 2F72 metastable state after very occasional collisionaldecay to that state Currently extended cavity lasers are forseen butdistributed-feedback lasers may be available in the near future

bull a high NA lens imaging system and photomultiplier detection system torecord the statistics of 369 nm fluorescence quantum jumps as a functionof clock laser frequency step

bull a fibre system to deliver the various cooling auxiliary and clock light fromsource to trap making use of achromatic doublets where necessary at thefibre-free space interface for launching into the trap

bull a monitoring and control processor which provides the primary coolingand clock laser pulse magnetic field and detection sequencing to observeand lock to the ion clock transition frequency The processor also monitorsfrequency and amplitude data necessary to determine normal laser andion operational conditions and initiate resetting and recovery algorithmswhere necessary and laser unit failure The actual magnetic field presenton the atomic volume is determined and actively stabilized so that aninaccuracy at the level of 1 times 10minus17 can be achieved

bull a redundancy level of at least two units for cooling clock and repumperlasers plus similar for frequency doubling crystals All redundancy unitsfor each wavelength will be fibre multiplexed as standard allowing redun-dant unit activation on determination of prior unit failure

Exp Astron

For an optimized quantum-projection-noise-limited performance of thesingle-ion clock Ramsey interrogation with a cycle time equal to the lifetimeof the metastable D32 level of 50 ms will be performed leading to an Allandeviation of about 27 times 10minus15 (τ s)minus12 This is sim45 times 10minus17 at 1 h satisfyingrequired stabilities for the science objectives Current laboratory results are astability of 1 times 10minus15 at 30 s averaging down to 3 times 10minus16 at 1000 s (PTBGermany) Also comparisons between two independent traps showed agree-ment at the level of several parts in 10minus16 over a series of eight measurementruns

The 171Yb+ ion clock offers a second option namely the 2S12ndash2 F72 octu-pole transition at 476 nm (see Table 1 and Fig 6) [48] It has an extremelylong upper state lifetime low electric quadrupole and second order Zeemansystematic shifts and therefore is a candidate for an even higher performanceclock if a clock laser system with lower instability than the above is providedApart from the clock laser it shares the laser system of the quadrupoleclock Thus with moderate additional resources the Yb+ ion apparatus couldprovide two optical clocks in a single package and specifically perform directfrequency comparisons between the quadrupole and the octupole transitionwith a large sensitivity Aα

The ion clock will operate under microprocessor control providing thecontrol algorithms for the integrated pulse sequencing for cooling repumpingclock interrogation magnetic field switching cooling fluorescence monitoringThe controller will also provide error flags for critical processes and systemreset and correction where necessary This will include ion re-loading andmicromotion reduction algorithms Automatic monitoring of laser power andspectral quality with signal re-optimisation or laser failure determination isrequired Redundant laser units will be pre-aligned and multiplexed into thefibre delivery to the trap so that redundant units can be readily activated in thecase of diode laser failure

423 Microwave-optical local oscillator (MOLO)

This oscillator provides two ultrastable frequency outputs one in the opticalregion the other in the microwave region the two frequencies being coher-ently related The MOLO is composed of three subunits the clock laser thereference cavity and the frequency comb see Fig 8

The clock laser subunit delivers the light for the excitation of the opticalclock transition It is generated by an extended-cavity diode laser at a wave-length of 871 nm Most of the output wave of this laser is sent to the ionclock The laser subunit containing two lasers for redundancy will require3 kg 3 l 15 W The frequency instability of the clock laser must be superiorto the minimum clock instability 27 times 10minus15 (τ s)minus12 for τ up to the timeconstant (sim10 s) of the servo system that locks the laser to the ionrsquos resonancesignal In order to achieve this the laser frequency is stabilized to the reference

Exp Astron

871 nm clock laser

EO

Detector and servo

AOM Femtosecond

frequency comb

To optical clock 10 GHz To FCDP

Microwave-optical local oscillator

ULE cavity

Fig 8 Microwave-optical local oscillator (MOLO) consisting of clock laser reference ULE cavityand femtosecond frequency comb (OFC) Radiofrequency inputs and outputs are not shown EOelectro-optic modulator ULE ultra-low expansion AOM frequency shifter

cavity where the PoundndashDreverndashHall method is used About 10 μW of thelaser output are used for this

The system will exhibit sim04 Hz (sim1 times 10minus15 relative) linewidth and afrequency drift lt01 Hzs The cavity consists of two high-reflectivity mirrorsoptically contacted to a cylinder of diameter 70 mm length 150 mm made ofULE (a glass exhibiting ultra low thermal expansion coefficient) with a cavityfinesse of about 300000 Residual vibrations of the satellite (arms lt 1 times 10minus6 g)should not be a limiting factor to the desired linewidth for safety marginan optimized cavity support can reduce the sensitivity to accelerations Foracoustic and thermal isolation the cavity resides inside an aluminium vacuumchamber equipped with a small (3 ls) ion-getter pump Two stages of polishedaluminium shields are implemented around the cavity which are activelytemperature stabilized so that the cavity is precisely kept at the zero crossingtemperature of the ULE thermal expansion coefficient Heat application andremoval is by thermoelectric elements between the shield and the vacuumchamber Additionally the entire vacuum chamber is temperature-stabilized

During transportation and launch the cavity will be rigidly fixed insidethe vacuum chamber by pressing on it from the sides by piezo-mechanicalactuators allowing accelerations of several g After bringing the satellite onthe orbit the actuators will be released

The optical setup for frequency stabilization will be a classic Pound-Drever-Hall scheme in reflection using an electro-optic modulator (EO) Couplingof the clock laser radiation to the cavity is via single-mode polarization main-taining optical fiber The fiber in- and out-couplers are mounted on miniaturepiezo-motor driven multiaxis translation stages Microprocessor control ofthe translation stages performs a laser-to-fiber incoupling stabilization andlaser-to-cavity mode-matching optimization (positioning inaccuracy lt10 μmalignment inaccuracy lt10 μrad)

The frequency stability of the clock laser will be limited by thermal 1f noisein the mirrors with an fractional Allan deviation slightly below 1 times 10minus15 from1 s to a few 10 s Resources including control electronics are 15 l 12 kg 8 W

Exp Astron

A femtosecond fiber laser based optical frequency comb generator (OFCFig 8) will be used for the clock frequency comparisons and to obtain aspectrally ultrapure microwave local oscillator output for the operation ofPHARAO for the microwave link and for the GNSS receiver The fem-tosecond laser is based on erbium-doped fiber pumped by fiber-coupled laserdiodes It emits 100 fs pulses at a repetition rate of 250 MHz with an averagepower of 100 mW The emission spectrum is centered around 1560 nm Spec-tral broadening to more than one octave is obtained via nonlinear interactionin a microstructure fibre so that the range from about 1050 to 2100 nm iscovered

Second harmonic generation of a portion of the broadened spectrum allowsobtaining the required light for the measurement of the clock laser frequencyat 871 nm The OFC will be referenced to the clock laser frequency (afterintroduction a controllable shift via the frequency shifter AOM) by makingthe carrier envelope offset frequency vanish so that the repetition rate fr isa high-order subharmonic of the optical clock reference frequency f1 = m1 frThe link between the optical and the microwave domain is then simply thedistribution of the repetition rate frequency or more conveniently a high(order 40) harmonic of the comb repetition rate 10 GHz

It has been demonstrated by several groups that frequency comparison andfrequency division with an OFC can be performed to very high accuracy andstability beyond the limitations imposed by the optical clocks themselves Forexample a fiber laser OFC with an instability (Allan deviation) of 6 times 10minus17

in 1 s has been demonstrated [51] OFCs based on fiber technology possessgood long-term reliability and high power efficiency The projection for a spacequalified OFC based on current industrial terrestrial technology is 16 kg mass20 l volume and 38 W power requirement

424 Frequency comparison and distribution package (FCDP)

This package distributes the ultra-stable 10 GHz reference frequency outputof the MOLO to the MWL performs frequency comparisons between MOLOand PHARAO and generates a variety of different reference signals for theindividual instruments including the 250 MHz reference clock to the OFCAll signals are coherent to the MOLO In the FCDP a USO is locked to theMOLO signal by a digital PLL A divider chain generates timing pulses (1 pps1000 pps etc) and keeps a continuous local time scale From the digital phasecomparison data the phase noise density spectrum between the USO and theMOLO is available by telemetry (TM) to assess the overall frequency purityin-flight

As the FCDP distributes the full performance reference from the MOLOdirectly to the MWL there is minimum impact of the FCDP on overall clockcomparison performance The USO-phase lock loop uses a similar designas for ACES-FCDP The digital PLL with linear phase detector allows theUSO to be slightly offset from its reference which is selectable by groundcommand This results in an accurate on-board time scale despite any offset

Exp Astron

of the MOLO reference frequency which can be synchronised to UTC Thefrequency synthesis chain re-uses modules developed for ACES-FCDP andtheir frequency inaccuracy Expected relative uncertainty (Allan deviation)of the frequency comparison between two signals provided to the FCDP is47 times 10minus17 at 1000 s 95 times 10minus18 at 10000 s

During non-availability of the MOLO ie during system start-up andcontingency conditions the USO runs as flywheel (ie it is unlocked withinstability lt15 times 10minus13 from 1 to 100 s) It will still provide useable butless accurate references to the instruments including the reference to OFCCommon view ground-to-ground time- and frequency comparison and supplyof the remaining instruments remain possible at the USO provided accuracy

Mass volume and power consumption are 8 kg 7 l 12 W Controldata rateis 10 kBday 50 kBday (FCDP alone)

425 Microwave link (MWL)

The microwave link is a two-way triple-frequency microwave link (Fig 9) Itis designed to perform accurate frequency comparison as well as time-transfer

Ku-Band Up-linkPower Tx 5 WCarrier 13475 GHzPN-Code 100 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

S-Band Down-linkPower Tx 1 WCarrier 2248 MHzPN-Code 1 MChips1pps 1 time marker sData 5 kBits

Ku-Band Down-linkPower Tx 1 WCarrier 1470333 GHzPN-Code 100 MChips1pps 1 time marker sData 5 kBits

EGE SC

S

Ku

KaKa-Band Up-linkPower Tx 5 WCarrier 345 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

Ka-Band Down-LinkPower Tx 1 WCarrier 32 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 5 kBits

GS Antenna diameterEGE terminal 1 mACES terminal 06 m

Fig 9 EGE microwave two-way link concept and characteristics See text for a description Thelink will be able to perform frequency comparisons with up to ten ground stations simultaneously(only one active comparison is shown) pps pulse per second Tx transmitter SC spacecraft

Exp Astron

between on-board and on-ground clocks Two-way operation allows for com-pensation of the effects due to spacecraft location and its motion relative toground Its two bi-directional links operate in Ka-band (32345 GHz) andKu-band (147135 GHz) The high carrier frequency of the up- and down-links allows for a noticeable reduction of the ionospheric delay A third down-link channel is in S-band (22 GHz) and is used to determine the ionospheretotal electron content MWL uses a combination of carrier phase and codephase data to measure with high precision the absolute ionospheric delayas well as ionospheric delay variations during a ground contact The useof three frequency bands allows to resolve third order ionospheric termsKey modulation parameters are Ka band Tone 250 MHz Ku band PN100 MChips S-band 1 MChips

The EGE-MWL concept is a considerably improved version of the ACES-MWL Microwave carrier generation is directly from the 10 GHz microwaveoutput available from the MOLO and having exceptionally low close-in phasenoise without intermediate frequency dividers or other complex synthesiserstages commonly used in RF transponders The high frequency carriers arekey to achieve the desired frequency comparison accuracy

The Ka-band carrier is generated by tripling the FCDP-generated signalThe Ku-band is subsequently generated by division by two using a verystable regenerative divider Finally the assigned radio frequency is obtainedby adding a low frequency signal using an image reject mixer A mixedanalog-digital receiver design will enhance its performance by use of a flexibledigital signal processing and digital control loops The number of simultaneousground contacts is increased from four (as in the ACES mission) to ten or moreusing the digital concept while the number of physical receiver channels perband is reduced from four to two one for operation to ground and one for in-flight calibration The concept allows adaptation of receiver loop parametersto optimise for the different portions of a highly elliptic orbit while reducingsystematic uncertainties due to high Doppler and Doppler rate Real-timedata exchange to ground terminals is by additional data modulation on all RFsignals The spacecraft carries two different small antennae for perigee and forapogee with gains of 8 dBi 20 dBi beamwidths 100 (full earth coverage) 14respectively power is 1 W for each

The MWL end-to-end performance between the attached clocks includingspace- and ground terminals is predicted to be a relative Allan deviation of1 times 10minus16 at 1000 s (perigee) and 1 times 10minus17 at 10000 s (apogee) Thesevalues are based on conservative implementation losses and other instrumentallosses assumed as 6 dB at apogee and 10 dB at perigee including additionallosses due to atmosphere An attitude knowledge of 1 is required

The MWL link will also support precise orbitographyndashcomplementary tothe GNSS receiver Generation of ranging data is simultaneous to two-waytime transfer and is available under all weather conditions Orbit determina-tion accuracy is expected to be better than 1 cm well matching EGE clockrequirements The value is expected based on demonstrated results fromother missions (METEOR ERS-2 TOPEXPOSEIDON) Laser ranging to

Exp Astron

the CCR array is used a few times during the mission to calibrate the MWLranging

In case of non-availability of the ultrapure signal from MOLO an additional10 GHz output is available from the FCDP to support ground-to-ground clockcomparison in common-view mode even after the nominal mission periodwhen the neither the clocks nor MOLO may be available any longer Basedon the PRARE experimentrsquos 12 years of operation this can be assumed forthe FCDPMWL combination as well The potential extended operational pe-riod makes the proposed configuration particularly attractive to metrologicallaboratories that rely on availability of long-lasting data in support of thegeneration of UTC

The resources of MWL are as follows mass 18 kg volume 18 l power55 W TC 2 kBday TM 120 MBday

426 Auxiliary units

The star camera assembly on the EGE spacecraft will provide real-timeattitude data primarily for the orientation of the spacecraft (with accuracy of1 in real-time) and the proper reduction of the GNSSSLR data

A GNSS receiver is used for the precise orbit determination time andfrequency transfer coarse positioning (lt1 m) for real time use by the Attitudeand Orbit Control as well as the back-up for time tagging of all payloaddata For the time and frequency transfer the receiver is nominally using theonboard frequency reference The receiver assembly includes three specialantennas By the year 2015 more than 100 GNSS satellites are expected tobe available for precise orbit determination permitting orbit inaccuracy of 1ndash2 cm RMS for altitudes up to approx 10000 km At apogee a combinationof GNSS signals and MWL ranging will be used with expected inaccuracyof 10 cm

5 Basic spacecraft specifications

The mission foresees a single spacecraft with a mission duration limited toapprox 3 years for cost reasons Attitude control along three axes withless than 1 error on a timescale of 1000 s is required The control can beperformed with cold gas thrusters magnetotorquers or reaction wheels Theselection will be done on the basis of the vibrations induced by these systems(desired level arms lt 10minus6 g)

Cold gas thrusters or monopropellant thrusters will be required to maintainthe orbit parameters during the mission by small orbit correction manoeuvresand will also guarantee a proper tuning of the orbit parameters after the releasefrom the launcher The ground segment (telemetry control and data transfer)could be via non-ESA stations user-operated to save costs

Further resources are indicated in Table 2

Exp Astron

The satellite bus design drivers are nadir-pointing microwave antennasthermal stability for the payload instruments no strong magnetic sources lowmicrovibrations harsh radiation environment due to crossing of the radiationbelts once per orbit power demand by science payload (with the possibility toreduce power demand by stand-by of some components during specific missionphases eg eclipse) mass limit due to required high altitude with low costlaunch

The satellite bus design can be based on established platforms such asFLEXBUS or PRIMA used on several past missions Their design philosophydiffers one having body-mounted solar arrays the other wings The latter canprovide larger amounts of power The vibration level induced by the panelswhen they are fixed and the duration of solar array drive periods per orbitwill have to be considered in the selection The FLEXBUS concept was usedfor a mission already requiring a low level of residual accelerations and theplatform is optimised for this parameter

For most of the satellite components one can consider using compo-nents and electronic units from navigation or communication satellites whichtypically cruise at altitudes between 15000 and 25000 km altitude LISAPathfinder which will cross the radiation belts during spiralling to theLagrange point will face similar problems concerning radiation as EGE as wellas XMM which is in a Molniya-type orbit with similar radiation characteristics

The instruments will therefore be located in the satellite structure such asto realise ionizing radiation shielding by the spacecraft structure and non-sensitive equipment to the maximum extent possible the need for additionalradiation shielding (including trapped proton radiation) for especially sensitiveparts of the instruments must be determined by ground tests It should bementioned that an orbit with higher perigee and less radiation dose is also anoption as it leads to a tolerable reduction of the science goals

The required plusmn3C temperature stability at the interface to the payloadenclosure is feasible The thermal control will employ the usual means (anenclosure with high thermal inertia heaters and a suitable radiator) Thelocation of the radiator will need a special study given the non-standard orbit

Magnetic field present during the mission can affect the properties of theonboard clock Using SPENVIS the magnetic field has been calculated alongthe spacecraft orbit the field intensity presents maxima around 02 G anda minimum value of the order of 10minus3 G As discussed above this is not alimitation to the instrument performance

6 Discussion and conclusion

The accuracy of the measurements depends crucially on the stability andaccuracy of the clocks While the performance of ground clocks (even thoseeventually deployed in new ground stations dedicated to EGE) will likelyimprove significantly from the present level the space clock design will beldquofrozenrdquo approx 10 years before data acquisition Thus their performance

Exp Astron

level will be significantly below the future ground station clock performanceAs specification we have chosen a performance level that has been demon-strated with laboratory breadboards recently We see no fundamental reasonthat this laboratory performance cannot also be achieved in a space ion opticalclock The experience with the PHARAO cold atom clock has shown thatan engineering model of a laser-based cold atom clock has a performancesimilar to that of a laboratory prototype The perturbation of the ion clock fre-quency due to magnetic fields along the orbit can be modelled and suppressedwith shielding and the residual effects measured and compensated for Themoderate temperature stability (plusmn 3C) to be provided in the experimentalbay can be achieved with thermal shields Accurate temperature controlof particular units is done by active stabilization of the units themselves

One important aspect is radiation which may affect all optical and trapcomponents as well as the lasers and reference cavity The radiation envi-ronment on the EGE orbit is significantly harsher than the near-earth ISSorbit on which ACES will operate The radiation effects must therefore becharacterized comprehensively with appropriate tests and then the radiationshield for the EGE instruments designed

In summary the Einstein Gravity Explorer mission (EGE) is a satellitemission aimed at a precise measurement of the properties of space-time usingatomic clocks By testing one of the most fundamental predictions of EinsteinrsquosTheory of General Relativity the gravitational redshift and thereby searchingfor quantum effects in gravity it explores one of the most important frontiers infundamental physics The EGE mission reaches beyond fundamental physicsproviding an important spin-off in geophysics Furthermore the technologicalaspects of the mission are ground-breaking for future time-keeping and ref-erence frame systems in space and thus are also expected to be of wide useto science and technology in general The payload of the mission has beenconceived as a compromise between cost and performance needing to fit thecost envelope of an ESA medium-class mission1

Acknowledgement We thank S Cesare for helpful information

References

1 Will CM The confrontation between general relativity and experiment Living Rev Relativ9 3 (2006)

2 Ni W-T Empirical foundations of the relativistic gravity Int J Mod Phys D 14 901 (2005)doi101142S0218271805007139

1Finally we point out that a spacecraft with a significantly reduced and thus lower-cost payload (noclocks only MOLO FCDP MWL) on a circular orbit of appropriate height could be used for along-duration earth science mission dedicated specifically to the measurement of the geopotential

Exp Astron

we may have a variation of the fundamental constants inversely proportionalto the distance from massive bodies

X (ϕ) = X (ϕ0 + ϕs) = X (ϕ0) + δX (R)

δX (R) = (dX

dϕ

)S

R

where a nonzero δX violates LPI and a small ϕs was assumedThe gravitational potential U is essentially proportional to the number of

baryons Z + N and inversely proportional to the distance R U = -GMR

Therefore the change of the fundamental constants near massive bodies canbe written

δX

X = K (X i) δ(Ui

c2

) (2)

where i refers to a particular composition of the mass M and where thecoefficients K(X i ) = minus(dXdϕ) c2SiGMi are not universal but are expectedto be different for eg the Sun and the Earth K(X Earth) = K(X Sun)Indeed the Sun consists mostly of hydrogen therefore the Sunrsquos scalar chargeis SSun Z (sp + se) The Earth contains heavier elements where the numberof neutrons exceeds the number of protons N 11 Z Therefore the scalarcharge of the Earth SEarth Z (11 sn + sp + se) and is sensitive to the neutronscalar charge (and also to the contribution of the nuclear binding) in contrastto the Sun

The most sensitive experiments suitable for a search of a space-time de-pendence in local or near-local experiments (such as satellite experiments) areclockndashclock comparisons In one type of terrestrial local experiment dissimilarclocks located nearby are compared for a duration spanning at least one yearDuring such periods the earth moves periodically closer and further awayfrom the sun so that the solar gravitational potential on Earth is modulatedThis enables a search for a dependence of the fundamental constants on thesunrsquos scalar charge [17 18] These experiments are able to set limits to K(XSun) for X = α meqQCD at the level of 10minus6 [12 13]

The proposed EGE satellite experiments are complementary to terrestrialexperiments allowing to set limits to K(X Earth) and combining with theresults on K(X Sun) from terrestrial comparisons to set limits on the depen-dence of the fundamental constants on the neutron scalar charge sn

In summary when the frequencies of two identical clocks (of nominalfrequency νa) that are located at different positions x1 x2 are comparedby measuring their frequency ratio at some position xrsquo the result may bewritten as

va2(xprime)

va1

(xprime) = 1 +

[1 +

sum

XAXa K (X i)

](Ui (x2) minus Ui (x1))

c2 (3)

The sensitivity factor AXa = X (d(ln νa)dX) is the sensitivity of the clockfrequency to a particular constant X and can be calculated using atomic and

Exp Astron

nuclear theory (see eg [12 13]) The measurement of this ratio which issensitive to K represents a test of General Relativityrsquos redshift

When the frequencies of two co-located but dissimilar clocks a b (ofnominal frequencies νa0 and νb 0) are compared in-situ at two locations x1 x2

(LPI test) the result is

[va1 (x1)

vb 1 (x1)

] [va2(x2)

vb 2 (x2)

] = 1 +[X

(AXa minus AXb

)K (X i)

](Ui (x2) minus Ui (x1))

c2 (4)

The result of the EGE mission may be given as a limits for (or nonzerovalues of) K(X Earth) for the three fundamental constants X = α meQCDand mqQCD

23 Measurement approaches for the gravitational frequency shift

The gravitational potential experienced by the ground and satellite clocksincludes the contributions from the earth and the sun (the contribution fromthe moon is an order smaller and may be neglected for the purposes of thediscussion but will be taken into account in data analysis)

Nonzero gravitational frequency shift signals (Eq 3) can be obtained inthree ways Method (1) below is the most sensitive and represents a strongreason to employ an optical clock which on short averaging times is ableto provide a higher frequency stability than a microwave clock Method (2)and (3) are less accurate but are complementary in terms of measurementapproach Finally method (3) addresses the solar gravitational frequencyshift while (1) and (2) measure the terrestrial effect The performance of theinstruments assumed in the following is shown in Fig 2

(1) Repeated measurements of satellite clock frequency variation betweenapogee and perigeeIn this measurement approach the difference (or ratio) between earthndashsatellite clock frequency ratio at perigee and at apogee is consideredThis difference is measured at each perigee and apogee passage andaveraged over the mission duration Systematic shifts (if not correlatedwith orbital motion) are expected to average out leading to a gain insensitivity of up to n12 where n = number of orbits This mode usesthe stability properties of the satellite clocks on the timescale fromperigee to apogee rather than their accuracy Assuming a specificationof le 3 times 10minus16 satellite optical clock fractional frequency instability forintegration times between 1000 and 25000 s n sim= 1000 comparisons(ie most orbits of the 3-year mission) and a perigee-apogee potentialdifference (U(apogee) minus U(perigee))c2 = Uc2 sim= 4 times 10minus10 allows a

Exp Astron

Fig 2 Performance of clocks and links on EGE Curve ldquooptical clocks EGErdquo is the specificationCurve ldquooptical clocks EGE (goal)rdquo refers to expected near-future performance of an opticalclock based on neutral atoms ldquoUpgraded PHARAOrdquo refers to the PHARAO microwave clockusing the high-quality microwave-optical local oscillator The microwave link (MWL) performanceexhibits a discontinuity since common-view contact time of ground stations with EGE is limited toseveral h The different slopes arise because on short time scales only white phase noise is relevantwhile on long time scales flicker noise also contributes The reproducibility is taken into account

measurement of the redshift at 25 ppb accuracy The ground clocks towhich the satellite clocks are compared to at perigee and at apogee willbe located in different ground stations The inaccuracy of these groundclocks must therefore be sufficiently smaller than 1 times 10minus17 in order notto contribute to the 25 ppb error Current state-of-the art inaccuracy is2 times 10minus17 and is likely to decrease further

(2) Absolute comparison between ground and satellite clocks with the satel-lite at apogeeHere the terrestrial gravitational potential difference is maximumU(apogee) minus U(perigee)c2 sim= 65 times 10minus10 Assuming a space optical clockinaccuracy of 1 times 10minus16 and a ground clock inaccuracy a factor 3lower (current state-of-the-art) allows a measurement of the redshift at150 ppb This measurement takes advantage of the repetitive nature ofthe orbit only to a small extent and relies on the accuracy of the opticalclocks in particular of the space clocks It is thus complementary to (1)albeit less accurateIn (1) and (2) the dominant contribution to the redshift is from Earthrsquosgravitational potential The contribution from the solar potential isstrongly reduced due to the fact that satellite and earth are an interactingsystem in free-fall toward the Sun [19] and can be modeled

(3) Comparisons between terrestrial clocksThe frequency comparison link to the EGE satellite allows terrestrialclock comparisons by a common-view method whereby the frequency

Exp Astron

signals from two distant ground clocks are sent to the satellite andcompared there The frequency difference between two identical groundclocks is determined if systematic instrumental errors are negligible bythe difference in gravitational potential at the two clock locations Thispotential difference includes a terrestrial and a solar contributionThe solar contribution varies according to the orientation of the clockpair with respect to the Sun with a 1 day-period For two clocks at asurface distance of sim6000 km and suitable orientation with respect tothe sun the normalized modulation amplitude is sim= 4 times 10minus13 This is thesame as the earth potential difference for 4 km height difference It ispossible that ground clocks will have reached an inaccuracy at the levelof 1 times 10minus18 at the time of the mission in 2015 Using the microwave linka comparison of ground clocks over sim3 h (typical common-view duration)with a resolution of 1 times 10minus17 can be performed limited by the instabilityof the link Repeating the clock comparison at each common-view win-dow (n = 1000 comparisons) will allow averaging and a measurementof the solar gravitational shift with fractional inaccuracy of 1 ppm Incomparison the best current results for the solar gravitational frequencyshift are at the few-percent level [20 21] A moderate averaging out ofground clock systematic shifts is assumed as well as modeling of the earthgravitational potential variations on the daily timescale at the appropriatelevel

(4) Comparisons of on-board clocksThe two on-board clocks are continuously compared these data areanalyzed for violation of LPI (Eq 4) Since the measurement is a nulltest the required accuracy of the gravitational potential at the satellitelocation is low and easily satisfied

24 Test of local position invariance and search for neutron scalar charge

The test of LPI is performed in conjunction with the measurement of thegravitational frequency shift

With approach 23 (1) the optical clock will be able to measure K (α Earth)with accuracy 30 ppb (spec)

Combining the results of the microwave and optical clocks will determineK(meQCD Earth) + εK(mqQCD Earth) with accuracy 85 ppb (spec)where ε asymp minus01

Simultaneously the on-board comparison 23 (4) yields the combination19 K(α Earth) +K(meQCD Earth) + εK(mqQCD Earth) with accuracy85 ppb (spec)

Note that these measurements are fully complementary to current andfuture terrestrial clock-clock comparisons which are sensitive to K(X Sun)and can be combined with the latter to determine or set limits to the neutronscalar charge

Exp Astron

25 Theoretical analysis of frequency comparison signals

The on-board clock-clock comparison signals can be analyzed straightfor-wardly in terms of a violation of LPI since signal propagation effects areabsent

The determination of the gravitational frequency dilation via ground-satellite clock comparisons is more complex (Eq 1 is strongly simplified) andrequires an accurate treatment of all relevant relativistic effects Gravitationin the solar system is described in the barycentric (BCRS) and geocentric(GCRS) non-rotating celestial reference systems by means of post-Newtoniansolutions of Einsteinrsquos equations for the metric tensor in harmonic gaugescodified in the conventions IAU2000 [22] In these conventions the relativisticstructure of Newton potential (order 1c2) and gravitomagnetic (order 1c3)

potentials are given For the Newton potential the multipolar expansion ofthe geopotential determined by gravity mapping is used The post-Newtonianeffects of the metric (described in terms of dimensionless parameters β and γ )

can be included Orbits of satellites are evaluated in the GCRS at the levelof a few cm by means of the Einstein-Infeld-Hoffmann equations at the order1c2 [23] using the relativistic Newton potential Instead the trajectories of theground clocks in the GCRS are evaluated from their positions fixed onthe Earth crust (ITRS International Terrestrial Reference System) by usingthe non-relativistic IERS2003 conventions [24 25]

For the propagation of light rays (here radio signals) between the satelliteand the ground stations one uses the null geodesics of the post-Newtoniansolution in the GCRS The timefrequency transfer properties have beentheoretically evaluated for an axisymmetric rotating body [26ndash29] at the order1c4 (ie with Newton and gravitomagnetic potentials developed in multipolarexpansions) These results have been developed for use in the ACES missionThe relative frequency dilation can be expressed as

ν

ν= 1 + 1

cL1 +

4sum

n = 2

1

cn(Ln + Gn)

where Ln describes special-relativistic Doppler effects and Gn the general-relativistic effects These quantities are time-dependent due to the orbitalmotion For concreteness we report the orders of the terms The kinematicaleffects are |L1c| lt 1 times 10minus5 |L2c2| lt 2 times 10minus10 |L3c3| lt 3 times 10minus15 |L4c4| lt

7 times 10minus20 The Newtonian contributions are G2 = G(M)2 + G(J2)

2 + G(J4)2 + G(J6)

2(M is the mass monopole Jn are the lowest multipoles) with estimates|G(M)

2 c2| lt 3 times 10minus11 |G(J2)2 c2| lt 2 times 10minus13 |G(J4)

2 c2| lt 3 times 10minus16 |G(J6)2 c2| lt

1 times 10minus16 For G3 = G(M)3 + G(J2)

3 the estimates are |G(M)3 c3| lt 2 times 10minus14

|G(J2)3 c3| lt 1 times 10minus18 Finally the G4 = G(M)

4 + G(S)4 contributions (G(S)

4 is thelowest gravitomagnetic effect) are of order |G(M)

4 c4| lt 1 times 10minus19 |G(S)4 c4| lt

1 times 10minus22 and thus not relevant for EGE

Exp Astron

The available theory is fairly complete for the data analysis in mode 23 (1)described below For modes 23(2 3) the theory will need to be generalizedto include the effect of the gravitational potential at the location of the groundclock(s) (including time-varying effects such as the tides) and of the solar andlunar potentials

The uncertainties of the various contributions above are minimized by ausing precise EGE orbit data obtained by satellite tracking and by usingcurrent and future accurate earth gravity information Required orbit positionknowledge is at 1 cm level near perigee and sim50 cm near apogee reachablealready today with the proposed orbitography approach described below Thevalidity of the special-relativistic Doppler shift has been and will continue to beindependently verified with increased accuracy by spectroscopy of relativisticatomic ion beams [30] Potential violations of this aspect of Lorentz Invari-ance are expected to be sufficiently bounded so as not to affect the signalinterpretation

26 Local Lorentz invariance tests

There has been an explosion of interest in Local Lorentz invariance in re-cent years [31 32] with numerous astronomical studies and high-precisionexperiments applied to search for violations of Lorentz invariance Signifi-cant developments have also occurred on the theoretical side and a theory(standard model extension SME [33]) has been worked out that provides aunified framework for describing and analyzing Lorentz Invariance violationsin a variety of systems Two aspects of Lorentz Invariance can be tested withEGE

261 Independence of the speed of light from the laboratory velocity

A possible dependence of the speed of light on the speed v of the laboratorycan be modeled according to the MansourindashSexl test theory [34] by

c (v) = c0(1 + Bv2

c2

)

Here v 377 kms is the velocity with respect to the cosmic microwavebackground the cosmologically preferred frame and B is a combinationof parameters describing deviations from the usual Lorentz transformationformulas B = 0 if Lorentz Invariance holds For a terrestrial experiment therotation of the Earth around its axis modulates v with a 300 ms amplitude Adependence c(v) can be searched for by measuring the frequency differencebetween a resonance frequency of a highly stable optical cavity (cavity fre-quencies are proportional to c) and an optical clock (Kennedy-Thorndike-typeexperiment [35]) and determining the modulation of the frequency differencecorrelated with the modulation of v The advantages of a space experimentare the high orbital velocity and strongly reduced cavity deformation thanksto microgravity [36 37] On the proposed elliptic orbit v varies between+4 and minus4 kms over approximately 1 h This variation is 13 times larger

Exp Astron

than for the case of a terrestrial experiment The shorter time scale is alsoadvantageous since the drift of the cavity is more predictable In the EGEmission the reference cavity of the clock laser is used The large numberof orbits improves the statistics An improvement of the accuracy of B by afactor 20 compared to the best current terrestrial results is realistic The dataanalysis can also be performed in the framework of the SME theory takinginto account complementary bounds obtained from terrestrial experimentsand astrophysical observations

262 Independence of Zeeman splitting frequency on the directionof the magnetic field

One class of tests of Lorentz Invariance consists in measuring the frequencysplitting between two levels of a quantum system induced by a static magneticfield as a function of orientation of the field direction with respect to thestars (HughesndashDrever-type experiments [31]) This class addresses the so-called matter sector and is complementary to combined-photonmatter-sectortests such as the above The most precise experiments are performed withatomic clocks eg masers or cold atom clocks [38] The EGE instruments allowsuch experiments since magnetic fields are applied continuously in the clockor repeatedly for calibration Compared to terrestrial experiments the highorbital velocity of EGE could lead to an improvement by a factor sim20

27 Application to geophysics

Using GPS coordinates of a point on the Earth can be obtained with aninaccuracy well below 1 cm in a well defined international terrestrial referencesystem The coordinates are purely geometrical and do not contain any gravityinformation A local gravitational potential UB is determined with respect toa reference gravitational potential U A by ldquolevellingrdquo the surface of the Earthie sequentially measuring gravity g and height differences every ca 50 m

UB minus U A = minusBint

A

g (r) middot d n

where n is the difference vector between subsequent locations The disadvan-tage of measuring height differences by the levelling method is a random walkeffect of accumulated errors

In the classical definition a geoid is defined as the particular equipotentialsurface nearest to mean sea level In these terms the geoid serves as areference surface for measuring the height and also to define a referencefor the gravitational potential Such a vertical reference frame historicallywas realized for a country or several countries by determining the mean ofsea level observations at tide gauge stations taken over long period of timeHowever modern satellite altimetry missions such as TOPEXPOSEIDON orENVISAT show that departure of the mean sea level from an equipotential

Exp Astron

surface may reach up to several meters on the global scale see eg [39]Therefore height systems based on different tide gauge stations may differin the realisation of the geoid and may differ with respect to the reference byseveral meters

The best gravity field models obtained from satellite data (eg missionGRACE) have reached a precision of about 1 cm over sim250 km half wave-length [40] Satellite data also reveals changes in time [41] However for typicalEarth topography with height variations of eg 1000 m over 30 km horizontaldistance one may expect variations in the geoid of about 80 cm Such ahigh-spatial-frequency signal in geoid variations cannot be detected by spacegravity missions and requires a combination of satellite and terrestrial gravitymeasurements like gravity anomalies deflections of vertical and GPSlevellingpoints The best combined global gravity field models are provided up to de-gree and order 360 in terms of spherical harmonic representation correspond-ing to a half-wavelength of about 55 km However in combining the satelliteand terrestrial gravity field measurements the problem of terrestrial data givenin different height systems between continents and different countries remainsand has to be tied to satellite measurements

The geodetic scientific community is currently establishing a Global Geo-detic Observing System (GGOS) [42] Its objectives are the early detectionof natural hazards the measurement of temporal changes of land ice andocean surfaces as well as the monitoring of mass transport processes in theEarth system Global change processes are small and therefore difficult toquantify Therefore the required precision relative to the Earthrsquos dimensionis 1 ppb GGOS will be established by the combination of geodetic spacetechniques (GPS Laser-Ranging very large baseline interferometry) andrealized by a very large number of terrestrial and space-borne observatoriesThe purely geometric terrestrial 3D coordinate system is in good shape andfully operational It has to be complemented however with a globally uniformheight system of similar precision The current precision level of regionalheight systems in terms of gravity potential differences is in the order of1 m2s2 (10 cm) with inconsistencies between these various systems up toseveral 10 m2s2 (several meters) The actual requirement in the context ofGGOS is 01 m2s2 (1 cm) with the need of a permanent ie dynamical controlThis requirement of high precision height control comes from the need tounderstand on a global scale processes such as sea level change global andcoastal dynamics of ocean circulation ice melting glacial isostatic adjustmentand land subsidence as well the interaction of these processes Only by meansof monitoring in terms of gravity potential changes at the above level ofprecision the change of ocean level can be understood as a global phenomenonand purely geometric height changes be complemented by information aboutthe associated density or mass changes

The geoid can also be defined in a relativistic way [43] as the surface whereaccurate clocks run with the same rate and where the surface is nearest tomean sea level The relation between the differences in the clock frequenciesand the gravitational potential is given in simplified form by Eq 1 At present

Exp Astron

there is no operational way to compare frequencies of the already availableoptical clocks on the global scale at the same level as their accuracy wouldallow With EGE and by then improved ground optical clocks it will becomepossible to obtain gravitational potential differences on a global scale by com-paring frequencies This would be a significant new dimension to gravity fielddetermination since such observables are given on the global scale and providein situ local gravity information at the same time EGE will allow clock-basedgravitational potential mapping with the same payload instruments used forthe fundamental physics experiments The on-board clocks are actually notrequired for this purpose MOLO MWL and FCDP suffice Thus EGE willallow establishing a global reference frame for the Earth gravitational fieldwith accuracy in the order of a few cm in terms of the geoid heights Thisassumes that by the time EGE is operational (mobile) ground clocks withfractional inaccuracy at the level of 10minus18 are available the measurementsare limited by MWL noise and thus require a long integration times to matchthe expected ground clock performance Such a reference frame will serve asreference for all gravity field modeling and clocks will be used to define theinternational atomic time scale and the Earth gravitational potential at thesame time

In summary with optical clocks geodesy will undergo a second revolutionafter geometry now being measured by clock-based GNSS systems also physi-cal heights and potential will be measured by (mobile) optical clocks using thegravitational redshift effect

28 Application to frequency standards and terrestrial fundamentalphysics studies

As has been described repeating ground clock comparisons and integratingover several months will provide global ground clock frequency comparisonsat the level of 1 times 10minus18 more than two orders of magnitude below the currentGPS and two-way satellite time and frequency transfer methods In the timeand frequency community the availability of such a global high performancemicrowave link will accelerate the process leading to a redefinition of the SIsecond based on clocks operating in the optical domain

The global character of the ground clock comparisons made possible byEGE can also contribute to the search for a time-variation of the fundamentalconstants

The comparisons do not require availability of the on-board clocks The riskof this type of measurements is therefore reduced Moreover the lifetime ofmany subunits required for this measurement (MWL and FCDP see below)can exceed 10 years (from previous experience with similar devices) andthus offers a motivation to increase the mission duration beyond the nominal3 years

Further applications to metrology are listed under the secondary goals inSection 21 above

Exp Astron

3 Mission profile

31 Orbit

The orbit must satisfy the following requirements

Primary 1 Large gravitational potential difference between apogee andperigee

2 sufficient contact time with at least two ground stations atperigee

Secondary Simultaneous visibility of satellite from distant (inter-continental) ground stations for several hours

To satisfy the primary objective the orbit must be highly elliptic The perigeeshould be low A very low perigee means a short contact time which decreasesthe accuracy of the frequency measurement A compromise is a perigeealtitude of sim2500 km For a sufficiently high apogee this yields a U of 60of the theoretical maximum between the earth surface and infinity

The baseline scenario is an orbit of approx 12 h period with the followingproperties

Apogee Altitude 37856 km (minus 10 deg latitude one over Atlantic andPacific resp)

Perigee Altitude 2500 km (+ 10 deg latitude one over Guatemala andMalaysia resp)

Inclination 634 argument of perigee 170RAAN 25 True Anomaly 220Apogee drift rate 70 kmday perigee drift rate sim 14 kmday

True anomaly and right ascension of the ascending node (RAAN) wereoptimized to minimize orbital precession Figure 3 shows the ground track andthe ground coverage at one of the perigees Common-view contact time forstations in Europe is about 6 h while between a station in Europe and in theUSA it is about 35 h Frequency comparison windows between a given groundstation and EGE repeat every 24 h

A final selection of the orbit requires an in-depth study of the costs of thepropellant and maneuvers as well as of the setting up of new ground stations

32 Ground segment

The mission requires two separate ground segments the science groundsegment and the standard satellite control segment which supports telemetryand telecommand (TMTC)

Exp Astron

Fig 3 Baseline EGE orbit One (out of two) apogee and one (out of two) perigee ground pointsare shown The ground coverage at one of the perigees is shown with circles at elevation anglesteps of 10 starting from 10

321 Science ground segment

As described above one part of the mission is to perform frequency com-parisons between ground-to-space and ground-to-ground (common-view andnon-common view) The use of a two-way microwave link (MWL) systemhas the advantage that a precise orbit or the precise relative velocity of thesatellite must not be known [4] The orbit determination requirements fora two-way system are significantly relaxed compared to the requirements tomodel the relativistic effects on the clocks The frequency comparison resultsare available both on ground and on the satellite

For ground-to-satellite clock comparisons five ground stations arefavourable two capable of viewing perigee three for viewing apogee Thisallows for the classic three-clock configuration where the performance of eachclock can be obtained individually The perigee stations will be observing thesatellite for sim25 min every perigee passage for the duration of the missionThe apogee stations will observe for sim3 h every apogee passage Of coursea larger number of stations as well as longer observation time are usefulfor robustness redundancy and orbit determination The stations could beinstalled on a temporary basis for the duration of the mission

The stations must have access to optical clocks of performance at least ashigh as todayrsquos state-of-the-art ground clocks but preferably higher Theywill include an optical frequency comb by means of which the optical clockfrequency is transformed to a radio frequency for comparison with the EGEclocks High-accuracy fiber-optic link technology already existing today allowsdistances of sim100 km between the ground station and the correspondingoptical clock allowing a certain freedom in the choice of the clock locationWith further increase of the fiber-optic link range currently being investigated

Exp Astron

the area over which the ground station can be positioned relative to the opticalclock will increase This option is of importance for the implementation of ageophysics measurement campaign for which a large number of clock locationsis of interest

The above ground stations will be complemented by some 15 additionalground terminals provided by the world-wide laboratories equipped with highperformance primary clocks at the time of the mission These will participate inorbit determination to 1 cm accuracy along the full orbit which can be achievedwith non-optical atomic clocks The EGE MWL is fully compatible to plannedACES ground terminals which might reduce the number of required stations

With a larger number of transportable optical clock ground stations de-ployed and moved across the continents a geophysics campaign could beimplemented to map the gravitational potential

Ground station equipment MWL uses dedicated small transportable termi-nals which can be easily installed at any location of interest Each terminalconsists of a steerable dish antenna of 1 m diameter with pointing accuracy of01 using a transmitted power of 5 W per band The high directivity antenna(beam aperture Ka-band 07 Ku-band 14 S-band 20) avoids multi-pathfrom nearby objects Antenna noise temperatures are Ka-band 450 K Ku-band 340 K Operation is at elevations above 10 and measurements areperformed above 20 For ranging purposes terminal location is determined byGPS surveying Each ground station computes its own ionospheric correctionsbased on a triple-band receiver The terminal has a built-in delay monitorwhich allows calibrated ranging during several months once its bias has beendetermined by laser ranging Its M and C interface is via local area networkto either a userrsquos computer or directly to the science network control centreby remote MampC The stationrsquos operation is fully automatic and unattendedSchedule and orbital data are received via LAN once per day Signal acquisi-tion is fully autonomous Data transmission is after each pass The terminal hasan uplink telecommand capability of 1 kBits and a downlink data capacity ofup to 10 kBitss For the local user there is a visual interface providing quick-look data from real-time comparison to the spacecraft clocks to verify theirgood health The final full performance product is calculated at the sciencedata centre

Local metrological data (temperature pressure humidity) are recorded tocorrect for tropospheric delay of the ranging data The main advantage ofMWL terminals compared to laser based systems is their all-weather capabil-ity although microwaves suffer higher propagation errors in the ionosphereand troposphere

The EGE ground terminal will profit from the ERS-2 mission PRARE(precise range and range rate equipment) [45] experiments whose terminalshad similar design and have been operated under extreme climatic conditionsincl Antarctica and high temperature regions from existing designs fromthe space segment (mission ACES) and no further special developments aredeemed necessary Cost per EGE master station is modest

Exp Astron

Satellite tracking via laser ranging Laser ranging is alternative to MWL rang-ing and also serves for absolute calibration of the latter For this purpose anon-board corner cube reflector (CCR) is provided Ranging will be done withthe International Laser Ranging Service Ideally the satellite will be rangedat every perigee passage and every apogee passage simultaneously with thefrequency comparison procedures Todayrsquos ranging precision is sufficient tosatisfy the science goals For apogee observations in the northern hemispherea large number of laser ranging stations is available The number of stationscapable of ranging at or near perigee in the southern hemisphere is six andthus is sufficient even considering unfavourable weather conditions at someof these The MWL ground stations are able to range even under thoseconditions

322 Control ground segment

Average data transfer rate is estimated at 300 MBday (approx 40 kBs onaverage) Data will be stored on board (2 GB capacity required) and will betransferred to a receiving station on the Earth once a day

XPLC

incl Data StorageIF to TMTC Link

HK amp TMTC

to ground

Opt Clock

AOM

MOLO

PHARAO

FCDP

10 GHz

100 MHz

250 MHz

GNSS Rx

MWLTampF Link

uplink Data

UTC(k)

TampF Space vs Ground

TM from GS

Tim

e S

cale

Orbit GNSS Time UTC

TampF Space vs Ground

MampC Time Scale

MampC

MampC

MampC

MampC

Mamp

C

PHARAO vs Opt Clock

PHARAO vs Opt Clock

10 GHz amp Epoch

10 MHz

amp Epoch

Epoch

10 MHz

amp Epoch

Opt Link

Fig 4 Schematic of the payload showing subunits and their signals The results of the frequencycomparison between the two on-board clocks (optical clock and microwave clock PHARAO) areobtained from a combination of the frequency values applied to the frequency shifters AOM (seeFigs 7 and 8) and those output from the frequency control and distribution package (FCDP)as indicated by the arrows ldquoPHARAO vs Opt Clockrdquo The satellite clock to ground clockcomparison results are output from the MWL XPLC is the on-board computer and is involvedin generation of signals for instrument control The optical link shown dashed is a possible add-onproviding an alternative to the MWL GS ground stations MampC monitoring and control HKhousekeeping TampF time and frequency

Exp Astron

4 Main payload instruments

41 Overview

The satellite payload consists of a pair of cold-atom clocks a microwave-optical local oscillator (MOLO) an on-board frequency distribution unit(FCDP) a satellite-to-ground clock comparison unit (MWL) and the onboardcomputercontrol system (XPLC) see Fig 4 Auxiliary units are the posi-tion determination GPSGalileo receiver (GNSS Rx) attitude monitor (startracker not shown) and the corner cube reflector array (CCR not shown)

The science goals require clocks of very high frequency stability on thetimescales of half the orbital period (6 h) as well as perigee passage (sim1000 s)see Fig 2 Only clocks based on cold atoms are suited for the science goalsTable 1 reports values for a subset of atomic clocks that have been investigatednot including clocks that require cryogenic operation or are too complex foruse in an medium-class mission The choice of a clock pair rather than a singleone fulfils several requirements it addresses more science goals improves thevalidity of some science results and it enables calibration accuracy evaluationand redundancy The validity of the science results is enhanced since theclocks differ not only in the employed atoms but also in terms of havingdifferent sensitivity to perturbations by electric and magnetic fields electronicserrors and misalignments A comparison of the results arising from each clockallows for an understanding of measurement errors Both the clocksrsquo and thefrequency comparison unitrsquos performances will be determined after launchand regularly verified during the course of the mission The only reliable

Table 1 Properties of different clock candidates for EGE

Clock Clock Fractional σ = Fractional Sensitivity Figure of Sensitivityatom frequency inaccuracy instability to α Aα merit |Aα| to meq

at 1000 s σ [1016] QCD

Sr+ 445 THz 3 times 10minus17 (p) 5 times 10minus17 (p) 042 08 (p) ndashYb+ (1) 688 THz 3 times 10minus16 (d) 3 times 10minus16 (d) 088 029 (d) ndash

3 times 10minus17 (p) 1 (p)Yb+ (2) 642 THz lt 1 times 10minus17 (p) 1 times 10minus17 (p) minus53 53 (p) ndashSr 435 THz 1 times 10minus16 (d) 1 times 10minus16 (d) 009 009 (d) ndash

lt 1 times 10minus17 (p) 09 (p)Yb 519 THz 7 times 10minus16 (d) 1 times 10minus16 (d) 031 03 (d) ndash

lt 1 times 10minus17 (p) 1 times 10minus17 (p) 31 (p)Cs 91 GHz 2 times 10minus16 (d) 1 times 10minus15 (d) 275 027 (d) 1Yb+ 126 GHz 1 times 10minus15 (p) 1 times 10minus15 (p) 34 034 (p) 1

A relevant figure of merit is the ratio of |AX | and fractional frequency instability σ on the perigeevisibility timescale The Cs clock is the only high performance clock with adequate sensitivity tothe particle masses me mq The indicated Cs clock performance was achieved using a microwavelocal oscillator based on a cryogenic oscillator On EGE the latter is to be replaced by an opticallocal oscillator-based device (MOLO) with similar short-term performance but without the needto operate at cryogenic temperature The clock instability at 10000 s is typically similar to theindicated inaccuracy(p) projected (d) demonstrated (1) quadropole (2) octupole

Exp Astron

way to achieve this is by having at least two clocks on board The on-boardclock-clock comparison provides performance evaluation independent of theground-satellite comparison unit Once the satellite clock performances areestablished the link performance can be established as well

The clocks are similar in basic structure but differ in the detailed implemen-tation They consist of

a source of atomsa preparation subunit which cools the atoms to near-standstill (millikelvintemperature or below)a trapping subunit that confines the atom(s) to a tiny volumean interrogation subunita clock control electronics package

The clocks share a common central unit the optical-microwave local oscil-lator which serves as a flywheel oscillator with outstanding stability on shorttimescales (01ndash10 s) for both clocks It consists of a laser stabilized to an ultra-low loss optical resonator made of ultra-low-expansion glass A microwavesignal is derived from the laser by phase-locking a frequency comb to the laserwave and using a multiple of the repetition rate

While the subunits of any optical clock have a similar technological basis(optics electrooptics acoustooptics thermal control vacuum systems) theydiffer in their specifications Depending on the choice of the atom species de-livery confinement and interrogation methods clocks of different perfor-mance result A number of laboratory optical clock demonstrators have beendeveloped worldwide This ensures availability of the know-how for the indus-trial implementation of the flight models

The first clock is a microwave clock based on the interrogation of slowlymoving ensembles of cold Cs atoms It is essentially based on the PHARAOclock [5 6] developed to engineering model level by CNES in the frame-work of the ESA ISS mission ACES In EGE this clock is upgraded usingthe microwave-optical local oscillator This improves significantly the clockperformance The specification is a relative instability (Allan deviation) of3 times 10minus14 (τ s)minus12 ie at the integration time τ = 1000 s it drops to 1 times 10minus15and to 3 times 10minus16 at 10000 s While this specification is at least a factor 3 lessstringent than for the optical clocks the significantly lower cost of obtaining aflight model and the sensitivity of its frequency to the particle masses makes ita suitable choice

The optical clock baseline instrument is a single-ion optical clock Amongvarious suitable ion species (Hg+ Al+ Sr+ Yb+) [44ndash47] the Yb+ ion hasbeen selected Its breadboard performance is established at the instability levelof 3 times 10minus16 over 1000 s it exhibits a large sensitivity coefficient Aα hassmall sensitivity to magnetic fields and exhibits significant potential for furtherimprovements (use of the octupole transition with sim6 times larger Aα andlower overall instability [48] see below) A trapped Yb+ ion is also a veryrobust system it has been demonstrated that a single Yb+ ion can remaintrapped uninterruptedly for many months Finally Yb+ is the only ion with

Exp Astron

which optical clocks demonstrators have been operated in not a single but twomajor metrology laboratories

Another attractive optical clock is based on ensembles of cold neutral Sror Yb atoms trapped in an optical lattice produced by laser beams This clocktype has a lower instability than a single-ion clock thanks to the large numberof atoms (sim105) used Its development is in progress in several laboratoriesin the world (in part supported by space agencies) and has to date reached6 times 10minus17 instability at 5000 s and 1 times 10minus16 uncertainty [49 50] with potentialfor even higher performance However it is more complex and costly thanthe Yb+ ion clock since the cold atom preparation and trapping subunitsinclude more components require higher laser powers and the respectiveoperating procedures are more complex It is therefore not considered for thepresent mission designed to be developed in the very near future with limitedresources but is an interesting candidate for missions without such constraintsIts estimated physical parameters are indicated in Table 2 for comparison

42 Main instruments

421 Microwave clock

The PHARAO cold atom clock is one of the two atomic clocks of the spacemission ACES managed by ESA It is proposed here to include a secondversion of this instrument as its costperformance ratio is very attractive forthe EGE scientific objectives Its concept is very similar to ground basedatomic fountains but with the major difference of zero-g operation Atomsslowly launched in free flight cross two microwave fields tuned to the transitionbetween the two hyperfine levels of the cesium ground state The interrogationmethod based on two separate oscillating fields (Ramsey scheme) allowsthe detection of an atomic line whose width is inversely proportional to thetransit time between the two microwave cavities The resonant microwavefield at 9192631770 GHz (SI definition of the second) is synthesized startingfrom an ultrastable quartz oscillator (USO) in the FCDP phaselocked to themicrowave output of the microwave-optical local oscillator and stabilized tothe clock line using the error signal generated by the cesium resonator In thisway the intrinsic qualities of the cesium hyperfine transition both in terms ofaccuracy and frequency stability are transferred to the USO In a microgravityenvironment the velocity of atoms along the ballistic trajectories is constantand can be changed continuously over almost two orders of magnitude (5ndash500 cms) Differently from atomic fountain clocks presently operated onground very long interaction times (up to few seconds) will be possible whilekeeping reasonable the size of the instrument

The instrument consists of 4 main elements a laser source which providesthe light required to cool and detect the cesium atoms (Fig 5 left) a frequencychain which generates the 92 GHz microwave signal a cesium vacuum tubewhere the interaction between the microwave field and the atoms occurs

Exp Astron

Table 2 Resources and technology readiness level (TRL) of the EGE spacecraft

Unit Volume Mass Power TRL Comment[l] [kg] [W]

Ion clock 100 70 60Atom source 4Atom preparation 6Atom trap 4Atom interrogation 6Clock control 6

Microwave clock 200 91 110 8 TRL by end 2008Local oscillator

Stabilized laser 18 15 23 6Frequency comb 20 16 38 6

Microwave link 18 18 55 8FCDP 7 8 12 8GPS receiver 2 5 7 9CCR 2 2 0 9Payload sum 367 225 305Thermal control 14 17 30 8OBDH 20 24 38 9Star tracker 1 1 8 9Radiation shield 10 40 0 9Support structure 170 204 0 9Power incl solar arr 50 58 10 9Propulsion 21 25 20 9AOCS 36 43 82 9TMTC 8 10 15 9Satellite bus sum 330 422 203 Flexbus conceptPayload plus bus incl 20 margin 8364 7764 6096Propellant 40Launch adapter 100SC wet masspower incl Margins 9164Lattice clock 135 105 152 Alternative clock

Atom source 4Atom preparation 6Atom trap 4Atom interrogation 6Clock control 6

Also included are the resources for an optical lattice clock based on neutral Strontium atoms

(Fig 5 right) and a computer The engineering model of PHARAO is fullyassembled at CNES Toulouse and undergoes functional and performancetests until the end of 2008 The operating temperature range is 10ndash33 andnon-operating temperatures are minus45 +60 The cesium tube consists in aUHV chamber pumped by getters and a 3 ls ion pump a cesium reservoira microwave cavity coils to provide a uniform magnetic field and three layersof magnetic shields Vacuum windows with fiber-optics collimators enable thetransmission of the cooling and detection light onto the cesium atoms

In autonomous mode PHARAO uses the USO as interrogation oscillatorand will provide a clock signal with fractional frequency stability below1 times 10minus13 (τs) minus12 and inaccuracy near 1 times 10minus16 When using the microwave-

Exp Astron

Fig 5 Left The PHARAO laser source engineering model with cover removed Dimensionsare 530 times 350 times 150 mm3 and the mass is 200 kg Ten polarization maintaining optical fibers(yellow) guide the laser beams to the cesium tube All diode lasers are mounted on Peltier coolersfor temperature regulation within 2 mK (courtesy EADS SODERN) Right The PHARAOengineering model cesium tube without the two external magnetic shields The volume is 990 times336 times 444 mm3 and the total mass is 44 kg

optical local oscillator PHARAO will provide a reduced frequency instabilityof 3 times 10minus14 (τs) minus12 This corresponds to 1 times 10minus15 at 1000 s and 3 times 10minus16

at 104 s

422 Ion optical clock

The 171Yb+ ion clock shown in Figs 6 and 7 provides an atomic frequencyreference based on a single 171Yb+ ion confined within an RF electric end-captrap (similar to a Paul trap) within an ultra high vacuum (UHV) chamber andlaser cooled to sim1 mK close to the Doppler cooling limit As a result the ionrsquosfirst order Doppler motion is completely removed and the ion experienceslittle collisional perturbation from its environment Under these conditionsthe narrow linewidth 2S12 (F = 0 mF = 0) minus 2 D32 (F = 2 mF = 0) quadrupole

Fig 6 Left Simplified level scheme of 171Yb+ showing relevant cooling (blue) auxiliary (red)quadrupole clock (green) and octupole clock (dark blue) transitions δνnat is the natural linewidthof the clock transition Middle close-up of Paul-type RF trap (PTB Braunschweig) the ion istrapped in the hole (diameter 14 mm) Right Complete ion trap with vacuum flange height 10 cm(NPL Teddington)

Exp Astron

935 nm D32-state repumper laser

638 nm F72-state repumper laser

739 nm extended cavity diode laser

AOM

PMT

Ion trap UHV chamber

vacuum pumps

Fibre-free space interface

Mu-metal shielding

EC

T

B

Magnetic field

Temperature

Ion fluorescence Processor

Clock laser frequency step and control

4355 nm

Frequengy doubling II

3695 nm cooling wave

Low-drift cavity

Low-drift cavity 171Yb+ 435 nm ion clock unit

876 nm clock laser wave from MOLO

Frequengy doubling

Fig 7 Simplified schematic of the single-ion optical clock unit AOM acousto-optic frequencyshifter PMT photomultiplier tube EC endcap T temperature stabilization system B magneticfield

clock transition at 4355 nm can be interrogated by spectrally narrow and stableclock laser light A diagram of the simplified energy level scheme for 171Yb+showing the relevant cooling auxiliary and clock transitions is given in Fig 6All these wavelengths are provided by diode lasers operated in fundamentalor frequency doubled mode The clock laser light is obtained from the 871 nmwave provided by the MOLO unit (see Section 423) after frequency-doublingin single pass by a KNbO3 nonlinear-optical crystal to 4355 nm The clocktransition spectral profile is observed by stepping the clock laser frequency bymeans of an acousto-optic frequency shifter (AOM in Fig 7) and recordingthe statistics of the quantum jumps in cooling laser induced fluorescence asa function of this frequency Under clock operation the AOM repeatedlysteps back and forward between the half-intensity points on the transitionprofile monitoring the signal imbalance between these points Any detectedimbalance is servo-corrected to zero by feedback to the AOM

In detail the optical clock architecture will comprise (Fig 7)

bull an RF end-cap trap for ionising and confining a single 171Yb+ ion withinan ultra-high vacuum chamber pumped by a small ion pump and non-evaporable getter pump The trap is driven by means of an ac voltage ofa few hundred volts at a drive frequency of sim10 MHz A small oven withseveral mg of 171Yb isotope is heated to provide a low flux of Yb atomsfrom which single ions can be ionised by electron bombardment from ahot wire filament within the trap potential well Low voltage dc is applied

Exp Astron

to additional electrodes to position the ion precisely at trap potentialcentre in order to minimise the ionrsquos micromotion at the drive frequencyThis needs automatic periodic monitoring and correction especially in theperiod after re-loading an ion into the trap External magnetic field coilsin three orthogonal axes allow the nulling of external fields and settingof a fixed field of sim1 μT The trap and coils are surrounded by mu-metalshielding to minimise external field changes during the orbit A level oftemperature control of 1 K of the region surrounding the trap is requiredfor maintaining the blackbody-induced frequency shift uncertainty withtemperature at the 10minus17 level

bull a laser platform to provide frequency doubling of an extended cavity739 nm diode laser for Doppler cooling of the ion on the 2S12(F = 1) minus2 P12 (F = 0) dipole transition at 3695 nm Light from an amplified diodelaser device frequency-doubled in a periodically poled LiTaO3 crystal insingle-pass will be used Typical laser powers used for driving the coolingtransition below saturation are about 2 μW for a beam waist in the trapof about 50 μm By modulation of the injection current of the 739 nmdiode laser at 147 GHz a sideband is generated that excites the 2S12 (F =0) rarr2 P12 (F = 1) ldquorepumperrdquo transition in order to avoid optical pump-ing between the ground hyperfine states

bull The platform also houses the auxiliary lasers at 935 and 638 nm The935 nm diode provides repumping of the ion from the 2 D32(F = 1)metastable level after occasional branching decays to this level during thecooling sequence The 638 nm diode allows fast recovery of the ion fromthe very-long-lived 2F72 metastable state after very occasional collisionaldecay to that state Currently extended cavity lasers are forseen butdistributed-feedback lasers may be available in the near future

bull a high NA lens imaging system and photomultiplier detection system torecord the statistics of 369 nm fluorescence quantum jumps as a functionof clock laser frequency step

bull a fibre system to deliver the various cooling auxiliary and clock light fromsource to trap making use of achromatic doublets where necessary at thefibre-free space interface for launching into the trap

bull a monitoring and control processor which provides the primary coolingand clock laser pulse magnetic field and detection sequencing to observeand lock to the ion clock transition frequency The processor also monitorsfrequency and amplitude data necessary to determine normal laser andion operational conditions and initiate resetting and recovery algorithmswhere necessary and laser unit failure The actual magnetic field presenton the atomic volume is determined and actively stabilized so that aninaccuracy at the level of 1 times 10minus17 can be achieved

bull a redundancy level of at least two units for cooling clock and repumperlasers plus similar for frequency doubling crystals All redundancy unitsfor each wavelength will be fibre multiplexed as standard allowing redun-dant unit activation on determination of prior unit failure

Exp Astron

For an optimized quantum-projection-noise-limited performance of thesingle-ion clock Ramsey interrogation with a cycle time equal to the lifetimeof the metastable D32 level of 50 ms will be performed leading to an Allandeviation of about 27 times 10minus15 (τ s)minus12 This is sim45 times 10minus17 at 1 h satisfyingrequired stabilities for the science objectives Current laboratory results are astability of 1 times 10minus15 at 30 s averaging down to 3 times 10minus16 at 1000 s (PTBGermany) Also comparisons between two independent traps showed agree-ment at the level of several parts in 10minus16 over a series of eight measurementruns

The 171Yb+ ion clock offers a second option namely the 2S12ndash2 F72 octu-pole transition at 476 nm (see Table 1 and Fig 6) [48] It has an extremelylong upper state lifetime low electric quadrupole and second order Zeemansystematic shifts and therefore is a candidate for an even higher performanceclock if a clock laser system with lower instability than the above is providedApart from the clock laser it shares the laser system of the quadrupoleclock Thus with moderate additional resources the Yb+ ion apparatus couldprovide two optical clocks in a single package and specifically perform directfrequency comparisons between the quadrupole and the octupole transitionwith a large sensitivity Aα

The ion clock will operate under microprocessor control providing thecontrol algorithms for the integrated pulse sequencing for cooling repumpingclock interrogation magnetic field switching cooling fluorescence monitoringThe controller will also provide error flags for critical processes and systemreset and correction where necessary This will include ion re-loading andmicromotion reduction algorithms Automatic monitoring of laser power andspectral quality with signal re-optimisation or laser failure determination isrequired Redundant laser units will be pre-aligned and multiplexed into thefibre delivery to the trap so that redundant units can be readily activated in thecase of diode laser failure

423 Microwave-optical local oscillator (MOLO)

This oscillator provides two ultrastable frequency outputs one in the opticalregion the other in the microwave region the two frequencies being coher-ently related The MOLO is composed of three subunits the clock laser thereference cavity and the frequency comb see Fig 8

The clock laser subunit delivers the light for the excitation of the opticalclock transition It is generated by an extended-cavity diode laser at a wave-length of 871 nm Most of the output wave of this laser is sent to the ionclock The laser subunit containing two lasers for redundancy will require3 kg 3 l 15 W The frequency instability of the clock laser must be superiorto the minimum clock instability 27 times 10minus15 (τ s)minus12 for τ up to the timeconstant (sim10 s) of the servo system that locks the laser to the ionrsquos resonancesignal In order to achieve this the laser frequency is stabilized to the reference

Exp Astron

871 nm clock laser

EO

Detector and servo

AOM Femtosecond

frequency comb

To optical clock 10 GHz To FCDP

Microwave-optical local oscillator

ULE cavity

Fig 8 Microwave-optical local oscillator (MOLO) consisting of clock laser reference ULE cavityand femtosecond frequency comb (OFC) Radiofrequency inputs and outputs are not shown EOelectro-optic modulator ULE ultra-low expansion AOM frequency shifter

cavity where the PoundndashDreverndashHall method is used About 10 μW of thelaser output are used for this

The system will exhibit sim04 Hz (sim1 times 10minus15 relative) linewidth and afrequency drift lt01 Hzs The cavity consists of two high-reflectivity mirrorsoptically contacted to a cylinder of diameter 70 mm length 150 mm made ofULE (a glass exhibiting ultra low thermal expansion coefficient) with a cavityfinesse of about 300000 Residual vibrations of the satellite (arms lt 1 times 10minus6 g)should not be a limiting factor to the desired linewidth for safety marginan optimized cavity support can reduce the sensitivity to accelerations Foracoustic and thermal isolation the cavity resides inside an aluminium vacuumchamber equipped with a small (3 ls) ion-getter pump Two stages of polishedaluminium shields are implemented around the cavity which are activelytemperature stabilized so that the cavity is precisely kept at the zero crossingtemperature of the ULE thermal expansion coefficient Heat application andremoval is by thermoelectric elements between the shield and the vacuumchamber Additionally the entire vacuum chamber is temperature-stabilized

During transportation and launch the cavity will be rigidly fixed insidethe vacuum chamber by pressing on it from the sides by piezo-mechanicalactuators allowing accelerations of several g After bringing the satellite onthe orbit the actuators will be released

The optical setup for frequency stabilization will be a classic Pound-Drever-Hall scheme in reflection using an electro-optic modulator (EO) Couplingof the clock laser radiation to the cavity is via single-mode polarization main-taining optical fiber The fiber in- and out-couplers are mounted on miniaturepiezo-motor driven multiaxis translation stages Microprocessor control ofthe translation stages performs a laser-to-fiber incoupling stabilization andlaser-to-cavity mode-matching optimization (positioning inaccuracy lt10 μmalignment inaccuracy lt10 μrad)

The frequency stability of the clock laser will be limited by thermal 1f noisein the mirrors with an fractional Allan deviation slightly below 1 times 10minus15 from1 s to a few 10 s Resources including control electronics are 15 l 12 kg 8 W

Exp Astron

A femtosecond fiber laser based optical frequency comb generator (OFCFig 8) will be used for the clock frequency comparisons and to obtain aspectrally ultrapure microwave local oscillator output for the operation ofPHARAO for the microwave link and for the GNSS receiver The fem-tosecond laser is based on erbium-doped fiber pumped by fiber-coupled laserdiodes It emits 100 fs pulses at a repetition rate of 250 MHz with an averagepower of 100 mW The emission spectrum is centered around 1560 nm Spec-tral broadening to more than one octave is obtained via nonlinear interactionin a microstructure fibre so that the range from about 1050 to 2100 nm iscovered

Second harmonic generation of a portion of the broadened spectrum allowsobtaining the required light for the measurement of the clock laser frequencyat 871 nm The OFC will be referenced to the clock laser frequency (afterintroduction a controllable shift via the frequency shifter AOM) by makingthe carrier envelope offset frequency vanish so that the repetition rate fr isa high-order subharmonic of the optical clock reference frequency f1 = m1 frThe link between the optical and the microwave domain is then simply thedistribution of the repetition rate frequency or more conveniently a high(order 40) harmonic of the comb repetition rate 10 GHz

It has been demonstrated by several groups that frequency comparison andfrequency division with an OFC can be performed to very high accuracy andstability beyond the limitations imposed by the optical clocks themselves Forexample a fiber laser OFC with an instability (Allan deviation) of 6 times 10minus17

in 1 s has been demonstrated [51] OFCs based on fiber technology possessgood long-term reliability and high power efficiency The projection for a spacequalified OFC based on current industrial terrestrial technology is 16 kg mass20 l volume and 38 W power requirement

424 Frequency comparison and distribution package (FCDP)

This package distributes the ultra-stable 10 GHz reference frequency outputof the MOLO to the MWL performs frequency comparisons between MOLOand PHARAO and generates a variety of different reference signals for theindividual instruments including the 250 MHz reference clock to the OFCAll signals are coherent to the MOLO In the FCDP a USO is locked to theMOLO signal by a digital PLL A divider chain generates timing pulses (1 pps1000 pps etc) and keeps a continuous local time scale From the digital phasecomparison data the phase noise density spectrum between the USO and theMOLO is available by telemetry (TM) to assess the overall frequency purityin-flight

As the FCDP distributes the full performance reference from the MOLOdirectly to the MWL there is minimum impact of the FCDP on overall clockcomparison performance The USO-phase lock loop uses a similar designas for ACES-FCDP The digital PLL with linear phase detector allows theUSO to be slightly offset from its reference which is selectable by groundcommand This results in an accurate on-board time scale despite any offset

Exp Astron

of the MOLO reference frequency which can be synchronised to UTC Thefrequency synthesis chain re-uses modules developed for ACES-FCDP andtheir frequency inaccuracy Expected relative uncertainty (Allan deviation)of the frequency comparison between two signals provided to the FCDP is47 times 10minus17 at 1000 s 95 times 10minus18 at 10000 s

During non-availability of the MOLO ie during system start-up andcontingency conditions the USO runs as flywheel (ie it is unlocked withinstability lt15 times 10minus13 from 1 to 100 s) It will still provide useable butless accurate references to the instruments including the reference to OFCCommon view ground-to-ground time- and frequency comparison and supplyof the remaining instruments remain possible at the USO provided accuracy

Mass volume and power consumption are 8 kg 7 l 12 W Controldata rateis 10 kBday 50 kBday (FCDP alone)

425 Microwave link (MWL)

The microwave link is a two-way triple-frequency microwave link (Fig 9) Itis designed to perform accurate frequency comparison as well as time-transfer

Ku-Band Up-linkPower Tx 5 WCarrier 13475 GHzPN-Code 100 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

S-Band Down-linkPower Tx 1 WCarrier 2248 MHzPN-Code 1 MChips1pps 1 time marker sData 5 kBits

Ku-Band Down-linkPower Tx 1 WCarrier 1470333 GHzPN-Code 100 MChips1pps 1 time marker sData 5 kBits

EGE SC

S

Ku

KaKa-Band Up-linkPower Tx 5 WCarrier 345 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

Ka-Band Down-LinkPower Tx 1 WCarrier 32 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 5 kBits

GS Antenna diameterEGE terminal 1 mACES terminal 06 m

Fig 9 EGE microwave two-way link concept and characteristics See text for a description Thelink will be able to perform frequency comparisons with up to ten ground stations simultaneously(only one active comparison is shown) pps pulse per second Tx transmitter SC spacecraft

Exp Astron

between on-board and on-ground clocks Two-way operation allows for com-pensation of the effects due to spacecraft location and its motion relative toground Its two bi-directional links operate in Ka-band (32345 GHz) andKu-band (147135 GHz) The high carrier frequency of the up- and down-links allows for a noticeable reduction of the ionospheric delay A third down-link channel is in S-band (22 GHz) and is used to determine the ionospheretotal electron content MWL uses a combination of carrier phase and codephase data to measure with high precision the absolute ionospheric delayas well as ionospheric delay variations during a ground contact The useof three frequency bands allows to resolve third order ionospheric termsKey modulation parameters are Ka band Tone 250 MHz Ku band PN100 MChips S-band 1 MChips

The EGE-MWL concept is a considerably improved version of the ACES-MWL Microwave carrier generation is directly from the 10 GHz microwaveoutput available from the MOLO and having exceptionally low close-in phasenoise without intermediate frequency dividers or other complex synthesiserstages commonly used in RF transponders The high frequency carriers arekey to achieve the desired frequency comparison accuracy

The Ka-band carrier is generated by tripling the FCDP-generated signalThe Ku-band is subsequently generated by division by two using a verystable regenerative divider Finally the assigned radio frequency is obtainedby adding a low frequency signal using an image reject mixer A mixedanalog-digital receiver design will enhance its performance by use of a flexibledigital signal processing and digital control loops The number of simultaneousground contacts is increased from four (as in the ACES mission) to ten or moreusing the digital concept while the number of physical receiver channels perband is reduced from four to two one for operation to ground and one for in-flight calibration The concept allows adaptation of receiver loop parametersto optimise for the different portions of a highly elliptic orbit while reducingsystematic uncertainties due to high Doppler and Doppler rate Real-timedata exchange to ground terminals is by additional data modulation on all RFsignals The spacecraft carries two different small antennae for perigee and forapogee with gains of 8 dBi 20 dBi beamwidths 100 (full earth coverage) 14respectively power is 1 W for each

The MWL end-to-end performance between the attached clocks includingspace- and ground terminals is predicted to be a relative Allan deviation of1 times 10minus16 at 1000 s (perigee) and 1 times 10minus17 at 10000 s (apogee) Thesevalues are based on conservative implementation losses and other instrumentallosses assumed as 6 dB at apogee and 10 dB at perigee including additionallosses due to atmosphere An attitude knowledge of 1 is required

The MWL link will also support precise orbitographyndashcomplementary tothe GNSS receiver Generation of ranging data is simultaneous to two-waytime transfer and is available under all weather conditions Orbit determina-tion accuracy is expected to be better than 1 cm well matching EGE clockrequirements The value is expected based on demonstrated results fromother missions (METEOR ERS-2 TOPEXPOSEIDON) Laser ranging to

Exp Astron

the CCR array is used a few times during the mission to calibrate the MWLranging

In case of non-availability of the ultrapure signal from MOLO an additional10 GHz output is available from the FCDP to support ground-to-ground clockcomparison in common-view mode even after the nominal mission periodwhen the neither the clocks nor MOLO may be available any longer Basedon the PRARE experimentrsquos 12 years of operation this can be assumed forthe FCDPMWL combination as well The potential extended operational pe-riod makes the proposed configuration particularly attractive to metrologicallaboratories that rely on availability of long-lasting data in support of thegeneration of UTC

The resources of MWL are as follows mass 18 kg volume 18 l power55 W TC 2 kBday TM 120 MBday

426 Auxiliary units

The star camera assembly on the EGE spacecraft will provide real-timeattitude data primarily for the orientation of the spacecraft (with accuracy of1 in real-time) and the proper reduction of the GNSSSLR data

A GNSS receiver is used for the precise orbit determination time andfrequency transfer coarse positioning (lt1 m) for real time use by the Attitudeand Orbit Control as well as the back-up for time tagging of all payloaddata For the time and frequency transfer the receiver is nominally using theonboard frequency reference The receiver assembly includes three specialantennas By the year 2015 more than 100 GNSS satellites are expected tobe available for precise orbit determination permitting orbit inaccuracy of 1ndash2 cm RMS for altitudes up to approx 10000 km At apogee a combinationof GNSS signals and MWL ranging will be used with expected inaccuracyof 10 cm

5 Basic spacecraft specifications

The mission foresees a single spacecraft with a mission duration limited toapprox 3 years for cost reasons Attitude control along three axes withless than 1 error on a timescale of 1000 s is required The control can beperformed with cold gas thrusters magnetotorquers or reaction wheels Theselection will be done on the basis of the vibrations induced by these systems(desired level arms lt 10minus6 g)

Cold gas thrusters or monopropellant thrusters will be required to maintainthe orbit parameters during the mission by small orbit correction manoeuvresand will also guarantee a proper tuning of the orbit parameters after the releasefrom the launcher The ground segment (telemetry control and data transfer)could be via non-ESA stations user-operated to save costs

Further resources are indicated in Table 2

Exp Astron

The satellite bus design drivers are nadir-pointing microwave antennasthermal stability for the payload instruments no strong magnetic sources lowmicrovibrations harsh radiation environment due to crossing of the radiationbelts once per orbit power demand by science payload (with the possibility toreduce power demand by stand-by of some components during specific missionphases eg eclipse) mass limit due to required high altitude with low costlaunch

The satellite bus design can be based on established platforms such asFLEXBUS or PRIMA used on several past missions Their design philosophydiffers one having body-mounted solar arrays the other wings The latter canprovide larger amounts of power The vibration level induced by the panelswhen they are fixed and the duration of solar array drive periods per orbitwill have to be considered in the selection The FLEXBUS concept was usedfor a mission already requiring a low level of residual accelerations and theplatform is optimised for this parameter

For most of the satellite components one can consider using compo-nents and electronic units from navigation or communication satellites whichtypically cruise at altitudes between 15000 and 25000 km altitude LISAPathfinder which will cross the radiation belts during spiralling to theLagrange point will face similar problems concerning radiation as EGE as wellas XMM which is in a Molniya-type orbit with similar radiation characteristics

The instruments will therefore be located in the satellite structure such asto realise ionizing radiation shielding by the spacecraft structure and non-sensitive equipment to the maximum extent possible the need for additionalradiation shielding (including trapped proton radiation) for especially sensitiveparts of the instruments must be determined by ground tests It should bementioned that an orbit with higher perigee and less radiation dose is also anoption as it leads to a tolerable reduction of the science goals

The required plusmn3C temperature stability at the interface to the payloadenclosure is feasible The thermal control will employ the usual means (anenclosure with high thermal inertia heaters and a suitable radiator) Thelocation of the radiator will need a special study given the non-standard orbit

Magnetic field present during the mission can affect the properties of theonboard clock Using SPENVIS the magnetic field has been calculated alongthe spacecraft orbit the field intensity presents maxima around 02 G anda minimum value of the order of 10minus3 G As discussed above this is not alimitation to the instrument performance

6 Discussion and conclusion

The accuracy of the measurements depends crucially on the stability andaccuracy of the clocks While the performance of ground clocks (even thoseeventually deployed in new ground stations dedicated to EGE) will likelyimprove significantly from the present level the space clock design will beldquofrozenrdquo approx 10 years before data acquisition Thus their performance

Exp Astron

level will be significantly below the future ground station clock performanceAs specification we have chosen a performance level that has been demon-strated with laboratory breadboards recently We see no fundamental reasonthat this laboratory performance cannot also be achieved in a space ion opticalclock The experience with the PHARAO cold atom clock has shown thatan engineering model of a laser-based cold atom clock has a performancesimilar to that of a laboratory prototype The perturbation of the ion clock fre-quency due to magnetic fields along the orbit can be modelled and suppressedwith shielding and the residual effects measured and compensated for Themoderate temperature stability (plusmn 3C) to be provided in the experimentalbay can be achieved with thermal shields Accurate temperature controlof particular units is done by active stabilization of the units themselves

One important aspect is radiation which may affect all optical and trapcomponents as well as the lasers and reference cavity The radiation envi-ronment on the EGE orbit is significantly harsher than the near-earth ISSorbit on which ACES will operate The radiation effects must therefore becharacterized comprehensively with appropriate tests and then the radiationshield for the EGE instruments designed

In summary the Einstein Gravity Explorer mission (EGE) is a satellitemission aimed at a precise measurement of the properties of space-time usingatomic clocks By testing one of the most fundamental predictions of EinsteinrsquosTheory of General Relativity the gravitational redshift and thereby searchingfor quantum effects in gravity it explores one of the most important frontiers infundamental physics The EGE mission reaches beyond fundamental physicsproviding an important spin-off in geophysics Furthermore the technologicalaspects of the mission are ground-breaking for future time-keeping and ref-erence frame systems in space and thus are also expected to be of wide useto science and technology in general The payload of the mission has beenconceived as a compromise between cost and performance needing to fit thecost envelope of an ESA medium-class mission1

Acknowledgement We thank S Cesare for helpful information

References

1 Will CM The confrontation between general relativity and experiment Living Rev Relativ9 3 (2006)

2 Ni W-T Empirical foundations of the relativistic gravity Int J Mod Phys D 14 901 (2005)doi101142S0218271805007139

1Finally we point out that a spacecraft with a significantly reduced and thus lower-cost payload (noclocks only MOLO FCDP MWL) on a circular orbit of appropriate height could be used for along-duration earth science mission dedicated specifically to the measurement of the geopotential

Exp Astron

nuclear theory (see eg [12 13]) The measurement of this ratio which issensitive to K represents a test of General Relativityrsquos redshift

When the frequencies of two co-located but dissimilar clocks a b (ofnominal frequencies νa0 and νb 0) are compared in-situ at two locations x1 x2

(LPI test) the result is

[va1 (x1)

vb 1 (x1)

] [va2(x2)

vb 2 (x2)

] = 1 +[X

(AXa minus AXb

)K (X i)

](Ui (x2) minus Ui (x1))

c2 (4)

The result of the EGE mission may be given as a limits for (or nonzerovalues of) K(X Earth) for the three fundamental constants X = α meQCDand mqQCD

23 Measurement approaches for the gravitational frequency shift

The gravitational potential experienced by the ground and satellite clocksincludes the contributions from the earth and the sun (the contribution fromthe moon is an order smaller and may be neglected for the purposes of thediscussion but will be taken into account in data analysis)

Nonzero gravitational frequency shift signals (Eq 3) can be obtained inthree ways Method (1) below is the most sensitive and represents a strongreason to employ an optical clock which on short averaging times is ableto provide a higher frequency stability than a microwave clock Method (2)and (3) are less accurate but are complementary in terms of measurementapproach Finally method (3) addresses the solar gravitational frequencyshift while (1) and (2) measure the terrestrial effect The performance of theinstruments assumed in the following is shown in Fig 2

(1) Repeated measurements of satellite clock frequency variation betweenapogee and perigeeIn this measurement approach the difference (or ratio) between earthndashsatellite clock frequency ratio at perigee and at apogee is consideredThis difference is measured at each perigee and apogee passage andaveraged over the mission duration Systematic shifts (if not correlatedwith orbital motion) are expected to average out leading to a gain insensitivity of up to n12 where n = number of orbits This mode usesthe stability properties of the satellite clocks on the timescale fromperigee to apogee rather than their accuracy Assuming a specificationof le 3 times 10minus16 satellite optical clock fractional frequency instability forintegration times between 1000 and 25000 s n sim= 1000 comparisons(ie most orbits of the 3-year mission) and a perigee-apogee potentialdifference (U(apogee) minus U(perigee))c2 = Uc2 sim= 4 times 10minus10 allows a

Exp Astron

Fig 2 Performance of clocks and links on EGE Curve ldquooptical clocks EGErdquo is the specificationCurve ldquooptical clocks EGE (goal)rdquo refers to expected near-future performance of an opticalclock based on neutral atoms ldquoUpgraded PHARAOrdquo refers to the PHARAO microwave clockusing the high-quality microwave-optical local oscillator The microwave link (MWL) performanceexhibits a discontinuity since common-view contact time of ground stations with EGE is limited toseveral h The different slopes arise because on short time scales only white phase noise is relevantwhile on long time scales flicker noise also contributes The reproducibility is taken into account

measurement of the redshift at 25 ppb accuracy The ground clocks towhich the satellite clocks are compared to at perigee and at apogee willbe located in different ground stations The inaccuracy of these groundclocks must therefore be sufficiently smaller than 1 times 10minus17 in order notto contribute to the 25 ppb error Current state-of-the art inaccuracy is2 times 10minus17 and is likely to decrease further

(2) Absolute comparison between ground and satellite clocks with the satel-lite at apogeeHere the terrestrial gravitational potential difference is maximumU(apogee) minus U(perigee)c2 sim= 65 times 10minus10 Assuming a space optical clockinaccuracy of 1 times 10minus16 and a ground clock inaccuracy a factor 3lower (current state-of-the-art) allows a measurement of the redshift at150 ppb This measurement takes advantage of the repetitive nature ofthe orbit only to a small extent and relies on the accuracy of the opticalclocks in particular of the space clocks It is thus complementary to (1)albeit less accurateIn (1) and (2) the dominant contribution to the redshift is from Earthrsquosgravitational potential The contribution from the solar potential isstrongly reduced due to the fact that satellite and earth are an interactingsystem in free-fall toward the Sun [19] and can be modeled

(3) Comparisons between terrestrial clocksThe frequency comparison link to the EGE satellite allows terrestrialclock comparisons by a common-view method whereby the frequency

Exp Astron

signals from two distant ground clocks are sent to the satellite andcompared there The frequency difference between two identical groundclocks is determined if systematic instrumental errors are negligible bythe difference in gravitational potential at the two clock locations Thispotential difference includes a terrestrial and a solar contributionThe solar contribution varies according to the orientation of the clockpair with respect to the Sun with a 1 day-period For two clocks at asurface distance of sim6000 km and suitable orientation with respect tothe sun the normalized modulation amplitude is sim= 4 times 10minus13 This is thesame as the earth potential difference for 4 km height difference It ispossible that ground clocks will have reached an inaccuracy at the levelof 1 times 10minus18 at the time of the mission in 2015 Using the microwave linka comparison of ground clocks over sim3 h (typical common-view duration)with a resolution of 1 times 10minus17 can be performed limited by the instabilityof the link Repeating the clock comparison at each common-view win-dow (n = 1000 comparisons) will allow averaging and a measurementof the solar gravitational shift with fractional inaccuracy of 1 ppm Incomparison the best current results for the solar gravitational frequencyshift are at the few-percent level [20 21] A moderate averaging out ofground clock systematic shifts is assumed as well as modeling of the earthgravitational potential variations on the daily timescale at the appropriatelevel

(4) Comparisons of on-board clocksThe two on-board clocks are continuously compared these data areanalyzed for violation of LPI (Eq 4) Since the measurement is a nulltest the required accuracy of the gravitational potential at the satellitelocation is low and easily satisfied

24 Test of local position invariance and search for neutron scalar charge

The test of LPI is performed in conjunction with the measurement of thegravitational frequency shift

With approach 23 (1) the optical clock will be able to measure K (α Earth)with accuracy 30 ppb (spec)

Combining the results of the microwave and optical clocks will determineK(meQCD Earth) + εK(mqQCD Earth) with accuracy 85 ppb (spec)where ε asymp minus01

Simultaneously the on-board comparison 23 (4) yields the combination19 K(α Earth) +K(meQCD Earth) + εK(mqQCD Earth) with accuracy85 ppb (spec)

Note that these measurements are fully complementary to current andfuture terrestrial clock-clock comparisons which are sensitive to K(X Sun)and can be combined with the latter to determine or set limits to the neutronscalar charge

Exp Astron

25 Theoretical analysis of frequency comparison signals

The on-board clock-clock comparison signals can be analyzed straightfor-wardly in terms of a violation of LPI since signal propagation effects areabsent

The determination of the gravitational frequency dilation via ground-satellite clock comparisons is more complex (Eq 1 is strongly simplified) andrequires an accurate treatment of all relevant relativistic effects Gravitationin the solar system is described in the barycentric (BCRS) and geocentric(GCRS) non-rotating celestial reference systems by means of post-Newtoniansolutions of Einsteinrsquos equations for the metric tensor in harmonic gaugescodified in the conventions IAU2000 [22] In these conventions the relativisticstructure of Newton potential (order 1c2) and gravitomagnetic (order 1c3)

potentials are given For the Newton potential the multipolar expansion ofthe geopotential determined by gravity mapping is used The post-Newtonianeffects of the metric (described in terms of dimensionless parameters β and γ )

can be included Orbits of satellites are evaluated in the GCRS at the levelof a few cm by means of the Einstein-Infeld-Hoffmann equations at the order1c2 [23] using the relativistic Newton potential Instead the trajectories of theground clocks in the GCRS are evaluated from their positions fixed onthe Earth crust (ITRS International Terrestrial Reference System) by usingthe non-relativistic IERS2003 conventions [24 25]

For the propagation of light rays (here radio signals) between the satelliteand the ground stations one uses the null geodesics of the post-Newtoniansolution in the GCRS The timefrequency transfer properties have beentheoretically evaluated for an axisymmetric rotating body [26ndash29] at the order1c4 (ie with Newton and gravitomagnetic potentials developed in multipolarexpansions) These results have been developed for use in the ACES missionThe relative frequency dilation can be expressed as

ν

ν= 1 + 1

cL1 +

4sum

n = 2

1

cn(Ln + Gn)

where Ln describes special-relativistic Doppler effects and Gn the general-relativistic effects These quantities are time-dependent due to the orbitalmotion For concreteness we report the orders of the terms The kinematicaleffects are |L1c| lt 1 times 10minus5 |L2c2| lt 2 times 10minus10 |L3c3| lt 3 times 10minus15 |L4c4| lt

7 times 10minus20 The Newtonian contributions are G2 = G(M)2 + G(J2)

2 + G(J4)2 + G(J6)

2(M is the mass monopole Jn are the lowest multipoles) with estimates|G(M)

2 c2| lt 3 times 10minus11 |G(J2)2 c2| lt 2 times 10minus13 |G(J4)

2 c2| lt 3 times 10minus16 |G(J6)2 c2| lt

1 times 10minus16 For G3 = G(M)3 + G(J2)

3 the estimates are |G(M)3 c3| lt 2 times 10minus14

|G(J2)3 c3| lt 1 times 10minus18 Finally the G4 = G(M)

4 + G(S)4 contributions (G(S)

4 is thelowest gravitomagnetic effect) are of order |G(M)

4 c4| lt 1 times 10minus19 |G(S)4 c4| lt

1 times 10minus22 and thus not relevant for EGE

Exp Astron

The available theory is fairly complete for the data analysis in mode 23 (1)described below For modes 23(2 3) the theory will need to be generalizedto include the effect of the gravitational potential at the location of the groundclock(s) (including time-varying effects such as the tides) and of the solar andlunar potentials

The uncertainties of the various contributions above are minimized by ausing precise EGE orbit data obtained by satellite tracking and by usingcurrent and future accurate earth gravity information Required orbit positionknowledge is at 1 cm level near perigee and sim50 cm near apogee reachablealready today with the proposed orbitography approach described below Thevalidity of the special-relativistic Doppler shift has been and will continue to beindependently verified with increased accuracy by spectroscopy of relativisticatomic ion beams [30] Potential violations of this aspect of Lorentz Invari-ance are expected to be sufficiently bounded so as not to affect the signalinterpretation

26 Local Lorentz invariance tests

There has been an explosion of interest in Local Lorentz invariance in re-cent years [31 32] with numerous astronomical studies and high-precisionexperiments applied to search for violations of Lorentz invariance Signifi-cant developments have also occurred on the theoretical side and a theory(standard model extension SME [33]) has been worked out that provides aunified framework for describing and analyzing Lorentz Invariance violationsin a variety of systems Two aspects of Lorentz Invariance can be tested withEGE

261 Independence of the speed of light from the laboratory velocity

A possible dependence of the speed of light on the speed v of the laboratorycan be modeled according to the MansourindashSexl test theory [34] by

c (v) = c0(1 + Bv2

c2

)

Here v 377 kms is the velocity with respect to the cosmic microwavebackground the cosmologically preferred frame and B is a combinationof parameters describing deviations from the usual Lorentz transformationformulas B = 0 if Lorentz Invariance holds For a terrestrial experiment therotation of the Earth around its axis modulates v with a 300 ms amplitude Adependence c(v) can be searched for by measuring the frequency differencebetween a resonance frequency of a highly stable optical cavity (cavity fre-quencies are proportional to c) and an optical clock (Kennedy-Thorndike-typeexperiment [35]) and determining the modulation of the frequency differencecorrelated with the modulation of v The advantages of a space experimentare the high orbital velocity and strongly reduced cavity deformation thanksto microgravity [36 37] On the proposed elliptic orbit v varies between+4 and minus4 kms over approximately 1 h This variation is 13 times larger

Exp Astron

than for the case of a terrestrial experiment The shorter time scale is alsoadvantageous since the drift of the cavity is more predictable In the EGEmission the reference cavity of the clock laser is used The large numberof orbits improves the statistics An improvement of the accuracy of B by afactor 20 compared to the best current terrestrial results is realistic The dataanalysis can also be performed in the framework of the SME theory takinginto account complementary bounds obtained from terrestrial experimentsand astrophysical observations

262 Independence of Zeeman splitting frequency on the directionof the magnetic field

One class of tests of Lorentz Invariance consists in measuring the frequencysplitting between two levels of a quantum system induced by a static magneticfield as a function of orientation of the field direction with respect to thestars (HughesndashDrever-type experiments [31]) This class addresses the so-called matter sector and is complementary to combined-photonmatter-sectortests such as the above The most precise experiments are performed withatomic clocks eg masers or cold atom clocks [38] The EGE instruments allowsuch experiments since magnetic fields are applied continuously in the clockor repeatedly for calibration Compared to terrestrial experiments the highorbital velocity of EGE could lead to an improvement by a factor sim20

27 Application to geophysics

Using GPS coordinates of a point on the Earth can be obtained with aninaccuracy well below 1 cm in a well defined international terrestrial referencesystem The coordinates are purely geometrical and do not contain any gravityinformation A local gravitational potential UB is determined with respect toa reference gravitational potential U A by ldquolevellingrdquo the surface of the Earthie sequentially measuring gravity g and height differences every ca 50 m

UB minus U A = minusBint

A

g (r) middot d n

where n is the difference vector between subsequent locations The disadvan-tage of measuring height differences by the levelling method is a random walkeffect of accumulated errors

In the classical definition a geoid is defined as the particular equipotentialsurface nearest to mean sea level In these terms the geoid serves as areference surface for measuring the height and also to define a referencefor the gravitational potential Such a vertical reference frame historicallywas realized for a country or several countries by determining the mean ofsea level observations at tide gauge stations taken over long period of timeHowever modern satellite altimetry missions such as TOPEXPOSEIDON orENVISAT show that departure of the mean sea level from an equipotential

Exp Astron

surface may reach up to several meters on the global scale see eg [39]Therefore height systems based on different tide gauge stations may differin the realisation of the geoid and may differ with respect to the reference byseveral meters

The best gravity field models obtained from satellite data (eg missionGRACE) have reached a precision of about 1 cm over sim250 km half wave-length [40] Satellite data also reveals changes in time [41] However for typicalEarth topography with height variations of eg 1000 m over 30 km horizontaldistance one may expect variations in the geoid of about 80 cm Such ahigh-spatial-frequency signal in geoid variations cannot be detected by spacegravity missions and requires a combination of satellite and terrestrial gravitymeasurements like gravity anomalies deflections of vertical and GPSlevellingpoints The best combined global gravity field models are provided up to de-gree and order 360 in terms of spherical harmonic representation correspond-ing to a half-wavelength of about 55 km However in combining the satelliteand terrestrial gravity field measurements the problem of terrestrial data givenin different height systems between continents and different countries remainsand has to be tied to satellite measurements

The geodetic scientific community is currently establishing a Global Geo-detic Observing System (GGOS) [42] Its objectives are the early detectionof natural hazards the measurement of temporal changes of land ice andocean surfaces as well as the monitoring of mass transport processes in theEarth system Global change processes are small and therefore difficult toquantify Therefore the required precision relative to the Earthrsquos dimensionis 1 ppb GGOS will be established by the combination of geodetic spacetechniques (GPS Laser-Ranging very large baseline interferometry) andrealized by a very large number of terrestrial and space-borne observatoriesThe purely geometric terrestrial 3D coordinate system is in good shape andfully operational It has to be complemented however with a globally uniformheight system of similar precision The current precision level of regionalheight systems in terms of gravity potential differences is in the order of1 m2s2 (10 cm) with inconsistencies between these various systems up toseveral 10 m2s2 (several meters) The actual requirement in the context ofGGOS is 01 m2s2 (1 cm) with the need of a permanent ie dynamical controlThis requirement of high precision height control comes from the need tounderstand on a global scale processes such as sea level change global andcoastal dynamics of ocean circulation ice melting glacial isostatic adjustmentand land subsidence as well the interaction of these processes Only by meansof monitoring in terms of gravity potential changes at the above level ofprecision the change of ocean level can be understood as a global phenomenonand purely geometric height changes be complemented by information aboutthe associated density or mass changes

The geoid can also be defined in a relativistic way [43] as the surface whereaccurate clocks run with the same rate and where the surface is nearest tomean sea level The relation between the differences in the clock frequenciesand the gravitational potential is given in simplified form by Eq 1 At present

Exp Astron

there is no operational way to compare frequencies of the already availableoptical clocks on the global scale at the same level as their accuracy wouldallow With EGE and by then improved ground optical clocks it will becomepossible to obtain gravitational potential differences on a global scale by com-paring frequencies This would be a significant new dimension to gravity fielddetermination since such observables are given on the global scale and providein situ local gravity information at the same time EGE will allow clock-basedgravitational potential mapping with the same payload instruments used forthe fundamental physics experiments The on-board clocks are actually notrequired for this purpose MOLO MWL and FCDP suffice Thus EGE willallow establishing a global reference frame for the Earth gravitational fieldwith accuracy in the order of a few cm in terms of the geoid heights Thisassumes that by the time EGE is operational (mobile) ground clocks withfractional inaccuracy at the level of 10minus18 are available the measurementsare limited by MWL noise and thus require a long integration times to matchthe expected ground clock performance Such a reference frame will serve asreference for all gravity field modeling and clocks will be used to define theinternational atomic time scale and the Earth gravitational potential at thesame time

In summary with optical clocks geodesy will undergo a second revolutionafter geometry now being measured by clock-based GNSS systems also physi-cal heights and potential will be measured by (mobile) optical clocks using thegravitational redshift effect

28 Application to frequency standards and terrestrial fundamentalphysics studies

As has been described repeating ground clock comparisons and integratingover several months will provide global ground clock frequency comparisonsat the level of 1 times 10minus18 more than two orders of magnitude below the currentGPS and two-way satellite time and frequency transfer methods In the timeand frequency community the availability of such a global high performancemicrowave link will accelerate the process leading to a redefinition of the SIsecond based on clocks operating in the optical domain

The global character of the ground clock comparisons made possible byEGE can also contribute to the search for a time-variation of the fundamentalconstants

The comparisons do not require availability of the on-board clocks The riskof this type of measurements is therefore reduced Moreover the lifetime ofmany subunits required for this measurement (MWL and FCDP see below)can exceed 10 years (from previous experience with similar devices) andthus offers a motivation to increase the mission duration beyond the nominal3 years

Further applications to metrology are listed under the secondary goals inSection 21 above

Exp Astron

3 Mission profile

31 Orbit

The orbit must satisfy the following requirements

Primary 1 Large gravitational potential difference between apogee andperigee

2 sufficient contact time with at least two ground stations atperigee

Secondary Simultaneous visibility of satellite from distant (inter-continental) ground stations for several hours

To satisfy the primary objective the orbit must be highly elliptic The perigeeshould be low A very low perigee means a short contact time which decreasesthe accuracy of the frequency measurement A compromise is a perigeealtitude of sim2500 km For a sufficiently high apogee this yields a U of 60of the theoretical maximum between the earth surface and infinity

The baseline scenario is an orbit of approx 12 h period with the followingproperties

Apogee Altitude 37856 km (minus 10 deg latitude one over Atlantic andPacific resp)

Perigee Altitude 2500 km (+ 10 deg latitude one over Guatemala andMalaysia resp)

Inclination 634 argument of perigee 170RAAN 25 True Anomaly 220Apogee drift rate 70 kmday perigee drift rate sim 14 kmday

True anomaly and right ascension of the ascending node (RAAN) wereoptimized to minimize orbital precession Figure 3 shows the ground track andthe ground coverage at one of the perigees Common-view contact time forstations in Europe is about 6 h while between a station in Europe and in theUSA it is about 35 h Frequency comparison windows between a given groundstation and EGE repeat every 24 h

A final selection of the orbit requires an in-depth study of the costs of thepropellant and maneuvers as well as of the setting up of new ground stations

32 Ground segment

The mission requires two separate ground segments the science groundsegment and the standard satellite control segment which supports telemetryand telecommand (TMTC)

Exp Astron

Fig 3 Baseline EGE orbit One (out of two) apogee and one (out of two) perigee ground pointsare shown The ground coverage at one of the perigees is shown with circles at elevation anglesteps of 10 starting from 10

321 Science ground segment

As described above one part of the mission is to perform frequency com-parisons between ground-to-space and ground-to-ground (common-view andnon-common view) The use of a two-way microwave link (MWL) systemhas the advantage that a precise orbit or the precise relative velocity of thesatellite must not be known [4] The orbit determination requirements fora two-way system are significantly relaxed compared to the requirements tomodel the relativistic effects on the clocks The frequency comparison resultsare available both on ground and on the satellite

For ground-to-satellite clock comparisons five ground stations arefavourable two capable of viewing perigee three for viewing apogee Thisallows for the classic three-clock configuration where the performance of eachclock can be obtained individually The perigee stations will be observing thesatellite for sim25 min every perigee passage for the duration of the missionThe apogee stations will observe for sim3 h every apogee passage Of coursea larger number of stations as well as longer observation time are usefulfor robustness redundancy and orbit determination The stations could beinstalled on a temporary basis for the duration of the mission

The stations must have access to optical clocks of performance at least ashigh as todayrsquos state-of-the-art ground clocks but preferably higher Theywill include an optical frequency comb by means of which the optical clockfrequency is transformed to a radio frequency for comparison with the EGEclocks High-accuracy fiber-optic link technology already existing today allowsdistances of sim100 km between the ground station and the correspondingoptical clock allowing a certain freedom in the choice of the clock locationWith further increase of the fiber-optic link range currently being investigated

Exp Astron

the area over which the ground station can be positioned relative to the opticalclock will increase This option is of importance for the implementation of ageophysics measurement campaign for which a large number of clock locationsis of interest

The above ground stations will be complemented by some 15 additionalground terminals provided by the world-wide laboratories equipped with highperformance primary clocks at the time of the mission These will participate inorbit determination to 1 cm accuracy along the full orbit which can be achievedwith non-optical atomic clocks The EGE MWL is fully compatible to plannedACES ground terminals which might reduce the number of required stations

With a larger number of transportable optical clock ground stations de-ployed and moved across the continents a geophysics campaign could beimplemented to map the gravitational potential

Ground station equipment MWL uses dedicated small transportable termi-nals which can be easily installed at any location of interest Each terminalconsists of a steerable dish antenna of 1 m diameter with pointing accuracy of01 using a transmitted power of 5 W per band The high directivity antenna(beam aperture Ka-band 07 Ku-band 14 S-band 20) avoids multi-pathfrom nearby objects Antenna noise temperatures are Ka-band 450 K Ku-band 340 K Operation is at elevations above 10 and measurements areperformed above 20 For ranging purposes terminal location is determined byGPS surveying Each ground station computes its own ionospheric correctionsbased on a triple-band receiver The terminal has a built-in delay monitorwhich allows calibrated ranging during several months once its bias has beendetermined by laser ranging Its M and C interface is via local area networkto either a userrsquos computer or directly to the science network control centreby remote MampC The stationrsquos operation is fully automatic and unattendedSchedule and orbital data are received via LAN once per day Signal acquisi-tion is fully autonomous Data transmission is after each pass The terminal hasan uplink telecommand capability of 1 kBits and a downlink data capacity ofup to 10 kBitss For the local user there is a visual interface providing quick-look data from real-time comparison to the spacecraft clocks to verify theirgood health The final full performance product is calculated at the sciencedata centre

Local metrological data (temperature pressure humidity) are recorded tocorrect for tropospheric delay of the ranging data The main advantage ofMWL terminals compared to laser based systems is their all-weather capabil-ity although microwaves suffer higher propagation errors in the ionosphereand troposphere

The EGE ground terminal will profit from the ERS-2 mission PRARE(precise range and range rate equipment) [45] experiments whose terminalshad similar design and have been operated under extreme climatic conditionsincl Antarctica and high temperature regions from existing designs fromthe space segment (mission ACES) and no further special developments aredeemed necessary Cost per EGE master station is modest

Exp Astron

Satellite tracking via laser ranging Laser ranging is alternative to MWL rang-ing and also serves for absolute calibration of the latter For this purpose anon-board corner cube reflector (CCR) is provided Ranging will be done withthe International Laser Ranging Service Ideally the satellite will be rangedat every perigee passage and every apogee passage simultaneously with thefrequency comparison procedures Todayrsquos ranging precision is sufficient tosatisfy the science goals For apogee observations in the northern hemispherea large number of laser ranging stations is available The number of stationscapable of ranging at or near perigee in the southern hemisphere is six andthus is sufficient even considering unfavourable weather conditions at someof these The MWL ground stations are able to range even under thoseconditions

322 Control ground segment

Average data transfer rate is estimated at 300 MBday (approx 40 kBs onaverage) Data will be stored on board (2 GB capacity required) and will betransferred to a receiving station on the Earth once a day

XPLC

incl Data StorageIF to TMTC Link

HK amp TMTC

to ground

Opt Clock

AOM

MOLO

PHARAO

FCDP

10 GHz

100 MHz

250 MHz

GNSS Rx

MWLTampF Link

uplink Data

UTC(k)

TampF Space vs Ground

TM from GS

Tim

e S

cale

Orbit GNSS Time UTC

TampF Space vs Ground

MampC Time Scale

MampC

MampC

MampC

MampC

Mamp

C

PHARAO vs Opt Clock

PHARAO vs Opt Clock

10 GHz amp Epoch

10 MHz

amp Epoch

Epoch

10 MHz

amp Epoch

Opt Link

Fig 4 Schematic of the payload showing subunits and their signals The results of the frequencycomparison between the two on-board clocks (optical clock and microwave clock PHARAO) areobtained from a combination of the frequency values applied to the frequency shifters AOM (seeFigs 7 and 8) and those output from the frequency control and distribution package (FCDP)as indicated by the arrows ldquoPHARAO vs Opt Clockrdquo The satellite clock to ground clockcomparison results are output from the MWL XPLC is the on-board computer and is involvedin generation of signals for instrument control The optical link shown dashed is a possible add-onproviding an alternative to the MWL GS ground stations MampC monitoring and control HKhousekeeping TampF time and frequency

Exp Astron

4 Main payload instruments

41 Overview

The satellite payload consists of a pair of cold-atom clocks a microwave-optical local oscillator (MOLO) an on-board frequency distribution unit(FCDP) a satellite-to-ground clock comparison unit (MWL) and the onboardcomputercontrol system (XPLC) see Fig 4 Auxiliary units are the posi-tion determination GPSGalileo receiver (GNSS Rx) attitude monitor (startracker not shown) and the corner cube reflector array (CCR not shown)

The science goals require clocks of very high frequency stability on thetimescales of half the orbital period (6 h) as well as perigee passage (sim1000 s)see Fig 2 Only clocks based on cold atoms are suited for the science goalsTable 1 reports values for a subset of atomic clocks that have been investigatednot including clocks that require cryogenic operation or are too complex foruse in an medium-class mission The choice of a clock pair rather than a singleone fulfils several requirements it addresses more science goals improves thevalidity of some science results and it enables calibration accuracy evaluationand redundancy The validity of the science results is enhanced since theclocks differ not only in the employed atoms but also in terms of havingdifferent sensitivity to perturbations by electric and magnetic fields electronicserrors and misalignments A comparison of the results arising from each clockallows for an understanding of measurement errors Both the clocksrsquo and thefrequency comparison unitrsquos performances will be determined after launchand regularly verified during the course of the mission The only reliable

Table 1 Properties of different clock candidates for EGE

Clock Clock Fractional σ = Fractional Sensitivity Figure of Sensitivityatom frequency inaccuracy instability to α Aα merit |Aα| to meq

at 1000 s σ [1016] QCD

Sr+ 445 THz 3 times 10minus17 (p) 5 times 10minus17 (p) 042 08 (p) ndashYb+ (1) 688 THz 3 times 10minus16 (d) 3 times 10minus16 (d) 088 029 (d) ndash

3 times 10minus17 (p) 1 (p)Yb+ (2) 642 THz lt 1 times 10minus17 (p) 1 times 10minus17 (p) minus53 53 (p) ndashSr 435 THz 1 times 10minus16 (d) 1 times 10minus16 (d) 009 009 (d) ndash

lt 1 times 10minus17 (p) 09 (p)Yb 519 THz 7 times 10minus16 (d) 1 times 10minus16 (d) 031 03 (d) ndash

lt 1 times 10minus17 (p) 1 times 10minus17 (p) 31 (p)Cs 91 GHz 2 times 10minus16 (d) 1 times 10minus15 (d) 275 027 (d) 1Yb+ 126 GHz 1 times 10minus15 (p) 1 times 10minus15 (p) 34 034 (p) 1

A relevant figure of merit is the ratio of |AX | and fractional frequency instability σ on the perigeevisibility timescale The Cs clock is the only high performance clock with adequate sensitivity tothe particle masses me mq The indicated Cs clock performance was achieved using a microwavelocal oscillator based on a cryogenic oscillator On EGE the latter is to be replaced by an opticallocal oscillator-based device (MOLO) with similar short-term performance but without the needto operate at cryogenic temperature The clock instability at 10000 s is typically similar to theindicated inaccuracy(p) projected (d) demonstrated (1) quadropole (2) octupole

Exp Astron

way to achieve this is by having at least two clocks on board The on-boardclock-clock comparison provides performance evaluation independent of theground-satellite comparison unit Once the satellite clock performances areestablished the link performance can be established as well

The clocks are similar in basic structure but differ in the detailed implemen-tation They consist of

a source of atomsa preparation subunit which cools the atoms to near-standstill (millikelvintemperature or below)a trapping subunit that confines the atom(s) to a tiny volumean interrogation subunita clock control electronics package

The clocks share a common central unit the optical-microwave local oscil-lator which serves as a flywheel oscillator with outstanding stability on shorttimescales (01ndash10 s) for both clocks It consists of a laser stabilized to an ultra-low loss optical resonator made of ultra-low-expansion glass A microwavesignal is derived from the laser by phase-locking a frequency comb to the laserwave and using a multiple of the repetition rate

While the subunits of any optical clock have a similar technological basis(optics electrooptics acoustooptics thermal control vacuum systems) theydiffer in their specifications Depending on the choice of the atom species de-livery confinement and interrogation methods clocks of different perfor-mance result A number of laboratory optical clock demonstrators have beendeveloped worldwide This ensures availability of the know-how for the indus-trial implementation of the flight models

The first clock is a microwave clock based on the interrogation of slowlymoving ensembles of cold Cs atoms It is essentially based on the PHARAOclock [5 6] developed to engineering model level by CNES in the frame-work of the ESA ISS mission ACES In EGE this clock is upgraded usingthe microwave-optical local oscillator This improves significantly the clockperformance The specification is a relative instability (Allan deviation) of3 times 10minus14 (τ s)minus12 ie at the integration time τ = 1000 s it drops to 1 times 10minus15and to 3 times 10minus16 at 10000 s While this specification is at least a factor 3 lessstringent than for the optical clocks the significantly lower cost of obtaining aflight model and the sensitivity of its frequency to the particle masses makes ita suitable choice

The optical clock baseline instrument is a single-ion optical clock Amongvarious suitable ion species (Hg+ Al+ Sr+ Yb+) [44ndash47] the Yb+ ion hasbeen selected Its breadboard performance is established at the instability levelof 3 times 10minus16 over 1000 s it exhibits a large sensitivity coefficient Aα hassmall sensitivity to magnetic fields and exhibits significant potential for furtherimprovements (use of the octupole transition with sim6 times larger Aα andlower overall instability [48] see below) A trapped Yb+ ion is also a veryrobust system it has been demonstrated that a single Yb+ ion can remaintrapped uninterruptedly for many months Finally Yb+ is the only ion with

Exp Astron

which optical clocks demonstrators have been operated in not a single but twomajor metrology laboratories

Another attractive optical clock is based on ensembles of cold neutral Sror Yb atoms trapped in an optical lattice produced by laser beams This clocktype has a lower instability than a single-ion clock thanks to the large numberof atoms (sim105) used Its development is in progress in several laboratoriesin the world (in part supported by space agencies) and has to date reached6 times 10minus17 instability at 5000 s and 1 times 10minus16 uncertainty [49 50] with potentialfor even higher performance However it is more complex and costly thanthe Yb+ ion clock since the cold atom preparation and trapping subunitsinclude more components require higher laser powers and the respectiveoperating procedures are more complex It is therefore not considered for thepresent mission designed to be developed in the very near future with limitedresources but is an interesting candidate for missions without such constraintsIts estimated physical parameters are indicated in Table 2 for comparison

42 Main instruments

421 Microwave clock

The PHARAO cold atom clock is one of the two atomic clocks of the spacemission ACES managed by ESA It is proposed here to include a secondversion of this instrument as its costperformance ratio is very attractive forthe EGE scientific objectives Its concept is very similar to ground basedatomic fountains but with the major difference of zero-g operation Atomsslowly launched in free flight cross two microwave fields tuned to the transitionbetween the two hyperfine levels of the cesium ground state The interrogationmethod based on two separate oscillating fields (Ramsey scheme) allowsthe detection of an atomic line whose width is inversely proportional to thetransit time between the two microwave cavities The resonant microwavefield at 9192631770 GHz (SI definition of the second) is synthesized startingfrom an ultrastable quartz oscillator (USO) in the FCDP phaselocked to themicrowave output of the microwave-optical local oscillator and stabilized tothe clock line using the error signal generated by the cesium resonator In thisway the intrinsic qualities of the cesium hyperfine transition both in terms ofaccuracy and frequency stability are transferred to the USO In a microgravityenvironment the velocity of atoms along the ballistic trajectories is constantand can be changed continuously over almost two orders of magnitude (5ndash500 cms) Differently from atomic fountain clocks presently operated onground very long interaction times (up to few seconds) will be possible whilekeeping reasonable the size of the instrument

The instrument consists of 4 main elements a laser source which providesthe light required to cool and detect the cesium atoms (Fig 5 left) a frequencychain which generates the 92 GHz microwave signal a cesium vacuum tubewhere the interaction between the microwave field and the atoms occurs

Exp Astron

Table 2 Resources and technology readiness level (TRL) of the EGE spacecraft

Unit Volume Mass Power TRL Comment[l] [kg] [W]

Ion clock 100 70 60Atom source 4Atom preparation 6Atom trap 4Atom interrogation 6Clock control 6

Microwave clock 200 91 110 8 TRL by end 2008Local oscillator

Stabilized laser 18 15 23 6Frequency comb 20 16 38 6

Microwave link 18 18 55 8FCDP 7 8 12 8GPS receiver 2 5 7 9CCR 2 2 0 9Payload sum 367 225 305Thermal control 14 17 30 8OBDH 20 24 38 9Star tracker 1 1 8 9Radiation shield 10 40 0 9Support structure 170 204 0 9Power incl solar arr 50 58 10 9Propulsion 21 25 20 9AOCS 36 43 82 9TMTC 8 10 15 9Satellite bus sum 330 422 203 Flexbus conceptPayload plus bus incl 20 margin 8364 7764 6096Propellant 40Launch adapter 100SC wet masspower incl Margins 9164Lattice clock 135 105 152 Alternative clock

Atom source 4Atom preparation 6Atom trap 4Atom interrogation 6Clock control 6

Also included are the resources for an optical lattice clock based on neutral Strontium atoms

(Fig 5 right) and a computer The engineering model of PHARAO is fullyassembled at CNES Toulouse and undergoes functional and performancetests until the end of 2008 The operating temperature range is 10ndash33 andnon-operating temperatures are minus45 +60 The cesium tube consists in aUHV chamber pumped by getters and a 3 ls ion pump a cesium reservoira microwave cavity coils to provide a uniform magnetic field and three layersof magnetic shields Vacuum windows with fiber-optics collimators enable thetransmission of the cooling and detection light onto the cesium atoms

In autonomous mode PHARAO uses the USO as interrogation oscillatorand will provide a clock signal with fractional frequency stability below1 times 10minus13 (τs) minus12 and inaccuracy near 1 times 10minus16 When using the microwave-

Exp Astron

Fig 5 Left The PHARAO laser source engineering model with cover removed Dimensionsare 530 times 350 times 150 mm3 and the mass is 200 kg Ten polarization maintaining optical fibers(yellow) guide the laser beams to the cesium tube All diode lasers are mounted on Peltier coolersfor temperature regulation within 2 mK (courtesy EADS SODERN) Right The PHARAOengineering model cesium tube without the two external magnetic shields The volume is 990 times336 times 444 mm3 and the total mass is 44 kg

optical local oscillator PHARAO will provide a reduced frequency instabilityof 3 times 10minus14 (τs) minus12 This corresponds to 1 times 10minus15 at 1000 s and 3 times 10minus16

at 104 s

422 Ion optical clock

The 171Yb+ ion clock shown in Figs 6 and 7 provides an atomic frequencyreference based on a single 171Yb+ ion confined within an RF electric end-captrap (similar to a Paul trap) within an ultra high vacuum (UHV) chamber andlaser cooled to sim1 mK close to the Doppler cooling limit As a result the ionrsquosfirst order Doppler motion is completely removed and the ion experienceslittle collisional perturbation from its environment Under these conditionsthe narrow linewidth 2S12 (F = 0 mF = 0) minus 2 D32 (F = 2 mF = 0) quadrupole

Fig 6 Left Simplified level scheme of 171Yb+ showing relevant cooling (blue) auxiliary (red)quadrupole clock (green) and octupole clock (dark blue) transitions δνnat is the natural linewidthof the clock transition Middle close-up of Paul-type RF trap (PTB Braunschweig) the ion istrapped in the hole (diameter 14 mm) Right Complete ion trap with vacuum flange height 10 cm(NPL Teddington)

Exp Astron

935 nm D32-state repumper laser

638 nm F72-state repumper laser

739 nm extended cavity diode laser

AOM

PMT

Ion trap UHV chamber

vacuum pumps

Fibre-free space interface

Mu-metal shielding

EC

T

B

Magnetic field

Temperature

Ion fluorescence Processor

Clock laser frequency step and control

4355 nm

Frequengy doubling II

3695 nm cooling wave

Low-drift cavity

Low-drift cavity 171Yb+ 435 nm ion clock unit

876 nm clock laser wave from MOLO

Frequengy doubling

Fig 7 Simplified schematic of the single-ion optical clock unit AOM acousto-optic frequencyshifter PMT photomultiplier tube EC endcap T temperature stabilization system B magneticfield

clock transition at 4355 nm can be interrogated by spectrally narrow and stableclock laser light A diagram of the simplified energy level scheme for 171Yb+showing the relevant cooling auxiliary and clock transitions is given in Fig 6All these wavelengths are provided by diode lasers operated in fundamentalor frequency doubled mode The clock laser light is obtained from the 871 nmwave provided by the MOLO unit (see Section 423) after frequency-doublingin single pass by a KNbO3 nonlinear-optical crystal to 4355 nm The clocktransition spectral profile is observed by stepping the clock laser frequency bymeans of an acousto-optic frequency shifter (AOM in Fig 7) and recordingthe statistics of the quantum jumps in cooling laser induced fluorescence asa function of this frequency Under clock operation the AOM repeatedlysteps back and forward between the half-intensity points on the transitionprofile monitoring the signal imbalance between these points Any detectedimbalance is servo-corrected to zero by feedback to the AOM

In detail the optical clock architecture will comprise (Fig 7)

bull an RF end-cap trap for ionising and confining a single 171Yb+ ion withinan ultra-high vacuum chamber pumped by a small ion pump and non-evaporable getter pump The trap is driven by means of an ac voltage ofa few hundred volts at a drive frequency of sim10 MHz A small oven withseveral mg of 171Yb isotope is heated to provide a low flux of Yb atomsfrom which single ions can be ionised by electron bombardment from ahot wire filament within the trap potential well Low voltage dc is applied

Exp Astron

to additional electrodes to position the ion precisely at trap potentialcentre in order to minimise the ionrsquos micromotion at the drive frequencyThis needs automatic periodic monitoring and correction especially in theperiod after re-loading an ion into the trap External magnetic field coilsin three orthogonal axes allow the nulling of external fields and settingof a fixed field of sim1 μT The trap and coils are surrounded by mu-metalshielding to minimise external field changes during the orbit A level oftemperature control of 1 K of the region surrounding the trap is requiredfor maintaining the blackbody-induced frequency shift uncertainty withtemperature at the 10minus17 level

bull a laser platform to provide frequency doubling of an extended cavity739 nm diode laser for Doppler cooling of the ion on the 2S12(F = 1) minus2 P12 (F = 0) dipole transition at 3695 nm Light from an amplified diodelaser device frequency-doubled in a periodically poled LiTaO3 crystal insingle-pass will be used Typical laser powers used for driving the coolingtransition below saturation are about 2 μW for a beam waist in the trapof about 50 μm By modulation of the injection current of the 739 nmdiode laser at 147 GHz a sideband is generated that excites the 2S12 (F =0) rarr2 P12 (F = 1) ldquorepumperrdquo transition in order to avoid optical pump-ing between the ground hyperfine states

bull The platform also houses the auxiliary lasers at 935 and 638 nm The935 nm diode provides repumping of the ion from the 2 D32(F = 1)metastable level after occasional branching decays to this level during thecooling sequence The 638 nm diode allows fast recovery of the ion fromthe very-long-lived 2F72 metastable state after very occasional collisionaldecay to that state Currently extended cavity lasers are forseen butdistributed-feedback lasers may be available in the near future

bull a high NA lens imaging system and photomultiplier detection system torecord the statistics of 369 nm fluorescence quantum jumps as a functionof clock laser frequency step

bull a fibre system to deliver the various cooling auxiliary and clock light fromsource to trap making use of achromatic doublets where necessary at thefibre-free space interface for launching into the trap

bull a monitoring and control processor which provides the primary coolingand clock laser pulse magnetic field and detection sequencing to observeand lock to the ion clock transition frequency The processor also monitorsfrequency and amplitude data necessary to determine normal laser andion operational conditions and initiate resetting and recovery algorithmswhere necessary and laser unit failure The actual magnetic field presenton the atomic volume is determined and actively stabilized so that aninaccuracy at the level of 1 times 10minus17 can be achieved

bull a redundancy level of at least two units for cooling clock and repumperlasers plus similar for frequency doubling crystals All redundancy unitsfor each wavelength will be fibre multiplexed as standard allowing redun-dant unit activation on determination of prior unit failure

Exp Astron

For an optimized quantum-projection-noise-limited performance of thesingle-ion clock Ramsey interrogation with a cycle time equal to the lifetimeof the metastable D32 level of 50 ms will be performed leading to an Allandeviation of about 27 times 10minus15 (τ s)minus12 This is sim45 times 10minus17 at 1 h satisfyingrequired stabilities for the science objectives Current laboratory results are astability of 1 times 10minus15 at 30 s averaging down to 3 times 10minus16 at 1000 s (PTBGermany) Also comparisons between two independent traps showed agree-ment at the level of several parts in 10minus16 over a series of eight measurementruns

The 171Yb+ ion clock offers a second option namely the 2S12ndash2 F72 octu-pole transition at 476 nm (see Table 1 and Fig 6) [48] It has an extremelylong upper state lifetime low electric quadrupole and second order Zeemansystematic shifts and therefore is a candidate for an even higher performanceclock if a clock laser system with lower instability than the above is providedApart from the clock laser it shares the laser system of the quadrupoleclock Thus with moderate additional resources the Yb+ ion apparatus couldprovide two optical clocks in a single package and specifically perform directfrequency comparisons between the quadrupole and the octupole transitionwith a large sensitivity Aα

The ion clock will operate under microprocessor control providing thecontrol algorithms for the integrated pulse sequencing for cooling repumpingclock interrogation magnetic field switching cooling fluorescence monitoringThe controller will also provide error flags for critical processes and systemreset and correction where necessary This will include ion re-loading andmicromotion reduction algorithms Automatic monitoring of laser power andspectral quality with signal re-optimisation or laser failure determination isrequired Redundant laser units will be pre-aligned and multiplexed into thefibre delivery to the trap so that redundant units can be readily activated in thecase of diode laser failure

423 Microwave-optical local oscillator (MOLO)

This oscillator provides two ultrastable frequency outputs one in the opticalregion the other in the microwave region the two frequencies being coher-ently related The MOLO is composed of three subunits the clock laser thereference cavity and the frequency comb see Fig 8

The clock laser subunit delivers the light for the excitation of the opticalclock transition It is generated by an extended-cavity diode laser at a wave-length of 871 nm Most of the output wave of this laser is sent to the ionclock The laser subunit containing two lasers for redundancy will require3 kg 3 l 15 W The frequency instability of the clock laser must be superiorto the minimum clock instability 27 times 10minus15 (τ s)minus12 for τ up to the timeconstant (sim10 s) of the servo system that locks the laser to the ionrsquos resonancesignal In order to achieve this the laser frequency is stabilized to the reference

Exp Astron

871 nm clock laser

EO

Detector and servo

AOM Femtosecond

frequency comb

To optical clock 10 GHz To FCDP

Microwave-optical local oscillator

ULE cavity

Fig 8 Microwave-optical local oscillator (MOLO) consisting of clock laser reference ULE cavityand femtosecond frequency comb (OFC) Radiofrequency inputs and outputs are not shown EOelectro-optic modulator ULE ultra-low expansion AOM frequency shifter

cavity where the PoundndashDreverndashHall method is used About 10 μW of thelaser output are used for this

The system will exhibit sim04 Hz (sim1 times 10minus15 relative) linewidth and afrequency drift lt01 Hzs The cavity consists of two high-reflectivity mirrorsoptically contacted to a cylinder of diameter 70 mm length 150 mm made ofULE (a glass exhibiting ultra low thermal expansion coefficient) with a cavityfinesse of about 300000 Residual vibrations of the satellite (arms lt 1 times 10minus6 g)should not be a limiting factor to the desired linewidth for safety marginan optimized cavity support can reduce the sensitivity to accelerations Foracoustic and thermal isolation the cavity resides inside an aluminium vacuumchamber equipped with a small (3 ls) ion-getter pump Two stages of polishedaluminium shields are implemented around the cavity which are activelytemperature stabilized so that the cavity is precisely kept at the zero crossingtemperature of the ULE thermal expansion coefficient Heat application andremoval is by thermoelectric elements between the shield and the vacuumchamber Additionally the entire vacuum chamber is temperature-stabilized

During transportation and launch the cavity will be rigidly fixed insidethe vacuum chamber by pressing on it from the sides by piezo-mechanicalactuators allowing accelerations of several g After bringing the satellite onthe orbit the actuators will be released

The optical setup for frequency stabilization will be a classic Pound-Drever-Hall scheme in reflection using an electro-optic modulator (EO) Couplingof the clock laser radiation to the cavity is via single-mode polarization main-taining optical fiber The fiber in- and out-couplers are mounted on miniaturepiezo-motor driven multiaxis translation stages Microprocessor control ofthe translation stages performs a laser-to-fiber incoupling stabilization andlaser-to-cavity mode-matching optimization (positioning inaccuracy lt10 μmalignment inaccuracy lt10 μrad)

The frequency stability of the clock laser will be limited by thermal 1f noisein the mirrors with an fractional Allan deviation slightly below 1 times 10minus15 from1 s to a few 10 s Resources including control electronics are 15 l 12 kg 8 W

Exp Astron

A femtosecond fiber laser based optical frequency comb generator (OFCFig 8) will be used for the clock frequency comparisons and to obtain aspectrally ultrapure microwave local oscillator output for the operation ofPHARAO for the microwave link and for the GNSS receiver The fem-tosecond laser is based on erbium-doped fiber pumped by fiber-coupled laserdiodes It emits 100 fs pulses at a repetition rate of 250 MHz with an averagepower of 100 mW The emission spectrum is centered around 1560 nm Spec-tral broadening to more than one octave is obtained via nonlinear interactionin a microstructure fibre so that the range from about 1050 to 2100 nm iscovered

Second harmonic generation of a portion of the broadened spectrum allowsobtaining the required light for the measurement of the clock laser frequencyat 871 nm The OFC will be referenced to the clock laser frequency (afterintroduction a controllable shift via the frequency shifter AOM) by makingthe carrier envelope offset frequency vanish so that the repetition rate fr isa high-order subharmonic of the optical clock reference frequency f1 = m1 frThe link between the optical and the microwave domain is then simply thedistribution of the repetition rate frequency or more conveniently a high(order 40) harmonic of the comb repetition rate 10 GHz

It has been demonstrated by several groups that frequency comparison andfrequency division with an OFC can be performed to very high accuracy andstability beyond the limitations imposed by the optical clocks themselves Forexample a fiber laser OFC with an instability (Allan deviation) of 6 times 10minus17

in 1 s has been demonstrated [51] OFCs based on fiber technology possessgood long-term reliability and high power efficiency The projection for a spacequalified OFC based on current industrial terrestrial technology is 16 kg mass20 l volume and 38 W power requirement

424 Frequency comparison and distribution package (FCDP)

This package distributes the ultra-stable 10 GHz reference frequency outputof the MOLO to the MWL performs frequency comparisons between MOLOand PHARAO and generates a variety of different reference signals for theindividual instruments including the 250 MHz reference clock to the OFCAll signals are coherent to the MOLO In the FCDP a USO is locked to theMOLO signal by a digital PLL A divider chain generates timing pulses (1 pps1000 pps etc) and keeps a continuous local time scale From the digital phasecomparison data the phase noise density spectrum between the USO and theMOLO is available by telemetry (TM) to assess the overall frequency purityin-flight

As the FCDP distributes the full performance reference from the MOLOdirectly to the MWL there is minimum impact of the FCDP on overall clockcomparison performance The USO-phase lock loop uses a similar designas for ACES-FCDP The digital PLL with linear phase detector allows theUSO to be slightly offset from its reference which is selectable by groundcommand This results in an accurate on-board time scale despite any offset

Exp Astron

of the MOLO reference frequency which can be synchronised to UTC Thefrequency synthesis chain re-uses modules developed for ACES-FCDP andtheir frequency inaccuracy Expected relative uncertainty (Allan deviation)of the frequency comparison between two signals provided to the FCDP is47 times 10minus17 at 1000 s 95 times 10minus18 at 10000 s

During non-availability of the MOLO ie during system start-up andcontingency conditions the USO runs as flywheel (ie it is unlocked withinstability lt15 times 10minus13 from 1 to 100 s) It will still provide useable butless accurate references to the instruments including the reference to OFCCommon view ground-to-ground time- and frequency comparison and supplyof the remaining instruments remain possible at the USO provided accuracy

Mass volume and power consumption are 8 kg 7 l 12 W Controldata rateis 10 kBday 50 kBday (FCDP alone)

425 Microwave link (MWL)

The microwave link is a two-way triple-frequency microwave link (Fig 9) Itis designed to perform accurate frequency comparison as well as time-transfer

Ku-Band Up-linkPower Tx 5 WCarrier 13475 GHzPN-Code 100 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

S-Band Down-linkPower Tx 1 WCarrier 2248 MHzPN-Code 1 MChips1pps 1 time marker sData 5 kBits

Ku-Band Down-linkPower Tx 1 WCarrier 1470333 GHzPN-Code 100 MChips1pps 1 time marker sData 5 kBits

EGE SC

S

Ku

KaKa-Band Up-linkPower Tx 5 WCarrier 345 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

Ka-Band Down-LinkPower Tx 1 WCarrier 32 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 5 kBits

GS Antenna diameterEGE terminal 1 mACES terminal 06 m

Fig 9 EGE microwave two-way link concept and characteristics See text for a description Thelink will be able to perform frequency comparisons with up to ten ground stations simultaneously(only one active comparison is shown) pps pulse per second Tx transmitter SC spacecraft

Exp Astron

between on-board and on-ground clocks Two-way operation allows for com-pensation of the effects due to spacecraft location and its motion relative toground Its two bi-directional links operate in Ka-band (32345 GHz) andKu-band (147135 GHz) The high carrier frequency of the up- and down-links allows for a noticeable reduction of the ionospheric delay A third down-link channel is in S-band (22 GHz) and is used to determine the ionospheretotal electron content MWL uses a combination of carrier phase and codephase data to measure with high precision the absolute ionospheric delayas well as ionospheric delay variations during a ground contact The useof three frequency bands allows to resolve third order ionospheric termsKey modulation parameters are Ka band Tone 250 MHz Ku band PN100 MChips S-band 1 MChips

The EGE-MWL concept is a considerably improved version of the ACES-MWL Microwave carrier generation is directly from the 10 GHz microwaveoutput available from the MOLO and having exceptionally low close-in phasenoise without intermediate frequency dividers or other complex synthesiserstages commonly used in RF transponders The high frequency carriers arekey to achieve the desired frequency comparison accuracy

The Ka-band carrier is generated by tripling the FCDP-generated signalThe Ku-band is subsequently generated by division by two using a verystable regenerative divider Finally the assigned radio frequency is obtainedby adding a low frequency signal using an image reject mixer A mixedanalog-digital receiver design will enhance its performance by use of a flexibledigital signal processing and digital control loops The number of simultaneousground contacts is increased from four (as in the ACES mission) to ten or moreusing the digital concept while the number of physical receiver channels perband is reduced from four to two one for operation to ground and one for in-flight calibration The concept allows adaptation of receiver loop parametersto optimise for the different portions of a highly elliptic orbit while reducingsystematic uncertainties due to high Doppler and Doppler rate Real-timedata exchange to ground terminals is by additional data modulation on all RFsignals The spacecraft carries two different small antennae for perigee and forapogee with gains of 8 dBi 20 dBi beamwidths 100 (full earth coverage) 14respectively power is 1 W for each

The MWL end-to-end performance between the attached clocks includingspace- and ground terminals is predicted to be a relative Allan deviation of1 times 10minus16 at 1000 s (perigee) and 1 times 10minus17 at 10000 s (apogee) Thesevalues are based on conservative implementation losses and other instrumentallosses assumed as 6 dB at apogee and 10 dB at perigee including additionallosses due to atmosphere An attitude knowledge of 1 is required

The MWL link will also support precise orbitographyndashcomplementary tothe GNSS receiver Generation of ranging data is simultaneous to two-waytime transfer and is available under all weather conditions Orbit determina-tion accuracy is expected to be better than 1 cm well matching EGE clockrequirements The value is expected based on demonstrated results fromother missions (METEOR ERS-2 TOPEXPOSEIDON) Laser ranging to

Exp Astron

the CCR array is used a few times during the mission to calibrate the MWLranging

In case of non-availability of the ultrapure signal from MOLO an additional10 GHz output is available from the FCDP to support ground-to-ground clockcomparison in common-view mode even after the nominal mission periodwhen the neither the clocks nor MOLO may be available any longer Basedon the PRARE experimentrsquos 12 years of operation this can be assumed forthe FCDPMWL combination as well The potential extended operational pe-riod makes the proposed configuration particularly attractive to metrologicallaboratories that rely on availability of long-lasting data in support of thegeneration of UTC

The resources of MWL are as follows mass 18 kg volume 18 l power55 W TC 2 kBday TM 120 MBday

426 Auxiliary units

The star camera assembly on the EGE spacecraft will provide real-timeattitude data primarily for the orientation of the spacecraft (with accuracy of1 in real-time) and the proper reduction of the GNSSSLR data

A GNSS receiver is used for the precise orbit determination time andfrequency transfer coarse positioning (lt1 m) for real time use by the Attitudeand Orbit Control as well as the back-up for time tagging of all payloaddata For the time and frequency transfer the receiver is nominally using theonboard frequency reference The receiver assembly includes three specialantennas By the year 2015 more than 100 GNSS satellites are expected tobe available for precise orbit determination permitting orbit inaccuracy of 1ndash2 cm RMS for altitudes up to approx 10000 km At apogee a combinationof GNSS signals and MWL ranging will be used with expected inaccuracyof 10 cm

5 Basic spacecraft specifications

The mission foresees a single spacecraft with a mission duration limited toapprox 3 years for cost reasons Attitude control along three axes withless than 1 error on a timescale of 1000 s is required The control can beperformed with cold gas thrusters magnetotorquers or reaction wheels Theselection will be done on the basis of the vibrations induced by these systems(desired level arms lt 10minus6 g)

Cold gas thrusters or monopropellant thrusters will be required to maintainthe orbit parameters during the mission by small orbit correction manoeuvresand will also guarantee a proper tuning of the orbit parameters after the releasefrom the launcher The ground segment (telemetry control and data transfer)could be via non-ESA stations user-operated to save costs

Further resources are indicated in Table 2

Exp Astron

The satellite bus design drivers are nadir-pointing microwave antennasthermal stability for the payload instruments no strong magnetic sources lowmicrovibrations harsh radiation environment due to crossing of the radiationbelts once per orbit power demand by science payload (with the possibility toreduce power demand by stand-by of some components during specific missionphases eg eclipse) mass limit due to required high altitude with low costlaunch

The satellite bus design can be based on established platforms such asFLEXBUS or PRIMA used on several past missions Their design philosophydiffers one having body-mounted solar arrays the other wings The latter canprovide larger amounts of power The vibration level induced by the panelswhen they are fixed and the duration of solar array drive periods per orbitwill have to be considered in the selection The FLEXBUS concept was usedfor a mission already requiring a low level of residual accelerations and theplatform is optimised for this parameter

For most of the satellite components one can consider using compo-nents and electronic units from navigation or communication satellites whichtypically cruise at altitudes between 15000 and 25000 km altitude LISAPathfinder which will cross the radiation belts during spiralling to theLagrange point will face similar problems concerning radiation as EGE as wellas XMM which is in a Molniya-type orbit with similar radiation characteristics

The instruments will therefore be located in the satellite structure such asto realise ionizing radiation shielding by the spacecraft structure and non-sensitive equipment to the maximum extent possible the need for additionalradiation shielding (including trapped proton radiation) for especially sensitiveparts of the instruments must be determined by ground tests It should bementioned that an orbit with higher perigee and less radiation dose is also anoption as it leads to a tolerable reduction of the science goals

The required plusmn3C temperature stability at the interface to the payloadenclosure is feasible The thermal control will employ the usual means (anenclosure with high thermal inertia heaters and a suitable radiator) Thelocation of the radiator will need a special study given the non-standard orbit

Magnetic field present during the mission can affect the properties of theonboard clock Using SPENVIS the magnetic field has been calculated alongthe spacecraft orbit the field intensity presents maxima around 02 G anda minimum value of the order of 10minus3 G As discussed above this is not alimitation to the instrument performance

6 Discussion and conclusion

The accuracy of the measurements depends crucially on the stability andaccuracy of the clocks While the performance of ground clocks (even thoseeventually deployed in new ground stations dedicated to EGE) will likelyimprove significantly from the present level the space clock design will beldquofrozenrdquo approx 10 years before data acquisition Thus their performance

Exp Astron

level will be significantly below the future ground station clock performanceAs specification we have chosen a performance level that has been demon-strated with laboratory breadboards recently We see no fundamental reasonthat this laboratory performance cannot also be achieved in a space ion opticalclock The experience with the PHARAO cold atom clock has shown thatan engineering model of a laser-based cold atom clock has a performancesimilar to that of a laboratory prototype The perturbation of the ion clock fre-quency due to magnetic fields along the orbit can be modelled and suppressedwith shielding and the residual effects measured and compensated for Themoderate temperature stability (plusmn 3C) to be provided in the experimentalbay can be achieved with thermal shields Accurate temperature controlof particular units is done by active stabilization of the units themselves

One important aspect is radiation which may affect all optical and trapcomponents as well as the lasers and reference cavity The radiation envi-ronment on the EGE orbit is significantly harsher than the near-earth ISSorbit on which ACES will operate The radiation effects must therefore becharacterized comprehensively with appropriate tests and then the radiationshield for the EGE instruments designed

In summary the Einstein Gravity Explorer mission (EGE) is a satellitemission aimed at a precise measurement of the properties of space-time usingatomic clocks By testing one of the most fundamental predictions of EinsteinrsquosTheory of General Relativity the gravitational redshift and thereby searchingfor quantum effects in gravity it explores one of the most important frontiers infundamental physics The EGE mission reaches beyond fundamental physicsproviding an important spin-off in geophysics Furthermore the technologicalaspects of the mission are ground-breaking for future time-keeping and ref-erence frame systems in space and thus are also expected to be of wide useto science and technology in general The payload of the mission has beenconceived as a compromise between cost and performance needing to fit thecost envelope of an ESA medium-class mission1

Acknowledgement We thank S Cesare for helpful information

References

1 Will CM The confrontation between general relativity and experiment Living Rev Relativ9 3 (2006)

2 Ni W-T Empirical foundations of the relativistic gravity Int J Mod Phys D 14 901 (2005)doi101142S0218271805007139

1Finally we point out that a spacecraft with a significantly reduced and thus lower-cost payload (noclocks only MOLO FCDP MWL) on a circular orbit of appropriate height could be used for along-duration earth science mission dedicated specifically to the measurement of the geopotential

Exp Astron

Fig 2 Performance of clocks and links on EGE Curve ldquooptical clocks EGErdquo is the specificationCurve ldquooptical clocks EGE (goal)rdquo refers to expected near-future performance of an opticalclock based on neutral atoms ldquoUpgraded PHARAOrdquo refers to the PHARAO microwave clockusing the high-quality microwave-optical local oscillator The microwave link (MWL) performanceexhibits a discontinuity since common-view contact time of ground stations with EGE is limited toseveral h The different slopes arise because on short time scales only white phase noise is relevantwhile on long time scales flicker noise also contributes The reproducibility is taken into account

measurement of the redshift at 25 ppb accuracy The ground clocks towhich the satellite clocks are compared to at perigee and at apogee willbe located in different ground stations The inaccuracy of these groundclocks must therefore be sufficiently smaller than 1 times 10minus17 in order notto contribute to the 25 ppb error Current state-of-the art inaccuracy is2 times 10minus17 and is likely to decrease further

(2) Absolute comparison between ground and satellite clocks with the satel-lite at apogeeHere the terrestrial gravitational potential difference is maximumU(apogee) minus U(perigee)c2 sim= 65 times 10minus10 Assuming a space optical clockinaccuracy of 1 times 10minus16 and a ground clock inaccuracy a factor 3lower (current state-of-the-art) allows a measurement of the redshift at150 ppb This measurement takes advantage of the repetitive nature ofthe orbit only to a small extent and relies on the accuracy of the opticalclocks in particular of the space clocks It is thus complementary to (1)albeit less accurateIn (1) and (2) the dominant contribution to the redshift is from Earthrsquosgravitational potential The contribution from the solar potential isstrongly reduced due to the fact that satellite and earth are an interactingsystem in free-fall toward the Sun [19] and can be modeled

(3) Comparisons between terrestrial clocksThe frequency comparison link to the EGE satellite allows terrestrialclock comparisons by a common-view method whereby the frequency

Exp Astron

signals from two distant ground clocks are sent to the satellite andcompared there The frequency difference between two identical groundclocks is determined if systematic instrumental errors are negligible bythe difference in gravitational potential at the two clock locations Thispotential difference includes a terrestrial and a solar contributionThe solar contribution varies according to the orientation of the clockpair with respect to the Sun with a 1 day-period For two clocks at asurface distance of sim6000 km and suitable orientation with respect tothe sun the normalized modulation amplitude is sim= 4 times 10minus13 This is thesame as the earth potential difference for 4 km height difference It ispossible that ground clocks will have reached an inaccuracy at the levelof 1 times 10minus18 at the time of the mission in 2015 Using the microwave linka comparison of ground clocks over sim3 h (typical common-view duration)with a resolution of 1 times 10minus17 can be performed limited by the instabilityof the link Repeating the clock comparison at each common-view win-dow (n = 1000 comparisons) will allow averaging and a measurementof the solar gravitational shift with fractional inaccuracy of 1 ppm Incomparison the best current results for the solar gravitational frequencyshift are at the few-percent level [20 21] A moderate averaging out ofground clock systematic shifts is assumed as well as modeling of the earthgravitational potential variations on the daily timescale at the appropriatelevel

(4) Comparisons of on-board clocksThe two on-board clocks are continuously compared these data areanalyzed for violation of LPI (Eq 4) Since the measurement is a nulltest the required accuracy of the gravitational potential at the satellitelocation is low and easily satisfied

24 Test of local position invariance and search for neutron scalar charge

The test of LPI is performed in conjunction with the measurement of thegravitational frequency shift

With approach 23 (1) the optical clock will be able to measure K (α Earth)with accuracy 30 ppb (spec)

Combining the results of the microwave and optical clocks will determineK(meQCD Earth) + εK(mqQCD Earth) with accuracy 85 ppb (spec)where ε asymp minus01

Simultaneously the on-board comparison 23 (4) yields the combination19 K(α Earth) +K(meQCD Earth) + εK(mqQCD Earth) with accuracy85 ppb (spec)

Note that these measurements are fully complementary to current andfuture terrestrial clock-clock comparisons which are sensitive to K(X Sun)and can be combined with the latter to determine or set limits to the neutronscalar charge

Exp Astron

25 Theoretical analysis of frequency comparison signals

The on-board clock-clock comparison signals can be analyzed straightfor-wardly in terms of a violation of LPI since signal propagation effects areabsent

The determination of the gravitational frequency dilation via ground-satellite clock comparisons is more complex (Eq 1 is strongly simplified) andrequires an accurate treatment of all relevant relativistic effects Gravitationin the solar system is described in the barycentric (BCRS) and geocentric(GCRS) non-rotating celestial reference systems by means of post-Newtoniansolutions of Einsteinrsquos equations for the metric tensor in harmonic gaugescodified in the conventions IAU2000 [22] In these conventions the relativisticstructure of Newton potential (order 1c2) and gravitomagnetic (order 1c3)

potentials are given For the Newton potential the multipolar expansion ofthe geopotential determined by gravity mapping is used The post-Newtonianeffects of the metric (described in terms of dimensionless parameters β and γ )

can be included Orbits of satellites are evaluated in the GCRS at the levelof a few cm by means of the Einstein-Infeld-Hoffmann equations at the order1c2 [23] using the relativistic Newton potential Instead the trajectories of theground clocks in the GCRS are evaluated from their positions fixed onthe Earth crust (ITRS International Terrestrial Reference System) by usingthe non-relativistic IERS2003 conventions [24 25]

For the propagation of light rays (here radio signals) between the satelliteand the ground stations one uses the null geodesics of the post-Newtoniansolution in the GCRS The timefrequency transfer properties have beentheoretically evaluated for an axisymmetric rotating body [26ndash29] at the order1c4 (ie with Newton and gravitomagnetic potentials developed in multipolarexpansions) These results have been developed for use in the ACES missionThe relative frequency dilation can be expressed as

ν

ν= 1 + 1

cL1 +

4sum

n = 2

1

cn(Ln + Gn)

where Ln describes special-relativistic Doppler effects and Gn the general-relativistic effects These quantities are time-dependent due to the orbitalmotion For concreteness we report the orders of the terms The kinematicaleffects are |L1c| lt 1 times 10minus5 |L2c2| lt 2 times 10minus10 |L3c3| lt 3 times 10minus15 |L4c4| lt

7 times 10minus20 The Newtonian contributions are G2 = G(M)2 + G(J2)

2 + G(J4)2 + G(J6)

2(M is the mass monopole Jn are the lowest multipoles) with estimates|G(M)

2 c2| lt 3 times 10minus11 |G(J2)2 c2| lt 2 times 10minus13 |G(J4)

2 c2| lt 3 times 10minus16 |G(J6)2 c2| lt

1 times 10minus16 For G3 = G(M)3 + G(J2)

3 the estimates are |G(M)3 c3| lt 2 times 10minus14

|G(J2)3 c3| lt 1 times 10minus18 Finally the G4 = G(M)

4 + G(S)4 contributions (G(S)

4 is thelowest gravitomagnetic effect) are of order |G(M)

4 c4| lt 1 times 10minus19 |G(S)4 c4| lt

1 times 10minus22 and thus not relevant for EGE

Exp Astron

The available theory is fairly complete for the data analysis in mode 23 (1)described below For modes 23(2 3) the theory will need to be generalizedto include the effect of the gravitational potential at the location of the groundclock(s) (including time-varying effects such as the tides) and of the solar andlunar potentials

The uncertainties of the various contributions above are minimized by ausing precise EGE orbit data obtained by satellite tracking and by usingcurrent and future accurate earth gravity information Required orbit positionknowledge is at 1 cm level near perigee and sim50 cm near apogee reachablealready today with the proposed orbitography approach described below Thevalidity of the special-relativistic Doppler shift has been and will continue to beindependently verified with increased accuracy by spectroscopy of relativisticatomic ion beams [30] Potential violations of this aspect of Lorentz Invari-ance are expected to be sufficiently bounded so as not to affect the signalinterpretation

26 Local Lorentz invariance tests

There has been an explosion of interest in Local Lorentz invariance in re-cent years [31 32] with numerous astronomical studies and high-precisionexperiments applied to search for violations of Lorentz invariance Signifi-cant developments have also occurred on the theoretical side and a theory(standard model extension SME [33]) has been worked out that provides aunified framework for describing and analyzing Lorentz Invariance violationsin a variety of systems Two aspects of Lorentz Invariance can be tested withEGE

261 Independence of the speed of light from the laboratory velocity

A possible dependence of the speed of light on the speed v of the laboratorycan be modeled according to the MansourindashSexl test theory [34] by

c (v) = c0(1 + Bv2

c2

)

Here v 377 kms is the velocity with respect to the cosmic microwavebackground the cosmologically preferred frame and B is a combinationof parameters describing deviations from the usual Lorentz transformationformulas B = 0 if Lorentz Invariance holds For a terrestrial experiment therotation of the Earth around its axis modulates v with a 300 ms amplitude Adependence c(v) can be searched for by measuring the frequency differencebetween a resonance frequency of a highly stable optical cavity (cavity fre-quencies are proportional to c) and an optical clock (Kennedy-Thorndike-typeexperiment [35]) and determining the modulation of the frequency differencecorrelated with the modulation of v The advantages of a space experimentare the high orbital velocity and strongly reduced cavity deformation thanksto microgravity [36 37] On the proposed elliptic orbit v varies between+4 and minus4 kms over approximately 1 h This variation is 13 times larger

Exp Astron

than for the case of a terrestrial experiment The shorter time scale is alsoadvantageous since the drift of the cavity is more predictable In the EGEmission the reference cavity of the clock laser is used The large numberof orbits improves the statistics An improvement of the accuracy of B by afactor 20 compared to the best current terrestrial results is realistic The dataanalysis can also be performed in the framework of the SME theory takinginto account complementary bounds obtained from terrestrial experimentsand astrophysical observations

262 Independence of Zeeman splitting frequency on the directionof the magnetic field

One class of tests of Lorentz Invariance consists in measuring the frequencysplitting between two levels of a quantum system induced by a static magneticfield as a function of orientation of the field direction with respect to thestars (HughesndashDrever-type experiments [31]) This class addresses the so-called matter sector and is complementary to combined-photonmatter-sectortests such as the above The most precise experiments are performed withatomic clocks eg masers or cold atom clocks [38] The EGE instruments allowsuch experiments since magnetic fields are applied continuously in the clockor repeatedly for calibration Compared to terrestrial experiments the highorbital velocity of EGE could lead to an improvement by a factor sim20

27 Application to geophysics

Using GPS coordinates of a point on the Earth can be obtained with aninaccuracy well below 1 cm in a well defined international terrestrial referencesystem The coordinates are purely geometrical and do not contain any gravityinformation A local gravitational potential UB is determined with respect toa reference gravitational potential U A by ldquolevellingrdquo the surface of the Earthie sequentially measuring gravity g and height differences every ca 50 m

UB minus U A = minusBint

A

g (r) middot d n

where n is the difference vector between subsequent locations The disadvan-tage of measuring height differences by the levelling method is a random walkeffect of accumulated errors

In the classical definition a geoid is defined as the particular equipotentialsurface nearest to mean sea level In these terms the geoid serves as areference surface for measuring the height and also to define a referencefor the gravitational potential Such a vertical reference frame historicallywas realized for a country or several countries by determining the mean ofsea level observations at tide gauge stations taken over long period of timeHowever modern satellite altimetry missions such as TOPEXPOSEIDON orENVISAT show that departure of the mean sea level from an equipotential

Exp Astron

surface may reach up to several meters on the global scale see eg [39]Therefore height systems based on different tide gauge stations may differin the realisation of the geoid and may differ with respect to the reference byseveral meters

The best gravity field models obtained from satellite data (eg missionGRACE) have reached a precision of about 1 cm over sim250 km half wave-length [40] Satellite data also reveals changes in time [41] However for typicalEarth topography with height variations of eg 1000 m over 30 km horizontaldistance one may expect variations in the geoid of about 80 cm Such ahigh-spatial-frequency signal in geoid variations cannot be detected by spacegravity missions and requires a combination of satellite and terrestrial gravitymeasurements like gravity anomalies deflections of vertical and GPSlevellingpoints The best combined global gravity field models are provided up to de-gree and order 360 in terms of spherical harmonic representation correspond-ing to a half-wavelength of about 55 km However in combining the satelliteand terrestrial gravity field measurements the problem of terrestrial data givenin different height systems between continents and different countries remainsand has to be tied to satellite measurements

The geodetic scientific community is currently establishing a Global Geo-detic Observing System (GGOS) [42] Its objectives are the early detectionof natural hazards the measurement of temporal changes of land ice andocean surfaces as well as the monitoring of mass transport processes in theEarth system Global change processes are small and therefore difficult toquantify Therefore the required precision relative to the Earthrsquos dimensionis 1 ppb GGOS will be established by the combination of geodetic spacetechniques (GPS Laser-Ranging very large baseline interferometry) andrealized by a very large number of terrestrial and space-borne observatoriesThe purely geometric terrestrial 3D coordinate system is in good shape andfully operational It has to be complemented however with a globally uniformheight system of similar precision The current precision level of regionalheight systems in terms of gravity potential differences is in the order of1 m2s2 (10 cm) with inconsistencies between these various systems up toseveral 10 m2s2 (several meters) The actual requirement in the context ofGGOS is 01 m2s2 (1 cm) with the need of a permanent ie dynamical controlThis requirement of high precision height control comes from the need tounderstand on a global scale processes such as sea level change global andcoastal dynamics of ocean circulation ice melting glacial isostatic adjustmentand land subsidence as well the interaction of these processes Only by meansof monitoring in terms of gravity potential changes at the above level ofprecision the change of ocean level can be understood as a global phenomenonand purely geometric height changes be complemented by information aboutthe associated density or mass changes

The geoid can also be defined in a relativistic way [43] as the surface whereaccurate clocks run with the same rate and where the surface is nearest tomean sea level The relation between the differences in the clock frequenciesand the gravitational potential is given in simplified form by Eq 1 At present

Exp Astron

there is no operational way to compare frequencies of the already availableoptical clocks on the global scale at the same level as their accuracy wouldallow With EGE and by then improved ground optical clocks it will becomepossible to obtain gravitational potential differences on a global scale by com-paring frequencies This would be a significant new dimension to gravity fielddetermination since such observables are given on the global scale and providein situ local gravity information at the same time EGE will allow clock-basedgravitational potential mapping with the same payload instruments used forthe fundamental physics experiments The on-board clocks are actually notrequired for this purpose MOLO MWL and FCDP suffice Thus EGE willallow establishing a global reference frame for the Earth gravitational fieldwith accuracy in the order of a few cm in terms of the geoid heights Thisassumes that by the time EGE is operational (mobile) ground clocks withfractional inaccuracy at the level of 10minus18 are available the measurementsare limited by MWL noise and thus require a long integration times to matchthe expected ground clock performance Such a reference frame will serve asreference for all gravity field modeling and clocks will be used to define theinternational atomic time scale and the Earth gravitational potential at thesame time

In summary with optical clocks geodesy will undergo a second revolutionafter geometry now being measured by clock-based GNSS systems also physi-cal heights and potential will be measured by (mobile) optical clocks using thegravitational redshift effect

28 Application to frequency standards and terrestrial fundamentalphysics studies

As has been described repeating ground clock comparisons and integratingover several months will provide global ground clock frequency comparisonsat the level of 1 times 10minus18 more than two orders of magnitude below the currentGPS and two-way satellite time and frequency transfer methods In the timeand frequency community the availability of such a global high performancemicrowave link will accelerate the process leading to a redefinition of the SIsecond based on clocks operating in the optical domain

The global character of the ground clock comparisons made possible byEGE can also contribute to the search for a time-variation of the fundamentalconstants

The comparisons do not require availability of the on-board clocks The riskof this type of measurements is therefore reduced Moreover the lifetime ofmany subunits required for this measurement (MWL and FCDP see below)can exceed 10 years (from previous experience with similar devices) andthus offers a motivation to increase the mission duration beyond the nominal3 years

Further applications to metrology are listed under the secondary goals inSection 21 above

Exp Astron

3 Mission profile

31 Orbit

The orbit must satisfy the following requirements

Primary 1 Large gravitational potential difference between apogee andperigee

2 sufficient contact time with at least two ground stations atperigee

Secondary Simultaneous visibility of satellite from distant (inter-continental) ground stations for several hours

To satisfy the primary objective the orbit must be highly elliptic The perigeeshould be low A very low perigee means a short contact time which decreasesthe accuracy of the frequency measurement A compromise is a perigeealtitude of sim2500 km For a sufficiently high apogee this yields a U of 60of the theoretical maximum between the earth surface and infinity

The baseline scenario is an orbit of approx 12 h period with the followingproperties

Apogee Altitude 37856 km (minus 10 deg latitude one over Atlantic andPacific resp)

Perigee Altitude 2500 km (+ 10 deg latitude one over Guatemala andMalaysia resp)

Inclination 634 argument of perigee 170RAAN 25 True Anomaly 220Apogee drift rate 70 kmday perigee drift rate sim 14 kmday

True anomaly and right ascension of the ascending node (RAAN) wereoptimized to minimize orbital precession Figure 3 shows the ground track andthe ground coverage at one of the perigees Common-view contact time forstations in Europe is about 6 h while between a station in Europe and in theUSA it is about 35 h Frequency comparison windows between a given groundstation and EGE repeat every 24 h

A final selection of the orbit requires an in-depth study of the costs of thepropellant and maneuvers as well as of the setting up of new ground stations

32 Ground segment

The mission requires two separate ground segments the science groundsegment and the standard satellite control segment which supports telemetryand telecommand (TMTC)

Exp Astron

Fig 3 Baseline EGE orbit One (out of two) apogee and one (out of two) perigee ground pointsare shown The ground coverage at one of the perigees is shown with circles at elevation anglesteps of 10 starting from 10

321 Science ground segment

As described above one part of the mission is to perform frequency com-parisons between ground-to-space and ground-to-ground (common-view andnon-common view) The use of a two-way microwave link (MWL) systemhas the advantage that a precise orbit or the precise relative velocity of thesatellite must not be known [4] The orbit determination requirements fora two-way system are significantly relaxed compared to the requirements tomodel the relativistic effects on the clocks The frequency comparison resultsare available both on ground and on the satellite

For ground-to-satellite clock comparisons five ground stations arefavourable two capable of viewing perigee three for viewing apogee Thisallows for the classic three-clock configuration where the performance of eachclock can be obtained individually The perigee stations will be observing thesatellite for sim25 min every perigee passage for the duration of the missionThe apogee stations will observe for sim3 h every apogee passage Of coursea larger number of stations as well as longer observation time are usefulfor robustness redundancy and orbit determination The stations could beinstalled on a temporary basis for the duration of the mission

The stations must have access to optical clocks of performance at least ashigh as todayrsquos state-of-the-art ground clocks but preferably higher Theywill include an optical frequency comb by means of which the optical clockfrequency is transformed to a radio frequency for comparison with the EGEclocks High-accuracy fiber-optic link technology already existing today allowsdistances of sim100 km between the ground station and the correspondingoptical clock allowing a certain freedom in the choice of the clock locationWith further increase of the fiber-optic link range currently being investigated

Exp Astron

the area over which the ground station can be positioned relative to the opticalclock will increase This option is of importance for the implementation of ageophysics measurement campaign for which a large number of clock locationsis of interest

The above ground stations will be complemented by some 15 additionalground terminals provided by the world-wide laboratories equipped with highperformance primary clocks at the time of the mission These will participate inorbit determination to 1 cm accuracy along the full orbit which can be achievedwith non-optical atomic clocks The EGE MWL is fully compatible to plannedACES ground terminals which might reduce the number of required stations

With a larger number of transportable optical clock ground stations de-ployed and moved across the continents a geophysics campaign could beimplemented to map the gravitational potential

Ground station equipment MWL uses dedicated small transportable termi-nals which can be easily installed at any location of interest Each terminalconsists of a steerable dish antenna of 1 m diameter with pointing accuracy of01 using a transmitted power of 5 W per band The high directivity antenna(beam aperture Ka-band 07 Ku-band 14 S-band 20) avoids multi-pathfrom nearby objects Antenna noise temperatures are Ka-band 450 K Ku-band 340 K Operation is at elevations above 10 and measurements areperformed above 20 For ranging purposes terminal location is determined byGPS surveying Each ground station computes its own ionospheric correctionsbased on a triple-band receiver The terminal has a built-in delay monitorwhich allows calibrated ranging during several months once its bias has beendetermined by laser ranging Its M and C interface is via local area networkto either a userrsquos computer or directly to the science network control centreby remote MampC The stationrsquos operation is fully automatic and unattendedSchedule and orbital data are received via LAN once per day Signal acquisi-tion is fully autonomous Data transmission is after each pass The terminal hasan uplink telecommand capability of 1 kBits and a downlink data capacity ofup to 10 kBitss For the local user there is a visual interface providing quick-look data from real-time comparison to the spacecraft clocks to verify theirgood health The final full performance product is calculated at the sciencedata centre

Local metrological data (temperature pressure humidity) are recorded tocorrect for tropospheric delay of the ranging data The main advantage ofMWL terminals compared to laser based systems is their all-weather capabil-ity although microwaves suffer higher propagation errors in the ionosphereand troposphere

The EGE ground terminal will profit from the ERS-2 mission PRARE(precise range and range rate equipment) [45] experiments whose terminalshad similar design and have been operated under extreme climatic conditionsincl Antarctica and high temperature regions from existing designs fromthe space segment (mission ACES) and no further special developments aredeemed necessary Cost per EGE master station is modest

Exp Astron

Satellite tracking via laser ranging Laser ranging is alternative to MWL rang-ing and also serves for absolute calibration of the latter For this purpose anon-board corner cube reflector (CCR) is provided Ranging will be done withthe International Laser Ranging Service Ideally the satellite will be rangedat every perigee passage and every apogee passage simultaneously with thefrequency comparison procedures Todayrsquos ranging precision is sufficient tosatisfy the science goals For apogee observations in the northern hemispherea large number of laser ranging stations is available The number of stationscapable of ranging at or near perigee in the southern hemisphere is six andthus is sufficient even considering unfavourable weather conditions at someof these The MWL ground stations are able to range even under thoseconditions

322 Control ground segment

Average data transfer rate is estimated at 300 MBday (approx 40 kBs onaverage) Data will be stored on board (2 GB capacity required) and will betransferred to a receiving station on the Earth once a day

XPLC

incl Data StorageIF to TMTC Link

HK amp TMTC

to ground

Opt Clock

AOM

MOLO

PHARAO

FCDP

10 GHz

100 MHz

250 MHz

GNSS Rx

MWLTampF Link

uplink Data

UTC(k)

TampF Space vs Ground

TM from GS

Tim

e S

cale

Orbit GNSS Time UTC

TampF Space vs Ground

MampC Time Scale

MampC

MampC

MampC

MampC

Mamp

C

PHARAO vs Opt Clock

PHARAO vs Opt Clock

10 GHz amp Epoch

10 MHz

amp Epoch

Epoch

10 MHz

amp Epoch

Opt Link

Fig 4 Schematic of the payload showing subunits and their signals The results of the frequencycomparison between the two on-board clocks (optical clock and microwave clock PHARAO) areobtained from a combination of the frequency values applied to the frequency shifters AOM (seeFigs 7 and 8) and those output from the frequency control and distribution package (FCDP)as indicated by the arrows ldquoPHARAO vs Opt Clockrdquo The satellite clock to ground clockcomparison results are output from the MWL XPLC is the on-board computer and is involvedin generation of signals for instrument control The optical link shown dashed is a possible add-onproviding an alternative to the MWL GS ground stations MampC monitoring and control HKhousekeeping TampF time and frequency

Exp Astron

4 Main payload instruments

41 Overview

The satellite payload consists of a pair of cold-atom clocks a microwave-optical local oscillator (MOLO) an on-board frequency distribution unit(FCDP) a satellite-to-ground clock comparison unit (MWL) and the onboardcomputercontrol system (XPLC) see Fig 4 Auxiliary units are the posi-tion determination GPSGalileo receiver (GNSS Rx) attitude monitor (startracker not shown) and the corner cube reflector array (CCR not shown)

The science goals require clocks of very high frequency stability on thetimescales of half the orbital period (6 h) as well as perigee passage (sim1000 s)see Fig 2 Only clocks based on cold atoms are suited for the science goalsTable 1 reports values for a subset of atomic clocks that have been investigatednot including clocks that require cryogenic operation or are too complex foruse in an medium-class mission The choice of a clock pair rather than a singleone fulfils several requirements it addresses more science goals improves thevalidity of some science results and it enables calibration accuracy evaluationand redundancy The validity of the science results is enhanced since theclocks differ not only in the employed atoms but also in terms of havingdifferent sensitivity to perturbations by electric and magnetic fields electronicserrors and misalignments A comparison of the results arising from each clockallows for an understanding of measurement errors Both the clocksrsquo and thefrequency comparison unitrsquos performances will be determined after launchand regularly verified during the course of the mission The only reliable

Table 1 Properties of different clock candidates for EGE

Clock Clock Fractional σ = Fractional Sensitivity Figure of Sensitivityatom frequency inaccuracy instability to α Aα merit |Aα| to meq

at 1000 s σ [1016] QCD

Sr+ 445 THz 3 times 10minus17 (p) 5 times 10minus17 (p) 042 08 (p) ndashYb+ (1) 688 THz 3 times 10minus16 (d) 3 times 10minus16 (d) 088 029 (d) ndash

3 times 10minus17 (p) 1 (p)Yb+ (2) 642 THz lt 1 times 10minus17 (p) 1 times 10minus17 (p) minus53 53 (p) ndashSr 435 THz 1 times 10minus16 (d) 1 times 10minus16 (d) 009 009 (d) ndash

lt 1 times 10minus17 (p) 09 (p)Yb 519 THz 7 times 10minus16 (d) 1 times 10minus16 (d) 031 03 (d) ndash

lt 1 times 10minus17 (p) 1 times 10minus17 (p) 31 (p)Cs 91 GHz 2 times 10minus16 (d) 1 times 10minus15 (d) 275 027 (d) 1Yb+ 126 GHz 1 times 10minus15 (p) 1 times 10minus15 (p) 34 034 (p) 1

A relevant figure of merit is the ratio of |AX | and fractional frequency instability σ on the perigeevisibility timescale The Cs clock is the only high performance clock with adequate sensitivity tothe particle masses me mq The indicated Cs clock performance was achieved using a microwavelocal oscillator based on a cryogenic oscillator On EGE the latter is to be replaced by an opticallocal oscillator-based device (MOLO) with similar short-term performance but without the needto operate at cryogenic temperature The clock instability at 10000 s is typically similar to theindicated inaccuracy(p) projected (d) demonstrated (1) quadropole (2) octupole

Exp Astron

way to achieve this is by having at least two clocks on board The on-boardclock-clock comparison provides performance evaluation independent of theground-satellite comparison unit Once the satellite clock performances areestablished the link performance can be established as well

The clocks are similar in basic structure but differ in the detailed implemen-tation They consist of

a source of atomsa preparation subunit which cools the atoms to near-standstill (millikelvintemperature or below)a trapping subunit that confines the atom(s) to a tiny volumean interrogation subunita clock control electronics package

The clocks share a common central unit the optical-microwave local oscil-lator which serves as a flywheel oscillator with outstanding stability on shorttimescales (01ndash10 s) for both clocks It consists of a laser stabilized to an ultra-low loss optical resonator made of ultra-low-expansion glass A microwavesignal is derived from the laser by phase-locking a frequency comb to the laserwave and using a multiple of the repetition rate

While the subunits of any optical clock have a similar technological basis(optics electrooptics acoustooptics thermal control vacuum systems) theydiffer in their specifications Depending on the choice of the atom species de-livery confinement and interrogation methods clocks of different perfor-mance result A number of laboratory optical clock demonstrators have beendeveloped worldwide This ensures availability of the know-how for the indus-trial implementation of the flight models

The first clock is a microwave clock based on the interrogation of slowlymoving ensembles of cold Cs atoms It is essentially based on the PHARAOclock [5 6] developed to engineering model level by CNES in the frame-work of the ESA ISS mission ACES In EGE this clock is upgraded usingthe microwave-optical local oscillator This improves significantly the clockperformance The specification is a relative instability (Allan deviation) of3 times 10minus14 (τ s)minus12 ie at the integration time τ = 1000 s it drops to 1 times 10minus15and to 3 times 10minus16 at 10000 s While this specification is at least a factor 3 lessstringent than for the optical clocks the significantly lower cost of obtaining aflight model and the sensitivity of its frequency to the particle masses makes ita suitable choice

The optical clock baseline instrument is a single-ion optical clock Amongvarious suitable ion species (Hg+ Al+ Sr+ Yb+) [44ndash47] the Yb+ ion hasbeen selected Its breadboard performance is established at the instability levelof 3 times 10minus16 over 1000 s it exhibits a large sensitivity coefficient Aα hassmall sensitivity to magnetic fields and exhibits significant potential for furtherimprovements (use of the octupole transition with sim6 times larger Aα andlower overall instability [48] see below) A trapped Yb+ ion is also a veryrobust system it has been demonstrated that a single Yb+ ion can remaintrapped uninterruptedly for many months Finally Yb+ is the only ion with

Exp Astron

which optical clocks demonstrators have been operated in not a single but twomajor metrology laboratories

Another attractive optical clock is based on ensembles of cold neutral Sror Yb atoms trapped in an optical lattice produced by laser beams This clocktype has a lower instability than a single-ion clock thanks to the large numberof atoms (sim105) used Its development is in progress in several laboratoriesin the world (in part supported by space agencies) and has to date reached6 times 10minus17 instability at 5000 s and 1 times 10minus16 uncertainty [49 50] with potentialfor even higher performance However it is more complex and costly thanthe Yb+ ion clock since the cold atom preparation and trapping subunitsinclude more components require higher laser powers and the respectiveoperating procedures are more complex It is therefore not considered for thepresent mission designed to be developed in the very near future with limitedresources but is an interesting candidate for missions without such constraintsIts estimated physical parameters are indicated in Table 2 for comparison

42 Main instruments

421 Microwave clock

The PHARAO cold atom clock is one of the two atomic clocks of the spacemission ACES managed by ESA It is proposed here to include a secondversion of this instrument as its costperformance ratio is very attractive forthe EGE scientific objectives Its concept is very similar to ground basedatomic fountains but with the major difference of zero-g operation Atomsslowly launched in free flight cross two microwave fields tuned to the transitionbetween the two hyperfine levels of the cesium ground state The interrogationmethod based on two separate oscillating fields (Ramsey scheme) allowsthe detection of an atomic line whose width is inversely proportional to thetransit time between the two microwave cavities The resonant microwavefield at 9192631770 GHz (SI definition of the second) is synthesized startingfrom an ultrastable quartz oscillator (USO) in the FCDP phaselocked to themicrowave output of the microwave-optical local oscillator and stabilized tothe clock line using the error signal generated by the cesium resonator In thisway the intrinsic qualities of the cesium hyperfine transition both in terms ofaccuracy and frequency stability are transferred to the USO In a microgravityenvironment the velocity of atoms along the ballistic trajectories is constantand can be changed continuously over almost two orders of magnitude (5ndash500 cms) Differently from atomic fountain clocks presently operated onground very long interaction times (up to few seconds) will be possible whilekeeping reasonable the size of the instrument

The instrument consists of 4 main elements a laser source which providesthe light required to cool and detect the cesium atoms (Fig 5 left) a frequencychain which generates the 92 GHz microwave signal a cesium vacuum tubewhere the interaction between the microwave field and the atoms occurs

Exp Astron

Table 2 Resources and technology readiness level (TRL) of the EGE spacecraft

Unit Volume Mass Power TRL Comment[l] [kg] [W]

Ion clock 100 70 60Atom source 4Atom preparation 6Atom trap 4Atom interrogation 6Clock control 6

Microwave clock 200 91 110 8 TRL by end 2008Local oscillator

Stabilized laser 18 15 23 6Frequency comb 20 16 38 6

Microwave link 18 18 55 8FCDP 7 8 12 8GPS receiver 2 5 7 9CCR 2 2 0 9Payload sum 367 225 305Thermal control 14 17 30 8OBDH 20 24 38 9Star tracker 1 1 8 9Radiation shield 10 40 0 9Support structure 170 204 0 9Power incl solar arr 50 58 10 9Propulsion 21 25 20 9AOCS 36 43 82 9TMTC 8 10 15 9Satellite bus sum 330 422 203 Flexbus conceptPayload plus bus incl 20 margin 8364 7764 6096Propellant 40Launch adapter 100SC wet masspower incl Margins 9164Lattice clock 135 105 152 Alternative clock

Atom source 4Atom preparation 6Atom trap 4Atom interrogation 6Clock control 6

Also included are the resources for an optical lattice clock based on neutral Strontium atoms

(Fig 5 right) and a computer The engineering model of PHARAO is fullyassembled at CNES Toulouse and undergoes functional and performancetests until the end of 2008 The operating temperature range is 10ndash33 andnon-operating temperatures are minus45 +60 The cesium tube consists in aUHV chamber pumped by getters and a 3 ls ion pump a cesium reservoira microwave cavity coils to provide a uniform magnetic field and three layersof magnetic shields Vacuum windows with fiber-optics collimators enable thetransmission of the cooling and detection light onto the cesium atoms

In autonomous mode PHARAO uses the USO as interrogation oscillatorand will provide a clock signal with fractional frequency stability below1 times 10minus13 (τs) minus12 and inaccuracy near 1 times 10minus16 When using the microwave-

Exp Astron

Fig 5 Left The PHARAO laser source engineering model with cover removed Dimensionsare 530 times 350 times 150 mm3 and the mass is 200 kg Ten polarization maintaining optical fibers(yellow) guide the laser beams to the cesium tube All diode lasers are mounted on Peltier coolersfor temperature regulation within 2 mK (courtesy EADS SODERN) Right The PHARAOengineering model cesium tube without the two external magnetic shields The volume is 990 times336 times 444 mm3 and the total mass is 44 kg

optical local oscillator PHARAO will provide a reduced frequency instabilityof 3 times 10minus14 (τs) minus12 This corresponds to 1 times 10minus15 at 1000 s and 3 times 10minus16

at 104 s

422 Ion optical clock

The 171Yb+ ion clock shown in Figs 6 and 7 provides an atomic frequencyreference based on a single 171Yb+ ion confined within an RF electric end-captrap (similar to a Paul trap) within an ultra high vacuum (UHV) chamber andlaser cooled to sim1 mK close to the Doppler cooling limit As a result the ionrsquosfirst order Doppler motion is completely removed and the ion experienceslittle collisional perturbation from its environment Under these conditionsthe narrow linewidth 2S12 (F = 0 mF = 0) minus 2 D32 (F = 2 mF = 0) quadrupole

Fig 6 Left Simplified level scheme of 171Yb+ showing relevant cooling (blue) auxiliary (red)quadrupole clock (green) and octupole clock (dark blue) transitions δνnat is the natural linewidthof the clock transition Middle close-up of Paul-type RF trap (PTB Braunschweig) the ion istrapped in the hole (diameter 14 mm) Right Complete ion trap with vacuum flange height 10 cm(NPL Teddington)

Exp Astron

935 nm D32-state repumper laser

638 nm F72-state repumper laser

739 nm extended cavity diode laser

AOM

PMT

Ion trap UHV chamber

vacuum pumps

Fibre-free space interface

Mu-metal shielding

EC

T

B

Magnetic field

Temperature

Ion fluorescence Processor

Clock laser frequency step and control

4355 nm

Frequengy doubling II

3695 nm cooling wave

Low-drift cavity

Low-drift cavity 171Yb+ 435 nm ion clock unit

876 nm clock laser wave from MOLO

Frequengy doubling

Fig 7 Simplified schematic of the single-ion optical clock unit AOM acousto-optic frequencyshifter PMT photomultiplier tube EC endcap T temperature stabilization system B magneticfield

clock transition at 4355 nm can be interrogated by spectrally narrow and stableclock laser light A diagram of the simplified energy level scheme for 171Yb+showing the relevant cooling auxiliary and clock transitions is given in Fig 6All these wavelengths are provided by diode lasers operated in fundamentalor frequency doubled mode The clock laser light is obtained from the 871 nmwave provided by the MOLO unit (see Section 423) after frequency-doublingin single pass by a KNbO3 nonlinear-optical crystal to 4355 nm The clocktransition spectral profile is observed by stepping the clock laser frequency bymeans of an acousto-optic frequency shifter (AOM in Fig 7) and recordingthe statistics of the quantum jumps in cooling laser induced fluorescence asa function of this frequency Under clock operation the AOM repeatedlysteps back and forward between the half-intensity points on the transitionprofile monitoring the signal imbalance between these points Any detectedimbalance is servo-corrected to zero by feedback to the AOM

In detail the optical clock architecture will comprise (Fig 7)

bull an RF end-cap trap for ionising and confining a single 171Yb+ ion withinan ultra-high vacuum chamber pumped by a small ion pump and non-evaporable getter pump The trap is driven by means of an ac voltage ofa few hundred volts at a drive frequency of sim10 MHz A small oven withseveral mg of 171Yb isotope is heated to provide a low flux of Yb atomsfrom which single ions can be ionised by electron bombardment from ahot wire filament within the trap potential well Low voltage dc is applied

Exp Astron

to additional electrodes to position the ion precisely at trap potentialcentre in order to minimise the ionrsquos micromotion at the drive frequencyThis needs automatic periodic monitoring and correction especially in theperiod after re-loading an ion into the trap External magnetic field coilsin three orthogonal axes allow the nulling of external fields and settingof a fixed field of sim1 μT The trap and coils are surrounded by mu-metalshielding to minimise external field changes during the orbit A level oftemperature control of 1 K of the region surrounding the trap is requiredfor maintaining the blackbody-induced frequency shift uncertainty withtemperature at the 10minus17 level

bull a laser platform to provide frequency doubling of an extended cavity739 nm diode laser for Doppler cooling of the ion on the 2S12(F = 1) minus2 P12 (F = 0) dipole transition at 3695 nm Light from an amplified diodelaser device frequency-doubled in a periodically poled LiTaO3 crystal insingle-pass will be used Typical laser powers used for driving the coolingtransition below saturation are about 2 μW for a beam waist in the trapof about 50 μm By modulation of the injection current of the 739 nmdiode laser at 147 GHz a sideband is generated that excites the 2S12 (F =0) rarr2 P12 (F = 1) ldquorepumperrdquo transition in order to avoid optical pump-ing between the ground hyperfine states

bull The platform also houses the auxiliary lasers at 935 and 638 nm The935 nm diode provides repumping of the ion from the 2 D32(F = 1)metastable level after occasional branching decays to this level during thecooling sequence The 638 nm diode allows fast recovery of the ion fromthe very-long-lived 2F72 metastable state after very occasional collisionaldecay to that state Currently extended cavity lasers are forseen butdistributed-feedback lasers may be available in the near future

bull a high NA lens imaging system and photomultiplier detection system torecord the statistics of 369 nm fluorescence quantum jumps as a functionof clock laser frequency step

bull a fibre system to deliver the various cooling auxiliary and clock light fromsource to trap making use of achromatic doublets where necessary at thefibre-free space interface for launching into the trap

bull a monitoring and control processor which provides the primary coolingand clock laser pulse magnetic field and detection sequencing to observeand lock to the ion clock transition frequency The processor also monitorsfrequency and amplitude data necessary to determine normal laser andion operational conditions and initiate resetting and recovery algorithmswhere necessary and laser unit failure The actual magnetic field presenton the atomic volume is determined and actively stabilized so that aninaccuracy at the level of 1 times 10minus17 can be achieved

bull a redundancy level of at least two units for cooling clock and repumperlasers plus similar for frequency doubling crystals All redundancy unitsfor each wavelength will be fibre multiplexed as standard allowing redun-dant unit activation on determination of prior unit failure

Exp Astron

For an optimized quantum-projection-noise-limited performance of thesingle-ion clock Ramsey interrogation with a cycle time equal to the lifetimeof the metastable D32 level of 50 ms will be performed leading to an Allandeviation of about 27 times 10minus15 (τ s)minus12 This is sim45 times 10minus17 at 1 h satisfyingrequired stabilities for the science objectives Current laboratory results are astability of 1 times 10minus15 at 30 s averaging down to 3 times 10minus16 at 1000 s (PTBGermany) Also comparisons between two independent traps showed agree-ment at the level of several parts in 10minus16 over a series of eight measurementruns

The 171Yb+ ion clock offers a second option namely the 2S12ndash2 F72 octu-pole transition at 476 nm (see Table 1 and Fig 6) [48] It has an extremelylong upper state lifetime low electric quadrupole and second order Zeemansystematic shifts and therefore is a candidate for an even higher performanceclock if a clock laser system with lower instability than the above is providedApart from the clock laser it shares the laser system of the quadrupoleclock Thus with moderate additional resources the Yb+ ion apparatus couldprovide two optical clocks in a single package and specifically perform directfrequency comparisons between the quadrupole and the octupole transitionwith a large sensitivity Aα

The ion clock will operate under microprocessor control providing thecontrol algorithms for the integrated pulse sequencing for cooling repumpingclock interrogation magnetic field switching cooling fluorescence monitoringThe controller will also provide error flags for critical processes and systemreset and correction where necessary This will include ion re-loading andmicromotion reduction algorithms Automatic monitoring of laser power andspectral quality with signal re-optimisation or laser failure determination isrequired Redundant laser units will be pre-aligned and multiplexed into thefibre delivery to the trap so that redundant units can be readily activated in thecase of diode laser failure

423 Microwave-optical local oscillator (MOLO)

This oscillator provides two ultrastable frequency outputs one in the opticalregion the other in the microwave region the two frequencies being coher-ently related The MOLO is composed of three subunits the clock laser thereference cavity and the frequency comb see Fig 8

The clock laser subunit delivers the light for the excitation of the opticalclock transition It is generated by an extended-cavity diode laser at a wave-length of 871 nm Most of the output wave of this laser is sent to the ionclock The laser subunit containing two lasers for redundancy will require3 kg 3 l 15 W The frequency instability of the clock laser must be superiorto the minimum clock instability 27 times 10minus15 (τ s)minus12 for τ up to the timeconstant (sim10 s) of the servo system that locks the laser to the ionrsquos resonancesignal In order to achieve this the laser frequency is stabilized to the reference

Exp Astron

871 nm clock laser

EO

Detector and servo

AOM Femtosecond

frequency comb

To optical clock 10 GHz To FCDP

Microwave-optical local oscillator

ULE cavity

Fig 8 Microwave-optical local oscillator (MOLO) consisting of clock laser reference ULE cavityand femtosecond frequency comb (OFC) Radiofrequency inputs and outputs are not shown EOelectro-optic modulator ULE ultra-low expansion AOM frequency shifter

cavity where the PoundndashDreverndashHall method is used About 10 μW of thelaser output are used for this

The system will exhibit sim04 Hz (sim1 times 10minus15 relative) linewidth and afrequency drift lt01 Hzs The cavity consists of two high-reflectivity mirrorsoptically contacted to a cylinder of diameter 70 mm length 150 mm made ofULE (a glass exhibiting ultra low thermal expansion coefficient) with a cavityfinesse of about 300000 Residual vibrations of the satellite (arms lt 1 times 10minus6 g)should not be a limiting factor to the desired linewidth for safety marginan optimized cavity support can reduce the sensitivity to accelerations Foracoustic and thermal isolation the cavity resides inside an aluminium vacuumchamber equipped with a small (3 ls) ion-getter pump Two stages of polishedaluminium shields are implemented around the cavity which are activelytemperature stabilized so that the cavity is precisely kept at the zero crossingtemperature of the ULE thermal expansion coefficient Heat application andremoval is by thermoelectric elements between the shield and the vacuumchamber Additionally the entire vacuum chamber is temperature-stabilized

During transportation and launch the cavity will be rigidly fixed insidethe vacuum chamber by pressing on it from the sides by piezo-mechanicalactuators allowing accelerations of several g After bringing the satellite onthe orbit the actuators will be released

The optical setup for frequency stabilization will be a classic Pound-Drever-Hall scheme in reflection using an electro-optic modulator (EO) Couplingof the clock laser radiation to the cavity is via single-mode polarization main-taining optical fiber The fiber in- and out-couplers are mounted on miniaturepiezo-motor driven multiaxis translation stages Microprocessor control ofthe translation stages performs a laser-to-fiber incoupling stabilization andlaser-to-cavity mode-matching optimization (positioning inaccuracy lt10 μmalignment inaccuracy lt10 μrad)

The frequency stability of the clock laser will be limited by thermal 1f noisein the mirrors with an fractional Allan deviation slightly below 1 times 10minus15 from1 s to a few 10 s Resources including control electronics are 15 l 12 kg 8 W

Exp Astron

A femtosecond fiber laser based optical frequency comb generator (OFCFig 8) will be used for the clock frequency comparisons and to obtain aspectrally ultrapure microwave local oscillator output for the operation ofPHARAO for the microwave link and for the GNSS receiver The fem-tosecond laser is based on erbium-doped fiber pumped by fiber-coupled laserdiodes It emits 100 fs pulses at a repetition rate of 250 MHz with an averagepower of 100 mW The emission spectrum is centered around 1560 nm Spec-tral broadening to more than one octave is obtained via nonlinear interactionin a microstructure fibre so that the range from about 1050 to 2100 nm iscovered

Second harmonic generation of a portion of the broadened spectrum allowsobtaining the required light for the measurement of the clock laser frequencyat 871 nm The OFC will be referenced to the clock laser frequency (afterintroduction a controllable shift via the frequency shifter AOM) by makingthe carrier envelope offset frequency vanish so that the repetition rate fr isa high-order subharmonic of the optical clock reference frequency f1 = m1 frThe link between the optical and the microwave domain is then simply thedistribution of the repetition rate frequency or more conveniently a high(order 40) harmonic of the comb repetition rate 10 GHz

It has been demonstrated by several groups that frequency comparison andfrequency division with an OFC can be performed to very high accuracy andstability beyond the limitations imposed by the optical clocks themselves Forexample a fiber laser OFC with an instability (Allan deviation) of 6 times 10minus17

in 1 s has been demonstrated [51] OFCs based on fiber technology possessgood long-term reliability and high power efficiency The projection for a spacequalified OFC based on current industrial terrestrial technology is 16 kg mass20 l volume and 38 W power requirement

424 Frequency comparison and distribution package (FCDP)

This package distributes the ultra-stable 10 GHz reference frequency outputof the MOLO to the MWL performs frequency comparisons between MOLOand PHARAO and generates a variety of different reference signals for theindividual instruments including the 250 MHz reference clock to the OFCAll signals are coherent to the MOLO In the FCDP a USO is locked to theMOLO signal by a digital PLL A divider chain generates timing pulses (1 pps1000 pps etc) and keeps a continuous local time scale From the digital phasecomparison data the phase noise density spectrum between the USO and theMOLO is available by telemetry (TM) to assess the overall frequency purityin-flight

As the FCDP distributes the full performance reference from the MOLOdirectly to the MWL there is minimum impact of the FCDP on overall clockcomparison performance The USO-phase lock loop uses a similar designas for ACES-FCDP The digital PLL with linear phase detector allows theUSO to be slightly offset from its reference which is selectable by groundcommand This results in an accurate on-board time scale despite any offset

Exp Astron

of the MOLO reference frequency which can be synchronised to UTC Thefrequency synthesis chain re-uses modules developed for ACES-FCDP andtheir frequency inaccuracy Expected relative uncertainty (Allan deviation)of the frequency comparison between two signals provided to the FCDP is47 times 10minus17 at 1000 s 95 times 10minus18 at 10000 s

During non-availability of the MOLO ie during system start-up andcontingency conditions the USO runs as flywheel (ie it is unlocked withinstability lt15 times 10minus13 from 1 to 100 s) It will still provide useable butless accurate references to the instruments including the reference to OFCCommon view ground-to-ground time- and frequency comparison and supplyof the remaining instruments remain possible at the USO provided accuracy

Mass volume and power consumption are 8 kg 7 l 12 W Controldata rateis 10 kBday 50 kBday (FCDP alone)

425 Microwave link (MWL)

The microwave link is a two-way triple-frequency microwave link (Fig 9) Itis designed to perform accurate frequency comparison as well as time-transfer

Ku-Band Up-linkPower Tx 5 WCarrier 13475 GHzPN-Code 100 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

S-Band Down-linkPower Tx 1 WCarrier 2248 MHzPN-Code 1 MChips1pps 1 time marker sData 5 kBits

Ku-Band Down-linkPower Tx 1 WCarrier 1470333 GHzPN-Code 100 MChips1pps 1 time marker sData 5 kBits

EGE SC

S

Ku

KaKa-Band Up-linkPower Tx 5 WCarrier 345 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

Ka-Band Down-LinkPower Tx 1 WCarrier 32 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 5 kBits

GS Antenna diameterEGE terminal 1 mACES terminal 06 m

Fig 9 EGE microwave two-way link concept and characteristics See text for a description Thelink will be able to perform frequency comparisons with up to ten ground stations simultaneously(only one active comparison is shown) pps pulse per second Tx transmitter SC spacecraft

Exp Astron

between on-board and on-ground clocks Two-way operation allows for com-pensation of the effects due to spacecraft location and its motion relative toground Its two bi-directional links operate in Ka-band (32345 GHz) andKu-band (147135 GHz) The high carrier frequency of the up- and down-links allows for a noticeable reduction of the ionospheric delay A third down-link channel is in S-band (22 GHz) and is used to determine the ionospheretotal electron content MWL uses a combination of carrier phase and codephase data to measure with high precision the absolute ionospheric delayas well as ionospheric delay variations during a ground contact The useof three frequency bands allows to resolve third order ionospheric termsKey modulation parameters are Ka band Tone 250 MHz Ku band PN100 MChips S-band 1 MChips

The EGE-MWL concept is a considerably improved version of the ACES-MWL Microwave carrier generation is directly from the 10 GHz microwaveoutput available from the MOLO and having exceptionally low close-in phasenoise without intermediate frequency dividers or other complex synthesiserstages commonly used in RF transponders The high frequency carriers arekey to achieve the desired frequency comparison accuracy

The Ka-band carrier is generated by tripling the FCDP-generated signalThe Ku-band is subsequently generated by division by two using a verystable regenerative divider Finally the assigned radio frequency is obtainedby adding a low frequency signal using an image reject mixer A mixedanalog-digital receiver design will enhance its performance by use of a flexibledigital signal processing and digital control loops The number of simultaneousground contacts is increased from four (as in the ACES mission) to ten or moreusing the digital concept while the number of physical receiver channels perband is reduced from four to two one for operation to ground and one for in-flight calibration The concept allows adaptation of receiver loop parametersto optimise for the different portions of a highly elliptic orbit while reducingsystematic uncertainties due to high Doppler and Doppler rate Real-timedata exchange to ground terminals is by additional data modulation on all RFsignals The spacecraft carries two different small antennae for perigee and forapogee with gains of 8 dBi 20 dBi beamwidths 100 (full earth coverage) 14respectively power is 1 W for each

The MWL end-to-end performance between the attached clocks includingspace- and ground terminals is predicted to be a relative Allan deviation of1 times 10minus16 at 1000 s (perigee) and 1 times 10minus17 at 10000 s (apogee) Thesevalues are based on conservative implementation losses and other instrumentallosses assumed as 6 dB at apogee and 10 dB at perigee including additionallosses due to atmosphere An attitude knowledge of 1 is required

The MWL link will also support precise orbitographyndashcomplementary tothe GNSS receiver Generation of ranging data is simultaneous to two-waytime transfer and is available under all weather conditions Orbit determina-tion accuracy is expected to be better than 1 cm well matching EGE clockrequirements The value is expected based on demonstrated results fromother missions (METEOR ERS-2 TOPEXPOSEIDON) Laser ranging to

Exp Astron

the CCR array is used a few times during the mission to calibrate the MWLranging

In case of non-availability of the ultrapure signal from MOLO an additional10 GHz output is available from the FCDP to support ground-to-ground clockcomparison in common-view mode even after the nominal mission periodwhen the neither the clocks nor MOLO may be available any longer Basedon the PRARE experimentrsquos 12 years of operation this can be assumed forthe FCDPMWL combination as well The potential extended operational pe-riod makes the proposed configuration particularly attractive to metrologicallaboratories that rely on availability of long-lasting data in support of thegeneration of UTC

The resources of MWL are as follows mass 18 kg volume 18 l power55 W TC 2 kBday TM 120 MBday

426 Auxiliary units

The star camera assembly on the EGE spacecraft will provide real-timeattitude data primarily for the orientation of the spacecraft (with accuracy of1 in real-time) and the proper reduction of the GNSSSLR data

A GNSS receiver is used for the precise orbit determination time andfrequency transfer coarse positioning (lt1 m) for real time use by the Attitudeand Orbit Control as well as the back-up for time tagging of all payloaddata For the time and frequency transfer the receiver is nominally using theonboard frequency reference The receiver assembly includes three specialantennas By the year 2015 more than 100 GNSS satellites are expected tobe available for precise orbit determination permitting orbit inaccuracy of 1ndash2 cm RMS for altitudes up to approx 10000 km At apogee a combinationof GNSS signals and MWL ranging will be used with expected inaccuracyof 10 cm

5 Basic spacecraft specifications

The mission foresees a single spacecraft with a mission duration limited toapprox 3 years for cost reasons Attitude control along three axes withless than 1 error on a timescale of 1000 s is required The control can beperformed with cold gas thrusters magnetotorquers or reaction wheels Theselection will be done on the basis of the vibrations induced by these systems(desired level arms lt 10minus6 g)

Cold gas thrusters or monopropellant thrusters will be required to maintainthe orbit parameters during the mission by small orbit correction manoeuvresand will also guarantee a proper tuning of the orbit parameters after the releasefrom the launcher The ground segment (telemetry control and data transfer)could be via non-ESA stations user-operated to save costs

Further resources are indicated in Table 2

Exp Astron

The satellite bus design drivers are nadir-pointing microwave antennasthermal stability for the payload instruments no strong magnetic sources lowmicrovibrations harsh radiation environment due to crossing of the radiationbelts once per orbit power demand by science payload (with the possibility toreduce power demand by stand-by of some components during specific missionphases eg eclipse) mass limit due to required high altitude with low costlaunch

The satellite bus design can be based on established platforms such asFLEXBUS or PRIMA used on several past missions Their design philosophydiffers one having body-mounted solar arrays the other wings The latter canprovide larger amounts of power The vibration level induced by the panelswhen they are fixed and the duration of solar array drive periods per orbitwill have to be considered in the selection The FLEXBUS concept was usedfor a mission already requiring a low level of residual accelerations and theplatform is optimised for this parameter

For most of the satellite components one can consider using compo-nents and electronic units from navigation or communication satellites whichtypically cruise at altitudes between 15000 and 25000 km altitude LISAPathfinder which will cross the radiation belts during spiralling to theLagrange point will face similar problems concerning radiation as EGE as wellas XMM which is in a Molniya-type orbit with similar radiation characteristics

The instruments will therefore be located in the satellite structure such asto realise ionizing radiation shielding by the spacecraft structure and non-sensitive equipment to the maximum extent possible the need for additionalradiation shielding (including trapped proton radiation) for especially sensitiveparts of the instruments must be determined by ground tests It should bementioned that an orbit with higher perigee and less radiation dose is also anoption as it leads to a tolerable reduction of the science goals

The required plusmn3C temperature stability at the interface to the payloadenclosure is feasible The thermal control will employ the usual means (anenclosure with high thermal inertia heaters and a suitable radiator) Thelocation of the radiator will need a special study given the non-standard orbit

Magnetic field present during the mission can affect the properties of theonboard clock Using SPENVIS the magnetic field has been calculated alongthe spacecraft orbit the field intensity presents maxima around 02 G anda minimum value of the order of 10minus3 G As discussed above this is not alimitation to the instrument performance

6 Discussion and conclusion

The accuracy of the measurements depends crucially on the stability andaccuracy of the clocks While the performance of ground clocks (even thoseeventually deployed in new ground stations dedicated to EGE) will likelyimprove significantly from the present level the space clock design will beldquofrozenrdquo approx 10 years before data acquisition Thus their performance

Exp Astron

level will be significantly below the future ground station clock performanceAs specification we have chosen a performance level that has been demon-strated with laboratory breadboards recently We see no fundamental reasonthat this laboratory performance cannot also be achieved in a space ion opticalclock The experience with the PHARAO cold atom clock has shown thatan engineering model of a laser-based cold atom clock has a performancesimilar to that of a laboratory prototype The perturbation of the ion clock fre-quency due to magnetic fields along the orbit can be modelled and suppressedwith shielding and the residual effects measured and compensated for Themoderate temperature stability (plusmn 3C) to be provided in the experimentalbay can be achieved with thermal shields Accurate temperature controlof particular units is done by active stabilization of the units themselves

One important aspect is radiation which may affect all optical and trapcomponents as well as the lasers and reference cavity The radiation envi-ronment on the EGE orbit is significantly harsher than the near-earth ISSorbit on which ACES will operate The radiation effects must therefore becharacterized comprehensively with appropriate tests and then the radiationshield for the EGE instruments designed

In summary the Einstein Gravity Explorer mission (EGE) is a satellitemission aimed at a precise measurement of the properties of space-time usingatomic clocks By testing one of the most fundamental predictions of EinsteinrsquosTheory of General Relativity the gravitational redshift and thereby searchingfor quantum effects in gravity it explores one of the most important frontiers infundamental physics The EGE mission reaches beyond fundamental physicsproviding an important spin-off in geophysics Furthermore the technologicalaspects of the mission are ground-breaking for future time-keeping and ref-erence frame systems in space and thus are also expected to be of wide useto science and technology in general The payload of the mission has beenconceived as a compromise between cost and performance needing to fit thecost envelope of an ESA medium-class mission1

Acknowledgement We thank S Cesare for helpful information

References

1 Will CM The confrontation between general relativity and experiment Living Rev Relativ9 3 (2006)

2 Ni W-T Empirical foundations of the relativistic gravity Int J Mod Phys D 14 901 (2005)doi101142S0218271805007139

1Finally we point out that a spacecraft with a significantly reduced and thus lower-cost payload (noclocks only MOLO FCDP MWL) on a circular orbit of appropriate height could be used for along-duration earth science mission dedicated specifically to the measurement of the geopotential

Exp Astron

signals from two distant ground clocks are sent to the satellite andcompared there The frequency difference between two identical groundclocks is determined if systematic instrumental errors are negligible bythe difference in gravitational potential at the two clock locations Thispotential difference includes a terrestrial and a solar contributionThe solar contribution varies according to the orientation of the clockpair with respect to the Sun with a 1 day-period For two clocks at asurface distance of sim6000 km and suitable orientation with respect tothe sun the normalized modulation amplitude is sim= 4 times 10minus13 This is thesame as the earth potential difference for 4 km height difference It ispossible that ground clocks will have reached an inaccuracy at the levelof 1 times 10minus18 at the time of the mission in 2015 Using the microwave linka comparison of ground clocks over sim3 h (typical common-view duration)with a resolution of 1 times 10minus17 can be performed limited by the instabilityof the link Repeating the clock comparison at each common-view win-dow (n = 1000 comparisons) will allow averaging and a measurementof the solar gravitational shift with fractional inaccuracy of 1 ppm Incomparison the best current results for the solar gravitational frequencyshift are at the few-percent level [20 21] A moderate averaging out ofground clock systematic shifts is assumed as well as modeling of the earthgravitational potential variations on the daily timescale at the appropriatelevel

(4) Comparisons of on-board clocksThe two on-board clocks are continuously compared these data areanalyzed for violation of LPI (Eq 4) Since the measurement is a nulltest the required accuracy of the gravitational potential at the satellitelocation is low and easily satisfied

24 Test of local position invariance and search for neutron scalar charge

The test of LPI is performed in conjunction with the measurement of thegravitational frequency shift

With approach 23 (1) the optical clock will be able to measure K (α Earth)with accuracy 30 ppb (spec)

Combining the results of the microwave and optical clocks will determineK(meQCD Earth) + εK(mqQCD Earth) with accuracy 85 ppb (spec)where ε asymp minus01

Simultaneously the on-board comparison 23 (4) yields the combination19 K(α Earth) +K(meQCD Earth) + εK(mqQCD Earth) with accuracy85 ppb (spec)

Note that these measurements are fully complementary to current andfuture terrestrial clock-clock comparisons which are sensitive to K(X Sun)and can be combined with the latter to determine or set limits to the neutronscalar charge

Exp Astron

25 Theoretical analysis of frequency comparison signals

The on-board clock-clock comparison signals can be analyzed straightfor-wardly in terms of a violation of LPI since signal propagation effects areabsent

The determination of the gravitational frequency dilation via ground-satellite clock comparisons is more complex (Eq 1 is strongly simplified) andrequires an accurate treatment of all relevant relativistic effects Gravitationin the solar system is described in the barycentric (BCRS) and geocentric(GCRS) non-rotating celestial reference systems by means of post-Newtoniansolutions of Einsteinrsquos equations for the metric tensor in harmonic gaugescodified in the conventions IAU2000 [22] In these conventions the relativisticstructure of Newton potential (order 1c2) and gravitomagnetic (order 1c3)

potentials are given For the Newton potential the multipolar expansion ofthe geopotential determined by gravity mapping is used The post-Newtonianeffects of the metric (described in terms of dimensionless parameters β and γ )

can be included Orbits of satellites are evaluated in the GCRS at the levelof a few cm by means of the Einstein-Infeld-Hoffmann equations at the order1c2 [23] using the relativistic Newton potential Instead the trajectories of theground clocks in the GCRS are evaluated from their positions fixed onthe Earth crust (ITRS International Terrestrial Reference System) by usingthe non-relativistic IERS2003 conventions [24 25]

For the propagation of light rays (here radio signals) between the satelliteand the ground stations one uses the null geodesics of the post-Newtoniansolution in the GCRS The timefrequency transfer properties have beentheoretically evaluated for an axisymmetric rotating body [26ndash29] at the order1c4 (ie with Newton and gravitomagnetic potentials developed in multipolarexpansions) These results have been developed for use in the ACES missionThe relative frequency dilation can be expressed as

ν

ν= 1 + 1

cL1 +

4sum

n = 2

1

cn(Ln + Gn)

where Ln describes special-relativistic Doppler effects and Gn the general-relativistic effects These quantities are time-dependent due to the orbitalmotion For concreteness we report the orders of the terms The kinematicaleffects are |L1c| lt 1 times 10minus5 |L2c2| lt 2 times 10minus10 |L3c3| lt 3 times 10minus15 |L4c4| lt

7 times 10minus20 The Newtonian contributions are G2 = G(M)2 + G(J2)

2 + G(J4)2 + G(J6)

2(M is the mass monopole Jn are the lowest multipoles) with estimates|G(M)

2 c2| lt 3 times 10minus11 |G(J2)2 c2| lt 2 times 10minus13 |G(J4)

2 c2| lt 3 times 10minus16 |G(J6)2 c2| lt

1 times 10minus16 For G3 = G(M)3 + G(J2)

3 the estimates are |G(M)3 c3| lt 2 times 10minus14

|G(J2)3 c3| lt 1 times 10minus18 Finally the G4 = G(M)

4 + G(S)4 contributions (G(S)

4 is thelowest gravitomagnetic effect) are of order |G(M)

4 c4| lt 1 times 10minus19 |G(S)4 c4| lt

1 times 10minus22 and thus not relevant for EGE

Exp Astron

The available theory is fairly complete for the data analysis in mode 23 (1)described below For modes 23(2 3) the theory will need to be generalizedto include the effect of the gravitational potential at the location of the groundclock(s) (including time-varying effects such as the tides) and of the solar andlunar potentials

The uncertainties of the various contributions above are minimized by ausing precise EGE orbit data obtained by satellite tracking and by usingcurrent and future accurate earth gravity information Required orbit positionknowledge is at 1 cm level near perigee and sim50 cm near apogee reachablealready today with the proposed orbitography approach described below Thevalidity of the special-relativistic Doppler shift has been and will continue to beindependently verified with increased accuracy by spectroscopy of relativisticatomic ion beams [30] Potential violations of this aspect of Lorentz Invari-ance are expected to be sufficiently bounded so as not to affect the signalinterpretation

26 Local Lorentz invariance tests

There has been an explosion of interest in Local Lorentz invariance in re-cent years [31 32] with numerous astronomical studies and high-precisionexperiments applied to search for violations of Lorentz invariance Signifi-cant developments have also occurred on the theoretical side and a theory(standard model extension SME [33]) has been worked out that provides aunified framework for describing and analyzing Lorentz Invariance violationsin a variety of systems Two aspects of Lorentz Invariance can be tested withEGE

261 Independence of the speed of light from the laboratory velocity

A possible dependence of the speed of light on the speed v of the laboratorycan be modeled according to the MansourindashSexl test theory [34] by

c (v) = c0(1 + Bv2

c2

)

Here v 377 kms is the velocity with respect to the cosmic microwavebackground the cosmologically preferred frame and B is a combinationof parameters describing deviations from the usual Lorentz transformationformulas B = 0 if Lorentz Invariance holds For a terrestrial experiment therotation of the Earth around its axis modulates v with a 300 ms amplitude Adependence c(v) can be searched for by measuring the frequency differencebetween a resonance frequency of a highly stable optical cavity (cavity fre-quencies are proportional to c) and an optical clock (Kennedy-Thorndike-typeexperiment [35]) and determining the modulation of the frequency differencecorrelated with the modulation of v The advantages of a space experimentare the high orbital velocity and strongly reduced cavity deformation thanksto microgravity [36 37] On the proposed elliptic orbit v varies between+4 and minus4 kms over approximately 1 h This variation is 13 times larger

Exp Astron

than for the case of a terrestrial experiment The shorter time scale is alsoadvantageous since the drift of the cavity is more predictable In the EGEmission the reference cavity of the clock laser is used The large numberof orbits improves the statistics An improvement of the accuracy of B by afactor 20 compared to the best current terrestrial results is realistic The dataanalysis can also be performed in the framework of the SME theory takinginto account complementary bounds obtained from terrestrial experimentsand astrophysical observations

262 Independence of Zeeman splitting frequency on the directionof the magnetic field

One class of tests of Lorentz Invariance consists in measuring the frequencysplitting between two levels of a quantum system induced by a static magneticfield as a function of orientation of the field direction with respect to thestars (HughesndashDrever-type experiments [31]) This class addresses the so-called matter sector and is complementary to combined-photonmatter-sectortests such as the above The most precise experiments are performed withatomic clocks eg masers or cold atom clocks [38] The EGE instruments allowsuch experiments since magnetic fields are applied continuously in the clockor repeatedly for calibration Compared to terrestrial experiments the highorbital velocity of EGE could lead to an improvement by a factor sim20

27 Application to geophysics

Using GPS coordinates of a point on the Earth can be obtained with aninaccuracy well below 1 cm in a well defined international terrestrial referencesystem The coordinates are purely geometrical and do not contain any gravityinformation A local gravitational potential UB is determined with respect toa reference gravitational potential U A by ldquolevellingrdquo the surface of the Earthie sequentially measuring gravity g and height differences every ca 50 m

UB minus U A = minusBint

A

g (r) middot d n

where n is the difference vector between subsequent locations The disadvan-tage of measuring height differences by the levelling method is a random walkeffect of accumulated errors

In the classical definition a geoid is defined as the particular equipotentialsurface nearest to mean sea level In these terms the geoid serves as areference surface for measuring the height and also to define a referencefor the gravitational potential Such a vertical reference frame historicallywas realized for a country or several countries by determining the mean ofsea level observations at tide gauge stations taken over long period of timeHowever modern satellite altimetry missions such as TOPEXPOSEIDON orENVISAT show that departure of the mean sea level from an equipotential

Exp Astron

surface may reach up to several meters on the global scale see eg [39]Therefore height systems based on different tide gauge stations may differin the realisation of the geoid and may differ with respect to the reference byseveral meters

The best gravity field models obtained from satellite data (eg missionGRACE) have reached a precision of about 1 cm over sim250 km half wave-length [40] Satellite data also reveals changes in time [41] However for typicalEarth topography with height variations of eg 1000 m over 30 km horizontaldistance one may expect variations in the geoid of about 80 cm Such ahigh-spatial-frequency signal in geoid variations cannot be detected by spacegravity missions and requires a combination of satellite and terrestrial gravitymeasurements like gravity anomalies deflections of vertical and GPSlevellingpoints The best combined global gravity field models are provided up to de-gree and order 360 in terms of spherical harmonic representation correspond-ing to a half-wavelength of about 55 km However in combining the satelliteand terrestrial gravity field measurements the problem of terrestrial data givenin different height systems between continents and different countries remainsand has to be tied to satellite measurements

The geodetic scientific community is currently establishing a Global Geo-detic Observing System (GGOS) [42] Its objectives are the early detectionof natural hazards the measurement of temporal changes of land ice andocean surfaces as well as the monitoring of mass transport processes in theEarth system Global change processes are small and therefore difficult toquantify Therefore the required precision relative to the Earthrsquos dimensionis 1 ppb GGOS will be established by the combination of geodetic spacetechniques (GPS Laser-Ranging very large baseline interferometry) andrealized by a very large number of terrestrial and space-borne observatoriesThe purely geometric terrestrial 3D coordinate system is in good shape andfully operational It has to be complemented however with a globally uniformheight system of similar precision The current precision level of regionalheight systems in terms of gravity potential differences is in the order of1 m2s2 (10 cm) with inconsistencies between these various systems up toseveral 10 m2s2 (several meters) The actual requirement in the context ofGGOS is 01 m2s2 (1 cm) with the need of a permanent ie dynamical controlThis requirement of high precision height control comes from the need tounderstand on a global scale processes such as sea level change global andcoastal dynamics of ocean circulation ice melting glacial isostatic adjustmentand land subsidence as well the interaction of these processes Only by meansof monitoring in terms of gravity potential changes at the above level ofprecision the change of ocean level can be understood as a global phenomenonand purely geometric height changes be complemented by information aboutthe associated density or mass changes

The geoid can also be defined in a relativistic way [43] as the surface whereaccurate clocks run with the same rate and where the surface is nearest tomean sea level The relation between the differences in the clock frequenciesand the gravitational potential is given in simplified form by Eq 1 At present

Exp Astron

there is no operational way to compare frequencies of the already availableoptical clocks on the global scale at the same level as their accuracy wouldallow With EGE and by then improved ground optical clocks it will becomepossible to obtain gravitational potential differences on a global scale by com-paring frequencies This would be a significant new dimension to gravity fielddetermination since such observables are given on the global scale and providein situ local gravity information at the same time EGE will allow clock-basedgravitational potential mapping with the same payload instruments used forthe fundamental physics experiments The on-board clocks are actually notrequired for this purpose MOLO MWL and FCDP suffice Thus EGE willallow establishing a global reference frame for the Earth gravitational fieldwith accuracy in the order of a few cm in terms of the geoid heights Thisassumes that by the time EGE is operational (mobile) ground clocks withfractional inaccuracy at the level of 10minus18 are available the measurementsare limited by MWL noise and thus require a long integration times to matchthe expected ground clock performance Such a reference frame will serve asreference for all gravity field modeling and clocks will be used to define theinternational atomic time scale and the Earth gravitational potential at thesame time

In summary with optical clocks geodesy will undergo a second revolutionafter geometry now being measured by clock-based GNSS systems also physi-cal heights and potential will be measured by (mobile) optical clocks using thegravitational redshift effect

28 Application to frequency standards and terrestrial fundamentalphysics studies

As has been described repeating ground clock comparisons and integratingover several months will provide global ground clock frequency comparisonsat the level of 1 times 10minus18 more than two orders of magnitude below the currentGPS and two-way satellite time and frequency transfer methods In the timeand frequency community the availability of such a global high performancemicrowave link will accelerate the process leading to a redefinition of the SIsecond based on clocks operating in the optical domain

The global character of the ground clock comparisons made possible byEGE can also contribute to the search for a time-variation of the fundamentalconstants

The comparisons do not require availability of the on-board clocks The riskof this type of measurements is therefore reduced Moreover the lifetime ofmany subunits required for this measurement (MWL and FCDP see below)can exceed 10 years (from previous experience with similar devices) andthus offers a motivation to increase the mission duration beyond the nominal3 years

Further applications to metrology are listed under the secondary goals inSection 21 above

Exp Astron

3 Mission profile

31 Orbit

The orbit must satisfy the following requirements

Primary 1 Large gravitational potential difference between apogee andperigee

2 sufficient contact time with at least two ground stations atperigee

Secondary Simultaneous visibility of satellite from distant (inter-continental) ground stations for several hours

To satisfy the primary objective the orbit must be highly elliptic The perigeeshould be low A very low perigee means a short contact time which decreasesthe accuracy of the frequency measurement A compromise is a perigeealtitude of sim2500 km For a sufficiently high apogee this yields a U of 60of the theoretical maximum between the earth surface and infinity

The baseline scenario is an orbit of approx 12 h period with the followingproperties

Apogee Altitude 37856 km (minus 10 deg latitude one over Atlantic andPacific resp)

Perigee Altitude 2500 km (+ 10 deg latitude one over Guatemala andMalaysia resp)

Inclination 634 argument of perigee 170RAAN 25 True Anomaly 220Apogee drift rate 70 kmday perigee drift rate sim 14 kmday

True anomaly and right ascension of the ascending node (RAAN) wereoptimized to minimize orbital precession Figure 3 shows the ground track andthe ground coverage at one of the perigees Common-view contact time forstations in Europe is about 6 h while between a station in Europe and in theUSA it is about 35 h Frequency comparison windows between a given groundstation and EGE repeat every 24 h

A final selection of the orbit requires an in-depth study of the costs of thepropellant and maneuvers as well as of the setting up of new ground stations

32 Ground segment

The mission requires two separate ground segments the science groundsegment and the standard satellite control segment which supports telemetryand telecommand (TMTC)

Exp Astron

Fig 3 Baseline EGE orbit One (out of two) apogee and one (out of two) perigee ground pointsare shown The ground coverage at one of the perigees is shown with circles at elevation anglesteps of 10 starting from 10

321 Science ground segment

As described above one part of the mission is to perform frequency com-parisons between ground-to-space and ground-to-ground (common-view andnon-common view) The use of a two-way microwave link (MWL) systemhas the advantage that a precise orbit or the precise relative velocity of thesatellite must not be known [4] The orbit determination requirements fora two-way system are significantly relaxed compared to the requirements tomodel the relativistic effects on the clocks The frequency comparison resultsare available both on ground and on the satellite

For ground-to-satellite clock comparisons five ground stations arefavourable two capable of viewing perigee three for viewing apogee Thisallows for the classic three-clock configuration where the performance of eachclock can be obtained individually The perigee stations will be observing thesatellite for sim25 min every perigee passage for the duration of the missionThe apogee stations will observe for sim3 h every apogee passage Of coursea larger number of stations as well as longer observation time are usefulfor robustness redundancy and orbit determination The stations could beinstalled on a temporary basis for the duration of the mission

The stations must have access to optical clocks of performance at least ashigh as todayrsquos state-of-the-art ground clocks but preferably higher Theywill include an optical frequency comb by means of which the optical clockfrequency is transformed to a radio frequency for comparison with the EGEclocks High-accuracy fiber-optic link technology already existing today allowsdistances of sim100 km between the ground station and the correspondingoptical clock allowing a certain freedom in the choice of the clock locationWith further increase of the fiber-optic link range currently being investigated

Exp Astron

the area over which the ground station can be positioned relative to the opticalclock will increase This option is of importance for the implementation of ageophysics measurement campaign for which a large number of clock locationsis of interest

The above ground stations will be complemented by some 15 additionalground terminals provided by the world-wide laboratories equipped with highperformance primary clocks at the time of the mission These will participate inorbit determination to 1 cm accuracy along the full orbit which can be achievedwith non-optical atomic clocks The EGE MWL is fully compatible to plannedACES ground terminals which might reduce the number of required stations

With a larger number of transportable optical clock ground stations de-ployed and moved across the continents a geophysics campaign could beimplemented to map the gravitational potential

Ground station equipment MWL uses dedicated small transportable termi-nals which can be easily installed at any location of interest Each terminalconsists of a steerable dish antenna of 1 m diameter with pointing accuracy of01 using a transmitted power of 5 W per band The high directivity antenna(beam aperture Ka-band 07 Ku-band 14 S-band 20) avoids multi-pathfrom nearby objects Antenna noise temperatures are Ka-band 450 K Ku-band 340 K Operation is at elevations above 10 and measurements areperformed above 20 For ranging purposes terminal location is determined byGPS surveying Each ground station computes its own ionospheric correctionsbased on a triple-band receiver The terminal has a built-in delay monitorwhich allows calibrated ranging during several months once its bias has beendetermined by laser ranging Its M and C interface is via local area networkto either a userrsquos computer or directly to the science network control centreby remote MampC The stationrsquos operation is fully automatic and unattendedSchedule and orbital data are received via LAN once per day Signal acquisi-tion is fully autonomous Data transmission is after each pass The terminal hasan uplink telecommand capability of 1 kBits and a downlink data capacity ofup to 10 kBitss For the local user there is a visual interface providing quick-look data from real-time comparison to the spacecraft clocks to verify theirgood health The final full performance product is calculated at the sciencedata centre

Local metrological data (temperature pressure humidity) are recorded tocorrect for tropospheric delay of the ranging data The main advantage ofMWL terminals compared to laser based systems is their all-weather capabil-ity although microwaves suffer higher propagation errors in the ionosphereand troposphere

The EGE ground terminal will profit from the ERS-2 mission PRARE(precise range and range rate equipment) [45] experiments whose terminalshad similar design and have been operated under extreme climatic conditionsincl Antarctica and high temperature regions from existing designs fromthe space segment (mission ACES) and no further special developments aredeemed necessary Cost per EGE master station is modest

Exp Astron

Satellite tracking via laser ranging Laser ranging is alternative to MWL rang-ing and also serves for absolute calibration of the latter For this purpose anon-board corner cube reflector (CCR) is provided Ranging will be done withthe International Laser Ranging Service Ideally the satellite will be rangedat every perigee passage and every apogee passage simultaneously with thefrequency comparison procedures Todayrsquos ranging precision is sufficient tosatisfy the science goals For apogee observations in the northern hemispherea large number of laser ranging stations is available The number of stationscapable of ranging at or near perigee in the southern hemisphere is six andthus is sufficient even considering unfavourable weather conditions at someof these The MWL ground stations are able to range even under thoseconditions

322 Control ground segment

Average data transfer rate is estimated at 300 MBday (approx 40 kBs onaverage) Data will be stored on board (2 GB capacity required) and will betransferred to a receiving station on the Earth once a day

XPLC

incl Data StorageIF to TMTC Link

HK amp TMTC

to ground

Opt Clock

AOM

MOLO

PHARAO

FCDP

10 GHz

100 MHz

250 MHz

GNSS Rx

MWLTampF Link

uplink Data

UTC(k)

TampF Space vs Ground

TM from GS

Tim

e S

cale

Orbit GNSS Time UTC

TampF Space vs Ground

MampC Time Scale

MampC

MampC

MampC

MampC

Mamp

C

PHARAO vs Opt Clock

PHARAO vs Opt Clock

10 GHz amp Epoch

10 MHz

amp Epoch

Epoch

10 MHz

amp Epoch

Opt Link

Fig 4 Schematic of the payload showing subunits and their signals The results of the frequencycomparison between the two on-board clocks (optical clock and microwave clock PHARAO) areobtained from a combination of the frequency values applied to the frequency shifters AOM (seeFigs 7 and 8) and those output from the frequency control and distribution package (FCDP)as indicated by the arrows ldquoPHARAO vs Opt Clockrdquo The satellite clock to ground clockcomparison results are output from the MWL XPLC is the on-board computer and is involvedin generation of signals for instrument control The optical link shown dashed is a possible add-onproviding an alternative to the MWL GS ground stations MampC monitoring and control HKhousekeeping TampF time and frequency

Exp Astron

4 Main payload instruments

41 Overview

The satellite payload consists of a pair of cold-atom clocks a microwave-optical local oscillator (MOLO) an on-board frequency distribution unit(FCDP) a satellite-to-ground clock comparison unit (MWL) and the onboardcomputercontrol system (XPLC) see Fig 4 Auxiliary units are the posi-tion determination GPSGalileo receiver (GNSS Rx) attitude monitor (startracker not shown) and the corner cube reflector array (CCR not shown)

The science goals require clocks of very high frequency stability on thetimescales of half the orbital period (6 h) as well as perigee passage (sim1000 s)see Fig 2 Only clocks based on cold atoms are suited for the science goalsTable 1 reports values for a subset of atomic clocks that have been investigatednot including clocks that require cryogenic operation or are too complex foruse in an medium-class mission The choice of a clock pair rather than a singleone fulfils several requirements it addresses more science goals improves thevalidity of some science results and it enables calibration accuracy evaluationand redundancy The validity of the science results is enhanced since theclocks differ not only in the employed atoms but also in terms of havingdifferent sensitivity to perturbations by electric and magnetic fields electronicserrors and misalignments A comparison of the results arising from each clockallows for an understanding of measurement errors Both the clocksrsquo and thefrequency comparison unitrsquos performances will be determined after launchand regularly verified during the course of the mission The only reliable

Table 1 Properties of different clock candidates for EGE

Clock Clock Fractional σ = Fractional Sensitivity Figure of Sensitivityatom frequency inaccuracy instability to α Aα merit |Aα| to meq

at 1000 s σ [1016] QCD

Sr+ 445 THz 3 times 10minus17 (p) 5 times 10minus17 (p) 042 08 (p) ndashYb+ (1) 688 THz 3 times 10minus16 (d) 3 times 10minus16 (d) 088 029 (d) ndash

3 times 10minus17 (p) 1 (p)Yb+ (2) 642 THz lt 1 times 10minus17 (p) 1 times 10minus17 (p) minus53 53 (p) ndashSr 435 THz 1 times 10minus16 (d) 1 times 10minus16 (d) 009 009 (d) ndash

lt 1 times 10minus17 (p) 09 (p)Yb 519 THz 7 times 10minus16 (d) 1 times 10minus16 (d) 031 03 (d) ndash

lt 1 times 10minus17 (p) 1 times 10minus17 (p) 31 (p)Cs 91 GHz 2 times 10minus16 (d) 1 times 10minus15 (d) 275 027 (d) 1Yb+ 126 GHz 1 times 10minus15 (p) 1 times 10minus15 (p) 34 034 (p) 1

A relevant figure of merit is the ratio of |AX | and fractional frequency instability σ on the perigeevisibility timescale The Cs clock is the only high performance clock with adequate sensitivity tothe particle masses me mq The indicated Cs clock performance was achieved using a microwavelocal oscillator based on a cryogenic oscillator On EGE the latter is to be replaced by an opticallocal oscillator-based device (MOLO) with similar short-term performance but without the needto operate at cryogenic temperature The clock instability at 10000 s is typically similar to theindicated inaccuracy(p) projected (d) demonstrated (1) quadropole (2) octupole

Exp Astron

way to achieve this is by having at least two clocks on board The on-boardclock-clock comparison provides performance evaluation independent of theground-satellite comparison unit Once the satellite clock performances areestablished the link performance can be established as well

The clocks are similar in basic structure but differ in the detailed implemen-tation They consist of

a source of atomsa preparation subunit which cools the atoms to near-standstill (millikelvintemperature or below)a trapping subunit that confines the atom(s) to a tiny volumean interrogation subunita clock control electronics package

The clocks share a common central unit the optical-microwave local oscil-lator which serves as a flywheel oscillator with outstanding stability on shorttimescales (01ndash10 s) for both clocks It consists of a laser stabilized to an ultra-low loss optical resonator made of ultra-low-expansion glass A microwavesignal is derived from the laser by phase-locking a frequency comb to the laserwave and using a multiple of the repetition rate

While the subunits of any optical clock have a similar technological basis(optics electrooptics acoustooptics thermal control vacuum systems) theydiffer in their specifications Depending on the choice of the atom species de-livery confinement and interrogation methods clocks of different perfor-mance result A number of laboratory optical clock demonstrators have beendeveloped worldwide This ensures availability of the know-how for the indus-trial implementation of the flight models

The first clock is a microwave clock based on the interrogation of slowlymoving ensembles of cold Cs atoms It is essentially based on the PHARAOclock [5 6] developed to engineering model level by CNES in the frame-work of the ESA ISS mission ACES In EGE this clock is upgraded usingthe microwave-optical local oscillator This improves significantly the clockperformance The specification is a relative instability (Allan deviation) of3 times 10minus14 (τ s)minus12 ie at the integration time τ = 1000 s it drops to 1 times 10minus15and to 3 times 10minus16 at 10000 s While this specification is at least a factor 3 lessstringent than for the optical clocks the significantly lower cost of obtaining aflight model and the sensitivity of its frequency to the particle masses makes ita suitable choice

The optical clock baseline instrument is a single-ion optical clock Amongvarious suitable ion species (Hg+ Al+ Sr+ Yb+) [44ndash47] the Yb+ ion hasbeen selected Its breadboard performance is established at the instability levelof 3 times 10minus16 over 1000 s it exhibits a large sensitivity coefficient Aα hassmall sensitivity to magnetic fields and exhibits significant potential for furtherimprovements (use of the octupole transition with sim6 times larger Aα andlower overall instability [48] see below) A trapped Yb+ ion is also a veryrobust system it has been demonstrated that a single Yb+ ion can remaintrapped uninterruptedly for many months Finally Yb+ is the only ion with

Exp Astron

which optical clocks demonstrators have been operated in not a single but twomajor metrology laboratories

Another attractive optical clock is based on ensembles of cold neutral Sror Yb atoms trapped in an optical lattice produced by laser beams This clocktype has a lower instability than a single-ion clock thanks to the large numberof atoms (sim105) used Its development is in progress in several laboratoriesin the world (in part supported by space agencies) and has to date reached6 times 10minus17 instability at 5000 s and 1 times 10minus16 uncertainty [49 50] with potentialfor even higher performance However it is more complex and costly thanthe Yb+ ion clock since the cold atom preparation and trapping subunitsinclude more components require higher laser powers and the respectiveoperating procedures are more complex It is therefore not considered for thepresent mission designed to be developed in the very near future with limitedresources but is an interesting candidate for missions without such constraintsIts estimated physical parameters are indicated in Table 2 for comparison

42 Main instruments

421 Microwave clock

The PHARAO cold atom clock is one of the two atomic clocks of the spacemission ACES managed by ESA It is proposed here to include a secondversion of this instrument as its costperformance ratio is very attractive forthe EGE scientific objectives Its concept is very similar to ground basedatomic fountains but with the major difference of zero-g operation Atomsslowly launched in free flight cross two microwave fields tuned to the transitionbetween the two hyperfine levels of the cesium ground state The interrogationmethod based on two separate oscillating fields (Ramsey scheme) allowsthe detection of an atomic line whose width is inversely proportional to thetransit time between the two microwave cavities The resonant microwavefield at 9192631770 GHz (SI definition of the second) is synthesized startingfrom an ultrastable quartz oscillator (USO) in the FCDP phaselocked to themicrowave output of the microwave-optical local oscillator and stabilized tothe clock line using the error signal generated by the cesium resonator In thisway the intrinsic qualities of the cesium hyperfine transition both in terms ofaccuracy and frequency stability are transferred to the USO In a microgravityenvironment the velocity of atoms along the ballistic trajectories is constantand can be changed continuously over almost two orders of magnitude (5ndash500 cms) Differently from atomic fountain clocks presently operated onground very long interaction times (up to few seconds) will be possible whilekeeping reasonable the size of the instrument

The instrument consists of 4 main elements a laser source which providesthe light required to cool and detect the cesium atoms (Fig 5 left) a frequencychain which generates the 92 GHz microwave signal a cesium vacuum tubewhere the interaction between the microwave field and the atoms occurs

Exp Astron

Table 2 Resources and technology readiness level (TRL) of the EGE spacecraft

Unit Volume Mass Power TRL Comment[l] [kg] [W]

Ion clock 100 70 60Atom source 4Atom preparation 6Atom trap 4Atom interrogation 6Clock control 6

Microwave clock 200 91 110 8 TRL by end 2008Local oscillator

Stabilized laser 18 15 23 6Frequency comb 20 16 38 6

Microwave link 18 18 55 8FCDP 7 8 12 8GPS receiver 2 5 7 9CCR 2 2 0 9Payload sum 367 225 305Thermal control 14 17 30 8OBDH 20 24 38 9Star tracker 1 1 8 9Radiation shield 10 40 0 9Support structure 170 204 0 9Power incl solar arr 50 58 10 9Propulsion 21 25 20 9AOCS 36 43 82 9TMTC 8 10 15 9Satellite bus sum 330 422 203 Flexbus conceptPayload plus bus incl 20 margin 8364 7764 6096Propellant 40Launch adapter 100SC wet masspower incl Margins 9164Lattice clock 135 105 152 Alternative clock

Atom source 4Atom preparation 6Atom trap 4Atom interrogation 6Clock control 6

Also included are the resources for an optical lattice clock based on neutral Strontium atoms

(Fig 5 right) and a computer The engineering model of PHARAO is fullyassembled at CNES Toulouse and undergoes functional and performancetests until the end of 2008 The operating temperature range is 10ndash33 andnon-operating temperatures are minus45 +60 The cesium tube consists in aUHV chamber pumped by getters and a 3 ls ion pump a cesium reservoira microwave cavity coils to provide a uniform magnetic field and three layersof magnetic shields Vacuum windows with fiber-optics collimators enable thetransmission of the cooling and detection light onto the cesium atoms

In autonomous mode PHARAO uses the USO as interrogation oscillatorand will provide a clock signal with fractional frequency stability below1 times 10minus13 (τs) minus12 and inaccuracy near 1 times 10minus16 When using the microwave-

Exp Astron

Fig 5 Left The PHARAO laser source engineering model with cover removed Dimensionsare 530 times 350 times 150 mm3 and the mass is 200 kg Ten polarization maintaining optical fibers(yellow) guide the laser beams to the cesium tube All diode lasers are mounted on Peltier coolersfor temperature regulation within 2 mK (courtesy EADS SODERN) Right The PHARAOengineering model cesium tube without the two external magnetic shields The volume is 990 times336 times 444 mm3 and the total mass is 44 kg

optical local oscillator PHARAO will provide a reduced frequency instabilityof 3 times 10minus14 (τs) minus12 This corresponds to 1 times 10minus15 at 1000 s and 3 times 10minus16

at 104 s

422 Ion optical clock

The 171Yb+ ion clock shown in Figs 6 and 7 provides an atomic frequencyreference based on a single 171Yb+ ion confined within an RF electric end-captrap (similar to a Paul trap) within an ultra high vacuum (UHV) chamber andlaser cooled to sim1 mK close to the Doppler cooling limit As a result the ionrsquosfirst order Doppler motion is completely removed and the ion experienceslittle collisional perturbation from its environment Under these conditionsthe narrow linewidth 2S12 (F = 0 mF = 0) minus 2 D32 (F = 2 mF = 0) quadrupole

Fig 6 Left Simplified level scheme of 171Yb+ showing relevant cooling (blue) auxiliary (red)quadrupole clock (green) and octupole clock (dark blue) transitions δνnat is the natural linewidthof the clock transition Middle close-up of Paul-type RF trap (PTB Braunschweig) the ion istrapped in the hole (diameter 14 mm) Right Complete ion trap with vacuum flange height 10 cm(NPL Teddington)

Exp Astron

935 nm D32-state repumper laser

638 nm F72-state repumper laser

739 nm extended cavity diode laser

AOM

PMT

Ion trap UHV chamber

vacuum pumps

Fibre-free space interface

Mu-metal shielding

EC

T

B

Magnetic field

Temperature

Ion fluorescence Processor

Clock laser frequency step and control

4355 nm

Frequengy doubling II

3695 nm cooling wave

Low-drift cavity

Low-drift cavity 171Yb+ 435 nm ion clock unit

876 nm clock laser wave from MOLO

Frequengy doubling

Fig 7 Simplified schematic of the single-ion optical clock unit AOM acousto-optic frequencyshifter PMT photomultiplier tube EC endcap T temperature stabilization system B magneticfield

clock transition at 4355 nm can be interrogated by spectrally narrow and stableclock laser light A diagram of the simplified energy level scheme for 171Yb+showing the relevant cooling auxiliary and clock transitions is given in Fig 6All these wavelengths are provided by diode lasers operated in fundamentalor frequency doubled mode The clock laser light is obtained from the 871 nmwave provided by the MOLO unit (see Section 423) after frequency-doublingin single pass by a KNbO3 nonlinear-optical crystal to 4355 nm The clocktransition spectral profile is observed by stepping the clock laser frequency bymeans of an acousto-optic frequency shifter (AOM in Fig 7) and recordingthe statistics of the quantum jumps in cooling laser induced fluorescence asa function of this frequency Under clock operation the AOM repeatedlysteps back and forward between the half-intensity points on the transitionprofile monitoring the signal imbalance between these points Any detectedimbalance is servo-corrected to zero by feedback to the AOM

In detail the optical clock architecture will comprise (Fig 7)

bull an RF end-cap trap for ionising and confining a single 171Yb+ ion withinan ultra-high vacuum chamber pumped by a small ion pump and non-evaporable getter pump The trap is driven by means of an ac voltage ofa few hundred volts at a drive frequency of sim10 MHz A small oven withseveral mg of 171Yb isotope is heated to provide a low flux of Yb atomsfrom which single ions can be ionised by electron bombardment from ahot wire filament within the trap potential well Low voltage dc is applied

Exp Astron

to additional electrodes to position the ion precisely at trap potentialcentre in order to minimise the ionrsquos micromotion at the drive frequencyThis needs automatic periodic monitoring and correction especially in theperiod after re-loading an ion into the trap External magnetic field coilsin three orthogonal axes allow the nulling of external fields and settingof a fixed field of sim1 μT The trap and coils are surrounded by mu-metalshielding to minimise external field changes during the orbit A level oftemperature control of 1 K of the region surrounding the trap is requiredfor maintaining the blackbody-induced frequency shift uncertainty withtemperature at the 10minus17 level

bull a laser platform to provide frequency doubling of an extended cavity739 nm diode laser for Doppler cooling of the ion on the 2S12(F = 1) minus2 P12 (F = 0) dipole transition at 3695 nm Light from an amplified diodelaser device frequency-doubled in a periodically poled LiTaO3 crystal insingle-pass will be used Typical laser powers used for driving the coolingtransition below saturation are about 2 μW for a beam waist in the trapof about 50 μm By modulation of the injection current of the 739 nmdiode laser at 147 GHz a sideband is generated that excites the 2S12 (F =0) rarr2 P12 (F = 1) ldquorepumperrdquo transition in order to avoid optical pump-ing between the ground hyperfine states

bull The platform also houses the auxiliary lasers at 935 and 638 nm The935 nm diode provides repumping of the ion from the 2 D32(F = 1)metastable level after occasional branching decays to this level during thecooling sequence The 638 nm diode allows fast recovery of the ion fromthe very-long-lived 2F72 metastable state after very occasional collisionaldecay to that state Currently extended cavity lasers are forseen butdistributed-feedback lasers may be available in the near future

bull a high NA lens imaging system and photomultiplier detection system torecord the statistics of 369 nm fluorescence quantum jumps as a functionof clock laser frequency step

bull a fibre system to deliver the various cooling auxiliary and clock light fromsource to trap making use of achromatic doublets where necessary at thefibre-free space interface for launching into the trap

bull a monitoring and control processor which provides the primary coolingand clock laser pulse magnetic field and detection sequencing to observeand lock to the ion clock transition frequency The processor also monitorsfrequency and amplitude data necessary to determine normal laser andion operational conditions and initiate resetting and recovery algorithmswhere necessary and laser unit failure The actual magnetic field presenton the atomic volume is determined and actively stabilized so that aninaccuracy at the level of 1 times 10minus17 can be achieved

bull a redundancy level of at least two units for cooling clock and repumperlasers plus similar for frequency doubling crystals All redundancy unitsfor each wavelength will be fibre multiplexed as standard allowing redun-dant unit activation on determination of prior unit failure

Exp Astron

For an optimized quantum-projection-noise-limited performance of thesingle-ion clock Ramsey interrogation with a cycle time equal to the lifetimeof the metastable D32 level of 50 ms will be performed leading to an Allandeviation of about 27 times 10minus15 (τ s)minus12 This is sim45 times 10minus17 at 1 h satisfyingrequired stabilities for the science objectives Current laboratory results are astability of 1 times 10minus15 at 30 s averaging down to 3 times 10minus16 at 1000 s (PTBGermany) Also comparisons between two independent traps showed agree-ment at the level of several parts in 10minus16 over a series of eight measurementruns

The 171Yb+ ion clock offers a second option namely the 2S12ndash2 F72 octu-pole transition at 476 nm (see Table 1 and Fig 6) [48] It has an extremelylong upper state lifetime low electric quadrupole and second order Zeemansystematic shifts and therefore is a candidate for an even higher performanceclock if a clock laser system with lower instability than the above is providedApart from the clock laser it shares the laser system of the quadrupoleclock Thus with moderate additional resources the Yb+ ion apparatus couldprovide two optical clocks in a single package and specifically perform directfrequency comparisons between the quadrupole and the octupole transitionwith a large sensitivity Aα

The ion clock will operate under microprocessor control providing thecontrol algorithms for the integrated pulse sequencing for cooling repumpingclock interrogation magnetic field switching cooling fluorescence monitoringThe controller will also provide error flags for critical processes and systemreset and correction where necessary This will include ion re-loading andmicromotion reduction algorithms Automatic monitoring of laser power andspectral quality with signal re-optimisation or laser failure determination isrequired Redundant laser units will be pre-aligned and multiplexed into thefibre delivery to the trap so that redundant units can be readily activated in thecase of diode laser failure

423 Microwave-optical local oscillator (MOLO)

This oscillator provides two ultrastable frequency outputs one in the opticalregion the other in the microwave region the two frequencies being coher-ently related The MOLO is composed of three subunits the clock laser thereference cavity and the frequency comb see Fig 8

The clock laser subunit delivers the light for the excitation of the opticalclock transition It is generated by an extended-cavity diode laser at a wave-length of 871 nm Most of the output wave of this laser is sent to the ionclock The laser subunit containing two lasers for redundancy will require3 kg 3 l 15 W The frequency instability of the clock laser must be superiorto the minimum clock instability 27 times 10minus15 (τ s)minus12 for τ up to the timeconstant (sim10 s) of the servo system that locks the laser to the ionrsquos resonancesignal In order to achieve this the laser frequency is stabilized to the reference

Exp Astron

871 nm clock laser

EO

Detector and servo

AOM Femtosecond

frequency comb

To optical clock 10 GHz To FCDP

Microwave-optical local oscillator

ULE cavity

Fig 8 Microwave-optical local oscillator (MOLO) consisting of clock laser reference ULE cavityand femtosecond frequency comb (OFC) Radiofrequency inputs and outputs are not shown EOelectro-optic modulator ULE ultra-low expansion AOM frequency shifter

cavity where the PoundndashDreverndashHall method is used About 10 μW of thelaser output are used for this

The system will exhibit sim04 Hz (sim1 times 10minus15 relative) linewidth and afrequency drift lt01 Hzs The cavity consists of two high-reflectivity mirrorsoptically contacted to a cylinder of diameter 70 mm length 150 mm made ofULE (a glass exhibiting ultra low thermal expansion coefficient) with a cavityfinesse of about 300000 Residual vibrations of the satellite (arms lt 1 times 10minus6 g)should not be a limiting factor to the desired linewidth for safety marginan optimized cavity support can reduce the sensitivity to accelerations Foracoustic and thermal isolation the cavity resides inside an aluminium vacuumchamber equipped with a small (3 ls) ion-getter pump Two stages of polishedaluminium shields are implemented around the cavity which are activelytemperature stabilized so that the cavity is precisely kept at the zero crossingtemperature of the ULE thermal expansion coefficient Heat application andremoval is by thermoelectric elements between the shield and the vacuumchamber Additionally the entire vacuum chamber is temperature-stabilized

During transportation and launch the cavity will be rigidly fixed insidethe vacuum chamber by pressing on it from the sides by piezo-mechanicalactuators allowing accelerations of several g After bringing the satellite onthe orbit the actuators will be released

The optical setup for frequency stabilization will be a classic Pound-Drever-Hall scheme in reflection using an electro-optic modulator (EO) Couplingof the clock laser radiation to the cavity is via single-mode polarization main-taining optical fiber The fiber in- and out-couplers are mounted on miniaturepiezo-motor driven multiaxis translation stages Microprocessor control ofthe translation stages performs a laser-to-fiber incoupling stabilization andlaser-to-cavity mode-matching optimization (positioning inaccuracy lt10 μmalignment inaccuracy lt10 μrad)

The frequency stability of the clock laser will be limited by thermal 1f noisein the mirrors with an fractional Allan deviation slightly below 1 times 10minus15 from1 s to a few 10 s Resources including control electronics are 15 l 12 kg 8 W

Exp Astron

A femtosecond fiber laser based optical frequency comb generator (OFCFig 8) will be used for the clock frequency comparisons and to obtain aspectrally ultrapure microwave local oscillator output for the operation ofPHARAO for the microwave link and for the GNSS receiver The fem-tosecond laser is based on erbium-doped fiber pumped by fiber-coupled laserdiodes It emits 100 fs pulses at a repetition rate of 250 MHz with an averagepower of 100 mW The emission spectrum is centered around 1560 nm Spec-tral broadening to more than one octave is obtained via nonlinear interactionin a microstructure fibre so that the range from about 1050 to 2100 nm iscovered

Second harmonic generation of a portion of the broadened spectrum allowsobtaining the required light for the measurement of the clock laser frequencyat 871 nm The OFC will be referenced to the clock laser frequency (afterintroduction a controllable shift via the frequency shifter AOM) by makingthe carrier envelope offset frequency vanish so that the repetition rate fr isa high-order subharmonic of the optical clock reference frequency f1 = m1 frThe link between the optical and the microwave domain is then simply thedistribution of the repetition rate frequency or more conveniently a high(order 40) harmonic of the comb repetition rate 10 GHz

It has been demonstrated by several groups that frequency comparison andfrequency division with an OFC can be performed to very high accuracy andstability beyond the limitations imposed by the optical clocks themselves Forexample a fiber laser OFC with an instability (Allan deviation) of 6 times 10minus17

in 1 s has been demonstrated [51] OFCs based on fiber technology possessgood long-term reliability and high power efficiency The projection for a spacequalified OFC based on current industrial terrestrial technology is 16 kg mass20 l volume and 38 W power requirement

424 Frequency comparison and distribution package (FCDP)

This package distributes the ultra-stable 10 GHz reference frequency outputof the MOLO to the MWL performs frequency comparisons between MOLOand PHARAO and generates a variety of different reference signals for theindividual instruments including the 250 MHz reference clock to the OFCAll signals are coherent to the MOLO In the FCDP a USO is locked to theMOLO signal by a digital PLL A divider chain generates timing pulses (1 pps1000 pps etc) and keeps a continuous local time scale From the digital phasecomparison data the phase noise density spectrum between the USO and theMOLO is available by telemetry (TM) to assess the overall frequency purityin-flight

As the FCDP distributes the full performance reference from the MOLOdirectly to the MWL there is minimum impact of the FCDP on overall clockcomparison performance The USO-phase lock loop uses a similar designas for ACES-FCDP The digital PLL with linear phase detector allows theUSO to be slightly offset from its reference which is selectable by groundcommand This results in an accurate on-board time scale despite any offset

Exp Astron

of the MOLO reference frequency which can be synchronised to UTC Thefrequency synthesis chain re-uses modules developed for ACES-FCDP andtheir frequency inaccuracy Expected relative uncertainty (Allan deviation)of the frequency comparison between two signals provided to the FCDP is47 times 10minus17 at 1000 s 95 times 10minus18 at 10000 s

During non-availability of the MOLO ie during system start-up andcontingency conditions the USO runs as flywheel (ie it is unlocked withinstability lt15 times 10minus13 from 1 to 100 s) It will still provide useable butless accurate references to the instruments including the reference to OFCCommon view ground-to-ground time- and frequency comparison and supplyof the remaining instruments remain possible at the USO provided accuracy

Mass volume and power consumption are 8 kg 7 l 12 W Controldata rateis 10 kBday 50 kBday (FCDP alone)

425 Microwave link (MWL)

The microwave link is a two-way triple-frequency microwave link (Fig 9) Itis designed to perform accurate frequency comparison as well as time-transfer

Ku-Band Up-linkPower Tx 5 WCarrier 13475 GHzPN-Code 100 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

S-Band Down-linkPower Tx 1 WCarrier 2248 MHzPN-Code 1 MChips1pps 1 time marker sData 5 kBits

Ku-Band Down-linkPower Tx 1 WCarrier 1470333 GHzPN-Code 100 MChips1pps 1 time marker sData 5 kBits

EGE SC

S

Ku

KaKa-Band Up-linkPower Tx 5 WCarrier 345 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

Ka-Band Down-LinkPower Tx 1 WCarrier 32 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 5 kBits

GS Antenna diameterEGE terminal 1 mACES terminal 06 m

Fig 9 EGE microwave two-way link concept and characteristics See text for a description Thelink will be able to perform frequency comparisons with up to ten ground stations simultaneously(only one active comparison is shown) pps pulse per second Tx transmitter SC spacecraft

Exp Astron

between on-board and on-ground clocks Two-way operation allows for com-pensation of the effects due to spacecraft location and its motion relative toground Its two bi-directional links operate in Ka-band (32345 GHz) andKu-band (147135 GHz) The high carrier frequency of the up- and down-links allows for a noticeable reduction of the ionospheric delay A third down-link channel is in S-band (22 GHz) and is used to determine the ionospheretotal electron content MWL uses a combination of carrier phase and codephase data to measure with high precision the absolute ionospheric delayas well as ionospheric delay variations during a ground contact The useof three frequency bands allows to resolve third order ionospheric termsKey modulation parameters are Ka band Tone 250 MHz Ku band PN100 MChips S-band 1 MChips

The EGE-MWL concept is a considerably improved version of the ACES-MWL Microwave carrier generation is directly from the 10 GHz microwaveoutput available from the MOLO and having exceptionally low close-in phasenoise without intermediate frequency dividers or other complex synthesiserstages commonly used in RF transponders The high frequency carriers arekey to achieve the desired frequency comparison accuracy

The Ka-band carrier is generated by tripling the FCDP-generated signalThe Ku-band is subsequently generated by division by two using a verystable regenerative divider Finally the assigned radio frequency is obtainedby adding a low frequency signal using an image reject mixer A mixedanalog-digital receiver design will enhance its performance by use of a flexibledigital signal processing and digital control loops The number of simultaneousground contacts is increased from four (as in the ACES mission) to ten or moreusing the digital concept while the number of physical receiver channels perband is reduced from four to two one for operation to ground and one for in-flight calibration The concept allows adaptation of receiver loop parametersto optimise for the different portions of a highly elliptic orbit while reducingsystematic uncertainties due to high Doppler and Doppler rate Real-timedata exchange to ground terminals is by additional data modulation on all RFsignals The spacecraft carries two different small antennae for perigee and forapogee with gains of 8 dBi 20 dBi beamwidths 100 (full earth coverage) 14respectively power is 1 W for each

The MWL end-to-end performance between the attached clocks includingspace- and ground terminals is predicted to be a relative Allan deviation of1 times 10minus16 at 1000 s (perigee) and 1 times 10minus17 at 10000 s (apogee) Thesevalues are based on conservative implementation losses and other instrumentallosses assumed as 6 dB at apogee and 10 dB at perigee including additionallosses due to atmosphere An attitude knowledge of 1 is required

The MWL link will also support precise orbitographyndashcomplementary tothe GNSS receiver Generation of ranging data is simultaneous to two-waytime transfer and is available under all weather conditions Orbit determina-tion accuracy is expected to be better than 1 cm well matching EGE clockrequirements The value is expected based on demonstrated results fromother missions (METEOR ERS-2 TOPEXPOSEIDON) Laser ranging to

Exp Astron

the CCR array is used a few times during the mission to calibrate the MWLranging

In case of non-availability of the ultrapure signal from MOLO an additional10 GHz output is available from the FCDP to support ground-to-ground clockcomparison in common-view mode even after the nominal mission periodwhen the neither the clocks nor MOLO may be available any longer Basedon the PRARE experimentrsquos 12 years of operation this can be assumed forthe FCDPMWL combination as well The potential extended operational pe-riod makes the proposed configuration particularly attractive to metrologicallaboratories that rely on availability of long-lasting data in support of thegeneration of UTC

The resources of MWL are as follows mass 18 kg volume 18 l power55 W TC 2 kBday TM 120 MBday

426 Auxiliary units

The star camera assembly on the EGE spacecraft will provide real-timeattitude data primarily for the orientation of the spacecraft (with accuracy of1 in real-time) and the proper reduction of the GNSSSLR data

A GNSS receiver is used for the precise orbit determination time andfrequency transfer coarse positioning (lt1 m) for real time use by the Attitudeand Orbit Control as well as the back-up for time tagging of all payloaddata For the time and frequency transfer the receiver is nominally using theonboard frequency reference The receiver assembly includes three specialantennas By the year 2015 more than 100 GNSS satellites are expected tobe available for precise orbit determination permitting orbit inaccuracy of 1ndash2 cm RMS for altitudes up to approx 10000 km At apogee a combinationof GNSS signals and MWL ranging will be used with expected inaccuracyof 10 cm

5 Basic spacecraft specifications

The mission foresees a single spacecraft with a mission duration limited toapprox 3 years for cost reasons Attitude control along three axes withless than 1 error on a timescale of 1000 s is required The control can beperformed with cold gas thrusters magnetotorquers or reaction wheels Theselection will be done on the basis of the vibrations induced by these systems(desired level arms lt 10minus6 g)

Cold gas thrusters or monopropellant thrusters will be required to maintainthe orbit parameters during the mission by small orbit correction manoeuvresand will also guarantee a proper tuning of the orbit parameters after the releasefrom the launcher The ground segment (telemetry control and data transfer)could be via non-ESA stations user-operated to save costs

Further resources are indicated in Table 2

Exp Astron

The satellite bus design drivers are nadir-pointing microwave antennasthermal stability for the payload instruments no strong magnetic sources lowmicrovibrations harsh radiation environment due to crossing of the radiationbelts once per orbit power demand by science payload (with the possibility toreduce power demand by stand-by of some components during specific missionphases eg eclipse) mass limit due to required high altitude with low costlaunch

The satellite bus design can be based on established platforms such asFLEXBUS or PRIMA used on several past missions Their design philosophydiffers one having body-mounted solar arrays the other wings The latter canprovide larger amounts of power The vibration level induced by the panelswhen they are fixed and the duration of solar array drive periods per orbitwill have to be considered in the selection The FLEXBUS concept was usedfor a mission already requiring a low level of residual accelerations and theplatform is optimised for this parameter

For most of the satellite components one can consider using compo-nents and electronic units from navigation or communication satellites whichtypically cruise at altitudes between 15000 and 25000 km altitude LISAPathfinder which will cross the radiation belts during spiralling to theLagrange point will face similar problems concerning radiation as EGE as wellas XMM which is in a Molniya-type orbit with similar radiation characteristics

The instruments will therefore be located in the satellite structure such asto realise ionizing radiation shielding by the spacecraft structure and non-sensitive equipment to the maximum extent possible the need for additionalradiation shielding (including trapped proton radiation) for especially sensitiveparts of the instruments must be determined by ground tests It should bementioned that an orbit with higher perigee and less radiation dose is also anoption as it leads to a tolerable reduction of the science goals

The required plusmn3C temperature stability at the interface to the payloadenclosure is feasible The thermal control will employ the usual means (anenclosure with high thermal inertia heaters and a suitable radiator) Thelocation of the radiator will need a special study given the non-standard orbit

Magnetic field present during the mission can affect the properties of theonboard clock Using SPENVIS the magnetic field has been calculated alongthe spacecraft orbit the field intensity presents maxima around 02 G anda minimum value of the order of 10minus3 G As discussed above this is not alimitation to the instrument performance

6 Discussion and conclusion

The accuracy of the measurements depends crucially on the stability andaccuracy of the clocks While the performance of ground clocks (even thoseeventually deployed in new ground stations dedicated to EGE) will likelyimprove significantly from the present level the space clock design will beldquofrozenrdquo approx 10 years before data acquisition Thus their performance

Exp Astron

level will be significantly below the future ground station clock performanceAs specification we have chosen a performance level that has been demon-strated with laboratory breadboards recently We see no fundamental reasonthat this laboratory performance cannot also be achieved in a space ion opticalclock The experience with the PHARAO cold atom clock has shown thatan engineering model of a laser-based cold atom clock has a performancesimilar to that of a laboratory prototype The perturbation of the ion clock fre-quency due to magnetic fields along the orbit can be modelled and suppressedwith shielding and the residual effects measured and compensated for Themoderate temperature stability (plusmn 3C) to be provided in the experimentalbay can be achieved with thermal shields Accurate temperature controlof particular units is done by active stabilization of the units themselves

One important aspect is radiation which may affect all optical and trapcomponents as well as the lasers and reference cavity The radiation envi-ronment on the EGE orbit is significantly harsher than the near-earth ISSorbit on which ACES will operate The radiation effects must therefore becharacterized comprehensively with appropriate tests and then the radiationshield for the EGE instruments designed

In summary the Einstein Gravity Explorer mission (EGE) is a satellitemission aimed at a precise measurement of the properties of space-time usingatomic clocks By testing one of the most fundamental predictions of EinsteinrsquosTheory of General Relativity the gravitational redshift and thereby searchingfor quantum effects in gravity it explores one of the most important frontiers infundamental physics The EGE mission reaches beyond fundamental physicsproviding an important spin-off in geophysics Furthermore the technologicalaspects of the mission are ground-breaking for future time-keeping and ref-erence frame systems in space and thus are also expected to be of wide useto science and technology in general The payload of the mission has beenconceived as a compromise between cost and performance needing to fit thecost envelope of an ESA medium-class mission1

Acknowledgement We thank S Cesare for helpful information

References

1 Will CM The confrontation between general relativity and experiment Living Rev Relativ9 3 (2006)

2 Ni W-T Empirical foundations of the relativistic gravity Int J Mod Phys D 14 901 (2005)doi101142S0218271805007139

1Finally we point out that a spacecraft with a significantly reduced and thus lower-cost payload (noclocks only MOLO FCDP MWL) on a circular orbit of appropriate height could be used for along-duration earth science mission dedicated specifically to the measurement of the geopotential

Exp Astron

25 Theoretical analysis of frequency comparison signals

The on-board clock-clock comparison signals can be analyzed straightfor-wardly in terms of a violation of LPI since signal propagation effects areabsent

The determination of the gravitational frequency dilation via ground-satellite clock comparisons is more complex (Eq 1 is strongly simplified) andrequires an accurate treatment of all relevant relativistic effects Gravitationin the solar system is described in the barycentric (BCRS) and geocentric(GCRS) non-rotating celestial reference systems by means of post-Newtoniansolutions of Einsteinrsquos equations for the metric tensor in harmonic gaugescodified in the conventions IAU2000 [22] In these conventions the relativisticstructure of Newton potential (order 1c2) and gravitomagnetic (order 1c3)

potentials are given For the Newton potential the multipolar expansion ofthe geopotential determined by gravity mapping is used The post-Newtonianeffects of the metric (described in terms of dimensionless parameters β and γ )

can be included Orbits of satellites are evaluated in the GCRS at the levelof a few cm by means of the Einstein-Infeld-Hoffmann equations at the order1c2 [23] using the relativistic Newton potential Instead the trajectories of theground clocks in the GCRS are evaluated from their positions fixed onthe Earth crust (ITRS International Terrestrial Reference System) by usingthe non-relativistic IERS2003 conventions [24 25]

For the propagation of light rays (here radio signals) between the satelliteand the ground stations one uses the null geodesics of the post-Newtoniansolution in the GCRS The timefrequency transfer properties have beentheoretically evaluated for an axisymmetric rotating body [26ndash29] at the order1c4 (ie with Newton and gravitomagnetic potentials developed in multipolarexpansions) These results have been developed for use in the ACES missionThe relative frequency dilation can be expressed as

ν

ν= 1 + 1

cL1 +

4sum

n = 2

1

cn(Ln + Gn)

where Ln describes special-relativistic Doppler effects and Gn the general-relativistic effects These quantities are time-dependent due to the orbitalmotion For concreteness we report the orders of the terms The kinematicaleffects are |L1c| lt 1 times 10minus5 |L2c2| lt 2 times 10minus10 |L3c3| lt 3 times 10minus15 |L4c4| lt

7 times 10minus20 The Newtonian contributions are G2 = G(M)2 + G(J2)

2 + G(J4)2 + G(J6)

2(M is the mass monopole Jn are the lowest multipoles) with estimates|G(M)

2 c2| lt 3 times 10minus11 |G(J2)2 c2| lt 2 times 10minus13 |G(J4)

2 c2| lt 3 times 10minus16 |G(J6)2 c2| lt

1 times 10minus16 For G3 = G(M)3 + G(J2)

3 the estimates are |G(M)3 c3| lt 2 times 10minus14

|G(J2)3 c3| lt 1 times 10minus18 Finally the G4 = G(M)

4 + G(S)4 contributions (G(S)

4 is thelowest gravitomagnetic effect) are of order |G(M)

4 c4| lt 1 times 10minus19 |G(S)4 c4| lt

1 times 10minus22 and thus not relevant for EGE

Exp Astron

The available theory is fairly complete for the data analysis in mode 23 (1)described below For modes 23(2 3) the theory will need to be generalizedto include the effect of the gravitational potential at the location of the groundclock(s) (including time-varying effects such as the tides) and of the solar andlunar potentials

The uncertainties of the various contributions above are minimized by ausing precise EGE orbit data obtained by satellite tracking and by usingcurrent and future accurate earth gravity information Required orbit positionknowledge is at 1 cm level near perigee and sim50 cm near apogee reachablealready today with the proposed orbitography approach described below Thevalidity of the special-relativistic Doppler shift has been and will continue to beindependently verified with increased accuracy by spectroscopy of relativisticatomic ion beams [30] Potential violations of this aspect of Lorentz Invari-ance are expected to be sufficiently bounded so as not to affect the signalinterpretation

26 Local Lorentz invariance tests

There has been an explosion of interest in Local Lorentz invariance in re-cent years [31 32] with numerous astronomical studies and high-precisionexperiments applied to search for violations of Lorentz invariance Signifi-cant developments have also occurred on the theoretical side and a theory(standard model extension SME [33]) has been worked out that provides aunified framework for describing and analyzing Lorentz Invariance violationsin a variety of systems Two aspects of Lorentz Invariance can be tested withEGE

261 Independence of the speed of light from the laboratory velocity

A possible dependence of the speed of light on the speed v of the laboratorycan be modeled according to the MansourindashSexl test theory [34] by

c (v) = c0(1 + Bv2

c2

)

Here v 377 kms is the velocity with respect to the cosmic microwavebackground the cosmologically preferred frame and B is a combinationof parameters describing deviations from the usual Lorentz transformationformulas B = 0 if Lorentz Invariance holds For a terrestrial experiment therotation of the Earth around its axis modulates v with a 300 ms amplitude Adependence c(v) can be searched for by measuring the frequency differencebetween a resonance frequency of a highly stable optical cavity (cavity fre-quencies are proportional to c) and an optical clock (Kennedy-Thorndike-typeexperiment [35]) and determining the modulation of the frequency differencecorrelated with the modulation of v The advantages of a space experimentare the high orbital velocity and strongly reduced cavity deformation thanksto microgravity [36 37] On the proposed elliptic orbit v varies between+4 and minus4 kms over approximately 1 h This variation is 13 times larger

Exp Astron

than for the case of a terrestrial experiment The shorter time scale is alsoadvantageous since the drift of the cavity is more predictable In the EGEmission the reference cavity of the clock laser is used The large numberof orbits improves the statistics An improvement of the accuracy of B by afactor 20 compared to the best current terrestrial results is realistic The dataanalysis can also be performed in the framework of the SME theory takinginto account complementary bounds obtained from terrestrial experimentsand astrophysical observations

262 Independence of Zeeman splitting frequency on the directionof the magnetic field

One class of tests of Lorentz Invariance consists in measuring the frequencysplitting between two levels of a quantum system induced by a static magneticfield as a function of orientation of the field direction with respect to thestars (HughesndashDrever-type experiments [31]) This class addresses the so-called matter sector and is complementary to combined-photonmatter-sectortests such as the above The most precise experiments are performed withatomic clocks eg masers or cold atom clocks [38] The EGE instruments allowsuch experiments since magnetic fields are applied continuously in the clockor repeatedly for calibration Compared to terrestrial experiments the highorbital velocity of EGE could lead to an improvement by a factor sim20

27 Application to geophysics

Using GPS coordinates of a point on the Earth can be obtained with aninaccuracy well below 1 cm in a well defined international terrestrial referencesystem The coordinates are purely geometrical and do not contain any gravityinformation A local gravitational potential UB is determined with respect toa reference gravitational potential U A by ldquolevellingrdquo the surface of the Earthie sequentially measuring gravity g and height differences every ca 50 m

UB minus U A = minusBint

A

g (r) middot d n

where n is the difference vector between subsequent locations The disadvan-tage of measuring height differences by the levelling method is a random walkeffect of accumulated errors

In the classical definition a geoid is defined as the particular equipotentialsurface nearest to mean sea level In these terms the geoid serves as areference surface for measuring the height and also to define a referencefor the gravitational potential Such a vertical reference frame historicallywas realized for a country or several countries by determining the mean ofsea level observations at tide gauge stations taken over long period of timeHowever modern satellite altimetry missions such as TOPEXPOSEIDON orENVISAT show that departure of the mean sea level from an equipotential

Exp Astron

surface may reach up to several meters on the global scale see eg [39]Therefore height systems based on different tide gauge stations may differin the realisation of the geoid and may differ with respect to the reference byseveral meters

The best gravity field models obtained from satellite data (eg missionGRACE) have reached a precision of about 1 cm over sim250 km half wave-length [40] Satellite data also reveals changes in time [41] However for typicalEarth topography with height variations of eg 1000 m over 30 km horizontaldistance one may expect variations in the geoid of about 80 cm Such ahigh-spatial-frequency signal in geoid variations cannot be detected by spacegravity missions and requires a combination of satellite and terrestrial gravitymeasurements like gravity anomalies deflections of vertical and GPSlevellingpoints The best combined global gravity field models are provided up to de-gree and order 360 in terms of spherical harmonic representation correspond-ing to a half-wavelength of about 55 km However in combining the satelliteand terrestrial gravity field measurements the problem of terrestrial data givenin different height systems between continents and different countries remainsand has to be tied to satellite measurements

The geodetic scientific community is currently establishing a Global Geo-detic Observing System (GGOS) [42] Its objectives are the early detectionof natural hazards the measurement of temporal changes of land ice andocean surfaces as well as the monitoring of mass transport processes in theEarth system Global change processes are small and therefore difficult toquantify Therefore the required precision relative to the Earthrsquos dimensionis 1 ppb GGOS will be established by the combination of geodetic spacetechniques (GPS Laser-Ranging very large baseline interferometry) andrealized by a very large number of terrestrial and space-borne observatoriesThe purely geometric terrestrial 3D coordinate system is in good shape andfully operational It has to be complemented however with a globally uniformheight system of similar precision The current precision level of regionalheight systems in terms of gravity potential differences is in the order of1 m2s2 (10 cm) with inconsistencies between these various systems up toseveral 10 m2s2 (several meters) The actual requirement in the context ofGGOS is 01 m2s2 (1 cm) with the need of a permanent ie dynamical controlThis requirement of high precision height control comes from the need tounderstand on a global scale processes such as sea level change global andcoastal dynamics of ocean circulation ice melting glacial isostatic adjustmentand land subsidence as well the interaction of these processes Only by meansof monitoring in terms of gravity potential changes at the above level ofprecision the change of ocean level can be understood as a global phenomenonand purely geometric height changes be complemented by information aboutthe associated density or mass changes

The geoid can also be defined in a relativistic way [43] as the surface whereaccurate clocks run with the same rate and where the surface is nearest tomean sea level The relation between the differences in the clock frequenciesand the gravitational potential is given in simplified form by Eq 1 At present

Exp Astron

there is no operational way to compare frequencies of the already availableoptical clocks on the global scale at the same level as their accuracy wouldallow With EGE and by then improved ground optical clocks it will becomepossible to obtain gravitational potential differences on a global scale by com-paring frequencies This would be a significant new dimension to gravity fielddetermination since such observables are given on the global scale and providein situ local gravity information at the same time EGE will allow clock-basedgravitational potential mapping with the same payload instruments used forthe fundamental physics experiments The on-board clocks are actually notrequired for this purpose MOLO MWL and FCDP suffice Thus EGE willallow establishing a global reference frame for the Earth gravitational fieldwith accuracy in the order of a few cm in terms of the geoid heights Thisassumes that by the time EGE is operational (mobile) ground clocks withfractional inaccuracy at the level of 10minus18 are available the measurementsare limited by MWL noise and thus require a long integration times to matchthe expected ground clock performance Such a reference frame will serve asreference for all gravity field modeling and clocks will be used to define theinternational atomic time scale and the Earth gravitational potential at thesame time

In summary with optical clocks geodesy will undergo a second revolutionafter geometry now being measured by clock-based GNSS systems also physi-cal heights and potential will be measured by (mobile) optical clocks using thegravitational redshift effect

28 Application to frequency standards and terrestrial fundamentalphysics studies

As has been described repeating ground clock comparisons and integratingover several months will provide global ground clock frequency comparisonsat the level of 1 times 10minus18 more than two orders of magnitude below the currentGPS and two-way satellite time and frequency transfer methods In the timeand frequency community the availability of such a global high performancemicrowave link will accelerate the process leading to a redefinition of the SIsecond based on clocks operating in the optical domain

The global character of the ground clock comparisons made possible byEGE can also contribute to the search for a time-variation of the fundamentalconstants

The comparisons do not require availability of the on-board clocks The riskof this type of measurements is therefore reduced Moreover the lifetime ofmany subunits required for this measurement (MWL and FCDP see below)can exceed 10 years (from previous experience with similar devices) andthus offers a motivation to increase the mission duration beyond the nominal3 years

Further applications to metrology are listed under the secondary goals inSection 21 above

Exp Astron

3 Mission profile

31 Orbit

The orbit must satisfy the following requirements

Primary 1 Large gravitational potential difference between apogee andperigee

2 sufficient contact time with at least two ground stations atperigee

Secondary Simultaneous visibility of satellite from distant (inter-continental) ground stations for several hours

To satisfy the primary objective the orbit must be highly elliptic The perigeeshould be low A very low perigee means a short contact time which decreasesthe accuracy of the frequency measurement A compromise is a perigeealtitude of sim2500 km For a sufficiently high apogee this yields a U of 60of the theoretical maximum between the earth surface and infinity

The baseline scenario is an orbit of approx 12 h period with the followingproperties

Apogee Altitude 37856 km (minus 10 deg latitude one over Atlantic andPacific resp)

Perigee Altitude 2500 km (+ 10 deg latitude one over Guatemala andMalaysia resp)

Inclination 634 argument of perigee 170RAAN 25 True Anomaly 220Apogee drift rate 70 kmday perigee drift rate sim 14 kmday

True anomaly and right ascension of the ascending node (RAAN) wereoptimized to minimize orbital precession Figure 3 shows the ground track andthe ground coverage at one of the perigees Common-view contact time forstations in Europe is about 6 h while between a station in Europe and in theUSA it is about 35 h Frequency comparison windows between a given groundstation and EGE repeat every 24 h

A final selection of the orbit requires an in-depth study of the costs of thepropellant and maneuvers as well as of the setting up of new ground stations

32 Ground segment

The mission requires two separate ground segments the science groundsegment and the standard satellite control segment which supports telemetryand telecommand (TMTC)

Exp Astron

Fig 3 Baseline EGE orbit One (out of two) apogee and one (out of two) perigee ground pointsare shown The ground coverage at one of the perigees is shown with circles at elevation anglesteps of 10 starting from 10

321 Science ground segment

As described above one part of the mission is to perform frequency com-parisons between ground-to-space and ground-to-ground (common-view andnon-common view) The use of a two-way microwave link (MWL) systemhas the advantage that a precise orbit or the precise relative velocity of thesatellite must not be known [4] The orbit determination requirements fora two-way system are significantly relaxed compared to the requirements tomodel the relativistic effects on the clocks The frequency comparison resultsare available both on ground and on the satellite

For ground-to-satellite clock comparisons five ground stations arefavourable two capable of viewing perigee three for viewing apogee Thisallows for the classic three-clock configuration where the performance of eachclock can be obtained individually The perigee stations will be observing thesatellite for sim25 min every perigee passage for the duration of the missionThe apogee stations will observe for sim3 h every apogee passage Of coursea larger number of stations as well as longer observation time are usefulfor robustness redundancy and orbit determination The stations could beinstalled on a temporary basis for the duration of the mission

The stations must have access to optical clocks of performance at least ashigh as todayrsquos state-of-the-art ground clocks but preferably higher Theywill include an optical frequency comb by means of which the optical clockfrequency is transformed to a radio frequency for comparison with the EGEclocks High-accuracy fiber-optic link technology already existing today allowsdistances of sim100 km between the ground station and the correspondingoptical clock allowing a certain freedom in the choice of the clock locationWith further increase of the fiber-optic link range currently being investigated

Exp Astron

the area over which the ground station can be positioned relative to the opticalclock will increase This option is of importance for the implementation of ageophysics measurement campaign for which a large number of clock locationsis of interest

The above ground stations will be complemented by some 15 additionalground terminals provided by the world-wide laboratories equipped with highperformance primary clocks at the time of the mission These will participate inorbit determination to 1 cm accuracy along the full orbit which can be achievedwith non-optical atomic clocks The EGE MWL is fully compatible to plannedACES ground terminals which might reduce the number of required stations

With a larger number of transportable optical clock ground stations de-ployed and moved across the continents a geophysics campaign could beimplemented to map the gravitational potential

Ground station equipment MWL uses dedicated small transportable termi-nals which can be easily installed at any location of interest Each terminalconsists of a steerable dish antenna of 1 m diameter with pointing accuracy of01 using a transmitted power of 5 W per band The high directivity antenna(beam aperture Ka-band 07 Ku-band 14 S-band 20) avoids multi-pathfrom nearby objects Antenna noise temperatures are Ka-band 450 K Ku-band 340 K Operation is at elevations above 10 and measurements areperformed above 20 For ranging purposes terminal location is determined byGPS surveying Each ground station computes its own ionospheric correctionsbased on a triple-band receiver The terminal has a built-in delay monitorwhich allows calibrated ranging during several months once its bias has beendetermined by laser ranging Its M and C interface is via local area networkto either a userrsquos computer or directly to the science network control centreby remote MampC The stationrsquos operation is fully automatic and unattendedSchedule and orbital data are received via LAN once per day Signal acquisi-tion is fully autonomous Data transmission is after each pass The terminal hasan uplink telecommand capability of 1 kBits and a downlink data capacity ofup to 10 kBitss For the local user there is a visual interface providing quick-look data from real-time comparison to the spacecraft clocks to verify theirgood health The final full performance product is calculated at the sciencedata centre

Local metrological data (temperature pressure humidity) are recorded tocorrect for tropospheric delay of the ranging data The main advantage ofMWL terminals compared to laser based systems is their all-weather capabil-ity although microwaves suffer higher propagation errors in the ionosphereand troposphere

The EGE ground terminal will profit from the ERS-2 mission PRARE(precise range and range rate equipment) [45] experiments whose terminalshad similar design and have been operated under extreme climatic conditionsincl Antarctica and high temperature regions from existing designs fromthe space segment (mission ACES) and no further special developments aredeemed necessary Cost per EGE master station is modest

Exp Astron

Satellite tracking via laser ranging Laser ranging is alternative to MWL rang-ing and also serves for absolute calibration of the latter For this purpose anon-board corner cube reflector (CCR) is provided Ranging will be done withthe International Laser Ranging Service Ideally the satellite will be rangedat every perigee passage and every apogee passage simultaneously with thefrequency comparison procedures Todayrsquos ranging precision is sufficient tosatisfy the science goals For apogee observations in the northern hemispherea large number of laser ranging stations is available The number of stationscapable of ranging at or near perigee in the southern hemisphere is six andthus is sufficient even considering unfavourable weather conditions at someof these The MWL ground stations are able to range even under thoseconditions

322 Control ground segment

Average data transfer rate is estimated at 300 MBday (approx 40 kBs onaverage) Data will be stored on board (2 GB capacity required) and will betransferred to a receiving station on the Earth once a day

XPLC

incl Data StorageIF to TMTC Link

HK amp TMTC

to ground

Opt Clock

AOM

MOLO

PHARAO

FCDP

10 GHz

100 MHz

250 MHz

GNSS Rx

MWLTampF Link

uplink Data

UTC(k)

TampF Space vs Ground

TM from GS

Tim

e S

cale

Orbit GNSS Time UTC

TampF Space vs Ground

MampC Time Scale

MampC

MampC

MampC

MampC

Mamp

C

PHARAO vs Opt Clock

PHARAO vs Opt Clock

10 GHz amp Epoch

10 MHz

amp Epoch

Epoch

10 MHz

amp Epoch

Opt Link

Fig 4 Schematic of the payload showing subunits and their signals The results of the frequencycomparison between the two on-board clocks (optical clock and microwave clock PHARAO) areobtained from a combination of the frequency values applied to the frequency shifters AOM (seeFigs 7 and 8) and those output from the frequency control and distribution package (FCDP)as indicated by the arrows ldquoPHARAO vs Opt Clockrdquo The satellite clock to ground clockcomparison results are output from the MWL XPLC is the on-board computer and is involvedin generation of signals for instrument control The optical link shown dashed is a possible add-onproviding an alternative to the MWL GS ground stations MampC monitoring and control HKhousekeeping TampF time and frequency

Exp Astron

4 Main payload instruments

41 Overview

The satellite payload consists of a pair of cold-atom clocks a microwave-optical local oscillator (MOLO) an on-board frequency distribution unit(FCDP) a satellite-to-ground clock comparison unit (MWL) and the onboardcomputercontrol system (XPLC) see Fig 4 Auxiliary units are the posi-tion determination GPSGalileo receiver (GNSS Rx) attitude monitor (startracker not shown) and the corner cube reflector array (CCR not shown)

The science goals require clocks of very high frequency stability on thetimescales of half the orbital period (6 h) as well as perigee passage (sim1000 s)see Fig 2 Only clocks based on cold atoms are suited for the science goalsTable 1 reports values for a subset of atomic clocks that have been investigatednot including clocks that require cryogenic operation or are too complex foruse in an medium-class mission The choice of a clock pair rather than a singleone fulfils several requirements it addresses more science goals improves thevalidity of some science results and it enables calibration accuracy evaluationand redundancy The validity of the science results is enhanced since theclocks differ not only in the employed atoms but also in terms of havingdifferent sensitivity to perturbations by electric and magnetic fields electronicserrors and misalignments A comparison of the results arising from each clockallows for an understanding of measurement errors Both the clocksrsquo and thefrequency comparison unitrsquos performances will be determined after launchand regularly verified during the course of the mission The only reliable

Table 1 Properties of different clock candidates for EGE

Clock Clock Fractional σ = Fractional Sensitivity Figure of Sensitivityatom frequency inaccuracy instability to α Aα merit |Aα| to meq

at 1000 s σ [1016] QCD

Sr+ 445 THz 3 times 10minus17 (p) 5 times 10minus17 (p) 042 08 (p) ndashYb+ (1) 688 THz 3 times 10minus16 (d) 3 times 10minus16 (d) 088 029 (d) ndash

3 times 10minus17 (p) 1 (p)Yb+ (2) 642 THz lt 1 times 10minus17 (p) 1 times 10minus17 (p) minus53 53 (p) ndashSr 435 THz 1 times 10minus16 (d) 1 times 10minus16 (d) 009 009 (d) ndash

lt 1 times 10minus17 (p) 09 (p)Yb 519 THz 7 times 10minus16 (d) 1 times 10minus16 (d) 031 03 (d) ndash

lt 1 times 10minus17 (p) 1 times 10minus17 (p) 31 (p)Cs 91 GHz 2 times 10minus16 (d) 1 times 10minus15 (d) 275 027 (d) 1Yb+ 126 GHz 1 times 10minus15 (p) 1 times 10minus15 (p) 34 034 (p) 1

A relevant figure of merit is the ratio of |AX | and fractional frequency instability σ on the perigeevisibility timescale The Cs clock is the only high performance clock with adequate sensitivity tothe particle masses me mq The indicated Cs clock performance was achieved using a microwavelocal oscillator based on a cryogenic oscillator On EGE the latter is to be replaced by an opticallocal oscillator-based device (MOLO) with similar short-term performance but without the needto operate at cryogenic temperature The clock instability at 10000 s is typically similar to theindicated inaccuracy(p) projected (d) demonstrated (1) quadropole (2) octupole

Exp Astron

way to achieve this is by having at least two clocks on board The on-boardclock-clock comparison provides performance evaluation independent of theground-satellite comparison unit Once the satellite clock performances areestablished the link performance can be established as well

The clocks are similar in basic structure but differ in the detailed implemen-tation They consist of

a source of atomsa preparation subunit which cools the atoms to near-standstill (millikelvintemperature or below)a trapping subunit that confines the atom(s) to a tiny volumean interrogation subunita clock control electronics package

The clocks share a common central unit the optical-microwave local oscil-lator which serves as a flywheel oscillator with outstanding stability on shorttimescales (01ndash10 s) for both clocks It consists of a laser stabilized to an ultra-low loss optical resonator made of ultra-low-expansion glass A microwavesignal is derived from the laser by phase-locking a frequency comb to the laserwave and using a multiple of the repetition rate

While the subunits of any optical clock have a similar technological basis(optics electrooptics acoustooptics thermal control vacuum systems) theydiffer in their specifications Depending on the choice of the atom species de-livery confinement and interrogation methods clocks of different perfor-mance result A number of laboratory optical clock demonstrators have beendeveloped worldwide This ensures availability of the know-how for the indus-trial implementation of the flight models

The first clock is a microwave clock based on the interrogation of slowlymoving ensembles of cold Cs atoms It is essentially based on the PHARAOclock [5 6] developed to engineering model level by CNES in the frame-work of the ESA ISS mission ACES In EGE this clock is upgraded usingthe microwave-optical local oscillator This improves significantly the clockperformance The specification is a relative instability (Allan deviation) of3 times 10minus14 (τ s)minus12 ie at the integration time τ = 1000 s it drops to 1 times 10minus15and to 3 times 10minus16 at 10000 s While this specification is at least a factor 3 lessstringent than for the optical clocks the significantly lower cost of obtaining aflight model and the sensitivity of its frequency to the particle masses makes ita suitable choice

The optical clock baseline instrument is a single-ion optical clock Amongvarious suitable ion species (Hg+ Al+ Sr+ Yb+) [44ndash47] the Yb+ ion hasbeen selected Its breadboard performance is established at the instability levelof 3 times 10minus16 over 1000 s it exhibits a large sensitivity coefficient Aα hassmall sensitivity to magnetic fields and exhibits significant potential for furtherimprovements (use of the octupole transition with sim6 times larger Aα andlower overall instability [48] see below) A trapped Yb+ ion is also a veryrobust system it has been demonstrated that a single Yb+ ion can remaintrapped uninterruptedly for many months Finally Yb+ is the only ion with

Exp Astron

which optical clocks demonstrators have been operated in not a single but twomajor metrology laboratories

Another attractive optical clock is based on ensembles of cold neutral Sror Yb atoms trapped in an optical lattice produced by laser beams This clocktype has a lower instability than a single-ion clock thanks to the large numberof atoms (sim105) used Its development is in progress in several laboratoriesin the world (in part supported by space agencies) and has to date reached6 times 10minus17 instability at 5000 s and 1 times 10minus16 uncertainty [49 50] with potentialfor even higher performance However it is more complex and costly thanthe Yb+ ion clock since the cold atom preparation and trapping subunitsinclude more components require higher laser powers and the respectiveoperating procedures are more complex It is therefore not considered for thepresent mission designed to be developed in the very near future with limitedresources but is an interesting candidate for missions without such constraintsIts estimated physical parameters are indicated in Table 2 for comparison

42 Main instruments

421 Microwave clock

The PHARAO cold atom clock is one of the two atomic clocks of the spacemission ACES managed by ESA It is proposed here to include a secondversion of this instrument as its costperformance ratio is very attractive forthe EGE scientific objectives Its concept is very similar to ground basedatomic fountains but with the major difference of zero-g operation Atomsslowly launched in free flight cross two microwave fields tuned to the transitionbetween the two hyperfine levels of the cesium ground state The interrogationmethod based on two separate oscillating fields (Ramsey scheme) allowsthe detection of an atomic line whose width is inversely proportional to thetransit time between the two microwave cavities The resonant microwavefield at 9192631770 GHz (SI definition of the second) is synthesized startingfrom an ultrastable quartz oscillator (USO) in the FCDP phaselocked to themicrowave output of the microwave-optical local oscillator and stabilized tothe clock line using the error signal generated by the cesium resonator In thisway the intrinsic qualities of the cesium hyperfine transition both in terms ofaccuracy and frequency stability are transferred to the USO In a microgravityenvironment the velocity of atoms along the ballistic trajectories is constantand can be changed continuously over almost two orders of magnitude (5ndash500 cms) Differently from atomic fountain clocks presently operated onground very long interaction times (up to few seconds) will be possible whilekeeping reasonable the size of the instrument

The instrument consists of 4 main elements a laser source which providesthe light required to cool and detect the cesium atoms (Fig 5 left) a frequencychain which generates the 92 GHz microwave signal a cesium vacuum tubewhere the interaction between the microwave field and the atoms occurs

Exp Astron

Table 2 Resources and technology readiness level (TRL) of the EGE spacecraft

Unit Volume Mass Power TRL Comment[l] [kg] [W]

Ion clock 100 70 60Atom source 4Atom preparation 6Atom trap 4Atom interrogation 6Clock control 6

Microwave clock 200 91 110 8 TRL by end 2008Local oscillator

Stabilized laser 18 15 23 6Frequency comb 20 16 38 6

Microwave link 18 18 55 8FCDP 7 8 12 8GPS receiver 2 5 7 9CCR 2 2 0 9Payload sum 367 225 305Thermal control 14 17 30 8OBDH 20 24 38 9Star tracker 1 1 8 9Radiation shield 10 40 0 9Support structure 170 204 0 9Power incl solar arr 50 58 10 9Propulsion 21 25 20 9AOCS 36 43 82 9TMTC 8 10 15 9Satellite bus sum 330 422 203 Flexbus conceptPayload plus bus incl 20 margin 8364 7764 6096Propellant 40Launch adapter 100SC wet masspower incl Margins 9164Lattice clock 135 105 152 Alternative clock

Atom source 4Atom preparation 6Atom trap 4Atom interrogation 6Clock control 6

Also included are the resources for an optical lattice clock based on neutral Strontium atoms

(Fig 5 right) and a computer The engineering model of PHARAO is fullyassembled at CNES Toulouse and undergoes functional and performancetests until the end of 2008 The operating temperature range is 10ndash33 andnon-operating temperatures are minus45 +60 The cesium tube consists in aUHV chamber pumped by getters and a 3 ls ion pump a cesium reservoira microwave cavity coils to provide a uniform magnetic field and three layersof magnetic shields Vacuum windows with fiber-optics collimators enable thetransmission of the cooling and detection light onto the cesium atoms

In autonomous mode PHARAO uses the USO as interrogation oscillatorand will provide a clock signal with fractional frequency stability below1 times 10minus13 (τs) minus12 and inaccuracy near 1 times 10minus16 When using the microwave-

Exp Astron

Fig 5 Left The PHARAO laser source engineering model with cover removed Dimensionsare 530 times 350 times 150 mm3 and the mass is 200 kg Ten polarization maintaining optical fibers(yellow) guide the laser beams to the cesium tube All diode lasers are mounted on Peltier coolersfor temperature regulation within 2 mK (courtesy EADS SODERN) Right The PHARAOengineering model cesium tube without the two external magnetic shields The volume is 990 times336 times 444 mm3 and the total mass is 44 kg

optical local oscillator PHARAO will provide a reduced frequency instabilityof 3 times 10minus14 (τs) minus12 This corresponds to 1 times 10minus15 at 1000 s and 3 times 10minus16

at 104 s

422 Ion optical clock

The 171Yb+ ion clock shown in Figs 6 and 7 provides an atomic frequencyreference based on a single 171Yb+ ion confined within an RF electric end-captrap (similar to a Paul trap) within an ultra high vacuum (UHV) chamber andlaser cooled to sim1 mK close to the Doppler cooling limit As a result the ionrsquosfirst order Doppler motion is completely removed and the ion experienceslittle collisional perturbation from its environment Under these conditionsthe narrow linewidth 2S12 (F = 0 mF = 0) minus 2 D32 (F = 2 mF = 0) quadrupole

Fig 6 Left Simplified level scheme of 171Yb+ showing relevant cooling (blue) auxiliary (red)quadrupole clock (green) and octupole clock (dark blue) transitions δνnat is the natural linewidthof the clock transition Middle close-up of Paul-type RF trap (PTB Braunschweig) the ion istrapped in the hole (diameter 14 mm) Right Complete ion trap with vacuum flange height 10 cm(NPL Teddington)

Exp Astron

935 nm D32-state repumper laser

638 nm F72-state repumper laser

739 nm extended cavity diode laser

AOM

PMT

Ion trap UHV chamber

vacuum pumps

Fibre-free space interface

Mu-metal shielding

EC

T

B

Magnetic field

Temperature

Ion fluorescence Processor

Clock laser frequency step and control

4355 nm

Frequengy doubling II

3695 nm cooling wave

Low-drift cavity

Low-drift cavity 171Yb+ 435 nm ion clock unit

876 nm clock laser wave from MOLO

Frequengy doubling

Fig 7 Simplified schematic of the single-ion optical clock unit AOM acousto-optic frequencyshifter PMT photomultiplier tube EC endcap T temperature stabilization system B magneticfield

clock transition at 4355 nm can be interrogated by spectrally narrow and stableclock laser light A diagram of the simplified energy level scheme for 171Yb+showing the relevant cooling auxiliary and clock transitions is given in Fig 6All these wavelengths are provided by diode lasers operated in fundamentalor frequency doubled mode The clock laser light is obtained from the 871 nmwave provided by the MOLO unit (see Section 423) after frequency-doublingin single pass by a KNbO3 nonlinear-optical crystal to 4355 nm The clocktransition spectral profile is observed by stepping the clock laser frequency bymeans of an acousto-optic frequency shifter (AOM in Fig 7) and recordingthe statistics of the quantum jumps in cooling laser induced fluorescence asa function of this frequency Under clock operation the AOM repeatedlysteps back and forward between the half-intensity points on the transitionprofile monitoring the signal imbalance between these points Any detectedimbalance is servo-corrected to zero by feedback to the AOM

In detail the optical clock architecture will comprise (Fig 7)

bull an RF end-cap trap for ionising and confining a single 171Yb+ ion withinan ultra-high vacuum chamber pumped by a small ion pump and non-evaporable getter pump The trap is driven by means of an ac voltage ofa few hundred volts at a drive frequency of sim10 MHz A small oven withseveral mg of 171Yb isotope is heated to provide a low flux of Yb atomsfrom which single ions can be ionised by electron bombardment from ahot wire filament within the trap potential well Low voltage dc is applied

Exp Astron

to additional electrodes to position the ion precisely at trap potentialcentre in order to minimise the ionrsquos micromotion at the drive frequencyThis needs automatic periodic monitoring and correction especially in theperiod after re-loading an ion into the trap External magnetic field coilsin three orthogonal axes allow the nulling of external fields and settingof a fixed field of sim1 μT The trap and coils are surrounded by mu-metalshielding to minimise external field changes during the orbit A level oftemperature control of 1 K of the region surrounding the trap is requiredfor maintaining the blackbody-induced frequency shift uncertainty withtemperature at the 10minus17 level

bull a laser platform to provide frequency doubling of an extended cavity739 nm diode laser for Doppler cooling of the ion on the 2S12(F = 1) minus2 P12 (F = 0) dipole transition at 3695 nm Light from an amplified diodelaser device frequency-doubled in a periodically poled LiTaO3 crystal insingle-pass will be used Typical laser powers used for driving the coolingtransition below saturation are about 2 μW for a beam waist in the trapof about 50 μm By modulation of the injection current of the 739 nmdiode laser at 147 GHz a sideband is generated that excites the 2S12 (F =0) rarr2 P12 (F = 1) ldquorepumperrdquo transition in order to avoid optical pump-ing between the ground hyperfine states

bull The platform also houses the auxiliary lasers at 935 and 638 nm The935 nm diode provides repumping of the ion from the 2 D32(F = 1)metastable level after occasional branching decays to this level during thecooling sequence The 638 nm diode allows fast recovery of the ion fromthe very-long-lived 2F72 metastable state after very occasional collisionaldecay to that state Currently extended cavity lasers are forseen butdistributed-feedback lasers may be available in the near future

bull a high NA lens imaging system and photomultiplier detection system torecord the statistics of 369 nm fluorescence quantum jumps as a functionof clock laser frequency step

bull a fibre system to deliver the various cooling auxiliary and clock light fromsource to trap making use of achromatic doublets where necessary at thefibre-free space interface for launching into the trap

bull a monitoring and control processor which provides the primary coolingand clock laser pulse magnetic field and detection sequencing to observeand lock to the ion clock transition frequency The processor also monitorsfrequency and amplitude data necessary to determine normal laser andion operational conditions and initiate resetting and recovery algorithmswhere necessary and laser unit failure The actual magnetic field presenton the atomic volume is determined and actively stabilized so that aninaccuracy at the level of 1 times 10minus17 can be achieved

bull a redundancy level of at least two units for cooling clock and repumperlasers plus similar for frequency doubling crystals All redundancy unitsfor each wavelength will be fibre multiplexed as standard allowing redun-dant unit activation on determination of prior unit failure

Exp Astron

For an optimized quantum-projection-noise-limited performance of thesingle-ion clock Ramsey interrogation with a cycle time equal to the lifetimeof the metastable D32 level of 50 ms will be performed leading to an Allandeviation of about 27 times 10minus15 (τ s)minus12 This is sim45 times 10minus17 at 1 h satisfyingrequired stabilities for the science objectives Current laboratory results are astability of 1 times 10minus15 at 30 s averaging down to 3 times 10minus16 at 1000 s (PTBGermany) Also comparisons between two independent traps showed agree-ment at the level of several parts in 10minus16 over a series of eight measurementruns

The 171Yb+ ion clock offers a second option namely the 2S12ndash2 F72 octu-pole transition at 476 nm (see Table 1 and Fig 6) [48] It has an extremelylong upper state lifetime low electric quadrupole and second order Zeemansystematic shifts and therefore is a candidate for an even higher performanceclock if a clock laser system with lower instability than the above is providedApart from the clock laser it shares the laser system of the quadrupoleclock Thus with moderate additional resources the Yb+ ion apparatus couldprovide two optical clocks in a single package and specifically perform directfrequency comparisons between the quadrupole and the octupole transitionwith a large sensitivity Aα

The ion clock will operate under microprocessor control providing thecontrol algorithms for the integrated pulse sequencing for cooling repumpingclock interrogation magnetic field switching cooling fluorescence monitoringThe controller will also provide error flags for critical processes and systemreset and correction where necessary This will include ion re-loading andmicromotion reduction algorithms Automatic monitoring of laser power andspectral quality with signal re-optimisation or laser failure determination isrequired Redundant laser units will be pre-aligned and multiplexed into thefibre delivery to the trap so that redundant units can be readily activated in thecase of diode laser failure

423 Microwave-optical local oscillator (MOLO)

This oscillator provides two ultrastable frequency outputs one in the opticalregion the other in the microwave region the two frequencies being coher-ently related The MOLO is composed of three subunits the clock laser thereference cavity and the frequency comb see Fig 8

The clock laser subunit delivers the light for the excitation of the opticalclock transition It is generated by an extended-cavity diode laser at a wave-length of 871 nm Most of the output wave of this laser is sent to the ionclock The laser subunit containing two lasers for redundancy will require3 kg 3 l 15 W The frequency instability of the clock laser must be superiorto the minimum clock instability 27 times 10minus15 (τ s)minus12 for τ up to the timeconstant (sim10 s) of the servo system that locks the laser to the ionrsquos resonancesignal In order to achieve this the laser frequency is stabilized to the reference

Exp Astron

871 nm clock laser

EO

Detector and servo

AOM Femtosecond

frequency comb

To optical clock 10 GHz To FCDP

Microwave-optical local oscillator

ULE cavity

Fig 8 Microwave-optical local oscillator (MOLO) consisting of clock laser reference ULE cavityand femtosecond frequency comb (OFC) Radiofrequency inputs and outputs are not shown EOelectro-optic modulator ULE ultra-low expansion AOM frequency shifter

cavity where the PoundndashDreverndashHall method is used About 10 μW of thelaser output are used for this

The system will exhibit sim04 Hz (sim1 times 10minus15 relative) linewidth and afrequency drift lt01 Hzs The cavity consists of two high-reflectivity mirrorsoptically contacted to a cylinder of diameter 70 mm length 150 mm made ofULE (a glass exhibiting ultra low thermal expansion coefficient) with a cavityfinesse of about 300000 Residual vibrations of the satellite (arms lt 1 times 10minus6 g)should not be a limiting factor to the desired linewidth for safety marginan optimized cavity support can reduce the sensitivity to accelerations Foracoustic and thermal isolation the cavity resides inside an aluminium vacuumchamber equipped with a small (3 ls) ion-getter pump Two stages of polishedaluminium shields are implemented around the cavity which are activelytemperature stabilized so that the cavity is precisely kept at the zero crossingtemperature of the ULE thermal expansion coefficient Heat application andremoval is by thermoelectric elements between the shield and the vacuumchamber Additionally the entire vacuum chamber is temperature-stabilized

During transportation and launch the cavity will be rigidly fixed insidethe vacuum chamber by pressing on it from the sides by piezo-mechanicalactuators allowing accelerations of several g After bringing the satellite onthe orbit the actuators will be released

The optical setup for frequency stabilization will be a classic Pound-Drever-Hall scheme in reflection using an electro-optic modulator (EO) Couplingof the clock laser radiation to the cavity is via single-mode polarization main-taining optical fiber The fiber in- and out-couplers are mounted on miniaturepiezo-motor driven multiaxis translation stages Microprocessor control ofthe translation stages performs a laser-to-fiber incoupling stabilization andlaser-to-cavity mode-matching optimization (positioning inaccuracy lt10 μmalignment inaccuracy lt10 μrad)

The frequency stability of the clock laser will be limited by thermal 1f noisein the mirrors with an fractional Allan deviation slightly below 1 times 10minus15 from1 s to a few 10 s Resources including control electronics are 15 l 12 kg 8 W

Exp Astron

A femtosecond fiber laser based optical frequency comb generator (OFCFig 8) will be used for the clock frequency comparisons and to obtain aspectrally ultrapure microwave local oscillator output for the operation ofPHARAO for the microwave link and for the GNSS receiver The fem-tosecond laser is based on erbium-doped fiber pumped by fiber-coupled laserdiodes It emits 100 fs pulses at a repetition rate of 250 MHz with an averagepower of 100 mW The emission spectrum is centered around 1560 nm Spec-tral broadening to more than one octave is obtained via nonlinear interactionin a microstructure fibre so that the range from about 1050 to 2100 nm iscovered

Second harmonic generation of a portion of the broadened spectrum allowsobtaining the required light for the measurement of the clock laser frequencyat 871 nm The OFC will be referenced to the clock laser frequency (afterintroduction a controllable shift via the frequency shifter AOM) by makingthe carrier envelope offset frequency vanish so that the repetition rate fr isa high-order subharmonic of the optical clock reference frequency f1 = m1 frThe link between the optical and the microwave domain is then simply thedistribution of the repetition rate frequency or more conveniently a high(order 40) harmonic of the comb repetition rate 10 GHz

It has been demonstrated by several groups that frequency comparison andfrequency division with an OFC can be performed to very high accuracy andstability beyond the limitations imposed by the optical clocks themselves Forexample a fiber laser OFC with an instability (Allan deviation) of 6 times 10minus17

in 1 s has been demonstrated [51] OFCs based on fiber technology possessgood long-term reliability and high power efficiency The projection for a spacequalified OFC based on current industrial terrestrial technology is 16 kg mass20 l volume and 38 W power requirement

424 Frequency comparison and distribution package (FCDP)

This package distributes the ultra-stable 10 GHz reference frequency outputof the MOLO to the MWL performs frequency comparisons between MOLOand PHARAO and generates a variety of different reference signals for theindividual instruments including the 250 MHz reference clock to the OFCAll signals are coherent to the MOLO In the FCDP a USO is locked to theMOLO signal by a digital PLL A divider chain generates timing pulses (1 pps1000 pps etc) and keeps a continuous local time scale From the digital phasecomparison data the phase noise density spectrum between the USO and theMOLO is available by telemetry (TM) to assess the overall frequency purityin-flight

As the FCDP distributes the full performance reference from the MOLOdirectly to the MWL there is minimum impact of the FCDP on overall clockcomparison performance The USO-phase lock loop uses a similar designas for ACES-FCDP The digital PLL with linear phase detector allows theUSO to be slightly offset from its reference which is selectable by groundcommand This results in an accurate on-board time scale despite any offset

Exp Astron

of the MOLO reference frequency which can be synchronised to UTC Thefrequency synthesis chain re-uses modules developed for ACES-FCDP andtheir frequency inaccuracy Expected relative uncertainty (Allan deviation)of the frequency comparison between two signals provided to the FCDP is47 times 10minus17 at 1000 s 95 times 10minus18 at 10000 s

During non-availability of the MOLO ie during system start-up andcontingency conditions the USO runs as flywheel (ie it is unlocked withinstability lt15 times 10minus13 from 1 to 100 s) It will still provide useable butless accurate references to the instruments including the reference to OFCCommon view ground-to-ground time- and frequency comparison and supplyof the remaining instruments remain possible at the USO provided accuracy

Mass volume and power consumption are 8 kg 7 l 12 W Controldata rateis 10 kBday 50 kBday (FCDP alone)

425 Microwave link (MWL)

The microwave link is a two-way triple-frequency microwave link (Fig 9) Itis designed to perform accurate frequency comparison as well as time-transfer

Ku-Band Up-linkPower Tx 5 WCarrier 13475 GHzPN-Code 100 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

S-Band Down-linkPower Tx 1 WCarrier 2248 MHzPN-Code 1 MChips1pps 1 time marker sData 5 kBits

Ku-Band Down-linkPower Tx 1 WCarrier 1470333 GHzPN-Code 100 MChips1pps 1 time marker sData 5 kBits

EGE SC

S

Ku

KaKa-Band Up-linkPower Tx 5 WCarrier 345 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

Ka-Band Down-LinkPower Tx 1 WCarrier 32 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 5 kBits

GS Antenna diameterEGE terminal 1 mACES terminal 06 m

Fig 9 EGE microwave two-way link concept and characteristics See text for a description Thelink will be able to perform frequency comparisons with up to ten ground stations simultaneously(only one active comparison is shown) pps pulse per second Tx transmitter SC spacecraft

Exp Astron

between on-board and on-ground clocks Two-way operation allows for com-pensation of the effects due to spacecraft location and its motion relative toground Its two bi-directional links operate in Ka-band (32345 GHz) andKu-band (147135 GHz) The high carrier frequency of the up- and down-links allows for a noticeable reduction of the ionospheric delay A third down-link channel is in S-band (22 GHz) and is used to determine the ionospheretotal electron content MWL uses a combination of carrier phase and codephase data to measure with high precision the absolute ionospheric delayas well as ionospheric delay variations during a ground contact The useof three frequency bands allows to resolve third order ionospheric termsKey modulation parameters are Ka band Tone 250 MHz Ku band PN100 MChips S-band 1 MChips

The EGE-MWL concept is a considerably improved version of the ACES-MWL Microwave carrier generation is directly from the 10 GHz microwaveoutput available from the MOLO and having exceptionally low close-in phasenoise without intermediate frequency dividers or other complex synthesiserstages commonly used in RF transponders The high frequency carriers arekey to achieve the desired frequency comparison accuracy

The Ka-band carrier is generated by tripling the FCDP-generated signalThe Ku-band is subsequently generated by division by two using a verystable regenerative divider Finally the assigned radio frequency is obtainedby adding a low frequency signal using an image reject mixer A mixedanalog-digital receiver design will enhance its performance by use of a flexibledigital signal processing and digital control loops The number of simultaneousground contacts is increased from four (as in the ACES mission) to ten or moreusing the digital concept while the number of physical receiver channels perband is reduced from four to two one for operation to ground and one for in-flight calibration The concept allows adaptation of receiver loop parametersto optimise for the different portions of a highly elliptic orbit while reducingsystematic uncertainties due to high Doppler and Doppler rate Real-timedata exchange to ground terminals is by additional data modulation on all RFsignals The spacecraft carries two different small antennae for perigee and forapogee with gains of 8 dBi 20 dBi beamwidths 100 (full earth coverage) 14respectively power is 1 W for each

The MWL end-to-end performance between the attached clocks includingspace- and ground terminals is predicted to be a relative Allan deviation of1 times 10minus16 at 1000 s (perigee) and 1 times 10minus17 at 10000 s (apogee) Thesevalues are based on conservative implementation losses and other instrumentallosses assumed as 6 dB at apogee and 10 dB at perigee including additionallosses due to atmosphere An attitude knowledge of 1 is required

The MWL link will also support precise orbitographyndashcomplementary tothe GNSS receiver Generation of ranging data is simultaneous to two-waytime transfer and is available under all weather conditions Orbit determina-tion accuracy is expected to be better than 1 cm well matching EGE clockrequirements The value is expected based on demonstrated results fromother missions (METEOR ERS-2 TOPEXPOSEIDON) Laser ranging to

Exp Astron

the CCR array is used a few times during the mission to calibrate the MWLranging

In case of non-availability of the ultrapure signal from MOLO an additional10 GHz output is available from the FCDP to support ground-to-ground clockcomparison in common-view mode even after the nominal mission periodwhen the neither the clocks nor MOLO may be available any longer Basedon the PRARE experimentrsquos 12 years of operation this can be assumed forthe FCDPMWL combination as well The potential extended operational pe-riod makes the proposed configuration particularly attractive to metrologicallaboratories that rely on availability of long-lasting data in support of thegeneration of UTC

The resources of MWL are as follows mass 18 kg volume 18 l power55 W TC 2 kBday TM 120 MBday

426 Auxiliary units

The star camera assembly on the EGE spacecraft will provide real-timeattitude data primarily for the orientation of the spacecraft (with accuracy of1 in real-time) and the proper reduction of the GNSSSLR data

A GNSS receiver is used for the precise orbit determination time andfrequency transfer coarse positioning (lt1 m) for real time use by the Attitudeand Orbit Control as well as the back-up for time tagging of all payloaddata For the time and frequency transfer the receiver is nominally using theonboard frequency reference The receiver assembly includes three specialantennas By the year 2015 more than 100 GNSS satellites are expected tobe available for precise orbit determination permitting orbit inaccuracy of 1ndash2 cm RMS for altitudes up to approx 10000 km At apogee a combinationof GNSS signals and MWL ranging will be used with expected inaccuracyof 10 cm

5 Basic spacecraft specifications

The mission foresees a single spacecraft with a mission duration limited toapprox 3 years for cost reasons Attitude control along three axes withless than 1 error on a timescale of 1000 s is required The control can beperformed with cold gas thrusters magnetotorquers or reaction wheels Theselection will be done on the basis of the vibrations induced by these systems(desired level arms lt 10minus6 g)

Cold gas thrusters or monopropellant thrusters will be required to maintainthe orbit parameters during the mission by small orbit correction manoeuvresand will also guarantee a proper tuning of the orbit parameters after the releasefrom the launcher The ground segment (telemetry control and data transfer)could be via non-ESA stations user-operated to save costs

Further resources are indicated in Table 2

Exp Astron

The satellite bus design drivers are nadir-pointing microwave antennasthermal stability for the payload instruments no strong magnetic sources lowmicrovibrations harsh radiation environment due to crossing of the radiationbelts once per orbit power demand by science payload (with the possibility toreduce power demand by stand-by of some components during specific missionphases eg eclipse) mass limit due to required high altitude with low costlaunch

The satellite bus design can be based on established platforms such asFLEXBUS or PRIMA used on several past missions Their design philosophydiffers one having body-mounted solar arrays the other wings The latter canprovide larger amounts of power The vibration level induced by the panelswhen they are fixed and the duration of solar array drive periods per orbitwill have to be considered in the selection The FLEXBUS concept was usedfor a mission already requiring a low level of residual accelerations and theplatform is optimised for this parameter

For most of the satellite components one can consider using compo-nents and electronic units from navigation or communication satellites whichtypically cruise at altitudes between 15000 and 25000 km altitude LISAPathfinder which will cross the radiation belts during spiralling to theLagrange point will face similar problems concerning radiation as EGE as wellas XMM which is in a Molniya-type orbit with similar radiation characteristics

The instruments will therefore be located in the satellite structure such asto realise ionizing radiation shielding by the spacecraft structure and non-sensitive equipment to the maximum extent possible the need for additionalradiation shielding (including trapped proton radiation) for especially sensitiveparts of the instruments must be determined by ground tests It should bementioned that an orbit with higher perigee and less radiation dose is also anoption as it leads to a tolerable reduction of the science goals

The required plusmn3C temperature stability at the interface to the payloadenclosure is feasible The thermal control will employ the usual means (anenclosure with high thermal inertia heaters and a suitable radiator) Thelocation of the radiator will need a special study given the non-standard orbit

Magnetic field present during the mission can affect the properties of theonboard clock Using SPENVIS the magnetic field has been calculated alongthe spacecraft orbit the field intensity presents maxima around 02 G anda minimum value of the order of 10minus3 G As discussed above this is not alimitation to the instrument performance

6 Discussion and conclusion

The accuracy of the measurements depends crucially on the stability andaccuracy of the clocks While the performance of ground clocks (even thoseeventually deployed in new ground stations dedicated to EGE) will likelyimprove significantly from the present level the space clock design will beldquofrozenrdquo approx 10 years before data acquisition Thus their performance

Exp Astron

level will be significantly below the future ground station clock performanceAs specification we have chosen a performance level that has been demon-strated with laboratory breadboards recently We see no fundamental reasonthat this laboratory performance cannot also be achieved in a space ion opticalclock The experience with the PHARAO cold atom clock has shown thatan engineering model of a laser-based cold atom clock has a performancesimilar to that of a laboratory prototype The perturbation of the ion clock fre-quency due to magnetic fields along the orbit can be modelled and suppressedwith shielding and the residual effects measured and compensated for Themoderate temperature stability (plusmn 3C) to be provided in the experimentalbay can be achieved with thermal shields Accurate temperature controlof particular units is done by active stabilization of the units themselves

One important aspect is radiation which may affect all optical and trapcomponents as well as the lasers and reference cavity The radiation envi-ronment on the EGE orbit is significantly harsher than the near-earth ISSorbit on which ACES will operate The radiation effects must therefore becharacterized comprehensively with appropriate tests and then the radiationshield for the EGE instruments designed

In summary the Einstein Gravity Explorer mission (EGE) is a satellitemission aimed at a precise measurement of the properties of space-time usingatomic clocks By testing one of the most fundamental predictions of EinsteinrsquosTheory of General Relativity the gravitational redshift and thereby searchingfor quantum effects in gravity it explores one of the most important frontiers infundamental physics The EGE mission reaches beyond fundamental physicsproviding an important spin-off in geophysics Furthermore the technologicalaspects of the mission are ground-breaking for future time-keeping and ref-erence frame systems in space and thus are also expected to be of wide useto science and technology in general The payload of the mission has beenconceived as a compromise between cost and performance needing to fit thecost envelope of an ESA medium-class mission1

Acknowledgement We thank S Cesare for helpful information

References

1 Will CM The confrontation between general relativity and experiment Living Rev Relativ9 3 (2006)

2 Ni W-T Empirical foundations of the relativistic gravity Int J Mod Phys D 14 901 (2005)doi101142S0218271805007139

1Finally we point out that a spacecraft with a significantly reduced and thus lower-cost payload (noclocks only MOLO FCDP MWL) on a circular orbit of appropriate height could be used for along-duration earth science mission dedicated specifically to the measurement of the geopotential

Exp Astron

The available theory is fairly complete for the data analysis in mode 23 (1)described below For modes 23(2 3) the theory will need to be generalizedto include the effect of the gravitational potential at the location of the groundclock(s) (including time-varying effects such as the tides) and of the solar andlunar potentials

The uncertainties of the various contributions above are minimized by ausing precise EGE orbit data obtained by satellite tracking and by usingcurrent and future accurate earth gravity information Required orbit positionknowledge is at 1 cm level near perigee and sim50 cm near apogee reachablealready today with the proposed orbitography approach described below Thevalidity of the special-relativistic Doppler shift has been and will continue to beindependently verified with increased accuracy by spectroscopy of relativisticatomic ion beams [30] Potential violations of this aspect of Lorentz Invari-ance are expected to be sufficiently bounded so as not to affect the signalinterpretation

26 Local Lorentz invariance tests

There has been an explosion of interest in Local Lorentz invariance in re-cent years [31 32] with numerous astronomical studies and high-precisionexperiments applied to search for violations of Lorentz invariance Signifi-cant developments have also occurred on the theoretical side and a theory(standard model extension SME [33]) has been worked out that provides aunified framework for describing and analyzing Lorentz Invariance violationsin a variety of systems Two aspects of Lorentz Invariance can be tested withEGE

261 Independence of the speed of light from the laboratory velocity

A possible dependence of the speed of light on the speed v of the laboratorycan be modeled according to the MansourindashSexl test theory [34] by

c (v) = c0(1 + Bv2

c2

)

Here v 377 kms is the velocity with respect to the cosmic microwavebackground the cosmologically preferred frame and B is a combinationof parameters describing deviations from the usual Lorentz transformationformulas B = 0 if Lorentz Invariance holds For a terrestrial experiment therotation of the Earth around its axis modulates v with a 300 ms amplitude Adependence c(v) can be searched for by measuring the frequency differencebetween a resonance frequency of a highly stable optical cavity (cavity fre-quencies are proportional to c) and an optical clock (Kennedy-Thorndike-typeexperiment [35]) and determining the modulation of the frequency differencecorrelated with the modulation of v The advantages of a space experimentare the high orbital velocity and strongly reduced cavity deformation thanksto microgravity [36 37] On the proposed elliptic orbit v varies between+4 and minus4 kms over approximately 1 h This variation is 13 times larger

Exp Astron

than for the case of a terrestrial experiment The shorter time scale is alsoadvantageous since the drift of the cavity is more predictable In the EGEmission the reference cavity of the clock laser is used The large numberof orbits improves the statistics An improvement of the accuracy of B by afactor 20 compared to the best current terrestrial results is realistic The dataanalysis can also be performed in the framework of the SME theory takinginto account complementary bounds obtained from terrestrial experimentsand astrophysical observations

262 Independence of Zeeman splitting frequency on the directionof the magnetic field

One class of tests of Lorentz Invariance consists in measuring the frequencysplitting between two levels of a quantum system induced by a static magneticfield as a function of orientation of the field direction with respect to thestars (HughesndashDrever-type experiments [31]) This class addresses the so-called matter sector and is complementary to combined-photonmatter-sectortests such as the above The most precise experiments are performed withatomic clocks eg masers or cold atom clocks [38] The EGE instruments allowsuch experiments since magnetic fields are applied continuously in the clockor repeatedly for calibration Compared to terrestrial experiments the highorbital velocity of EGE could lead to an improvement by a factor sim20

27 Application to geophysics

Using GPS coordinates of a point on the Earth can be obtained with aninaccuracy well below 1 cm in a well defined international terrestrial referencesystem The coordinates are purely geometrical and do not contain any gravityinformation A local gravitational potential UB is determined with respect toa reference gravitational potential U A by ldquolevellingrdquo the surface of the Earthie sequentially measuring gravity g and height differences every ca 50 m

UB minus U A = minusBint

A

g (r) middot d n

where n is the difference vector between subsequent locations The disadvan-tage of measuring height differences by the levelling method is a random walkeffect of accumulated errors

In the classical definition a geoid is defined as the particular equipotentialsurface nearest to mean sea level In these terms the geoid serves as areference surface for measuring the height and also to define a referencefor the gravitational potential Such a vertical reference frame historicallywas realized for a country or several countries by determining the mean ofsea level observations at tide gauge stations taken over long period of timeHowever modern satellite altimetry missions such as TOPEXPOSEIDON orENVISAT show that departure of the mean sea level from an equipotential

Exp Astron

surface may reach up to several meters on the global scale see eg [39]Therefore height systems based on different tide gauge stations may differin the realisation of the geoid and may differ with respect to the reference byseveral meters

The best gravity field models obtained from satellite data (eg missionGRACE) have reached a precision of about 1 cm over sim250 km half wave-length [40] Satellite data also reveals changes in time [41] However for typicalEarth topography with height variations of eg 1000 m over 30 km horizontaldistance one may expect variations in the geoid of about 80 cm Such ahigh-spatial-frequency signal in geoid variations cannot be detected by spacegravity missions and requires a combination of satellite and terrestrial gravitymeasurements like gravity anomalies deflections of vertical and GPSlevellingpoints The best combined global gravity field models are provided up to de-gree and order 360 in terms of spherical harmonic representation correspond-ing to a half-wavelength of about 55 km However in combining the satelliteand terrestrial gravity field measurements the problem of terrestrial data givenin different height systems between continents and different countries remainsand has to be tied to satellite measurements

The geodetic scientific community is currently establishing a Global Geo-detic Observing System (GGOS) [42] Its objectives are the early detectionof natural hazards the measurement of temporal changes of land ice andocean surfaces as well as the monitoring of mass transport processes in theEarth system Global change processes are small and therefore difficult toquantify Therefore the required precision relative to the Earthrsquos dimensionis 1 ppb GGOS will be established by the combination of geodetic spacetechniques (GPS Laser-Ranging very large baseline interferometry) andrealized by a very large number of terrestrial and space-borne observatoriesThe purely geometric terrestrial 3D coordinate system is in good shape andfully operational It has to be complemented however with a globally uniformheight system of similar precision The current precision level of regionalheight systems in terms of gravity potential differences is in the order of1 m2s2 (10 cm) with inconsistencies between these various systems up toseveral 10 m2s2 (several meters) The actual requirement in the context ofGGOS is 01 m2s2 (1 cm) with the need of a permanent ie dynamical controlThis requirement of high precision height control comes from the need tounderstand on a global scale processes such as sea level change global andcoastal dynamics of ocean circulation ice melting glacial isostatic adjustmentand land subsidence as well the interaction of these processes Only by meansof monitoring in terms of gravity potential changes at the above level ofprecision the change of ocean level can be understood as a global phenomenonand purely geometric height changes be complemented by information aboutthe associated density or mass changes

The geoid can also be defined in a relativistic way [43] as the surface whereaccurate clocks run with the same rate and where the surface is nearest tomean sea level The relation between the differences in the clock frequenciesand the gravitational potential is given in simplified form by Eq 1 At present

Exp Astron

there is no operational way to compare frequencies of the already availableoptical clocks on the global scale at the same level as their accuracy wouldallow With EGE and by then improved ground optical clocks it will becomepossible to obtain gravitational potential differences on a global scale by com-paring frequencies This would be a significant new dimension to gravity fielddetermination since such observables are given on the global scale and providein situ local gravity information at the same time EGE will allow clock-basedgravitational potential mapping with the same payload instruments used forthe fundamental physics experiments The on-board clocks are actually notrequired for this purpose MOLO MWL and FCDP suffice Thus EGE willallow establishing a global reference frame for the Earth gravitational fieldwith accuracy in the order of a few cm in terms of the geoid heights Thisassumes that by the time EGE is operational (mobile) ground clocks withfractional inaccuracy at the level of 10minus18 are available the measurementsare limited by MWL noise and thus require a long integration times to matchthe expected ground clock performance Such a reference frame will serve asreference for all gravity field modeling and clocks will be used to define theinternational atomic time scale and the Earth gravitational potential at thesame time

In summary with optical clocks geodesy will undergo a second revolutionafter geometry now being measured by clock-based GNSS systems also physi-cal heights and potential will be measured by (mobile) optical clocks using thegravitational redshift effect

28 Application to frequency standards and terrestrial fundamentalphysics studies

As has been described repeating ground clock comparisons and integratingover several months will provide global ground clock frequency comparisonsat the level of 1 times 10minus18 more than two orders of magnitude below the currentGPS and two-way satellite time and frequency transfer methods In the timeand frequency community the availability of such a global high performancemicrowave link will accelerate the process leading to a redefinition of the SIsecond based on clocks operating in the optical domain

The global character of the ground clock comparisons made possible byEGE can also contribute to the search for a time-variation of the fundamentalconstants

The comparisons do not require availability of the on-board clocks The riskof this type of measurements is therefore reduced Moreover the lifetime ofmany subunits required for this measurement (MWL and FCDP see below)can exceed 10 years (from previous experience with similar devices) andthus offers a motivation to increase the mission duration beyond the nominal3 years

Further applications to metrology are listed under the secondary goals inSection 21 above

Exp Astron

3 Mission profile

31 Orbit

The orbit must satisfy the following requirements

Primary 1 Large gravitational potential difference between apogee andperigee

2 sufficient contact time with at least two ground stations atperigee

Secondary Simultaneous visibility of satellite from distant (inter-continental) ground stations for several hours

To satisfy the primary objective the orbit must be highly elliptic The perigeeshould be low A very low perigee means a short contact time which decreasesthe accuracy of the frequency measurement A compromise is a perigeealtitude of sim2500 km For a sufficiently high apogee this yields a U of 60of the theoretical maximum between the earth surface and infinity

The baseline scenario is an orbit of approx 12 h period with the followingproperties

Apogee Altitude 37856 km (minus 10 deg latitude one over Atlantic andPacific resp)

Perigee Altitude 2500 km (+ 10 deg latitude one over Guatemala andMalaysia resp)

Inclination 634 argument of perigee 170RAAN 25 True Anomaly 220Apogee drift rate 70 kmday perigee drift rate sim 14 kmday

True anomaly and right ascension of the ascending node (RAAN) wereoptimized to minimize orbital precession Figure 3 shows the ground track andthe ground coverage at one of the perigees Common-view contact time forstations in Europe is about 6 h while between a station in Europe and in theUSA it is about 35 h Frequency comparison windows between a given groundstation and EGE repeat every 24 h

A final selection of the orbit requires an in-depth study of the costs of thepropellant and maneuvers as well as of the setting up of new ground stations

32 Ground segment

The mission requires two separate ground segments the science groundsegment and the standard satellite control segment which supports telemetryand telecommand (TMTC)

Exp Astron

Fig 3 Baseline EGE orbit One (out of two) apogee and one (out of two) perigee ground pointsare shown The ground coverage at one of the perigees is shown with circles at elevation anglesteps of 10 starting from 10

321 Science ground segment

As described above one part of the mission is to perform frequency com-parisons between ground-to-space and ground-to-ground (common-view andnon-common view) The use of a two-way microwave link (MWL) systemhas the advantage that a precise orbit or the precise relative velocity of thesatellite must not be known [4] The orbit determination requirements fora two-way system are significantly relaxed compared to the requirements tomodel the relativistic effects on the clocks The frequency comparison resultsare available both on ground and on the satellite

For ground-to-satellite clock comparisons five ground stations arefavourable two capable of viewing perigee three for viewing apogee Thisallows for the classic three-clock configuration where the performance of eachclock can be obtained individually The perigee stations will be observing thesatellite for sim25 min every perigee passage for the duration of the missionThe apogee stations will observe for sim3 h every apogee passage Of coursea larger number of stations as well as longer observation time are usefulfor robustness redundancy and orbit determination The stations could beinstalled on a temporary basis for the duration of the mission

The stations must have access to optical clocks of performance at least ashigh as todayrsquos state-of-the-art ground clocks but preferably higher Theywill include an optical frequency comb by means of which the optical clockfrequency is transformed to a radio frequency for comparison with the EGEclocks High-accuracy fiber-optic link technology already existing today allowsdistances of sim100 km between the ground station and the correspondingoptical clock allowing a certain freedom in the choice of the clock locationWith further increase of the fiber-optic link range currently being investigated

Exp Astron

the area over which the ground station can be positioned relative to the opticalclock will increase This option is of importance for the implementation of ageophysics measurement campaign for which a large number of clock locationsis of interest

The above ground stations will be complemented by some 15 additionalground terminals provided by the world-wide laboratories equipped with highperformance primary clocks at the time of the mission These will participate inorbit determination to 1 cm accuracy along the full orbit which can be achievedwith non-optical atomic clocks The EGE MWL is fully compatible to plannedACES ground terminals which might reduce the number of required stations

With a larger number of transportable optical clock ground stations de-ployed and moved across the continents a geophysics campaign could beimplemented to map the gravitational potential

Ground station equipment MWL uses dedicated small transportable termi-nals which can be easily installed at any location of interest Each terminalconsists of a steerable dish antenna of 1 m diameter with pointing accuracy of01 using a transmitted power of 5 W per band The high directivity antenna(beam aperture Ka-band 07 Ku-band 14 S-band 20) avoids multi-pathfrom nearby objects Antenna noise temperatures are Ka-band 450 K Ku-band 340 K Operation is at elevations above 10 and measurements areperformed above 20 For ranging purposes terminal location is determined byGPS surveying Each ground station computes its own ionospheric correctionsbased on a triple-band receiver The terminal has a built-in delay monitorwhich allows calibrated ranging during several months once its bias has beendetermined by laser ranging Its M and C interface is via local area networkto either a userrsquos computer or directly to the science network control centreby remote MampC The stationrsquos operation is fully automatic and unattendedSchedule and orbital data are received via LAN once per day Signal acquisi-tion is fully autonomous Data transmission is after each pass The terminal hasan uplink telecommand capability of 1 kBits and a downlink data capacity ofup to 10 kBitss For the local user there is a visual interface providing quick-look data from real-time comparison to the spacecraft clocks to verify theirgood health The final full performance product is calculated at the sciencedata centre

Local metrological data (temperature pressure humidity) are recorded tocorrect for tropospheric delay of the ranging data The main advantage ofMWL terminals compared to laser based systems is their all-weather capabil-ity although microwaves suffer higher propagation errors in the ionosphereand troposphere

The EGE ground terminal will profit from the ERS-2 mission PRARE(precise range and range rate equipment) [45] experiments whose terminalshad similar design and have been operated under extreme climatic conditionsincl Antarctica and high temperature regions from existing designs fromthe space segment (mission ACES) and no further special developments aredeemed necessary Cost per EGE master station is modest

Exp Astron

Satellite tracking via laser ranging Laser ranging is alternative to MWL rang-ing and also serves for absolute calibration of the latter For this purpose anon-board corner cube reflector (CCR) is provided Ranging will be done withthe International Laser Ranging Service Ideally the satellite will be rangedat every perigee passage and every apogee passage simultaneously with thefrequency comparison procedures Todayrsquos ranging precision is sufficient tosatisfy the science goals For apogee observations in the northern hemispherea large number of laser ranging stations is available The number of stationscapable of ranging at or near perigee in the southern hemisphere is six andthus is sufficient even considering unfavourable weather conditions at someof these The MWL ground stations are able to range even under thoseconditions

322 Control ground segment

Average data transfer rate is estimated at 300 MBday (approx 40 kBs onaverage) Data will be stored on board (2 GB capacity required) and will betransferred to a receiving station on the Earth once a day

XPLC

incl Data StorageIF to TMTC Link

HK amp TMTC

to ground

Opt Clock

AOM

MOLO

PHARAO

FCDP

10 GHz

100 MHz

250 MHz

GNSS Rx

MWLTampF Link

uplink Data

UTC(k)

TampF Space vs Ground

TM from GS

Tim

e S

cale

Orbit GNSS Time UTC

TampF Space vs Ground

MampC Time Scale

MampC

MampC

MampC

MampC

Mamp

C

PHARAO vs Opt Clock

PHARAO vs Opt Clock

10 GHz amp Epoch

10 MHz

amp Epoch

Epoch

10 MHz

amp Epoch

Opt Link

Fig 4 Schematic of the payload showing subunits and their signals The results of the frequencycomparison between the two on-board clocks (optical clock and microwave clock PHARAO) areobtained from a combination of the frequency values applied to the frequency shifters AOM (seeFigs 7 and 8) and those output from the frequency control and distribution package (FCDP)as indicated by the arrows ldquoPHARAO vs Opt Clockrdquo The satellite clock to ground clockcomparison results are output from the MWL XPLC is the on-board computer and is involvedin generation of signals for instrument control The optical link shown dashed is a possible add-onproviding an alternative to the MWL GS ground stations MampC monitoring and control HKhousekeeping TampF time and frequency

Exp Astron

4 Main payload instruments

41 Overview

The satellite payload consists of a pair of cold-atom clocks a microwave-optical local oscillator (MOLO) an on-board frequency distribution unit(FCDP) a satellite-to-ground clock comparison unit (MWL) and the onboardcomputercontrol system (XPLC) see Fig 4 Auxiliary units are the posi-tion determination GPSGalileo receiver (GNSS Rx) attitude monitor (startracker not shown) and the corner cube reflector array (CCR not shown)

The science goals require clocks of very high frequency stability on thetimescales of half the orbital period (6 h) as well as perigee passage (sim1000 s)see Fig 2 Only clocks based on cold atoms are suited for the science goalsTable 1 reports values for a subset of atomic clocks that have been investigatednot including clocks that require cryogenic operation or are too complex foruse in an medium-class mission The choice of a clock pair rather than a singleone fulfils several requirements it addresses more science goals improves thevalidity of some science results and it enables calibration accuracy evaluationand redundancy The validity of the science results is enhanced since theclocks differ not only in the employed atoms but also in terms of havingdifferent sensitivity to perturbations by electric and magnetic fields electronicserrors and misalignments A comparison of the results arising from each clockallows for an understanding of measurement errors Both the clocksrsquo and thefrequency comparison unitrsquos performances will be determined after launchand regularly verified during the course of the mission The only reliable

Table 1 Properties of different clock candidates for EGE

Clock Clock Fractional σ = Fractional Sensitivity Figure of Sensitivityatom frequency inaccuracy instability to α Aα merit |Aα| to meq

at 1000 s σ [1016] QCD

Sr+ 445 THz 3 times 10minus17 (p) 5 times 10minus17 (p) 042 08 (p) ndashYb+ (1) 688 THz 3 times 10minus16 (d) 3 times 10minus16 (d) 088 029 (d) ndash

3 times 10minus17 (p) 1 (p)Yb+ (2) 642 THz lt 1 times 10minus17 (p) 1 times 10minus17 (p) minus53 53 (p) ndashSr 435 THz 1 times 10minus16 (d) 1 times 10minus16 (d) 009 009 (d) ndash

lt 1 times 10minus17 (p) 09 (p)Yb 519 THz 7 times 10minus16 (d) 1 times 10minus16 (d) 031 03 (d) ndash

lt 1 times 10minus17 (p) 1 times 10minus17 (p) 31 (p)Cs 91 GHz 2 times 10minus16 (d) 1 times 10minus15 (d) 275 027 (d) 1Yb+ 126 GHz 1 times 10minus15 (p) 1 times 10minus15 (p) 34 034 (p) 1

A relevant figure of merit is the ratio of |AX | and fractional frequency instability σ on the perigeevisibility timescale The Cs clock is the only high performance clock with adequate sensitivity tothe particle masses me mq The indicated Cs clock performance was achieved using a microwavelocal oscillator based on a cryogenic oscillator On EGE the latter is to be replaced by an opticallocal oscillator-based device (MOLO) with similar short-term performance but without the needto operate at cryogenic temperature The clock instability at 10000 s is typically similar to theindicated inaccuracy(p) projected (d) demonstrated (1) quadropole (2) octupole

Exp Astron

way to achieve this is by having at least two clocks on board The on-boardclock-clock comparison provides performance evaluation independent of theground-satellite comparison unit Once the satellite clock performances areestablished the link performance can be established as well

The clocks are similar in basic structure but differ in the detailed implemen-tation They consist of

a source of atomsa preparation subunit which cools the atoms to near-standstill (millikelvintemperature or below)a trapping subunit that confines the atom(s) to a tiny volumean interrogation subunita clock control electronics package

The clocks share a common central unit the optical-microwave local oscil-lator which serves as a flywheel oscillator with outstanding stability on shorttimescales (01ndash10 s) for both clocks It consists of a laser stabilized to an ultra-low loss optical resonator made of ultra-low-expansion glass A microwavesignal is derived from the laser by phase-locking a frequency comb to the laserwave and using a multiple of the repetition rate

While the subunits of any optical clock have a similar technological basis(optics electrooptics acoustooptics thermal control vacuum systems) theydiffer in their specifications Depending on the choice of the atom species de-livery confinement and interrogation methods clocks of different perfor-mance result A number of laboratory optical clock demonstrators have beendeveloped worldwide This ensures availability of the know-how for the indus-trial implementation of the flight models

The first clock is a microwave clock based on the interrogation of slowlymoving ensembles of cold Cs atoms It is essentially based on the PHARAOclock [5 6] developed to engineering model level by CNES in the frame-work of the ESA ISS mission ACES In EGE this clock is upgraded usingthe microwave-optical local oscillator This improves significantly the clockperformance The specification is a relative instability (Allan deviation) of3 times 10minus14 (τ s)minus12 ie at the integration time τ = 1000 s it drops to 1 times 10minus15and to 3 times 10minus16 at 10000 s While this specification is at least a factor 3 lessstringent than for the optical clocks the significantly lower cost of obtaining aflight model and the sensitivity of its frequency to the particle masses makes ita suitable choice

The optical clock baseline instrument is a single-ion optical clock Amongvarious suitable ion species (Hg+ Al+ Sr+ Yb+) [44ndash47] the Yb+ ion hasbeen selected Its breadboard performance is established at the instability levelof 3 times 10minus16 over 1000 s it exhibits a large sensitivity coefficient Aα hassmall sensitivity to magnetic fields and exhibits significant potential for furtherimprovements (use of the octupole transition with sim6 times larger Aα andlower overall instability [48] see below) A trapped Yb+ ion is also a veryrobust system it has been demonstrated that a single Yb+ ion can remaintrapped uninterruptedly for many months Finally Yb+ is the only ion with

Exp Astron

which optical clocks demonstrators have been operated in not a single but twomajor metrology laboratories

Another attractive optical clock is based on ensembles of cold neutral Sror Yb atoms trapped in an optical lattice produced by laser beams This clocktype has a lower instability than a single-ion clock thanks to the large numberof atoms (sim105) used Its development is in progress in several laboratoriesin the world (in part supported by space agencies) and has to date reached6 times 10minus17 instability at 5000 s and 1 times 10minus16 uncertainty [49 50] with potentialfor even higher performance However it is more complex and costly thanthe Yb+ ion clock since the cold atom preparation and trapping subunitsinclude more components require higher laser powers and the respectiveoperating procedures are more complex It is therefore not considered for thepresent mission designed to be developed in the very near future with limitedresources but is an interesting candidate for missions without such constraintsIts estimated physical parameters are indicated in Table 2 for comparison

42 Main instruments

421 Microwave clock

The PHARAO cold atom clock is one of the two atomic clocks of the spacemission ACES managed by ESA It is proposed here to include a secondversion of this instrument as its costperformance ratio is very attractive forthe EGE scientific objectives Its concept is very similar to ground basedatomic fountains but with the major difference of zero-g operation Atomsslowly launched in free flight cross two microwave fields tuned to the transitionbetween the two hyperfine levels of the cesium ground state The interrogationmethod based on two separate oscillating fields (Ramsey scheme) allowsthe detection of an atomic line whose width is inversely proportional to thetransit time between the two microwave cavities The resonant microwavefield at 9192631770 GHz (SI definition of the second) is synthesized startingfrom an ultrastable quartz oscillator (USO) in the FCDP phaselocked to themicrowave output of the microwave-optical local oscillator and stabilized tothe clock line using the error signal generated by the cesium resonator In thisway the intrinsic qualities of the cesium hyperfine transition both in terms ofaccuracy and frequency stability are transferred to the USO In a microgravityenvironment the velocity of atoms along the ballistic trajectories is constantand can be changed continuously over almost two orders of magnitude (5ndash500 cms) Differently from atomic fountain clocks presently operated onground very long interaction times (up to few seconds) will be possible whilekeeping reasonable the size of the instrument

The instrument consists of 4 main elements a laser source which providesthe light required to cool and detect the cesium atoms (Fig 5 left) a frequencychain which generates the 92 GHz microwave signal a cesium vacuum tubewhere the interaction between the microwave field and the atoms occurs

Exp Astron

Table 2 Resources and technology readiness level (TRL) of the EGE spacecraft

Unit Volume Mass Power TRL Comment[l] [kg] [W]

Ion clock 100 70 60Atom source 4Atom preparation 6Atom trap 4Atom interrogation 6Clock control 6

Microwave clock 200 91 110 8 TRL by end 2008Local oscillator

Stabilized laser 18 15 23 6Frequency comb 20 16 38 6

Microwave link 18 18 55 8FCDP 7 8 12 8GPS receiver 2 5 7 9CCR 2 2 0 9Payload sum 367 225 305Thermal control 14 17 30 8OBDH 20 24 38 9Star tracker 1 1 8 9Radiation shield 10 40 0 9Support structure 170 204 0 9Power incl solar arr 50 58 10 9Propulsion 21 25 20 9AOCS 36 43 82 9TMTC 8 10 15 9Satellite bus sum 330 422 203 Flexbus conceptPayload plus bus incl 20 margin 8364 7764 6096Propellant 40Launch adapter 100SC wet masspower incl Margins 9164Lattice clock 135 105 152 Alternative clock

Atom source 4Atom preparation 6Atom trap 4Atom interrogation 6Clock control 6

Also included are the resources for an optical lattice clock based on neutral Strontium atoms

(Fig 5 right) and a computer The engineering model of PHARAO is fullyassembled at CNES Toulouse and undergoes functional and performancetests until the end of 2008 The operating temperature range is 10ndash33 andnon-operating temperatures are minus45 +60 The cesium tube consists in aUHV chamber pumped by getters and a 3 ls ion pump a cesium reservoira microwave cavity coils to provide a uniform magnetic field and three layersof magnetic shields Vacuum windows with fiber-optics collimators enable thetransmission of the cooling and detection light onto the cesium atoms

In autonomous mode PHARAO uses the USO as interrogation oscillatorand will provide a clock signal with fractional frequency stability below1 times 10minus13 (τs) minus12 and inaccuracy near 1 times 10minus16 When using the microwave-

Exp Astron

Fig 5 Left The PHARAO laser source engineering model with cover removed Dimensionsare 530 times 350 times 150 mm3 and the mass is 200 kg Ten polarization maintaining optical fibers(yellow) guide the laser beams to the cesium tube All diode lasers are mounted on Peltier coolersfor temperature regulation within 2 mK (courtesy EADS SODERN) Right The PHARAOengineering model cesium tube without the two external magnetic shields The volume is 990 times336 times 444 mm3 and the total mass is 44 kg

optical local oscillator PHARAO will provide a reduced frequency instabilityof 3 times 10minus14 (τs) minus12 This corresponds to 1 times 10minus15 at 1000 s and 3 times 10minus16

at 104 s

422 Ion optical clock

The 171Yb+ ion clock shown in Figs 6 and 7 provides an atomic frequencyreference based on a single 171Yb+ ion confined within an RF electric end-captrap (similar to a Paul trap) within an ultra high vacuum (UHV) chamber andlaser cooled to sim1 mK close to the Doppler cooling limit As a result the ionrsquosfirst order Doppler motion is completely removed and the ion experienceslittle collisional perturbation from its environment Under these conditionsthe narrow linewidth 2S12 (F = 0 mF = 0) minus 2 D32 (F = 2 mF = 0) quadrupole

Fig 6 Left Simplified level scheme of 171Yb+ showing relevant cooling (blue) auxiliary (red)quadrupole clock (green) and octupole clock (dark blue) transitions δνnat is the natural linewidthof the clock transition Middle close-up of Paul-type RF trap (PTB Braunschweig) the ion istrapped in the hole (diameter 14 mm) Right Complete ion trap with vacuum flange height 10 cm(NPL Teddington)

Exp Astron

935 nm D32-state repumper laser

638 nm F72-state repumper laser

739 nm extended cavity diode laser

AOM

PMT

Ion trap UHV chamber

vacuum pumps

Fibre-free space interface

Mu-metal shielding

EC

T

B

Magnetic field

Temperature

Ion fluorescence Processor

Clock laser frequency step and control

4355 nm

Frequengy doubling II

3695 nm cooling wave

Low-drift cavity

Low-drift cavity 171Yb+ 435 nm ion clock unit

876 nm clock laser wave from MOLO

Frequengy doubling

Fig 7 Simplified schematic of the single-ion optical clock unit AOM acousto-optic frequencyshifter PMT photomultiplier tube EC endcap T temperature stabilization system B magneticfield

clock transition at 4355 nm can be interrogated by spectrally narrow and stableclock laser light A diagram of the simplified energy level scheme for 171Yb+showing the relevant cooling auxiliary and clock transitions is given in Fig 6All these wavelengths are provided by diode lasers operated in fundamentalor frequency doubled mode The clock laser light is obtained from the 871 nmwave provided by the MOLO unit (see Section 423) after frequency-doublingin single pass by a KNbO3 nonlinear-optical crystal to 4355 nm The clocktransition spectral profile is observed by stepping the clock laser frequency bymeans of an acousto-optic frequency shifter (AOM in Fig 7) and recordingthe statistics of the quantum jumps in cooling laser induced fluorescence asa function of this frequency Under clock operation the AOM repeatedlysteps back and forward between the half-intensity points on the transitionprofile monitoring the signal imbalance between these points Any detectedimbalance is servo-corrected to zero by feedback to the AOM

In detail the optical clock architecture will comprise (Fig 7)

bull an RF end-cap trap for ionising and confining a single 171Yb+ ion withinan ultra-high vacuum chamber pumped by a small ion pump and non-evaporable getter pump The trap is driven by means of an ac voltage ofa few hundred volts at a drive frequency of sim10 MHz A small oven withseveral mg of 171Yb isotope is heated to provide a low flux of Yb atomsfrom which single ions can be ionised by electron bombardment from ahot wire filament within the trap potential well Low voltage dc is applied

Exp Astron

to additional electrodes to position the ion precisely at trap potentialcentre in order to minimise the ionrsquos micromotion at the drive frequencyThis needs automatic periodic monitoring and correction especially in theperiod after re-loading an ion into the trap External magnetic field coilsin three orthogonal axes allow the nulling of external fields and settingof a fixed field of sim1 μT The trap and coils are surrounded by mu-metalshielding to minimise external field changes during the orbit A level oftemperature control of 1 K of the region surrounding the trap is requiredfor maintaining the blackbody-induced frequency shift uncertainty withtemperature at the 10minus17 level

bull a laser platform to provide frequency doubling of an extended cavity739 nm diode laser for Doppler cooling of the ion on the 2S12(F = 1) minus2 P12 (F = 0) dipole transition at 3695 nm Light from an amplified diodelaser device frequency-doubled in a periodically poled LiTaO3 crystal insingle-pass will be used Typical laser powers used for driving the coolingtransition below saturation are about 2 μW for a beam waist in the trapof about 50 μm By modulation of the injection current of the 739 nmdiode laser at 147 GHz a sideband is generated that excites the 2S12 (F =0) rarr2 P12 (F = 1) ldquorepumperrdquo transition in order to avoid optical pump-ing between the ground hyperfine states

bull The platform also houses the auxiliary lasers at 935 and 638 nm The935 nm diode provides repumping of the ion from the 2 D32(F = 1)metastable level after occasional branching decays to this level during thecooling sequence The 638 nm diode allows fast recovery of the ion fromthe very-long-lived 2F72 metastable state after very occasional collisionaldecay to that state Currently extended cavity lasers are forseen butdistributed-feedback lasers may be available in the near future

bull a high NA lens imaging system and photomultiplier detection system torecord the statistics of 369 nm fluorescence quantum jumps as a functionof clock laser frequency step

bull a fibre system to deliver the various cooling auxiliary and clock light fromsource to trap making use of achromatic doublets where necessary at thefibre-free space interface for launching into the trap

bull a monitoring and control processor which provides the primary coolingand clock laser pulse magnetic field and detection sequencing to observeand lock to the ion clock transition frequency The processor also monitorsfrequency and amplitude data necessary to determine normal laser andion operational conditions and initiate resetting and recovery algorithmswhere necessary and laser unit failure The actual magnetic field presenton the atomic volume is determined and actively stabilized so that aninaccuracy at the level of 1 times 10minus17 can be achieved

bull a redundancy level of at least two units for cooling clock and repumperlasers plus similar for frequency doubling crystals All redundancy unitsfor each wavelength will be fibre multiplexed as standard allowing redun-dant unit activation on determination of prior unit failure

Exp Astron

For an optimized quantum-projection-noise-limited performance of thesingle-ion clock Ramsey interrogation with a cycle time equal to the lifetimeof the metastable D32 level of 50 ms will be performed leading to an Allandeviation of about 27 times 10minus15 (τ s)minus12 This is sim45 times 10minus17 at 1 h satisfyingrequired stabilities for the science objectives Current laboratory results are astability of 1 times 10minus15 at 30 s averaging down to 3 times 10minus16 at 1000 s (PTBGermany) Also comparisons between two independent traps showed agree-ment at the level of several parts in 10minus16 over a series of eight measurementruns

The 171Yb+ ion clock offers a second option namely the 2S12ndash2 F72 octu-pole transition at 476 nm (see Table 1 and Fig 6) [48] It has an extremelylong upper state lifetime low electric quadrupole and second order Zeemansystematic shifts and therefore is a candidate for an even higher performanceclock if a clock laser system with lower instability than the above is providedApart from the clock laser it shares the laser system of the quadrupoleclock Thus with moderate additional resources the Yb+ ion apparatus couldprovide two optical clocks in a single package and specifically perform directfrequency comparisons between the quadrupole and the octupole transitionwith a large sensitivity Aα

The ion clock will operate under microprocessor control providing thecontrol algorithms for the integrated pulse sequencing for cooling repumpingclock interrogation magnetic field switching cooling fluorescence monitoringThe controller will also provide error flags for critical processes and systemreset and correction where necessary This will include ion re-loading andmicromotion reduction algorithms Automatic monitoring of laser power andspectral quality with signal re-optimisation or laser failure determination isrequired Redundant laser units will be pre-aligned and multiplexed into thefibre delivery to the trap so that redundant units can be readily activated in thecase of diode laser failure

423 Microwave-optical local oscillator (MOLO)

This oscillator provides two ultrastable frequency outputs one in the opticalregion the other in the microwave region the two frequencies being coher-ently related The MOLO is composed of three subunits the clock laser thereference cavity and the frequency comb see Fig 8

The clock laser subunit delivers the light for the excitation of the opticalclock transition It is generated by an extended-cavity diode laser at a wave-length of 871 nm Most of the output wave of this laser is sent to the ionclock The laser subunit containing two lasers for redundancy will require3 kg 3 l 15 W The frequency instability of the clock laser must be superiorto the minimum clock instability 27 times 10minus15 (τ s)minus12 for τ up to the timeconstant (sim10 s) of the servo system that locks the laser to the ionrsquos resonancesignal In order to achieve this the laser frequency is stabilized to the reference

Exp Astron

871 nm clock laser

EO

Detector and servo

AOM Femtosecond

frequency comb

To optical clock 10 GHz To FCDP

Microwave-optical local oscillator

ULE cavity

Fig 8 Microwave-optical local oscillator (MOLO) consisting of clock laser reference ULE cavityand femtosecond frequency comb (OFC) Radiofrequency inputs and outputs are not shown EOelectro-optic modulator ULE ultra-low expansion AOM frequency shifter

cavity where the PoundndashDreverndashHall method is used About 10 μW of thelaser output are used for this

The system will exhibit sim04 Hz (sim1 times 10minus15 relative) linewidth and afrequency drift lt01 Hzs The cavity consists of two high-reflectivity mirrorsoptically contacted to a cylinder of diameter 70 mm length 150 mm made ofULE (a glass exhibiting ultra low thermal expansion coefficient) with a cavityfinesse of about 300000 Residual vibrations of the satellite (arms lt 1 times 10minus6 g)should not be a limiting factor to the desired linewidth for safety marginan optimized cavity support can reduce the sensitivity to accelerations Foracoustic and thermal isolation the cavity resides inside an aluminium vacuumchamber equipped with a small (3 ls) ion-getter pump Two stages of polishedaluminium shields are implemented around the cavity which are activelytemperature stabilized so that the cavity is precisely kept at the zero crossingtemperature of the ULE thermal expansion coefficient Heat application andremoval is by thermoelectric elements between the shield and the vacuumchamber Additionally the entire vacuum chamber is temperature-stabilized

During transportation and launch the cavity will be rigidly fixed insidethe vacuum chamber by pressing on it from the sides by piezo-mechanicalactuators allowing accelerations of several g After bringing the satellite onthe orbit the actuators will be released

The optical setup for frequency stabilization will be a classic Pound-Drever-Hall scheme in reflection using an electro-optic modulator (EO) Couplingof the clock laser radiation to the cavity is via single-mode polarization main-taining optical fiber The fiber in- and out-couplers are mounted on miniaturepiezo-motor driven multiaxis translation stages Microprocessor control ofthe translation stages performs a laser-to-fiber incoupling stabilization andlaser-to-cavity mode-matching optimization (positioning inaccuracy lt10 μmalignment inaccuracy lt10 μrad)

The frequency stability of the clock laser will be limited by thermal 1f noisein the mirrors with an fractional Allan deviation slightly below 1 times 10minus15 from1 s to a few 10 s Resources including control electronics are 15 l 12 kg 8 W

Exp Astron

A femtosecond fiber laser based optical frequency comb generator (OFCFig 8) will be used for the clock frequency comparisons and to obtain aspectrally ultrapure microwave local oscillator output for the operation ofPHARAO for the microwave link and for the GNSS receiver The fem-tosecond laser is based on erbium-doped fiber pumped by fiber-coupled laserdiodes It emits 100 fs pulses at a repetition rate of 250 MHz with an averagepower of 100 mW The emission spectrum is centered around 1560 nm Spec-tral broadening to more than one octave is obtained via nonlinear interactionin a microstructure fibre so that the range from about 1050 to 2100 nm iscovered

Second harmonic generation of a portion of the broadened spectrum allowsobtaining the required light for the measurement of the clock laser frequencyat 871 nm The OFC will be referenced to the clock laser frequency (afterintroduction a controllable shift via the frequency shifter AOM) by makingthe carrier envelope offset frequency vanish so that the repetition rate fr isa high-order subharmonic of the optical clock reference frequency f1 = m1 frThe link between the optical and the microwave domain is then simply thedistribution of the repetition rate frequency or more conveniently a high(order 40) harmonic of the comb repetition rate 10 GHz

It has been demonstrated by several groups that frequency comparison andfrequency division with an OFC can be performed to very high accuracy andstability beyond the limitations imposed by the optical clocks themselves Forexample a fiber laser OFC with an instability (Allan deviation) of 6 times 10minus17

in 1 s has been demonstrated [51] OFCs based on fiber technology possessgood long-term reliability and high power efficiency The projection for a spacequalified OFC based on current industrial terrestrial technology is 16 kg mass20 l volume and 38 W power requirement

424 Frequency comparison and distribution package (FCDP)

This package distributes the ultra-stable 10 GHz reference frequency outputof the MOLO to the MWL performs frequency comparisons between MOLOand PHARAO and generates a variety of different reference signals for theindividual instruments including the 250 MHz reference clock to the OFCAll signals are coherent to the MOLO In the FCDP a USO is locked to theMOLO signal by a digital PLL A divider chain generates timing pulses (1 pps1000 pps etc) and keeps a continuous local time scale From the digital phasecomparison data the phase noise density spectrum between the USO and theMOLO is available by telemetry (TM) to assess the overall frequency purityin-flight

As the FCDP distributes the full performance reference from the MOLOdirectly to the MWL there is minimum impact of the FCDP on overall clockcomparison performance The USO-phase lock loop uses a similar designas for ACES-FCDP The digital PLL with linear phase detector allows theUSO to be slightly offset from its reference which is selectable by groundcommand This results in an accurate on-board time scale despite any offset

Exp Astron

of the MOLO reference frequency which can be synchronised to UTC Thefrequency synthesis chain re-uses modules developed for ACES-FCDP andtheir frequency inaccuracy Expected relative uncertainty (Allan deviation)of the frequency comparison between two signals provided to the FCDP is47 times 10minus17 at 1000 s 95 times 10minus18 at 10000 s

During non-availability of the MOLO ie during system start-up andcontingency conditions the USO runs as flywheel (ie it is unlocked withinstability lt15 times 10minus13 from 1 to 100 s) It will still provide useable butless accurate references to the instruments including the reference to OFCCommon view ground-to-ground time- and frequency comparison and supplyof the remaining instruments remain possible at the USO provided accuracy

Mass volume and power consumption are 8 kg 7 l 12 W Controldata rateis 10 kBday 50 kBday (FCDP alone)

425 Microwave link (MWL)

The microwave link is a two-way triple-frequency microwave link (Fig 9) Itis designed to perform accurate frequency comparison as well as time-transfer

Ku-Band Up-linkPower Tx 5 WCarrier 13475 GHzPN-Code 100 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

S-Band Down-linkPower Tx 1 WCarrier 2248 MHzPN-Code 1 MChips1pps 1 time marker sData 5 kBits

Ku-Band Down-linkPower Tx 1 WCarrier 1470333 GHzPN-Code 100 MChips1pps 1 time marker sData 5 kBits

EGE SC

S

Ku

KaKa-Band Up-linkPower Tx 5 WCarrier 345 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

Ka-Band Down-LinkPower Tx 1 WCarrier 32 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 5 kBits

GS Antenna diameterEGE terminal 1 mACES terminal 06 m

Fig 9 EGE microwave two-way link concept and characteristics See text for a description Thelink will be able to perform frequency comparisons with up to ten ground stations simultaneously(only one active comparison is shown) pps pulse per second Tx transmitter SC spacecraft

Exp Astron

between on-board and on-ground clocks Two-way operation allows for com-pensation of the effects due to spacecraft location and its motion relative toground Its two bi-directional links operate in Ka-band (32345 GHz) andKu-band (147135 GHz) The high carrier frequency of the up- and down-links allows for a noticeable reduction of the ionospheric delay A third down-link channel is in S-band (22 GHz) and is used to determine the ionospheretotal electron content MWL uses a combination of carrier phase and codephase data to measure with high precision the absolute ionospheric delayas well as ionospheric delay variations during a ground contact The useof three frequency bands allows to resolve third order ionospheric termsKey modulation parameters are Ka band Tone 250 MHz Ku band PN100 MChips S-band 1 MChips

The EGE-MWL concept is a considerably improved version of the ACES-MWL Microwave carrier generation is directly from the 10 GHz microwaveoutput available from the MOLO and having exceptionally low close-in phasenoise without intermediate frequency dividers or other complex synthesiserstages commonly used in RF transponders The high frequency carriers arekey to achieve the desired frequency comparison accuracy

The Ka-band carrier is generated by tripling the FCDP-generated signalThe Ku-band is subsequently generated by division by two using a verystable regenerative divider Finally the assigned radio frequency is obtainedby adding a low frequency signal using an image reject mixer A mixedanalog-digital receiver design will enhance its performance by use of a flexibledigital signal processing and digital control loops The number of simultaneousground contacts is increased from four (as in the ACES mission) to ten or moreusing the digital concept while the number of physical receiver channels perband is reduced from four to two one for operation to ground and one for in-flight calibration The concept allows adaptation of receiver loop parametersto optimise for the different portions of a highly elliptic orbit while reducingsystematic uncertainties due to high Doppler and Doppler rate Real-timedata exchange to ground terminals is by additional data modulation on all RFsignals The spacecraft carries two different small antennae for perigee and forapogee with gains of 8 dBi 20 dBi beamwidths 100 (full earth coverage) 14respectively power is 1 W for each

The MWL end-to-end performance between the attached clocks includingspace- and ground terminals is predicted to be a relative Allan deviation of1 times 10minus16 at 1000 s (perigee) and 1 times 10minus17 at 10000 s (apogee) Thesevalues are based on conservative implementation losses and other instrumentallosses assumed as 6 dB at apogee and 10 dB at perigee including additionallosses due to atmosphere An attitude knowledge of 1 is required

The MWL link will also support precise orbitographyndashcomplementary tothe GNSS receiver Generation of ranging data is simultaneous to two-waytime transfer and is available under all weather conditions Orbit determina-tion accuracy is expected to be better than 1 cm well matching EGE clockrequirements The value is expected based on demonstrated results fromother missions (METEOR ERS-2 TOPEXPOSEIDON) Laser ranging to

Exp Astron

the CCR array is used a few times during the mission to calibrate the MWLranging

In case of non-availability of the ultrapure signal from MOLO an additional10 GHz output is available from the FCDP to support ground-to-ground clockcomparison in common-view mode even after the nominal mission periodwhen the neither the clocks nor MOLO may be available any longer Basedon the PRARE experimentrsquos 12 years of operation this can be assumed forthe FCDPMWL combination as well The potential extended operational pe-riod makes the proposed configuration particularly attractive to metrologicallaboratories that rely on availability of long-lasting data in support of thegeneration of UTC

The resources of MWL are as follows mass 18 kg volume 18 l power55 W TC 2 kBday TM 120 MBday

426 Auxiliary units

The star camera assembly on the EGE spacecraft will provide real-timeattitude data primarily for the orientation of the spacecraft (with accuracy of1 in real-time) and the proper reduction of the GNSSSLR data

A GNSS receiver is used for the precise orbit determination time andfrequency transfer coarse positioning (lt1 m) for real time use by the Attitudeand Orbit Control as well as the back-up for time tagging of all payloaddata For the time and frequency transfer the receiver is nominally using theonboard frequency reference The receiver assembly includes three specialantennas By the year 2015 more than 100 GNSS satellites are expected tobe available for precise orbit determination permitting orbit inaccuracy of 1ndash2 cm RMS for altitudes up to approx 10000 km At apogee a combinationof GNSS signals and MWL ranging will be used with expected inaccuracyof 10 cm

5 Basic spacecraft specifications

The mission foresees a single spacecraft with a mission duration limited toapprox 3 years for cost reasons Attitude control along three axes withless than 1 error on a timescale of 1000 s is required The control can beperformed with cold gas thrusters magnetotorquers or reaction wheels Theselection will be done on the basis of the vibrations induced by these systems(desired level arms lt 10minus6 g)

Cold gas thrusters or monopropellant thrusters will be required to maintainthe orbit parameters during the mission by small orbit correction manoeuvresand will also guarantee a proper tuning of the orbit parameters after the releasefrom the launcher The ground segment (telemetry control and data transfer)could be via non-ESA stations user-operated to save costs

Further resources are indicated in Table 2

Exp Astron

The satellite bus design drivers are nadir-pointing microwave antennasthermal stability for the payload instruments no strong magnetic sources lowmicrovibrations harsh radiation environment due to crossing of the radiationbelts once per orbit power demand by science payload (with the possibility toreduce power demand by stand-by of some components during specific missionphases eg eclipse) mass limit due to required high altitude with low costlaunch

The satellite bus design can be based on established platforms such asFLEXBUS or PRIMA used on several past missions Their design philosophydiffers one having body-mounted solar arrays the other wings The latter canprovide larger amounts of power The vibration level induced by the panelswhen they are fixed and the duration of solar array drive periods per orbitwill have to be considered in the selection The FLEXBUS concept was usedfor a mission already requiring a low level of residual accelerations and theplatform is optimised for this parameter

For most of the satellite components one can consider using compo-nents and electronic units from navigation or communication satellites whichtypically cruise at altitudes between 15000 and 25000 km altitude LISAPathfinder which will cross the radiation belts during spiralling to theLagrange point will face similar problems concerning radiation as EGE as wellas XMM which is in a Molniya-type orbit with similar radiation characteristics

The instruments will therefore be located in the satellite structure such asto realise ionizing radiation shielding by the spacecraft structure and non-sensitive equipment to the maximum extent possible the need for additionalradiation shielding (including trapped proton radiation) for especially sensitiveparts of the instruments must be determined by ground tests It should bementioned that an orbit with higher perigee and less radiation dose is also anoption as it leads to a tolerable reduction of the science goals

The required plusmn3C temperature stability at the interface to the payloadenclosure is feasible The thermal control will employ the usual means (anenclosure with high thermal inertia heaters and a suitable radiator) Thelocation of the radiator will need a special study given the non-standard orbit

Magnetic field present during the mission can affect the properties of theonboard clock Using SPENVIS the magnetic field has been calculated alongthe spacecraft orbit the field intensity presents maxima around 02 G anda minimum value of the order of 10minus3 G As discussed above this is not alimitation to the instrument performance

6 Discussion and conclusion

The accuracy of the measurements depends crucially on the stability andaccuracy of the clocks While the performance of ground clocks (even thoseeventually deployed in new ground stations dedicated to EGE) will likelyimprove significantly from the present level the space clock design will beldquofrozenrdquo approx 10 years before data acquisition Thus their performance

Exp Astron

level will be significantly below the future ground station clock performanceAs specification we have chosen a performance level that has been demon-strated with laboratory breadboards recently We see no fundamental reasonthat this laboratory performance cannot also be achieved in a space ion opticalclock The experience with the PHARAO cold atom clock has shown thatan engineering model of a laser-based cold atom clock has a performancesimilar to that of a laboratory prototype The perturbation of the ion clock fre-quency due to magnetic fields along the orbit can be modelled and suppressedwith shielding and the residual effects measured and compensated for Themoderate temperature stability (plusmn 3C) to be provided in the experimentalbay can be achieved with thermal shields Accurate temperature controlof particular units is done by active stabilization of the units themselves

One important aspect is radiation which may affect all optical and trapcomponents as well as the lasers and reference cavity The radiation envi-ronment on the EGE orbit is significantly harsher than the near-earth ISSorbit on which ACES will operate The radiation effects must therefore becharacterized comprehensively with appropriate tests and then the radiationshield for the EGE instruments designed

In summary the Einstein Gravity Explorer mission (EGE) is a satellitemission aimed at a precise measurement of the properties of space-time usingatomic clocks By testing one of the most fundamental predictions of EinsteinrsquosTheory of General Relativity the gravitational redshift and thereby searchingfor quantum effects in gravity it explores one of the most important frontiers infundamental physics The EGE mission reaches beyond fundamental physicsproviding an important spin-off in geophysics Furthermore the technologicalaspects of the mission are ground-breaking for future time-keeping and ref-erence frame systems in space and thus are also expected to be of wide useto science and technology in general The payload of the mission has beenconceived as a compromise between cost and performance needing to fit thecost envelope of an ESA medium-class mission1

Acknowledgement We thank S Cesare for helpful information

References

1 Will CM The confrontation between general relativity and experiment Living Rev Relativ9 3 (2006)

2 Ni W-T Empirical foundations of the relativistic gravity Int J Mod Phys D 14 901 (2005)doi101142S0218271805007139

1Finally we point out that a spacecraft with a significantly reduced and thus lower-cost payload (noclocks only MOLO FCDP MWL) on a circular orbit of appropriate height could be used for along-duration earth science mission dedicated specifically to the measurement of the geopotential

Exp Astron

than for the case of a terrestrial experiment The shorter time scale is alsoadvantageous since the drift of the cavity is more predictable In the EGEmission the reference cavity of the clock laser is used The large numberof orbits improves the statistics An improvement of the accuracy of B by afactor 20 compared to the best current terrestrial results is realistic The dataanalysis can also be performed in the framework of the SME theory takinginto account complementary bounds obtained from terrestrial experimentsand astrophysical observations

262 Independence of Zeeman splitting frequency on the directionof the magnetic field

One class of tests of Lorentz Invariance consists in measuring the frequencysplitting between two levels of a quantum system induced by a static magneticfield as a function of orientation of the field direction with respect to thestars (HughesndashDrever-type experiments [31]) This class addresses the so-called matter sector and is complementary to combined-photonmatter-sectortests such as the above The most precise experiments are performed withatomic clocks eg masers or cold atom clocks [38] The EGE instruments allowsuch experiments since magnetic fields are applied continuously in the clockor repeatedly for calibration Compared to terrestrial experiments the highorbital velocity of EGE could lead to an improvement by a factor sim20

27 Application to geophysics

Using GPS coordinates of a point on the Earth can be obtained with aninaccuracy well below 1 cm in a well defined international terrestrial referencesystem The coordinates are purely geometrical and do not contain any gravityinformation A local gravitational potential UB is determined with respect toa reference gravitational potential U A by ldquolevellingrdquo the surface of the Earthie sequentially measuring gravity g and height differences every ca 50 m

UB minus U A = minusBint

A

g (r) middot d n

where n is the difference vector between subsequent locations The disadvan-tage of measuring height differences by the levelling method is a random walkeffect of accumulated errors

In the classical definition a geoid is defined as the particular equipotentialsurface nearest to mean sea level In these terms the geoid serves as areference surface for measuring the height and also to define a referencefor the gravitational potential Such a vertical reference frame historicallywas realized for a country or several countries by determining the mean ofsea level observations at tide gauge stations taken over long period of timeHowever modern satellite altimetry missions such as TOPEXPOSEIDON orENVISAT show that departure of the mean sea level from an equipotential

Exp Astron

surface may reach up to several meters on the global scale see eg [39]Therefore height systems based on different tide gauge stations may differin the realisation of the geoid and may differ with respect to the reference byseveral meters

The best gravity field models obtained from satellite data (eg missionGRACE) have reached a precision of about 1 cm over sim250 km half wave-length [40] Satellite data also reveals changes in time [41] However for typicalEarth topography with height variations of eg 1000 m over 30 km horizontaldistance one may expect variations in the geoid of about 80 cm Such ahigh-spatial-frequency signal in geoid variations cannot be detected by spacegravity missions and requires a combination of satellite and terrestrial gravitymeasurements like gravity anomalies deflections of vertical and GPSlevellingpoints The best combined global gravity field models are provided up to de-gree and order 360 in terms of spherical harmonic representation correspond-ing to a half-wavelength of about 55 km However in combining the satelliteand terrestrial gravity field measurements the problem of terrestrial data givenin different height systems between continents and different countries remainsand has to be tied to satellite measurements

The geodetic scientific community is currently establishing a Global Geo-detic Observing System (GGOS) [42] Its objectives are the early detectionof natural hazards the measurement of temporal changes of land ice andocean surfaces as well as the monitoring of mass transport processes in theEarth system Global change processes are small and therefore difficult toquantify Therefore the required precision relative to the Earthrsquos dimensionis 1 ppb GGOS will be established by the combination of geodetic spacetechniques (GPS Laser-Ranging very large baseline interferometry) andrealized by a very large number of terrestrial and space-borne observatoriesThe purely geometric terrestrial 3D coordinate system is in good shape andfully operational It has to be complemented however with a globally uniformheight system of similar precision The current precision level of regionalheight systems in terms of gravity potential differences is in the order of1 m2s2 (10 cm) with inconsistencies between these various systems up toseveral 10 m2s2 (several meters) The actual requirement in the context ofGGOS is 01 m2s2 (1 cm) with the need of a permanent ie dynamical controlThis requirement of high precision height control comes from the need tounderstand on a global scale processes such as sea level change global andcoastal dynamics of ocean circulation ice melting glacial isostatic adjustmentand land subsidence as well the interaction of these processes Only by meansof monitoring in terms of gravity potential changes at the above level ofprecision the change of ocean level can be understood as a global phenomenonand purely geometric height changes be complemented by information aboutthe associated density or mass changes

The geoid can also be defined in a relativistic way [43] as the surface whereaccurate clocks run with the same rate and where the surface is nearest tomean sea level The relation between the differences in the clock frequenciesand the gravitational potential is given in simplified form by Eq 1 At present

Exp Astron

there is no operational way to compare frequencies of the already availableoptical clocks on the global scale at the same level as their accuracy wouldallow With EGE and by then improved ground optical clocks it will becomepossible to obtain gravitational potential differences on a global scale by com-paring frequencies This would be a significant new dimension to gravity fielddetermination since such observables are given on the global scale and providein situ local gravity information at the same time EGE will allow clock-basedgravitational potential mapping with the same payload instruments used forthe fundamental physics experiments The on-board clocks are actually notrequired for this purpose MOLO MWL and FCDP suffice Thus EGE willallow establishing a global reference frame for the Earth gravitational fieldwith accuracy in the order of a few cm in terms of the geoid heights Thisassumes that by the time EGE is operational (mobile) ground clocks withfractional inaccuracy at the level of 10minus18 are available the measurementsare limited by MWL noise and thus require a long integration times to matchthe expected ground clock performance Such a reference frame will serve asreference for all gravity field modeling and clocks will be used to define theinternational atomic time scale and the Earth gravitational potential at thesame time

In summary with optical clocks geodesy will undergo a second revolutionafter geometry now being measured by clock-based GNSS systems also physi-cal heights and potential will be measured by (mobile) optical clocks using thegravitational redshift effect

28 Application to frequency standards and terrestrial fundamentalphysics studies

As has been described repeating ground clock comparisons and integratingover several months will provide global ground clock frequency comparisonsat the level of 1 times 10minus18 more than two orders of magnitude below the currentGPS and two-way satellite time and frequency transfer methods In the timeand frequency community the availability of such a global high performancemicrowave link will accelerate the process leading to a redefinition of the SIsecond based on clocks operating in the optical domain

The global character of the ground clock comparisons made possible byEGE can also contribute to the search for a time-variation of the fundamentalconstants

The comparisons do not require availability of the on-board clocks The riskof this type of measurements is therefore reduced Moreover the lifetime ofmany subunits required for this measurement (MWL and FCDP see below)can exceed 10 years (from previous experience with similar devices) andthus offers a motivation to increase the mission duration beyond the nominal3 years

Further applications to metrology are listed under the secondary goals inSection 21 above

Exp Astron

3 Mission profile

31 Orbit

The orbit must satisfy the following requirements

Primary 1 Large gravitational potential difference between apogee andperigee

2 sufficient contact time with at least two ground stations atperigee

Secondary Simultaneous visibility of satellite from distant (inter-continental) ground stations for several hours

To satisfy the primary objective the orbit must be highly elliptic The perigeeshould be low A very low perigee means a short contact time which decreasesthe accuracy of the frequency measurement A compromise is a perigeealtitude of sim2500 km For a sufficiently high apogee this yields a U of 60of the theoretical maximum between the earth surface and infinity

The baseline scenario is an orbit of approx 12 h period with the followingproperties

Apogee Altitude 37856 km (minus 10 deg latitude one over Atlantic andPacific resp)

Perigee Altitude 2500 km (+ 10 deg latitude one over Guatemala andMalaysia resp)

Inclination 634 argument of perigee 170RAAN 25 True Anomaly 220Apogee drift rate 70 kmday perigee drift rate sim 14 kmday

True anomaly and right ascension of the ascending node (RAAN) wereoptimized to minimize orbital precession Figure 3 shows the ground track andthe ground coverage at one of the perigees Common-view contact time forstations in Europe is about 6 h while between a station in Europe and in theUSA it is about 35 h Frequency comparison windows between a given groundstation and EGE repeat every 24 h

A final selection of the orbit requires an in-depth study of the costs of thepropellant and maneuvers as well as of the setting up of new ground stations

32 Ground segment

The mission requires two separate ground segments the science groundsegment and the standard satellite control segment which supports telemetryand telecommand (TMTC)

Exp Astron

Fig 3 Baseline EGE orbit One (out of two) apogee and one (out of two) perigee ground pointsare shown The ground coverage at one of the perigees is shown with circles at elevation anglesteps of 10 starting from 10

321 Science ground segment

As described above one part of the mission is to perform frequency com-parisons between ground-to-space and ground-to-ground (common-view andnon-common view) The use of a two-way microwave link (MWL) systemhas the advantage that a precise orbit or the precise relative velocity of thesatellite must not be known [4] The orbit determination requirements fora two-way system are significantly relaxed compared to the requirements tomodel the relativistic effects on the clocks The frequency comparison resultsare available both on ground and on the satellite

For ground-to-satellite clock comparisons five ground stations arefavourable two capable of viewing perigee three for viewing apogee Thisallows for the classic three-clock configuration where the performance of eachclock can be obtained individually The perigee stations will be observing thesatellite for sim25 min every perigee passage for the duration of the missionThe apogee stations will observe for sim3 h every apogee passage Of coursea larger number of stations as well as longer observation time are usefulfor robustness redundancy and orbit determination The stations could beinstalled on a temporary basis for the duration of the mission

The stations must have access to optical clocks of performance at least ashigh as todayrsquos state-of-the-art ground clocks but preferably higher Theywill include an optical frequency comb by means of which the optical clockfrequency is transformed to a radio frequency for comparison with the EGEclocks High-accuracy fiber-optic link technology already existing today allowsdistances of sim100 km between the ground station and the correspondingoptical clock allowing a certain freedom in the choice of the clock locationWith further increase of the fiber-optic link range currently being investigated

Exp Astron

the area over which the ground station can be positioned relative to the opticalclock will increase This option is of importance for the implementation of ageophysics measurement campaign for which a large number of clock locationsis of interest

The above ground stations will be complemented by some 15 additionalground terminals provided by the world-wide laboratories equipped with highperformance primary clocks at the time of the mission These will participate inorbit determination to 1 cm accuracy along the full orbit which can be achievedwith non-optical atomic clocks The EGE MWL is fully compatible to plannedACES ground terminals which might reduce the number of required stations

With a larger number of transportable optical clock ground stations de-ployed and moved across the continents a geophysics campaign could beimplemented to map the gravitational potential

Ground station equipment MWL uses dedicated small transportable termi-nals which can be easily installed at any location of interest Each terminalconsists of a steerable dish antenna of 1 m diameter with pointing accuracy of01 using a transmitted power of 5 W per band The high directivity antenna(beam aperture Ka-band 07 Ku-band 14 S-band 20) avoids multi-pathfrom nearby objects Antenna noise temperatures are Ka-band 450 K Ku-band 340 K Operation is at elevations above 10 and measurements areperformed above 20 For ranging purposes terminal location is determined byGPS surveying Each ground station computes its own ionospheric correctionsbased on a triple-band receiver The terminal has a built-in delay monitorwhich allows calibrated ranging during several months once its bias has beendetermined by laser ranging Its M and C interface is via local area networkto either a userrsquos computer or directly to the science network control centreby remote MampC The stationrsquos operation is fully automatic and unattendedSchedule and orbital data are received via LAN once per day Signal acquisi-tion is fully autonomous Data transmission is after each pass The terminal hasan uplink telecommand capability of 1 kBits and a downlink data capacity ofup to 10 kBitss For the local user there is a visual interface providing quick-look data from real-time comparison to the spacecraft clocks to verify theirgood health The final full performance product is calculated at the sciencedata centre

Local metrological data (temperature pressure humidity) are recorded tocorrect for tropospheric delay of the ranging data The main advantage ofMWL terminals compared to laser based systems is their all-weather capabil-ity although microwaves suffer higher propagation errors in the ionosphereand troposphere

The EGE ground terminal will profit from the ERS-2 mission PRARE(precise range and range rate equipment) [45] experiments whose terminalshad similar design and have been operated under extreme climatic conditionsincl Antarctica and high temperature regions from existing designs fromthe space segment (mission ACES) and no further special developments aredeemed necessary Cost per EGE master station is modest

Exp Astron

Satellite tracking via laser ranging Laser ranging is alternative to MWL rang-ing and also serves for absolute calibration of the latter For this purpose anon-board corner cube reflector (CCR) is provided Ranging will be done withthe International Laser Ranging Service Ideally the satellite will be rangedat every perigee passage and every apogee passage simultaneously with thefrequency comparison procedures Todayrsquos ranging precision is sufficient tosatisfy the science goals For apogee observations in the northern hemispherea large number of laser ranging stations is available The number of stationscapable of ranging at or near perigee in the southern hemisphere is six andthus is sufficient even considering unfavourable weather conditions at someof these The MWL ground stations are able to range even under thoseconditions

322 Control ground segment

Average data transfer rate is estimated at 300 MBday (approx 40 kBs onaverage) Data will be stored on board (2 GB capacity required) and will betransferred to a receiving station on the Earth once a day

XPLC

incl Data StorageIF to TMTC Link

HK amp TMTC

to ground

Opt Clock

AOM

MOLO

PHARAO

FCDP

10 GHz

100 MHz

250 MHz

GNSS Rx

MWLTampF Link

uplink Data

UTC(k)

TampF Space vs Ground

TM from GS

Tim

e S

cale

Orbit GNSS Time UTC

TampF Space vs Ground

MampC Time Scale

MampC

MampC

MampC

MampC

Mamp

C

PHARAO vs Opt Clock

PHARAO vs Opt Clock

10 GHz amp Epoch

10 MHz

amp Epoch

Epoch

10 MHz

amp Epoch

Opt Link

Fig 4 Schematic of the payload showing subunits and their signals The results of the frequencycomparison between the two on-board clocks (optical clock and microwave clock PHARAO) areobtained from a combination of the frequency values applied to the frequency shifters AOM (seeFigs 7 and 8) and those output from the frequency control and distribution package (FCDP)as indicated by the arrows ldquoPHARAO vs Opt Clockrdquo The satellite clock to ground clockcomparison results are output from the MWL XPLC is the on-board computer and is involvedin generation of signals for instrument control The optical link shown dashed is a possible add-onproviding an alternative to the MWL GS ground stations MampC monitoring and control HKhousekeeping TampF time and frequency

Exp Astron

4 Main payload instruments

41 Overview

The satellite payload consists of a pair of cold-atom clocks a microwave-optical local oscillator (MOLO) an on-board frequency distribution unit(FCDP) a satellite-to-ground clock comparison unit (MWL) and the onboardcomputercontrol system (XPLC) see Fig 4 Auxiliary units are the posi-tion determination GPSGalileo receiver (GNSS Rx) attitude monitor (startracker not shown) and the corner cube reflector array (CCR not shown)

The science goals require clocks of very high frequency stability on thetimescales of half the orbital period (6 h) as well as perigee passage (sim1000 s)see Fig 2 Only clocks based on cold atoms are suited for the science goalsTable 1 reports values for a subset of atomic clocks that have been investigatednot including clocks that require cryogenic operation or are too complex foruse in an medium-class mission The choice of a clock pair rather than a singleone fulfils several requirements it addresses more science goals improves thevalidity of some science results and it enables calibration accuracy evaluationand redundancy The validity of the science results is enhanced since theclocks differ not only in the employed atoms but also in terms of havingdifferent sensitivity to perturbations by electric and magnetic fields electronicserrors and misalignments A comparison of the results arising from each clockallows for an understanding of measurement errors Both the clocksrsquo and thefrequency comparison unitrsquos performances will be determined after launchand regularly verified during the course of the mission The only reliable

Table 1 Properties of different clock candidates for EGE

Clock Clock Fractional σ = Fractional Sensitivity Figure of Sensitivityatom frequency inaccuracy instability to α Aα merit |Aα| to meq

at 1000 s σ [1016] QCD

Sr+ 445 THz 3 times 10minus17 (p) 5 times 10minus17 (p) 042 08 (p) ndashYb+ (1) 688 THz 3 times 10minus16 (d) 3 times 10minus16 (d) 088 029 (d) ndash

3 times 10minus17 (p) 1 (p)Yb+ (2) 642 THz lt 1 times 10minus17 (p) 1 times 10minus17 (p) minus53 53 (p) ndashSr 435 THz 1 times 10minus16 (d) 1 times 10minus16 (d) 009 009 (d) ndash

lt 1 times 10minus17 (p) 09 (p)Yb 519 THz 7 times 10minus16 (d) 1 times 10minus16 (d) 031 03 (d) ndash

lt 1 times 10minus17 (p) 1 times 10minus17 (p) 31 (p)Cs 91 GHz 2 times 10minus16 (d) 1 times 10minus15 (d) 275 027 (d) 1Yb+ 126 GHz 1 times 10minus15 (p) 1 times 10minus15 (p) 34 034 (p) 1

A relevant figure of merit is the ratio of |AX | and fractional frequency instability σ on the perigeevisibility timescale The Cs clock is the only high performance clock with adequate sensitivity tothe particle masses me mq The indicated Cs clock performance was achieved using a microwavelocal oscillator based on a cryogenic oscillator On EGE the latter is to be replaced by an opticallocal oscillator-based device (MOLO) with similar short-term performance but without the needto operate at cryogenic temperature The clock instability at 10000 s is typically similar to theindicated inaccuracy(p) projected (d) demonstrated (1) quadropole (2) octupole

Exp Astron

way to achieve this is by having at least two clocks on board The on-boardclock-clock comparison provides performance evaluation independent of theground-satellite comparison unit Once the satellite clock performances areestablished the link performance can be established as well

The clocks are similar in basic structure but differ in the detailed implemen-tation They consist of

a source of atomsa preparation subunit which cools the atoms to near-standstill (millikelvintemperature or below)a trapping subunit that confines the atom(s) to a tiny volumean interrogation subunita clock control electronics package

The clocks share a common central unit the optical-microwave local oscil-lator which serves as a flywheel oscillator with outstanding stability on shorttimescales (01ndash10 s) for both clocks It consists of a laser stabilized to an ultra-low loss optical resonator made of ultra-low-expansion glass A microwavesignal is derived from the laser by phase-locking a frequency comb to the laserwave and using a multiple of the repetition rate

While the subunits of any optical clock have a similar technological basis(optics electrooptics acoustooptics thermal control vacuum systems) theydiffer in their specifications Depending on the choice of the atom species de-livery confinement and interrogation methods clocks of different perfor-mance result A number of laboratory optical clock demonstrators have beendeveloped worldwide This ensures availability of the know-how for the indus-trial implementation of the flight models

The first clock is a microwave clock based on the interrogation of slowlymoving ensembles of cold Cs atoms It is essentially based on the PHARAOclock [5 6] developed to engineering model level by CNES in the frame-work of the ESA ISS mission ACES In EGE this clock is upgraded usingthe microwave-optical local oscillator This improves significantly the clockperformance The specification is a relative instability (Allan deviation) of3 times 10minus14 (τ s)minus12 ie at the integration time τ = 1000 s it drops to 1 times 10minus15and to 3 times 10minus16 at 10000 s While this specification is at least a factor 3 lessstringent than for the optical clocks the significantly lower cost of obtaining aflight model and the sensitivity of its frequency to the particle masses makes ita suitable choice

The optical clock baseline instrument is a single-ion optical clock Amongvarious suitable ion species (Hg+ Al+ Sr+ Yb+) [44ndash47] the Yb+ ion hasbeen selected Its breadboard performance is established at the instability levelof 3 times 10minus16 over 1000 s it exhibits a large sensitivity coefficient Aα hassmall sensitivity to magnetic fields and exhibits significant potential for furtherimprovements (use of the octupole transition with sim6 times larger Aα andlower overall instability [48] see below) A trapped Yb+ ion is also a veryrobust system it has been demonstrated that a single Yb+ ion can remaintrapped uninterruptedly for many months Finally Yb+ is the only ion with

Exp Astron

which optical clocks demonstrators have been operated in not a single but twomajor metrology laboratories

Another attractive optical clock is based on ensembles of cold neutral Sror Yb atoms trapped in an optical lattice produced by laser beams This clocktype has a lower instability than a single-ion clock thanks to the large numberof atoms (sim105) used Its development is in progress in several laboratoriesin the world (in part supported by space agencies) and has to date reached6 times 10minus17 instability at 5000 s and 1 times 10minus16 uncertainty [49 50] with potentialfor even higher performance However it is more complex and costly thanthe Yb+ ion clock since the cold atom preparation and trapping subunitsinclude more components require higher laser powers and the respectiveoperating procedures are more complex It is therefore not considered for thepresent mission designed to be developed in the very near future with limitedresources but is an interesting candidate for missions without such constraintsIts estimated physical parameters are indicated in Table 2 for comparison

42 Main instruments

421 Microwave clock

The PHARAO cold atom clock is one of the two atomic clocks of the spacemission ACES managed by ESA It is proposed here to include a secondversion of this instrument as its costperformance ratio is very attractive forthe EGE scientific objectives Its concept is very similar to ground basedatomic fountains but with the major difference of zero-g operation Atomsslowly launched in free flight cross two microwave fields tuned to the transitionbetween the two hyperfine levels of the cesium ground state The interrogationmethod based on two separate oscillating fields (Ramsey scheme) allowsthe detection of an atomic line whose width is inversely proportional to thetransit time between the two microwave cavities The resonant microwavefield at 9192631770 GHz (SI definition of the second) is synthesized startingfrom an ultrastable quartz oscillator (USO) in the FCDP phaselocked to themicrowave output of the microwave-optical local oscillator and stabilized tothe clock line using the error signal generated by the cesium resonator In thisway the intrinsic qualities of the cesium hyperfine transition both in terms ofaccuracy and frequency stability are transferred to the USO In a microgravityenvironment the velocity of atoms along the ballistic trajectories is constantand can be changed continuously over almost two orders of magnitude (5ndash500 cms) Differently from atomic fountain clocks presently operated onground very long interaction times (up to few seconds) will be possible whilekeeping reasonable the size of the instrument

The instrument consists of 4 main elements a laser source which providesthe light required to cool and detect the cesium atoms (Fig 5 left) a frequencychain which generates the 92 GHz microwave signal a cesium vacuum tubewhere the interaction between the microwave field and the atoms occurs

Exp Astron

Table 2 Resources and technology readiness level (TRL) of the EGE spacecraft

Unit Volume Mass Power TRL Comment[l] [kg] [W]

Ion clock 100 70 60Atom source 4Atom preparation 6Atom trap 4Atom interrogation 6Clock control 6

Microwave clock 200 91 110 8 TRL by end 2008Local oscillator

Stabilized laser 18 15 23 6Frequency comb 20 16 38 6

Microwave link 18 18 55 8FCDP 7 8 12 8GPS receiver 2 5 7 9CCR 2 2 0 9Payload sum 367 225 305Thermal control 14 17 30 8OBDH 20 24 38 9Star tracker 1 1 8 9Radiation shield 10 40 0 9Support structure 170 204 0 9Power incl solar arr 50 58 10 9Propulsion 21 25 20 9AOCS 36 43 82 9TMTC 8 10 15 9Satellite bus sum 330 422 203 Flexbus conceptPayload plus bus incl 20 margin 8364 7764 6096Propellant 40Launch adapter 100SC wet masspower incl Margins 9164Lattice clock 135 105 152 Alternative clock

Atom source 4Atom preparation 6Atom trap 4Atom interrogation 6Clock control 6

Also included are the resources for an optical lattice clock based on neutral Strontium atoms

(Fig 5 right) and a computer The engineering model of PHARAO is fullyassembled at CNES Toulouse and undergoes functional and performancetests until the end of 2008 The operating temperature range is 10ndash33 andnon-operating temperatures are minus45 +60 The cesium tube consists in aUHV chamber pumped by getters and a 3 ls ion pump a cesium reservoira microwave cavity coils to provide a uniform magnetic field and three layersof magnetic shields Vacuum windows with fiber-optics collimators enable thetransmission of the cooling and detection light onto the cesium atoms

In autonomous mode PHARAO uses the USO as interrogation oscillatorand will provide a clock signal with fractional frequency stability below1 times 10minus13 (τs) minus12 and inaccuracy near 1 times 10minus16 When using the microwave-

Exp Astron

Fig 5 Left The PHARAO laser source engineering model with cover removed Dimensionsare 530 times 350 times 150 mm3 and the mass is 200 kg Ten polarization maintaining optical fibers(yellow) guide the laser beams to the cesium tube All diode lasers are mounted on Peltier coolersfor temperature regulation within 2 mK (courtesy EADS SODERN) Right The PHARAOengineering model cesium tube without the two external magnetic shields The volume is 990 times336 times 444 mm3 and the total mass is 44 kg

optical local oscillator PHARAO will provide a reduced frequency instabilityof 3 times 10minus14 (τs) minus12 This corresponds to 1 times 10minus15 at 1000 s and 3 times 10minus16

at 104 s

422 Ion optical clock

The 171Yb+ ion clock shown in Figs 6 and 7 provides an atomic frequencyreference based on a single 171Yb+ ion confined within an RF electric end-captrap (similar to a Paul trap) within an ultra high vacuum (UHV) chamber andlaser cooled to sim1 mK close to the Doppler cooling limit As a result the ionrsquosfirst order Doppler motion is completely removed and the ion experienceslittle collisional perturbation from its environment Under these conditionsthe narrow linewidth 2S12 (F = 0 mF = 0) minus 2 D32 (F = 2 mF = 0) quadrupole

Fig 6 Left Simplified level scheme of 171Yb+ showing relevant cooling (blue) auxiliary (red)quadrupole clock (green) and octupole clock (dark blue) transitions δνnat is the natural linewidthof the clock transition Middle close-up of Paul-type RF trap (PTB Braunschweig) the ion istrapped in the hole (diameter 14 mm) Right Complete ion trap with vacuum flange height 10 cm(NPL Teddington)

Exp Astron

935 nm D32-state repumper laser

638 nm F72-state repumper laser

739 nm extended cavity diode laser

AOM

PMT

Ion trap UHV chamber

vacuum pumps

Fibre-free space interface

Mu-metal shielding

EC

T

B

Magnetic field

Temperature

Ion fluorescence Processor

Clock laser frequency step and control

4355 nm

Frequengy doubling II

3695 nm cooling wave

Low-drift cavity

Low-drift cavity 171Yb+ 435 nm ion clock unit

876 nm clock laser wave from MOLO

Frequengy doubling

Fig 7 Simplified schematic of the single-ion optical clock unit AOM acousto-optic frequencyshifter PMT photomultiplier tube EC endcap T temperature stabilization system B magneticfield

clock transition at 4355 nm can be interrogated by spectrally narrow and stableclock laser light A diagram of the simplified energy level scheme for 171Yb+showing the relevant cooling auxiliary and clock transitions is given in Fig 6All these wavelengths are provided by diode lasers operated in fundamentalor frequency doubled mode The clock laser light is obtained from the 871 nmwave provided by the MOLO unit (see Section 423) after frequency-doublingin single pass by a KNbO3 nonlinear-optical crystal to 4355 nm The clocktransition spectral profile is observed by stepping the clock laser frequency bymeans of an acousto-optic frequency shifter (AOM in Fig 7) and recordingthe statistics of the quantum jumps in cooling laser induced fluorescence asa function of this frequency Under clock operation the AOM repeatedlysteps back and forward between the half-intensity points on the transitionprofile monitoring the signal imbalance between these points Any detectedimbalance is servo-corrected to zero by feedback to the AOM

In detail the optical clock architecture will comprise (Fig 7)

bull an RF end-cap trap for ionising and confining a single 171Yb+ ion withinan ultra-high vacuum chamber pumped by a small ion pump and non-evaporable getter pump The trap is driven by means of an ac voltage ofa few hundred volts at a drive frequency of sim10 MHz A small oven withseveral mg of 171Yb isotope is heated to provide a low flux of Yb atomsfrom which single ions can be ionised by electron bombardment from ahot wire filament within the trap potential well Low voltage dc is applied

Exp Astron

to additional electrodes to position the ion precisely at trap potentialcentre in order to minimise the ionrsquos micromotion at the drive frequencyThis needs automatic periodic monitoring and correction especially in theperiod after re-loading an ion into the trap External magnetic field coilsin three orthogonal axes allow the nulling of external fields and settingof a fixed field of sim1 μT The trap and coils are surrounded by mu-metalshielding to minimise external field changes during the orbit A level oftemperature control of 1 K of the region surrounding the trap is requiredfor maintaining the blackbody-induced frequency shift uncertainty withtemperature at the 10minus17 level

bull a laser platform to provide frequency doubling of an extended cavity739 nm diode laser for Doppler cooling of the ion on the 2S12(F = 1) minus2 P12 (F = 0) dipole transition at 3695 nm Light from an amplified diodelaser device frequency-doubled in a periodically poled LiTaO3 crystal insingle-pass will be used Typical laser powers used for driving the coolingtransition below saturation are about 2 μW for a beam waist in the trapof about 50 μm By modulation of the injection current of the 739 nmdiode laser at 147 GHz a sideband is generated that excites the 2S12 (F =0) rarr2 P12 (F = 1) ldquorepumperrdquo transition in order to avoid optical pump-ing between the ground hyperfine states

bull The platform also houses the auxiliary lasers at 935 and 638 nm The935 nm diode provides repumping of the ion from the 2 D32(F = 1)metastable level after occasional branching decays to this level during thecooling sequence The 638 nm diode allows fast recovery of the ion fromthe very-long-lived 2F72 metastable state after very occasional collisionaldecay to that state Currently extended cavity lasers are forseen butdistributed-feedback lasers may be available in the near future

bull a high NA lens imaging system and photomultiplier detection system torecord the statistics of 369 nm fluorescence quantum jumps as a functionof clock laser frequency step

bull a fibre system to deliver the various cooling auxiliary and clock light fromsource to trap making use of achromatic doublets where necessary at thefibre-free space interface for launching into the trap

bull a monitoring and control processor which provides the primary coolingand clock laser pulse magnetic field and detection sequencing to observeand lock to the ion clock transition frequency The processor also monitorsfrequency and amplitude data necessary to determine normal laser andion operational conditions and initiate resetting and recovery algorithmswhere necessary and laser unit failure The actual magnetic field presenton the atomic volume is determined and actively stabilized so that aninaccuracy at the level of 1 times 10minus17 can be achieved

bull a redundancy level of at least two units for cooling clock and repumperlasers plus similar for frequency doubling crystals All redundancy unitsfor each wavelength will be fibre multiplexed as standard allowing redun-dant unit activation on determination of prior unit failure

Exp Astron

For an optimized quantum-projection-noise-limited performance of thesingle-ion clock Ramsey interrogation with a cycle time equal to the lifetimeof the metastable D32 level of 50 ms will be performed leading to an Allandeviation of about 27 times 10minus15 (τ s)minus12 This is sim45 times 10minus17 at 1 h satisfyingrequired stabilities for the science objectives Current laboratory results are astability of 1 times 10minus15 at 30 s averaging down to 3 times 10minus16 at 1000 s (PTBGermany) Also comparisons between two independent traps showed agree-ment at the level of several parts in 10minus16 over a series of eight measurementruns

The 171Yb+ ion clock offers a second option namely the 2S12ndash2 F72 octu-pole transition at 476 nm (see Table 1 and Fig 6) [48] It has an extremelylong upper state lifetime low electric quadrupole and second order Zeemansystematic shifts and therefore is a candidate for an even higher performanceclock if a clock laser system with lower instability than the above is providedApart from the clock laser it shares the laser system of the quadrupoleclock Thus with moderate additional resources the Yb+ ion apparatus couldprovide two optical clocks in a single package and specifically perform directfrequency comparisons between the quadrupole and the octupole transitionwith a large sensitivity Aα

The ion clock will operate under microprocessor control providing thecontrol algorithms for the integrated pulse sequencing for cooling repumpingclock interrogation magnetic field switching cooling fluorescence monitoringThe controller will also provide error flags for critical processes and systemreset and correction where necessary This will include ion re-loading andmicromotion reduction algorithms Automatic monitoring of laser power andspectral quality with signal re-optimisation or laser failure determination isrequired Redundant laser units will be pre-aligned and multiplexed into thefibre delivery to the trap so that redundant units can be readily activated in thecase of diode laser failure

423 Microwave-optical local oscillator (MOLO)

This oscillator provides two ultrastable frequency outputs one in the opticalregion the other in the microwave region the two frequencies being coher-ently related The MOLO is composed of three subunits the clock laser thereference cavity and the frequency comb see Fig 8

The clock laser subunit delivers the light for the excitation of the opticalclock transition It is generated by an extended-cavity diode laser at a wave-length of 871 nm Most of the output wave of this laser is sent to the ionclock The laser subunit containing two lasers for redundancy will require3 kg 3 l 15 W The frequency instability of the clock laser must be superiorto the minimum clock instability 27 times 10minus15 (τ s)minus12 for τ up to the timeconstant (sim10 s) of the servo system that locks the laser to the ionrsquos resonancesignal In order to achieve this the laser frequency is stabilized to the reference

Exp Astron

871 nm clock laser

EO

Detector and servo

AOM Femtosecond

frequency comb

To optical clock 10 GHz To FCDP

Microwave-optical local oscillator

ULE cavity

Fig 8 Microwave-optical local oscillator (MOLO) consisting of clock laser reference ULE cavityand femtosecond frequency comb (OFC) Radiofrequency inputs and outputs are not shown EOelectro-optic modulator ULE ultra-low expansion AOM frequency shifter

cavity where the PoundndashDreverndashHall method is used About 10 μW of thelaser output are used for this

The system will exhibit sim04 Hz (sim1 times 10minus15 relative) linewidth and afrequency drift lt01 Hzs The cavity consists of two high-reflectivity mirrorsoptically contacted to a cylinder of diameter 70 mm length 150 mm made ofULE (a glass exhibiting ultra low thermal expansion coefficient) with a cavityfinesse of about 300000 Residual vibrations of the satellite (arms lt 1 times 10minus6 g)should not be a limiting factor to the desired linewidth for safety marginan optimized cavity support can reduce the sensitivity to accelerations Foracoustic and thermal isolation the cavity resides inside an aluminium vacuumchamber equipped with a small (3 ls) ion-getter pump Two stages of polishedaluminium shields are implemented around the cavity which are activelytemperature stabilized so that the cavity is precisely kept at the zero crossingtemperature of the ULE thermal expansion coefficient Heat application andremoval is by thermoelectric elements between the shield and the vacuumchamber Additionally the entire vacuum chamber is temperature-stabilized

During transportation and launch the cavity will be rigidly fixed insidethe vacuum chamber by pressing on it from the sides by piezo-mechanicalactuators allowing accelerations of several g After bringing the satellite onthe orbit the actuators will be released

The optical setup for frequency stabilization will be a classic Pound-Drever-Hall scheme in reflection using an electro-optic modulator (EO) Couplingof the clock laser radiation to the cavity is via single-mode polarization main-taining optical fiber The fiber in- and out-couplers are mounted on miniaturepiezo-motor driven multiaxis translation stages Microprocessor control ofthe translation stages performs a laser-to-fiber incoupling stabilization andlaser-to-cavity mode-matching optimization (positioning inaccuracy lt10 μmalignment inaccuracy lt10 μrad)

The frequency stability of the clock laser will be limited by thermal 1f noisein the mirrors with an fractional Allan deviation slightly below 1 times 10minus15 from1 s to a few 10 s Resources including control electronics are 15 l 12 kg 8 W

Exp Astron

A femtosecond fiber laser based optical frequency comb generator (OFCFig 8) will be used for the clock frequency comparisons and to obtain aspectrally ultrapure microwave local oscillator output for the operation ofPHARAO for the microwave link and for the GNSS receiver The fem-tosecond laser is based on erbium-doped fiber pumped by fiber-coupled laserdiodes It emits 100 fs pulses at a repetition rate of 250 MHz with an averagepower of 100 mW The emission spectrum is centered around 1560 nm Spec-tral broadening to more than one octave is obtained via nonlinear interactionin a microstructure fibre so that the range from about 1050 to 2100 nm iscovered

Second harmonic generation of a portion of the broadened spectrum allowsobtaining the required light for the measurement of the clock laser frequencyat 871 nm The OFC will be referenced to the clock laser frequency (afterintroduction a controllable shift via the frequency shifter AOM) by makingthe carrier envelope offset frequency vanish so that the repetition rate fr isa high-order subharmonic of the optical clock reference frequency f1 = m1 frThe link between the optical and the microwave domain is then simply thedistribution of the repetition rate frequency or more conveniently a high(order 40) harmonic of the comb repetition rate 10 GHz

It has been demonstrated by several groups that frequency comparison andfrequency division with an OFC can be performed to very high accuracy andstability beyond the limitations imposed by the optical clocks themselves Forexample a fiber laser OFC with an instability (Allan deviation) of 6 times 10minus17

in 1 s has been demonstrated [51] OFCs based on fiber technology possessgood long-term reliability and high power efficiency The projection for a spacequalified OFC based on current industrial terrestrial technology is 16 kg mass20 l volume and 38 W power requirement

424 Frequency comparison and distribution package (FCDP)

This package distributes the ultra-stable 10 GHz reference frequency outputof the MOLO to the MWL performs frequency comparisons between MOLOand PHARAO and generates a variety of different reference signals for theindividual instruments including the 250 MHz reference clock to the OFCAll signals are coherent to the MOLO In the FCDP a USO is locked to theMOLO signal by a digital PLL A divider chain generates timing pulses (1 pps1000 pps etc) and keeps a continuous local time scale From the digital phasecomparison data the phase noise density spectrum between the USO and theMOLO is available by telemetry (TM) to assess the overall frequency purityin-flight

As the FCDP distributes the full performance reference from the MOLOdirectly to the MWL there is minimum impact of the FCDP on overall clockcomparison performance The USO-phase lock loop uses a similar designas for ACES-FCDP The digital PLL with linear phase detector allows theUSO to be slightly offset from its reference which is selectable by groundcommand This results in an accurate on-board time scale despite any offset

Exp Astron

of the MOLO reference frequency which can be synchronised to UTC Thefrequency synthesis chain re-uses modules developed for ACES-FCDP andtheir frequency inaccuracy Expected relative uncertainty (Allan deviation)of the frequency comparison between two signals provided to the FCDP is47 times 10minus17 at 1000 s 95 times 10minus18 at 10000 s

During non-availability of the MOLO ie during system start-up andcontingency conditions the USO runs as flywheel (ie it is unlocked withinstability lt15 times 10minus13 from 1 to 100 s) It will still provide useable butless accurate references to the instruments including the reference to OFCCommon view ground-to-ground time- and frequency comparison and supplyof the remaining instruments remain possible at the USO provided accuracy

Mass volume and power consumption are 8 kg 7 l 12 W Controldata rateis 10 kBday 50 kBday (FCDP alone)

425 Microwave link (MWL)

The microwave link is a two-way triple-frequency microwave link (Fig 9) Itis designed to perform accurate frequency comparison as well as time-transfer

Ku-Band Up-linkPower Tx 5 WCarrier 13475 GHzPN-Code 100 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

S-Band Down-linkPower Tx 1 WCarrier 2248 MHzPN-Code 1 MChips1pps 1 time marker sData 5 kBits

Ku-Band Down-linkPower Tx 1 WCarrier 1470333 GHzPN-Code 100 MChips1pps 1 time marker sData 5 kBits

EGE SC

S

Ku

KaKa-Band Up-linkPower Tx 5 WCarrier 345 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

Ka-Band Down-LinkPower Tx 1 WCarrier 32 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 5 kBits

GS Antenna diameterEGE terminal 1 mACES terminal 06 m

Fig 9 EGE microwave two-way link concept and characteristics See text for a description Thelink will be able to perform frequency comparisons with up to ten ground stations simultaneously(only one active comparison is shown) pps pulse per second Tx transmitter SC spacecraft

Exp Astron

between on-board and on-ground clocks Two-way operation allows for com-pensation of the effects due to spacecraft location and its motion relative toground Its two bi-directional links operate in Ka-band (32345 GHz) andKu-band (147135 GHz) The high carrier frequency of the up- and down-links allows for a noticeable reduction of the ionospheric delay A third down-link channel is in S-band (22 GHz) and is used to determine the ionospheretotal electron content MWL uses a combination of carrier phase and codephase data to measure with high precision the absolute ionospheric delayas well as ionospheric delay variations during a ground contact The useof three frequency bands allows to resolve third order ionospheric termsKey modulation parameters are Ka band Tone 250 MHz Ku band PN100 MChips S-band 1 MChips

The EGE-MWL concept is a considerably improved version of the ACES-MWL Microwave carrier generation is directly from the 10 GHz microwaveoutput available from the MOLO and having exceptionally low close-in phasenoise without intermediate frequency dividers or other complex synthesiserstages commonly used in RF transponders The high frequency carriers arekey to achieve the desired frequency comparison accuracy

The Ka-band carrier is generated by tripling the FCDP-generated signalThe Ku-band is subsequently generated by division by two using a verystable regenerative divider Finally the assigned radio frequency is obtainedby adding a low frequency signal using an image reject mixer A mixedanalog-digital receiver design will enhance its performance by use of a flexibledigital signal processing and digital control loops The number of simultaneousground contacts is increased from four (as in the ACES mission) to ten or moreusing the digital concept while the number of physical receiver channels perband is reduced from four to two one for operation to ground and one for in-flight calibration The concept allows adaptation of receiver loop parametersto optimise for the different portions of a highly elliptic orbit while reducingsystematic uncertainties due to high Doppler and Doppler rate Real-timedata exchange to ground terminals is by additional data modulation on all RFsignals The spacecraft carries two different small antennae for perigee and forapogee with gains of 8 dBi 20 dBi beamwidths 100 (full earth coverage) 14respectively power is 1 W for each

The MWL end-to-end performance between the attached clocks includingspace- and ground terminals is predicted to be a relative Allan deviation of1 times 10minus16 at 1000 s (perigee) and 1 times 10minus17 at 10000 s (apogee) Thesevalues are based on conservative implementation losses and other instrumentallosses assumed as 6 dB at apogee and 10 dB at perigee including additionallosses due to atmosphere An attitude knowledge of 1 is required

The MWL link will also support precise orbitographyndashcomplementary tothe GNSS receiver Generation of ranging data is simultaneous to two-waytime transfer and is available under all weather conditions Orbit determina-tion accuracy is expected to be better than 1 cm well matching EGE clockrequirements The value is expected based on demonstrated results fromother missions (METEOR ERS-2 TOPEXPOSEIDON) Laser ranging to

Exp Astron

the CCR array is used a few times during the mission to calibrate the MWLranging

In case of non-availability of the ultrapure signal from MOLO an additional10 GHz output is available from the FCDP to support ground-to-ground clockcomparison in common-view mode even after the nominal mission periodwhen the neither the clocks nor MOLO may be available any longer Basedon the PRARE experimentrsquos 12 years of operation this can be assumed forthe FCDPMWL combination as well The potential extended operational pe-riod makes the proposed configuration particularly attractive to metrologicallaboratories that rely on availability of long-lasting data in support of thegeneration of UTC

The resources of MWL are as follows mass 18 kg volume 18 l power55 W TC 2 kBday TM 120 MBday

426 Auxiliary units

The star camera assembly on the EGE spacecraft will provide real-timeattitude data primarily for the orientation of the spacecraft (with accuracy of1 in real-time) and the proper reduction of the GNSSSLR data

A GNSS receiver is used for the precise orbit determination time andfrequency transfer coarse positioning (lt1 m) for real time use by the Attitudeand Orbit Control as well as the back-up for time tagging of all payloaddata For the time and frequency transfer the receiver is nominally using theonboard frequency reference The receiver assembly includes three specialantennas By the year 2015 more than 100 GNSS satellites are expected tobe available for precise orbit determination permitting orbit inaccuracy of 1ndash2 cm RMS for altitudes up to approx 10000 km At apogee a combinationof GNSS signals and MWL ranging will be used with expected inaccuracyof 10 cm

5 Basic spacecraft specifications

The mission foresees a single spacecraft with a mission duration limited toapprox 3 years for cost reasons Attitude control along three axes withless than 1 error on a timescale of 1000 s is required The control can beperformed with cold gas thrusters magnetotorquers or reaction wheels Theselection will be done on the basis of the vibrations induced by these systems(desired level arms lt 10minus6 g)

Cold gas thrusters or monopropellant thrusters will be required to maintainthe orbit parameters during the mission by small orbit correction manoeuvresand will also guarantee a proper tuning of the orbit parameters after the releasefrom the launcher The ground segment (telemetry control and data transfer)could be via non-ESA stations user-operated to save costs

Further resources are indicated in Table 2

Exp Astron

The satellite bus design drivers are nadir-pointing microwave antennasthermal stability for the payload instruments no strong magnetic sources lowmicrovibrations harsh radiation environment due to crossing of the radiationbelts once per orbit power demand by science payload (with the possibility toreduce power demand by stand-by of some components during specific missionphases eg eclipse) mass limit due to required high altitude with low costlaunch

The satellite bus design can be based on established platforms such asFLEXBUS or PRIMA used on several past missions Their design philosophydiffers one having body-mounted solar arrays the other wings The latter canprovide larger amounts of power The vibration level induced by the panelswhen they are fixed and the duration of solar array drive periods per orbitwill have to be considered in the selection The FLEXBUS concept was usedfor a mission already requiring a low level of residual accelerations and theplatform is optimised for this parameter

For most of the satellite components one can consider using compo-nents and electronic units from navigation or communication satellites whichtypically cruise at altitudes between 15000 and 25000 km altitude LISAPathfinder which will cross the radiation belts during spiralling to theLagrange point will face similar problems concerning radiation as EGE as wellas XMM which is in a Molniya-type orbit with similar radiation characteristics

The instruments will therefore be located in the satellite structure such asto realise ionizing radiation shielding by the spacecraft structure and non-sensitive equipment to the maximum extent possible the need for additionalradiation shielding (including trapped proton radiation) for especially sensitiveparts of the instruments must be determined by ground tests It should bementioned that an orbit with higher perigee and less radiation dose is also anoption as it leads to a tolerable reduction of the science goals

The required plusmn3C temperature stability at the interface to the payloadenclosure is feasible The thermal control will employ the usual means (anenclosure with high thermal inertia heaters and a suitable radiator) Thelocation of the radiator will need a special study given the non-standard orbit

Magnetic field present during the mission can affect the properties of theonboard clock Using SPENVIS the magnetic field has been calculated alongthe spacecraft orbit the field intensity presents maxima around 02 G anda minimum value of the order of 10minus3 G As discussed above this is not alimitation to the instrument performance

6 Discussion and conclusion

The accuracy of the measurements depends crucially on the stability andaccuracy of the clocks While the performance of ground clocks (even thoseeventually deployed in new ground stations dedicated to EGE) will likelyimprove significantly from the present level the space clock design will beldquofrozenrdquo approx 10 years before data acquisition Thus their performance

Exp Astron

level will be significantly below the future ground station clock performanceAs specification we have chosen a performance level that has been demon-strated with laboratory breadboards recently We see no fundamental reasonthat this laboratory performance cannot also be achieved in a space ion opticalclock The experience with the PHARAO cold atom clock has shown thatan engineering model of a laser-based cold atom clock has a performancesimilar to that of a laboratory prototype The perturbation of the ion clock fre-quency due to magnetic fields along the orbit can be modelled and suppressedwith shielding and the residual effects measured and compensated for Themoderate temperature stability (plusmn 3C) to be provided in the experimentalbay can be achieved with thermal shields Accurate temperature controlof particular units is done by active stabilization of the units themselves

One important aspect is radiation which may affect all optical and trapcomponents as well as the lasers and reference cavity The radiation envi-ronment on the EGE orbit is significantly harsher than the near-earth ISSorbit on which ACES will operate The radiation effects must therefore becharacterized comprehensively with appropriate tests and then the radiationshield for the EGE instruments designed

In summary the Einstein Gravity Explorer mission (EGE) is a satellitemission aimed at a precise measurement of the properties of space-time usingatomic clocks By testing one of the most fundamental predictions of EinsteinrsquosTheory of General Relativity the gravitational redshift and thereby searchingfor quantum effects in gravity it explores one of the most important frontiers infundamental physics The EGE mission reaches beyond fundamental physicsproviding an important spin-off in geophysics Furthermore the technologicalaspects of the mission are ground-breaking for future time-keeping and ref-erence frame systems in space and thus are also expected to be of wide useto science and technology in general The payload of the mission has beenconceived as a compromise between cost and performance needing to fit thecost envelope of an ESA medium-class mission1

Acknowledgement We thank S Cesare for helpful information

References

1 Will CM The confrontation between general relativity and experiment Living Rev Relativ9 3 (2006)

2 Ni W-T Empirical foundations of the relativistic gravity Int J Mod Phys D 14 901 (2005)doi101142S0218271805007139

1Finally we point out that a spacecraft with a significantly reduced and thus lower-cost payload (noclocks only MOLO FCDP MWL) on a circular orbit of appropriate height could be used for along-duration earth science mission dedicated specifically to the measurement of the geopotential

Exp Astron

surface may reach up to several meters on the global scale see eg [39]Therefore height systems based on different tide gauge stations may differin the realisation of the geoid and may differ with respect to the reference byseveral meters

The best gravity field models obtained from satellite data (eg missionGRACE) have reached a precision of about 1 cm over sim250 km half wave-length [40] Satellite data also reveals changes in time [41] However for typicalEarth topography with height variations of eg 1000 m over 30 km horizontaldistance one may expect variations in the geoid of about 80 cm Such ahigh-spatial-frequency signal in geoid variations cannot be detected by spacegravity missions and requires a combination of satellite and terrestrial gravitymeasurements like gravity anomalies deflections of vertical and GPSlevellingpoints The best combined global gravity field models are provided up to de-gree and order 360 in terms of spherical harmonic representation correspond-ing to a half-wavelength of about 55 km However in combining the satelliteand terrestrial gravity field measurements the problem of terrestrial data givenin different height systems between continents and different countries remainsand has to be tied to satellite measurements

The geodetic scientific community is currently establishing a Global Geo-detic Observing System (GGOS) [42] Its objectives are the early detectionof natural hazards the measurement of temporal changes of land ice andocean surfaces as well as the monitoring of mass transport processes in theEarth system Global change processes are small and therefore difficult toquantify Therefore the required precision relative to the Earthrsquos dimensionis 1 ppb GGOS will be established by the combination of geodetic spacetechniques (GPS Laser-Ranging very large baseline interferometry) andrealized by a very large number of terrestrial and space-borne observatoriesThe purely geometric terrestrial 3D coordinate system is in good shape andfully operational It has to be complemented however with a globally uniformheight system of similar precision The current precision level of regionalheight systems in terms of gravity potential differences is in the order of1 m2s2 (10 cm) with inconsistencies between these various systems up toseveral 10 m2s2 (several meters) The actual requirement in the context ofGGOS is 01 m2s2 (1 cm) with the need of a permanent ie dynamical controlThis requirement of high precision height control comes from the need tounderstand on a global scale processes such as sea level change global andcoastal dynamics of ocean circulation ice melting glacial isostatic adjustmentand land subsidence as well the interaction of these processes Only by meansof monitoring in terms of gravity potential changes at the above level ofprecision the change of ocean level can be understood as a global phenomenonand purely geometric height changes be complemented by information aboutthe associated density or mass changes

The geoid can also be defined in a relativistic way [43] as the surface whereaccurate clocks run with the same rate and where the surface is nearest tomean sea level The relation between the differences in the clock frequenciesand the gravitational potential is given in simplified form by Eq 1 At present

Exp Astron

there is no operational way to compare frequencies of the already availableoptical clocks on the global scale at the same level as their accuracy wouldallow With EGE and by then improved ground optical clocks it will becomepossible to obtain gravitational potential differences on a global scale by com-paring frequencies This would be a significant new dimension to gravity fielddetermination since such observables are given on the global scale and providein situ local gravity information at the same time EGE will allow clock-basedgravitational potential mapping with the same payload instruments used forthe fundamental physics experiments The on-board clocks are actually notrequired for this purpose MOLO MWL and FCDP suffice Thus EGE willallow establishing a global reference frame for the Earth gravitational fieldwith accuracy in the order of a few cm in terms of the geoid heights Thisassumes that by the time EGE is operational (mobile) ground clocks withfractional inaccuracy at the level of 10minus18 are available the measurementsare limited by MWL noise and thus require a long integration times to matchthe expected ground clock performance Such a reference frame will serve asreference for all gravity field modeling and clocks will be used to define theinternational atomic time scale and the Earth gravitational potential at thesame time

In summary with optical clocks geodesy will undergo a second revolutionafter geometry now being measured by clock-based GNSS systems also physi-cal heights and potential will be measured by (mobile) optical clocks using thegravitational redshift effect

28 Application to frequency standards and terrestrial fundamentalphysics studies

As has been described repeating ground clock comparisons and integratingover several months will provide global ground clock frequency comparisonsat the level of 1 times 10minus18 more than two orders of magnitude below the currentGPS and two-way satellite time and frequency transfer methods In the timeand frequency community the availability of such a global high performancemicrowave link will accelerate the process leading to a redefinition of the SIsecond based on clocks operating in the optical domain

The global character of the ground clock comparisons made possible byEGE can also contribute to the search for a time-variation of the fundamentalconstants

The comparisons do not require availability of the on-board clocks The riskof this type of measurements is therefore reduced Moreover the lifetime ofmany subunits required for this measurement (MWL and FCDP see below)can exceed 10 years (from previous experience with similar devices) andthus offers a motivation to increase the mission duration beyond the nominal3 years

Further applications to metrology are listed under the secondary goals inSection 21 above

Exp Astron

3 Mission profile

31 Orbit

The orbit must satisfy the following requirements

Primary 1 Large gravitational potential difference between apogee andperigee

2 sufficient contact time with at least two ground stations atperigee

Secondary Simultaneous visibility of satellite from distant (inter-continental) ground stations for several hours

To satisfy the primary objective the orbit must be highly elliptic The perigeeshould be low A very low perigee means a short contact time which decreasesthe accuracy of the frequency measurement A compromise is a perigeealtitude of sim2500 km For a sufficiently high apogee this yields a U of 60of the theoretical maximum between the earth surface and infinity

The baseline scenario is an orbit of approx 12 h period with the followingproperties

Apogee Altitude 37856 km (minus 10 deg latitude one over Atlantic andPacific resp)

Perigee Altitude 2500 km (+ 10 deg latitude one over Guatemala andMalaysia resp)

Inclination 634 argument of perigee 170RAAN 25 True Anomaly 220Apogee drift rate 70 kmday perigee drift rate sim 14 kmday

True anomaly and right ascension of the ascending node (RAAN) wereoptimized to minimize orbital precession Figure 3 shows the ground track andthe ground coverage at one of the perigees Common-view contact time forstations in Europe is about 6 h while between a station in Europe and in theUSA it is about 35 h Frequency comparison windows between a given groundstation and EGE repeat every 24 h

A final selection of the orbit requires an in-depth study of the costs of thepropellant and maneuvers as well as of the setting up of new ground stations

32 Ground segment

The mission requires two separate ground segments the science groundsegment and the standard satellite control segment which supports telemetryand telecommand (TMTC)

Exp Astron

Fig 3 Baseline EGE orbit One (out of two) apogee and one (out of two) perigee ground pointsare shown The ground coverage at one of the perigees is shown with circles at elevation anglesteps of 10 starting from 10

321 Science ground segment

As described above one part of the mission is to perform frequency com-parisons between ground-to-space and ground-to-ground (common-view andnon-common view) The use of a two-way microwave link (MWL) systemhas the advantage that a precise orbit or the precise relative velocity of thesatellite must not be known [4] The orbit determination requirements fora two-way system are significantly relaxed compared to the requirements tomodel the relativistic effects on the clocks The frequency comparison resultsare available both on ground and on the satellite

For ground-to-satellite clock comparisons five ground stations arefavourable two capable of viewing perigee three for viewing apogee Thisallows for the classic three-clock configuration where the performance of eachclock can be obtained individually The perigee stations will be observing thesatellite for sim25 min every perigee passage for the duration of the missionThe apogee stations will observe for sim3 h every apogee passage Of coursea larger number of stations as well as longer observation time are usefulfor robustness redundancy and orbit determination The stations could beinstalled on a temporary basis for the duration of the mission

The stations must have access to optical clocks of performance at least ashigh as todayrsquos state-of-the-art ground clocks but preferably higher Theywill include an optical frequency comb by means of which the optical clockfrequency is transformed to a radio frequency for comparison with the EGEclocks High-accuracy fiber-optic link technology already existing today allowsdistances of sim100 km between the ground station and the correspondingoptical clock allowing a certain freedom in the choice of the clock locationWith further increase of the fiber-optic link range currently being investigated

Exp Astron

the area over which the ground station can be positioned relative to the opticalclock will increase This option is of importance for the implementation of ageophysics measurement campaign for which a large number of clock locationsis of interest

The above ground stations will be complemented by some 15 additionalground terminals provided by the world-wide laboratories equipped with highperformance primary clocks at the time of the mission These will participate inorbit determination to 1 cm accuracy along the full orbit which can be achievedwith non-optical atomic clocks The EGE MWL is fully compatible to plannedACES ground terminals which might reduce the number of required stations

With a larger number of transportable optical clock ground stations de-ployed and moved across the continents a geophysics campaign could beimplemented to map the gravitational potential

Ground station equipment MWL uses dedicated small transportable termi-nals which can be easily installed at any location of interest Each terminalconsists of a steerable dish antenna of 1 m diameter with pointing accuracy of01 using a transmitted power of 5 W per band The high directivity antenna(beam aperture Ka-band 07 Ku-band 14 S-band 20) avoids multi-pathfrom nearby objects Antenna noise temperatures are Ka-band 450 K Ku-band 340 K Operation is at elevations above 10 and measurements areperformed above 20 For ranging purposes terminal location is determined byGPS surveying Each ground station computes its own ionospheric correctionsbased on a triple-band receiver The terminal has a built-in delay monitorwhich allows calibrated ranging during several months once its bias has beendetermined by laser ranging Its M and C interface is via local area networkto either a userrsquos computer or directly to the science network control centreby remote MampC The stationrsquos operation is fully automatic and unattendedSchedule and orbital data are received via LAN once per day Signal acquisi-tion is fully autonomous Data transmission is after each pass The terminal hasan uplink telecommand capability of 1 kBits and a downlink data capacity ofup to 10 kBitss For the local user there is a visual interface providing quick-look data from real-time comparison to the spacecraft clocks to verify theirgood health The final full performance product is calculated at the sciencedata centre

Local metrological data (temperature pressure humidity) are recorded tocorrect for tropospheric delay of the ranging data The main advantage ofMWL terminals compared to laser based systems is their all-weather capabil-ity although microwaves suffer higher propagation errors in the ionosphereand troposphere

The EGE ground terminal will profit from the ERS-2 mission PRARE(precise range and range rate equipment) [45] experiments whose terminalshad similar design and have been operated under extreme climatic conditionsincl Antarctica and high temperature regions from existing designs fromthe space segment (mission ACES) and no further special developments aredeemed necessary Cost per EGE master station is modest

Exp Astron

Satellite tracking via laser ranging Laser ranging is alternative to MWL rang-ing and also serves for absolute calibration of the latter For this purpose anon-board corner cube reflector (CCR) is provided Ranging will be done withthe International Laser Ranging Service Ideally the satellite will be rangedat every perigee passage and every apogee passage simultaneously with thefrequency comparison procedures Todayrsquos ranging precision is sufficient tosatisfy the science goals For apogee observations in the northern hemispherea large number of laser ranging stations is available The number of stationscapable of ranging at or near perigee in the southern hemisphere is six andthus is sufficient even considering unfavourable weather conditions at someof these The MWL ground stations are able to range even under thoseconditions

322 Control ground segment

Average data transfer rate is estimated at 300 MBday (approx 40 kBs onaverage) Data will be stored on board (2 GB capacity required) and will betransferred to a receiving station on the Earth once a day

XPLC

incl Data StorageIF to TMTC Link

HK amp TMTC

to ground

Opt Clock

AOM

MOLO

PHARAO

FCDP

10 GHz

100 MHz

250 MHz

GNSS Rx

MWLTampF Link

uplink Data

UTC(k)

TampF Space vs Ground

TM from GS

Tim

e S

cale

Orbit GNSS Time UTC

TampF Space vs Ground

MampC Time Scale

MampC

MampC

MampC

MampC

Mamp

C

PHARAO vs Opt Clock

PHARAO vs Opt Clock

10 GHz amp Epoch

10 MHz

amp Epoch

Epoch

10 MHz

amp Epoch

Opt Link

Fig 4 Schematic of the payload showing subunits and their signals The results of the frequencycomparison between the two on-board clocks (optical clock and microwave clock PHARAO) areobtained from a combination of the frequency values applied to the frequency shifters AOM (seeFigs 7 and 8) and those output from the frequency control and distribution package (FCDP)as indicated by the arrows ldquoPHARAO vs Opt Clockrdquo The satellite clock to ground clockcomparison results are output from the MWL XPLC is the on-board computer and is involvedin generation of signals for instrument control The optical link shown dashed is a possible add-onproviding an alternative to the MWL GS ground stations MampC monitoring and control HKhousekeeping TampF time and frequency

Exp Astron

4 Main payload instruments

41 Overview

The satellite payload consists of a pair of cold-atom clocks a microwave-optical local oscillator (MOLO) an on-board frequency distribution unit(FCDP) a satellite-to-ground clock comparison unit (MWL) and the onboardcomputercontrol system (XPLC) see Fig 4 Auxiliary units are the posi-tion determination GPSGalileo receiver (GNSS Rx) attitude monitor (startracker not shown) and the corner cube reflector array (CCR not shown)

The science goals require clocks of very high frequency stability on thetimescales of half the orbital period (6 h) as well as perigee passage (sim1000 s)see Fig 2 Only clocks based on cold atoms are suited for the science goalsTable 1 reports values for a subset of atomic clocks that have been investigatednot including clocks that require cryogenic operation or are too complex foruse in an medium-class mission The choice of a clock pair rather than a singleone fulfils several requirements it addresses more science goals improves thevalidity of some science results and it enables calibration accuracy evaluationand redundancy The validity of the science results is enhanced since theclocks differ not only in the employed atoms but also in terms of havingdifferent sensitivity to perturbations by electric and magnetic fields electronicserrors and misalignments A comparison of the results arising from each clockallows for an understanding of measurement errors Both the clocksrsquo and thefrequency comparison unitrsquos performances will be determined after launchand regularly verified during the course of the mission The only reliable

Table 1 Properties of different clock candidates for EGE

Clock Clock Fractional σ = Fractional Sensitivity Figure of Sensitivityatom frequency inaccuracy instability to α Aα merit |Aα| to meq

at 1000 s σ [1016] QCD

Sr+ 445 THz 3 times 10minus17 (p) 5 times 10minus17 (p) 042 08 (p) ndashYb+ (1) 688 THz 3 times 10minus16 (d) 3 times 10minus16 (d) 088 029 (d) ndash

3 times 10minus17 (p) 1 (p)Yb+ (2) 642 THz lt 1 times 10minus17 (p) 1 times 10minus17 (p) minus53 53 (p) ndashSr 435 THz 1 times 10minus16 (d) 1 times 10minus16 (d) 009 009 (d) ndash

lt 1 times 10minus17 (p) 09 (p)Yb 519 THz 7 times 10minus16 (d) 1 times 10minus16 (d) 031 03 (d) ndash

lt 1 times 10minus17 (p) 1 times 10minus17 (p) 31 (p)Cs 91 GHz 2 times 10minus16 (d) 1 times 10minus15 (d) 275 027 (d) 1Yb+ 126 GHz 1 times 10minus15 (p) 1 times 10minus15 (p) 34 034 (p) 1

A relevant figure of merit is the ratio of |AX | and fractional frequency instability σ on the perigeevisibility timescale The Cs clock is the only high performance clock with adequate sensitivity tothe particle masses me mq The indicated Cs clock performance was achieved using a microwavelocal oscillator based on a cryogenic oscillator On EGE the latter is to be replaced by an opticallocal oscillator-based device (MOLO) with similar short-term performance but without the needto operate at cryogenic temperature The clock instability at 10000 s is typically similar to theindicated inaccuracy(p) projected (d) demonstrated (1) quadropole (2) octupole

Exp Astron

way to achieve this is by having at least two clocks on board The on-boardclock-clock comparison provides performance evaluation independent of theground-satellite comparison unit Once the satellite clock performances areestablished the link performance can be established as well

The clocks are similar in basic structure but differ in the detailed implemen-tation They consist of

a source of atomsa preparation subunit which cools the atoms to near-standstill (millikelvintemperature or below)a trapping subunit that confines the atom(s) to a tiny volumean interrogation subunita clock control electronics package

The clocks share a common central unit the optical-microwave local oscil-lator which serves as a flywheel oscillator with outstanding stability on shorttimescales (01ndash10 s) for both clocks It consists of a laser stabilized to an ultra-low loss optical resonator made of ultra-low-expansion glass A microwavesignal is derived from the laser by phase-locking a frequency comb to the laserwave and using a multiple of the repetition rate

While the subunits of any optical clock have a similar technological basis(optics electrooptics acoustooptics thermal control vacuum systems) theydiffer in their specifications Depending on the choice of the atom species de-livery confinement and interrogation methods clocks of different perfor-mance result A number of laboratory optical clock demonstrators have beendeveloped worldwide This ensures availability of the know-how for the indus-trial implementation of the flight models

The first clock is a microwave clock based on the interrogation of slowlymoving ensembles of cold Cs atoms It is essentially based on the PHARAOclock [5 6] developed to engineering model level by CNES in the frame-work of the ESA ISS mission ACES In EGE this clock is upgraded usingthe microwave-optical local oscillator This improves significantly the clockperformance The specification is a relative instability (Allan deviation) of3 times 10minus14 (τ s)minus12 ie at the integration time τ = 1000 s it drops to 1 times 10minus15and to 3 times 10minus16 at 10000 s While this specification is at least a factor 3 lessstringent than for the optical clocks the significantly lower cost of obtaining aflight model and the sensitivity of its frequency to the particle masses makes ita suitable choice

The optical clock baseline instrument is a single-ion optical clock Amongvarious suitable ion species (Hg+ Al+ Sr+ Yb+) [44ndash47] the Yb+ ion hasbeen selected Its breadboard performance is established at the instability levelof 3 times 10minus16 over 1000 s it exhibits a large sensitivity coefficient Aα hassmall sensitivity to magnetic fields and exhibits significant potential for furtherimprovements (use of the octupole transition with sim6 times larger Aα andlower overall instability [48] see below) A trapped Yb+ ion is also a veryrobust system it has been demonstrated that a single Yb+ ion can remaintrapped uninterruptedly for many months Finally Yb+ is the only ion with

Exp Astron

which optical clocks demonstrators have been operated in not a single but twomajor metrology laboratories

Another attractive optical clock is based on ensembles of cold neutral Sror Yb atoms trapped in an optical lattice produced by laser beams This clocktype has a lower instability than a single-ion clock thanks to the large numberof atoms (sim105) used Its development is in progress in several laboratoriesin the world (in part supported by space agencies) and has to date reached6 times 10minus17 instability at 5000 s and 1 times 10minus16 uncertainty [49 50] with potentialfor even higher performance However it is more complex and costly thanthe Yb+ ion clock since the cold atom preparation and trapping subunitsinclude more components require higher laser powers and the respectiveoperating procedures are more complex It is therefore not considered for thepresent mission designed to be developed in the very near future with limitedresources but is an interesting candidate for missions without such constraintsIts estimated physical parameters are indicated in Table 2 for comparison

42 Main instruments

421 Microwave clock

The PHARAO cold atom clock is one of the two atomic clocks of the spacemission ACES managed by ESA It is proposed here to include a secondversion of this instrument as its costperformance ratio is very attractive forthe EGE scientific objectives Its concept is very similar to ground basedatomic fountains but with the major difference of zero-g operation Atomsslowly launched in free flight cross two microwave fields tuned to the transitionbetween the two hyperfine levels of the cesium ground state The interrogationmethod based on two separate oscillating fields (Ramsey scheme) allowsthe detection of an atomic line whose width is inversely proportional to thetransit time between the two microwave cavities The resonant microwavefield at 9192631770 GHz (SI definition of the second) is synthesized startingfrom an ultrastable quartz oscillator (USO) in the FCDP phaselocked to themicrowave output of the microwave-optical local oscillator and stabilized tothe clock line using the error signal generated by the cesium resonator In thisway the intrinsic qualities of the cesium hyperfine transition both in terms ofaccuracy and frequency stability are transferred to the USO In a microgravityenvironment the velocity of atoms along the ballistic trajectories is constantand can be changed continuously over almost two orders of magnitude (5ndash500 cms) Differently from atomic fountain clocks presently operated onground very long interaction times (up to few seconds) will be possible whilekeeping reasonable the size of the instrument

The instrument consists of 4 main elements a laser source which providesthe light required to cool and detect the cesium atoms (Fig 5 left) a frequencychain which generates the 92 GHz microwave signal a cesium vacuum tubewhere the interaction between the microwave field and the atoms occurs

Exp Astron

Table 2 Resources and technology readiness level (TRL) of the EGE spacecraft

Unit Volume Mass Power TRL Comment[l] [kg] [W]

Ion clock 100 70 60Atom source 4Atom preparation 6Atom trap 4Atom interrogation 6Clock control 6

Microwave clock 200 91 110 8 TRL by end 2008Local oscillator

Stabilized laser 18 15 23 6Frequency comb 20 16 38 6

Microwave link 18 18 55 8FCDP 7 8 12 8GPS receiver 2 5 7 9CCR 2 2 0 9Payload sum 367 225 305Thermal control 14 17 30 8OBDH 20 24 38 9Star tracker 1 1 8 9Radiation shield 10 40 0 9Support structure 170 204 0 9Power incl solar arr 50 58 10 9Propulsion 21 25 20 9AOCS 36 43 82 9TMTC 8 10 15 9Satellite bus sum 330 422 203 Flexbus conceptPayload plus bus incl 20 margin 8364 7764 6096Propellant 40Launch adapter 100SC wet masspower incl Margins 9164Lattice clock 135 105 152 Alternative clock

Atom source 4Atom preparation 6Atom trap 4Atom interrogation 6Clock control 6

Also included are the resources for an optical lattice clock based on neutral Strontium atoms

(Fig 5 right) and a computer The engineering model of PHARAO is fullyassembled at CNES Toulouse and undergoes functional and performancetests until the end of 2008 The operating temperature range is 10ndash33 andnon-operating temperatures are minus45 +60 The cesium tube consists in aUHV chamber pumped by getters and a 3 ls ion pump a cesium reservoira microwave cavity coils to provide a uniform magnetic field and three layersof magnetic shields Vacuum windows with fiber-optics collimators enable thetransmission of the cooling and detection light onto the cesium atoms

In autonomous mode PHARAO uses the USO as interrogation oscillatorand will provide a clock signal with fractional frequency stability below1 times 10minus13 (τs) minus12 and inaccuracy near 1 times 10minus16 When using the microwave-

Exp Astron

Fig 5 Left The PHARAO laser source engineering model with cover removed Dimensionsare 530 times 350 times 150 mm3 and the mass is 200 kg Ten polarization maintaining optical fibers(yellow) guide the laser beams to the cesium tube All diode lasers are mounted on Peltier coolersfor temperature regulation within 2 mK (courtesy EADS SODERN) Right The PHARAOengineering model cesium tube without the two external magnetic shields The volume is 990 times336 times 444 mm3 and the total mass is 44 kg

optical local oscillator PHARAO will provide a reduced frequency instabilityof 3 times 10minus14 (τs) minus12 This corresponds to 1 times 10minus15 at 1000 s and 3 times 10minus16

at 104 s

422 Ion optical clock

The 171Yb+ ion clock shown in Figs 6 and 7 provides an atomic frequencyreference based on a single 171Yb+ ion confined within an RF electric end-captrap (similar to a Paul trap) within an ultra high vacuum (UHV) chamber andlaser cooled to sim1 mK close to the Doppler cooling limit As a result the ionrsquosfirst order Doppler motion is completely removed and the ion experienceslittle collisional perturbation from its environment Under these conditionsthe narrow linewidth 2S12 (F = 0 mF = 0) minus 2 D32 (F = 2 mF = 0) quadrupole

Fig 6 Left Simplified level scheme of 171Yb+ showing relevant cooling (blue) auxiliary (red)quadrupole clock (green) and octupole clock (dark blue) transitions δνnat is the natural linewidthof the clock transition Middle close-up of Paul-type RF trap (PTB Braunschweig) the ion istrapped in the hole (diameter 14 mm) Right Complete ion trap with vacuum flange height 10 cm(NPL Teddington)

Exp Astron

935 nm D32-state repumper laser

638 nm F72-state repumper laser

739 nm extended cavity diode laser

AOM

PMT

Ion trap UHV chamber

vacuum pumps

Fibre-free space interface

Mu-metal shielding

EC

T

B

Magnetic field

Temperature

Ion fluorescence Processor

Clock laser frequency step and control

4355 nm

Frequengy doubling II

3695 nm cooling wave

Low-drift cavity

Low-drift cavity 171Yb+ 435 nm ion clock unit

876 nm clock laser wave from MOLO

Frequengy doubling

Fig 7 Simplified schematic of the single-ion optical clock unit AOM acousto-optic frequencyshifter PMT photomultiplier tube EC endcap T temperature stabilization system B magneticfield

clock transition at 4355 nm can be interrogated by spectrally narrow and stableclock laser light A diagram of the simplified energy level scheme for 171Yb+showing the relevant cooling auxiliary and clock transitions is given in Fig 6All these wavelengths are provided by diode lasers operated in fundamentalor frequency doubled mode The clock laser light is obtained from the 871 nmwave provided by the MOLO unit (see Section 423) after frequency-doublingin single pass by a KNbO3 nonlinear-optical crystal to 4355 nm The clocktransition spectral profile is observed by stepping the clock laser frequency bymeans of an acousto-optic frequency shifter (AOM in Fig 7) and recordingthe statistics of the quantum jumps in cooling laser induced fluorescence asa function of this frequency Under clock operation the AOM repeatedlysteps back and forward between the half-intensity points on the transitionprofile monitoring the signal imbalance between these points Any detectedimbalance is servo-corrected to zero by feedback to the AOM

In detail the optical clock architecture will comprise (Fig 7)

bull an RF end-cap trap for ionising and confining a single 171Yb+ ion withinan ultra-high vacuum chamber pumped by a small ion pump and non-evaporable getter pump The trap is driven by means of an ac voltage ofa few hundred volts at a drive frequency of sim10 MHz A small oven withseveral mg of 171Yb isotope is heated to provide a low flux of Yb atomsfrom which single ions can be ionised by electron bombardment from ahot wire filament within the trap potential well Low voltage dc is applied

Exp Astron

to additional electrodes to position the ion precisely at trap potentialcentre in order to minimise the ionrsquos micromotion at the drive frequencyThis needs automatic periodic monitoring and correction especially in theperiod after re-loading an ion into the trap External magnetic field coilsin three orthogonal axes allow the nulling of external fields and settingof a fixed field of sim1 μT The trap and coils are surrounded by mu-metalshielding to minimise external field changes during the orbit A level oftemperature control of 1 K of the region surrounding the trap is requiredfor maintaining the blackbody-induced frequency shift uncertainty withtemperature at the 10minus17 level

bull a laser platform to provide frequency doubling of an extended cavity739 nm diode laser for Doppler cooling of the ion on the 2S12(F = 1) minus2 P12 (F = 0) dipole transition at 3695 nm Light from an amplified diodelaser device frequency-doubled in a periodically poled LiTaO3 crystal insingle-pass will be used Typical laser powers used for driving the coolingtransition below saturation are about 2 μW for a beam waist in the trapof about 50 μm By modulation of the injection current of the 739 nmdiode laser at 147 GHz a sideband is generated that excites the 2S12 (F =0) rarr2 P12 (F = 1) ldquorepumperrdquo transition in order to avoid optical pump-ing between the ground hyperfine states

bull The platform also houses the auxiliary lasers at 935 and 638 nm The935 nm diode provides repumping of the ion from the 2 D32(F = 1)metastable level after occasional branching decays to this level during thecooling sequence The 638 nm diode allows fast recovery of the ion fromthe very-long-lived 2F72 metastable state after very occasional collisionaldecay to that state Currently extended cavity lasers are forseen butdistributed-feedback lasers may be available in the near future

bull a high NA lens imaging system and photomultiplier detection system torecord the statistics of 369 nm fluorescence quantum jumps as a functionof clock laser frequency step

bull a fibre system to deliver the various cooling auxiliary and clock light fromsource to trap making use of achromatic doublets where necessary at thefibre-free space interface for launching into the trap

bull a monitoring and control processor which provides the primary coolingand clock laser pulse magnetic field and detection sequencing to observeand lock to the ion clock transition frequency The processor also monitorsfrequency and amplitude data necessary to determine normal laser andion operational conditions and initiate resetting and recovery algorithmswhere necessary and laser unit failure The actual magnetic field presenton the atomic volume is determined and actively stabilized so that aninaccuracy at the level of 1 times 10minus17 can be achieved

bull a redundancy level of at least two units for cooling clock and repumperlasers plus similar for frequency doubling crystals All redundancy unitsfor each wavelength will be fibre multiplexed as standard allowing redun-dant unit activation on determination of prior unit failure

Exp Astron

For an optimized quantum-projection-noise-limited performance of thesingle-ion clock Ramsey interrogation with a cycle time equal to the lifetimeof the metastable D32 level of 50 ms will be performed leading to an Allandeviation of about 27 times 10minus15 (τ s)minus12 This is sim45 times 10minus17 at 1 h satisfyingrequired stabilities for the science objectives Current laboratory results are astability of 1 times 10minus15 at 30 s averaging down to 3 times 10minus16 at 1000 s (PTBGermany) Also comparisons between two independent traps showed agree-ment at the level of several parts in 10minus16 over a series of eight measurementruns

The 171Yb+ ion clock offers a second option namely the 2S12ndash2 F72 octu-pole transition at 476 nm (see Table 1 and Fig 6) [48] It has an extremelylong upper state lifetime low electric quadrupole and second order Zeemansystematic shifts and therefore is a candidate for an even higher performanceclock if a clock laser system with lower instability than the above is providedApart from the clock laser it shares the laser system of the quadrupoleclock Thus with moderate additional resources the Yb+ ion apparatus couldprovide two optical clocks in a single package and specifically perform directfrequency comparisons between the quadrupole and the octupole transitionwith a large sensitivity Aα

The ion clock will operate under microprocessor control providing thecontrol algorithms for the integrated pulse sequencing for cooling repumpingclock interrogation magnetic field switching cooling fluorescence monitoringThe controller will also provide error flags for critical processes and systemreset and correction where necessary This will include ion re-loading andmicromotion reduction algorithms Automatic monitoring of laser power andspectral quality with signal re-optimisation or laser failure determination isrequired Redundant laser units will be pre-aligned and multiplexed into thefibre delivery to the trap so that redundant units can be readily activated in thecase of diode laser failure

423 Microwave-optical local oscillator (MOLO)

This oscillator provides two ultrastable frequency outputs one in the opticalregion the other in the microwave region the two frequencies being coher-ently related The MOLO is composed of three subunits the clock laser thereference cavity and the frequency comb see Fig 8

The clock laser subunit delivers the light for the excitation of the opticalclock transition It is generated by an extended-cavity diode laser at a wave-length of 871 nm Most of the output wave of this laser is sent to the ionclock The laser subunit containing two lasers for redundancy will require3 kg 3 l 15 W The frequency instability of the clock laser must be superiorto the minimum clock instability 27 times 10minus15 (τ s)minus12 for τ up to the timeconstant (sim10 s) of the servo system that locks the laser to the ionrsquos resonancesignal In order to achieve this the laser frequency is stabilized to the reference

Exp Astron

871 nm clock laser

EO

Detector and servo

AOM Femtosecond

frequency comb

To optical clock 10 GHz To FCDP

Microwave-optical local oscillator

ULE cavity

Fig 8 Microwave-optical local oscillator (MOLO) consisting of clock laser reference ULE cavityand femtosecond frequency comb (OFC) Radiofrequency inputs and outputs are not shown EOelectro-optic modulator ULE ultra-low expansion AOM frequency shifter

cavity where the PoundndashDreverndashHall method is used About 10 μW of thelaser output are used for this

The system will exhibit sim04 Hz (sim1 times 10minus15 relative) linewidth and afrequency drift lt01 Hzs The cavity consists of two high-reflectivity mirrorsoptically contacted to a cylinder of diameter 70 mm length 150 mm made ofULE (a glass exhibiting ultra low thermal expansion coefficient) with a cavityfinesse of about 300000 Residual vibrations of the satellite (arms lt 1 times 10minus6 g)should not be a limiting factor to the desired linewidth for safety marginan optimized cavity support can reduce the sensitivity to accelerations Foracoustic and thermal isolation the cavity resides inside an aluminium vacuumchamber equipped with a small (3 ls) ion-getter pump Two stages of polishedaluminium shields are implemented around the cavity which are activelytemperature stabilized so that the cavity is precisely kept at the zero crossingtemperature of the ULE thermal expansion coefficient Heat application andremoval is by thermoelectric elements between the shield and the vacuumchamber Additionally the entire vacuum chamber is temperature-stabilized

During transportation and launch the cavity will be rigidly fixed insidethe vacuum chamber by pressing on it from the sides by piezo-mechanicalactuators allowing accelerations of several g After bringing the satellite onthe orbit the actuators will be released

The optical setup for frequency stabilization will be a classic Pound-Drever-Hall scheme in reflection using an electro-optic modulator (EO) Couplingof the clock laser radiation to the cavity is via single-mode polarization main-taining optical fiber The fiber in- and out-couplers are mounted on miniaturepiezo-motor driven multiaxis translation stages Microprocessor control ofthe translation stages performs a laser-to-fiber incoupling stabilization andlaser-to-cavity mode-matching optimization (positioning inaccuracy lt10 μmalignment inaccuracy lt10 μrad)

The frequency stability of the clock laser will be limited by thermal 1f noisein the mirrors with an fractional Allan deviation slightly below 1 times 10minus15 from1 s to a few 10 s Resources including control electronics are 15 l 12 kg 8 W

Exp Astron

A femtosecond fiber laser based optical frequency comb generator (OFCFig 8) will be used for the clock frequency comparisons and to obtain aspectrally ultrapure microwave local oscillator output for the operation ofPHARAO for the microwave link and for the GNSS receiver The fem-tosecond laser is based on erbium-doped fiber pumped by fiber-coupled laserdiodes It emits 100 fs pulses at a repetition rate of 250 MHz with an averagepower of 100 mW The emission spectrum is centered around 1560 nm Spec-tral broadening to more than one octave is obtained via nonlinear interactionin a microstructure fibre so that the range from about 1050 to 2100 nm iscovered

Second harmonic generation of a portion of the broadened spectrum allowsobtaining the required light for the measurement of the clock laser frequencyat 871 nm The OFC will be referenced to the clock laser frequency (afterintroduction a controllable shift via the frequency shifter AOM) by makingthe carrier envelope offset frequency vanish so that the repetition rate fr isa high-order subharmonic of the optical clock reference frequency f1 = m1 frThe link between the optical and the microwave domain is then simply thedistribution of the repetition rate frequency or more conveniently a high(order 40) harmonic of the comb repetition rate 10 GHz

It has been demonstrated by several groups that frequency comparison andfrequency division with an OFC can be performed to very high accuracy andstability beyond the limitations imposed by the optical clocks themselves Forexample a fiber laser OFC with an instability (Allan deviation) of 6 times 10minus17

in 1 s has been demonstrated [51] OFCs based on fiber technology possessgood long-term reliability and high power efficiency The projection for a spacequalified OFC based on current industrial terrestrial technology is 16 kg mass20 l volume and 38 W power requirement

424 Frequency comparison and distribution package (FCDP)

This package distributes the ultra-stable 10 GHz reference frequency outputof the MOLO to the MWL performs frequency comparisons between MOLOand PHARAO and generates a variety of different reference signals for theindividual instruments including the 250 MHz reference clock to the OFCAll signals are coherent to the MOLO In the FCDP a USO is locked to theMOLO signal by a digital PLL A divider chain generates timing pulses (1 pps1000 pps etc) and keeps a continuous local time scale From the digital phasecomparison data the phase noise density spectrum between the USO and theMOLO is available by telemetry (TM) to assess the overall frequency purityin-flight

As the FCDP distributes the full performance reference from the MOLOdirectly to the MWL there is minimum impact of the FCDP on overall clockcomparison performance The USO-phase lock loop uses a similar designas for ACES-FCDP The digital PLL with linear phase detector allows theUSO to be slightly offset from its reference which is selectable by groundcommand This results in an accurate on-board time scale despite any offset

Exp Astron

of the MOLO reference frequency which can be synchronised to UTC Thefrequency synthesis chain re-uses modules developed for ACES-FCDP andtheir frequency inaccuracy Expected relative uncertainty (Allan deviation)of the frequency comparison between two signals provided to the FCDP is47 times 10minus17 at 1000 s 95 times 10minus18 at 10000 s

During non-availability of the MOLO ie during system start-up andcontingency conditions the USO runs as flywheel (ie it is unlocked withinstability lt15 times 10minus13 from 1 to 100 s) It will still provide useable butless accurate references to the instruments including the reference to OFCCommon view ground-to-ground time- and frequency comparison and supplyof the remaining instruments remain possible at the USO provided accuracy

Mass volume and power consumption are 8 kg 7 l 12 W Controldata rateis 10 kBday 50 kBday (FCDP alone)

425 Microwave link (MWL)

The microwave link is a two-way triple-frequency microwave link (Fig 9) Itis designed to perform accurate frequency comparison as well as time-transfer

Ku-Band Up-linkPower Tx 5 WCarrier 13475 GHzPN-Code 100 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

S-Band Down-linkPower Tx 1 WCarrier 2248 MHzPN-Code 1 MChips1pps 1 time marker sData 5 kBits

Ku-Band Down-linkPower Tx 1 WCarrier 1470333 GHzPN-Code 100 MChips1pps 1 time marker sData 5 kBits

EGE SC

S

Ku

KaKa-Band Up-linkPower Tx 5 WCarrier 345 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

Ka-Band Down-LinkPower Tx 1 WCarrier 32 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 5 kBits

GS Antenna diameterEGE terminal 1 mACES terminal 06 m

Fig 9 EGE microwave two-way link concept and characteristics See text for a description Thelink will be able to perform frequency comparisons with up to ten ground stations simultaneously(only one active comparison is shown) pps pulse per second Tx transmitter SC spacecraft

Exp Astron

between on-board and on-ground clocks Two-way operation allows for com-pensation of the effects due to spacecraft location and its motion relative toground Its two bi-directional links operate in Ka-band (32345 GHz) andKu-band (147135 GHz) The high carrier frequency of the up- and down-links allows for a noticeable reduction of the ionospheric delay A third down-link channel is in S-band (22 GHz) and is used to determine the ionospheretotal electron content MWL uses a combination of carrier phase and codephase data to measure with high precision the absolute ionospheric delayas well as ionospheric delay variations during a ground contact The useof three frequency bands allows to resolve third order ionospheric termsKey modulation parameters are Ka band Tone 250 MHz Ku band PN100 MChips S-band 1 MChips

The EGE-MWL concept is a considerably improved version of the ACES-MWL Microwave carrier generation is directly from the 10 GHz microwaveoutput available from the MOLO and having exceptionally low close-in phasenoise without intermediate frequency dividers or other complex synthesiserstages commonly used in RF transponders The high frequency carriers arekey to achieve the desired frequency comparison accuracy

The Ka-band carrier is generated by tripling the FCDP-generated signalThe Ku-band is subsequently generated by division by two using a verystable regenerative divider Finally the assigned radio frequency is obtainedby adding a low frequency signal using an image reject mixer A mixedanalog-digital receiver design will enhance its performance by use of a flexibledigital signal processing and digital control loops The number of simultaneousground contacts is increased from four (as in the ACES mission) to ten or moreusing the digital concept while the number of physical receiver channels perband is reduced from four to two one for operation to ground and one for in-flight calibration The concept allows adaptation of receiver loop parametersto optimise for the different portions of a highly elliptic orbit while reducingsystematic uncertainties due to high Doppler and Doppler rate Real-timedata exchange to ground terminals is by additional data modulation on all RFsignals The spacecraft carries two different small antennae for perigee and forapogee with gains of 8 dBi 20 dBi beamwidths 100 (full earth coverage) 14respectively power is 1 W for each

The MWL end-to-end performance between the attached clocks includingspace- and ground terminals is predicted to be a relative Allan deviation of1 times 10minus16 at 1000 s (perigee) and 1 times 10minus17 at 10000 s (apogee) Thesevalues are based on conservative implementation losses and other instrumentallosses assumed as 6 dB at apogee and 10 dB at perigee including additionallosses due to atmosphere An attitude knowledge of 1 is required

The MWL link will also support precise orbitographyndashcomplementary tothe GNSS receiver Generation of ranging data is simultaneous to two-waytime transfer and is available under all weather conditions Orbit determina-tion accuracy is expected to be better than 1 cm well matching EGE clockrequirements The value is expected based on demonstrated results fromother missions (METEOR ERS-2 TOPEXPOSEIDON) Laser ranging to

Exp Astron

the CCR array is used a few times during the mission to calibrate the MWLranging

In case of non-availability of the ultrapure signal from MOLO an additional10 GHz output is available from the FCDP to support ground-to-ground clockcomparison in common-view mode even after the nominal mission periodwhen the neither the clocks nor MOLO may be available any longer Basedon the PRARE experimentrsquos 12 years of operation this can be assumed forthe FCDPMWL combination as well The potential extended operational pe-riod makes the proposed configuration particularly attractive to metrologicallaboratories that rely on availability of long-lasting data in support of thegeneration of UTC

The resources of MWL are as follows mass 18 kg volume 18 l power55 W TC 2 kBday TM 120 MBday

426 Auxiliary units

The star camera assembly on the EGE spacecraft will provide real-timeattitude data primarily for the orientation of the spacecraft (with accuracy of1 in real-time) and the proper reduction of the GNSSSLR data

A GNSS receiver is used for the precise orbit determination time andfrequency transfer coarse positioning (lt1 m) for real time use by the Attitudeand Orbit Control as well as the back-up for time tagging of all payloaddata For the time and frequency transfer the receiver is nominally using theonboard frequency reference The receiver assembly includes three specialantennas By the year 2015 more than 100 GNSS satellites are expected tobe available for precise orbit determination permitting orbit inaccuracy of 1ndash2 cm RMS for altitudes up to approx 10000 km At apogee a combinationof GNSS signals and MWL ranging will be used with expected inaccuracyof 10 cm

5 Basic spacecraft specifications

The mission foresees a single spacecraft with a mission duration limited toapprox 3 years for cost reasons Attitude control along three axes withless than 1 error on a timescale of 1000 s is required The control can beperformed with cold gas thrusters magnetotorquers or reaction wheels Theselection will be done on the basis of the vibrations induced by these systems(desired level arms lt 10minus6 g)

Cold gas thrusters or monopropellant thrusters will be required to maintainthe orbit parameters during the mission by small orbit correction manoeuvresand will also guarantee a proper tuning of the orbit parameters after the releasefrom the launcher The ground segment (telemetry control and data transfer)could be via non-ESA stations user-operated to save costs

Further resources are indicated in Table 2

Exp Astron

The satellite bus design drivers are nadir-pointing microwave antennasthermal stability for the payload instruments no strong magnetic sources lowmicrovibrations harsh radiation environment due to crossing of the radiationbelts once per orbit power demand by science payload (with the possibility toreduce power demand by stand-by of some components during specific missionphases eg eclipse) mass limit due to required high altitude with low costlaunch

The satellite bus design can be based on established platforms such asFLEXBUS or PRIMA used on several past missions Their design philosophydiffers one having body-mounted solar arrays the other wings The latter canprovide larger amounts of power The vibration level induced by the panelswhen they are fixed and the duration of solar array drive periods per orbitwill have to be considered in the selection The FLEXBUS concept was usedfor a mission already requiring a low level of residual accelerations and theplatform is optimised for this parameter

For most of the satellite components one can consider using compo-nents and electronic units from navigation or communication satellites whichtypically cruise at altitudes between 15000 and 25000 km altitude LISAPathfinder which will cross the radiation belts during spiralling to theLagrange point will face similar problems concerning radiation as EGE as wellas XMM which is in a Molniya-type orbit with similar radiation characteristics

The instruments will therefore be located in the satellite structure such asto realise ionizing radiation shielding by the spacecraft structure and non-sensitive equipment to the maximum extent possible the need for additionalradiation shielding (including trapped proton radiation) for especially sensitiveparts of the instruments must be determined by ground tests It should bementioned that an orbit with higher perigee and less radiation dose is also anoption as it leads to a tolerable reduction of the science goals

The required plusmn3C temperature stability at the interface to the payloadenclosure is feasible The thermal control will employ the usual means (anenclosure with high thermal inertia heaters and a suitable radiator) Thelocation of the radiator will need a special study given the non-standard orbit

Magnetic field present during the mission can affect the properties of theonboard clock Using SPENVIS the magnetic field has been calculated alongthe spacecraft orbit the field intensity presents maxima around 02 G anda minimum value of the order of 10minus3 G As discussed above this is not alimitation to the instrument performance

6 Discussion and conclusion

The accuracy of the measurements depends crucially on the stability andaccuracy of the clocks While the performance of ground clocks (even thoseeventually deployed in new ground stations dedicated to EGE) will likelyimprove significantly from the present level the space clock design will beldquofrozenrdquo approx 10 years before data acquisition Thus their performance

Exp Astron

level will be significantly below the future ground station clock performanceAs specification we have chosen a performance level that has been demon-strated with laboratory breadboards recently We see no fundamental reasonthat this laboratory performance cannot also be achieved in a space ion opticalclock The experience with the PHARAO cold atom clock has shown thatan engineering model of a laser-based cold atom clock has a performancesimilar to that of a laboratory prototype The perturbation of the ion clock fre-quency due to magnetic fields along the orbit can be modelled and suppressedwith shielding and the residual effects measured and compensated for Themoderate temperature stability (plusmn 3C) to be provided in the experimentalbay can be achieved with thermal shields Accurate temperature controlof particular units is done by active stabilization of the units themselves

One important aspect is radiation which may affect all optical and trapcomponents as well as the lasers and reference cavity The radiation envi-ronment on the EGE orbit is significantly harsher than the near-earth ISSorbit on which ACES will operate The radiation effects must therefore becharacterized comprehensively with appropriate tests and then the radiationshield for the EGE instruments designed

In summary the Einstein Gravity Explorer mission (EGE) is a satellitemission aimed at a precise measurement of the properties of space-time usingatomic clocks By testing one of the most fundamental predictions of EinsteinrsquosTheory of General Relativity the gravitational redshift and thereby searchingfor quantum effects in gravity it explores one of the most important frontiers infundamental physics The EGE mission reaches beyond fundamental physicsproviding an important spin-off in geophysics Furthermore the technologicalaspects of the mission are ground-breaking for future time-keeping and ref-erence frame systems in space and thus are also expected to be of wide useto science and technology in general The payload of the mission has beenconceived as a compromise between cost and performance needing to fit thecost envelope of an ESA medium-class mission1

Acknowledgement We thank S Cesare for helpful information

References

1 Will CM The confrontation between general relativity and experiment Living Rev Relativ9 3 (2006)

2 Ni W-T Empirical foundations of the relativistic gravity Int J Mod Phys D 14 901 (2005)doi101142S0218271805007139

1Finally we point out that a spacecraft with a significantly reduced and thus lower-cost payload (noclocks only MOLO FCDP MWL) on a circular orbit of appropriate height could be used for along-duration earth science mission dedicated specifically to the measurement of the geopotential

Exp Astron

there is no operational way to compare frequencies of the already availableoptical clocks on the global scale at the same level as their accuracy wouldallow With EGE and by then improved ground optical clocks it will becomepossible to obtain gravitational potential differences on a global scale by com-paring frequencies This would be a significant new dimension to gravity fielddetermination since such observables are given on the global scale and providein situ local gravity information at the same time EGE will allow clock-basedgravitational potential mapping with the same payload instruments used forthe fundamental physics experiments The on-board clocks are actually notrequired for this purpose MOLO MWL and FCDP suffice Thus EGE willallow establishing a global reference frame for the Earth gravitational fieldwith accuracy in the order of a few cm in terms of the geoid heights Thisassumes that by the time EGE is operational (mobile) ground clocks withfractional inaccuracy at the level of 10minus18 are available the measurementsare limited by MWL noise and thus require a long integration times to matchthe expected ground clock performance Such a reference frame will serve asreference for all gravity field modeling and clocks will be used to define theinternational atomic time scale and the Earth gravitational potential at thesame time

In summary with optical clocks geodesy will undergo a second revolutionafter geometry now being measured by clock-based GNSS systems also physi-cal heights and potential will be measured by (mobile) optical clocks using thegravitational redshift effect

28 Application to frequency standards and terrestrial fundamentalphysics studies

As has been described repeating ground clock comparisons and integratingover several months will provide global ground clock frequency comparisonsat the level of 1 times 10minus18 more than two orders of magnitude below the currentGPS and two-way satellite time and frequency transfer methods In the timeand frequency community the availability of such a global high performancemicrowave link will accelerate the process leading to a redefinition of the SIsecond based on clocks operating in the optical domain

The global character of the ground clock comparisons made possible byEGE can also contribute to the search for a time-variation of the fundamentalconstants

The comparisons do not require availability of the on-board clocks The riskof this type of measurements is therefore reduced Moreover the lifetime ofmany subunits required for this measurement (MWL and FCDP see below)can exceed 10 years (from previous experience with similar devices) andthus offers a motivation to increase the mission duration beyond the nominal3 years

Further applications to metrology are listed under the secondary goals inSection 21 above

Exp Astron

3 Mission profile

31 Orbit

The orbit must satisfy the following requirements

Primary 1 Large gravitational potential difference between apogee andperigee

2 sufficient contact time with at least two ground stations atperigee

Secondary Simultaneous visibility of satellite from distant (inter-continental) ground stations for several hours

To satisfy the primary objective the orbit must be highly elliptic The perigeeshould be low A very low perigee means a short contact time which decreasesthe accuracy of the frequency measurement A compromise is a perigeealtitude of sim2500 km For a sufficiently high apogee this yields a U of 60of the theoretical maximum between the earth surface and infinity

The baseline scenario is an orbit of approx 12 h period with the followingproperties

Apogee Altitude 37856 km (minus 10 deg latitude one over Atlantic andPacific resp)

Perigee Altitude 2500 km (+ 10 deg latitude one over Guatemala andMalaysia resp)

Inclination 634 argument of perigee 170RAAN 25 True Anomaly 220Apogee drift rate 70 kmday perigee drift rate sim 14 kmday

True anomaly and right ascension of the ascending node (RAAN) wereoptimized to minimize orbital precession Figure 3 shows the ground track andthe ground coverage at one of the perigees Common-view contact time forstations in Europe is about 6 h while between a station in Europe and in theUSA it is about 35 h Frequency comparison windows between a given groundstation and EGE repeat every 24 h

A final selection of the orbit requires an in-depth study of the costs of thepropellant and maneuvers as well as of the setting up of new ground stations

32 Ground segment

The mission requires two separate ground segments the science groundsegment and the standard satellite control segment which supports telemetryand telecommand (TMTC)

Exp Astron

Fig 3 Baseline EGE orbit One (out of two) apogee and one (out of two) perigee ground pointsare shown The ground coverage at one of the perigees is shown with circles at elevation anglesteps of 10 starting from 10

321 Science ground segment

As described above one part of the mission is to perform frequency com-parisons between ground-to-space and ground-to-ground (common-view andnon-common view) The use of a two-way microwave link (MWL) systemhas the advantage that a precise orbit or the precise relative velocity of thesatellite must not be known [4] The orbit determination requirements fora two-way system are significantly relaxed compared to the requirements tomodel the relativistic effects on the clocks The frequency comparison resultsare available both on ground and on the satellite

For ground-to-satellite clock comparisons five ground stations arefavourable two capable of viewing perigee three for viewing apogee Thisallows for the classic three-clock configuration where the performance of eachclock can be obtained individually The perigee stations will be observing thesatellite for sim25 min every perigee passage for the duration of the missionThe apogee stations will observe for sim3 h every apogee passage Of coursea larger number of stations as well as longer observation time are usefulfor robustness redundancy and orbit determination The stations could beinstalled on a temporary basis for the duration of the mission

The stations must have access to optical clocks of performance at least ashigh as todayrsquos state-of-the-art ground clocks but preferably higher Theywill include an optical frequency comb by means of which the optical clockfrequency is transformed to a radio frequency for comparison with the EGEclocks High-accuracy fiber-optic link technology already existing today allowsdistances of sim100 km between the ground station and the correspondingoptical clock allowing a certain freedom in the choice of the clock locationWith further increase of the fiber-optic link range currently being investigated

Exp Astron

the area over which the ground station can be positioned relative to the opticalclock will increase This option is of importance for the implementation of ageophysics measurement campaign for which a large number of clock locationsis of interest

The above ground stations will be complemented by some 15 additionalground terminals provided by the world-wide laboratories equipped with highperformance primary clocks at the time of the mission These will participate inorbit determination to 1 cm accuracy along the full orbit which can be achievedwith non-optical atomic clocks The EGE MWL is fully compatible to plannedACES ground terminals which might reduce the number of required stations

With a larger number of transportable optical clock ground stations de-ployed and moved across the continents a geophysics campaign could beimplemented to map the gravitational potential

Ground station equipment MWL uses dedicated small transportable termi-nals which can be easily installed at any location of interest Each terminalconsists of a steerable dish antenna of 1 m diameter with pointing accuracy of01 using a transmitted power of 5 W per band The high directivity antenna(beam aperture Ka-band 07 Ku-band 14 S-band 20) avoids multi-pathfrom nearby objects Antenna noise temperatures are Ka-band 450 K Ku-band 340 K Operation is at elevations above 10 and measurements areperformed above 20 For ranging purposes terminal location is determined byGPS surveying Each ground station computes its own ionospheric correctionsbased on a triple-band receiver The terminal has a built-in delay monitorwhich allows calibrated ranging during several months once its bias has beendetermined by laser ranging Its M and C interface is via local area networkto either a userrsquos computer or directly to the science network control centreby remote MampC The stationrsquos operation is fully automatic and unattendedSchedule and orbital data are received via LAN once per day Signal acquisi-tion is fully autonomous Data transmission is after each pass The terminal hasan uplink telecommand capability of 1 kBits and a downlink data capacity ofup to 10 kBitss For the local user there is a visual interface providing quick-look data from real-time comparison to the spacecraft clocks to verify theirgood health The final full performance product is calculated at the sciencedata centre

Local metrological data (temperature pressure humidity) are recorded tocorrect for tropospheric delay of the ranging data The main advantage ofMWL terminals compared to laser based systems is their all-weather capabil-ity although microwaves suffer higher propagation errors in the ionosphereand troposphere

The EGE ground terminal will profit from the ERS-2 mission PRARE(precise range and range rate equipment) [45] experiments whose terminalshad similar design and have been operated under extreme climatic conditionsincl Antarctica and high temperature regions from existing designs fromthe space segment (mission ACES) and no further special developments aredeemed necessary Cost per EGE master station is modest

Exp Astron

Satellite tracking via laser ranging Laser ranging is alternative to MWL rang-ing and also serves for absolute calibration of the latter For this purpose anon-board corner cube reflector (CCR) is provided Ranging will be done withthe International Laser Ranging Service Ideally the satellite will be rangedat every perigee passage and every apogee passage simultaneously with thefrequency comparison procedures Todayrsquos ranging precision is sufficient tosatisfy the science goals For apogee observations in the northern hemispherea large number of laser ranging stations is available The number of stationscapable of ranging at or near perigee in the southern hemisphere is six andthus is sufficient even considering unfavourable weather conditions at someof these The MWL ground stations are able to range even under thoseconditions

322 Control ground segment

Average data transfer rate is estimated at 300 MBday (approx 40 kBs onaverage) Data will be stored on board (2 GB capacity required) and will betransferred to a receiving station on the Earth once a day

XPLC

incl Data StorageIF to TMTC Link

HK amp TMTC

to ground

Opt Clock

AOM

MOLO

PHARAO

FCDP

10 GHz

100 MHz

250 MHz

GNSS Rx

MWLTampF Link

uplink Data

UTC(k)

TampF Space vs Ground

TM from GS

Tim

e S

cale

Orbit GNSS Time UTC

TampF Space vs Ground

MampC Time Scale

MampC

MampC

MampC

MampC

Mamp

C

PHARAO vs Opt Clock

PHARAO vs Opt Clock

10 GHz amp Epoch

10 MHz

amp Epoch

Epoch

10 MHz

amp Epoch

Opt Link

Fig 4 Schematic of the payload showing subunits and their signals The results of the frequencycomparison between the two on-board clocks (optical clock and microwave clock PHARAO) areobtained from a combination of the frequency values applied to the frequency shifters AOM (seeFigs 7 and 8) and those output from the frequency control and distribution package (FCDP)as indicated by the arrows ldquoPHARAO vs Opt Clockrdquo The satellite clock to ground clockcomparison results are output from the MWL XPLC is the on-board computer and is involvedin generation of signals for instrument control The optical link shown dashed is a possible add-onproviding an alternative to the MWL GS ground stations MampC monitoring and control HKhousekeeping TampF time and frequency

Exp Astron

4 Main payload instruments

41 Overview

The satellite payload consists of a pair of cold-atom clocks a microwave-optical local oscillator (MOLO) an on-board frequency distribution unit(FCDP) a satellite-to-ground clock comparison unit (MWL) and the onboardcomputercontrol system (XPLC) see Fig 4 Auxiliary units are the posi-tion determination GPSGalileo receiver (GNSS Rx) attitude monitor (startracker not shown) and the corner cube reflector array (CCR not shown)

The science goals require clocks of very high frequency stability on thetimescales of half the orbital period (6 h) as well as perigee passage (sim1000 s)see Fig 2 Only clocks based on cold atoms are suited for the science goalsTable 1 reports values for a subset of atomic clocks that have been investigatednot including clocks that require cryogenic operation or are too complex foruse in an medium-class mission The choice of a clock pair rather than a singleone fulfils several requirements it addresses more science goals improves thevalidity of some science results and it enables calibration accuracy evaluationand redundancy The validity of the science results is enhanced since theclocks differ not only in the employed atoms but also in terms of havingdifferent sensitivity to perturbations by electric and magnetic fields electronicserrors and misalignments A comparison of the results arising from each clockallows for an understanding of measurement errors Both the clocksrsquo and thefrequency comparison unitrsquos performances will be determined after launchand regularly verified during the course of the mission The only reliable

Table 1 Properties of different clock candidates for EGE

Clock Clock Fractional σ = Fractional Sensitivity Figure of Sensitivityatom frequency inaccuracy instability to α Aα merit |Aα| to meq

at 1000 s σ [1016] QCD

Sr+ 445 THz 3 times 10minus17 (p) 5 times 10minus17 (p) 042 08 (p) ndashYb+ (1) 688 THz 3 times 10minus16 (d) 3 times 10minus16 (d) 088 029 (d) ndash

3 times 10minus17 (p) 1 (p)Yb+ (2) 642 THz lt 1 times 10minus17 (p) 1 times 10minus17 (p) minus53 53 (p) ndashSr 435 THz 1 times 10minus16 (d) 1 times 10minus16 (d) 009 009 (d) ndash

lt 1 times 10minus17 (p) 09 (p)Yb 519 THz 7 times 10minus16 (d) 1 times 10minus16 (d) 031 03 (d) ndash

lt 1 times 10minus17 (p) 1 times 10minus17 (p) 31 (p)Cs 91 GHz 2 times 10minus16 (d) 1 times 10minus15 (d) 275 027 (d) 1Yb+ 126 GHz 1 times 10minus15 (p) 1 times 10minus15 (p) 34 034 (p) 1

A relevant figure of merit is the ratio of |AX | and fractional frequency instability σ on the perigeevisibility timescale The Cs clock is the only high performance clock with adequate sensitivity tothe particle masses me mq The indicated Cs clock performance was achieved using a microwavelocal oscillator based on a cryogenic oscillator On EGE the latter is to be replaced by an opticallocal oscillator-based device (MOLO) with similar short-term performance but without the needto operate at cryogenic temperature The clock instability at 10000 s is typically similar to theindicated inaccuracy(p) projected (d) demonstrated (1) quadropole (2) octupole

Exp Astron

way to achieve this is by having at least two clocks on board The on-boardclock-clock comparison provides performance evaluation independent of theground-satellite comparison unit Once the satellite clock performances areestablished the link performance can be established as well

The clocks are similar in basic structure but differ in the detailed implemen-tation They consist of

a source of atomsa preparation subunit which cools the atoms to near-standstill (millikelvintemperature or below)a trapping subunit that confines the atom(s) to a tiny volumean interrogation subunita clock control electronics package

The clocks share a common central unit the optical-microwave local oscil-lator which serves as a flywheel oscillator with outstanding stability on shorttimescales (01ndash10 s) for both clocks It consists of a laser stabilized to an ultra-low loss optical resonator made of ultra-low-expansion glass A microwavesignal is derived from the laser by phase-locking a frequency comb to the laserwave and using a multiple of the repetition rate

While the subunits of any optical clock have a similar technological basis(optics electrooptics acoustooptics thermal control vacuum systems) theydiffer in their specifications Depending on the choice of the atom species de-livery confinement and interrogation methods clocks of different perfor-mance result A number of laboratory optical clock demonstrators have beendeveloped worldwide This ensures availability of the know-how for the indus-trial implementation of the flight models

The first clock is a microwave clock based on the interrogation of slowlymoving ensembles of cold Cs atoms It is essentially based on the PHARAOclock [5 6] developed to engineering model level by CNES in the frame-work of the ESA ISS mission ACES In EGE this clock is upgraded usingthe microwave-optical local oscillator This improves significantly the clockperformance The specification is a relative instability (Allan deviation) of3 times 10minus14 (τ s)minus12 ie at the integration time τ = 1000 s it drops to 1 times 10minus15and to 3 times 10minus16 at 10000 s While this specification is at least a factor 3 lessstringent than for the optical clocks the significantly lower cost of obtaining aflight model and the sensitivity of its frequency to the particle masses makes ita suitable choice

The optical clock baseline instrument is a single-ion optical clock Amongvarious suitable ion species (Hg+ Al+ Sr+ Yb+) [44ndash47] the Yb+ ion hasbeen selected Its breadboard performance is established at the instability levelof 3 times 10minus16 over 1000 s it exhibits a large sensitivity coefficient Aα hassmall sensitivity to magnetic fields and exhibits significant potential for furtherimprovements (use of the octupole transition with sim6 times larger Aα andlower overall instability [48] see below) A trapped Yb+ ion is also a veryrobust system it has been demonstrated that a single Yb+ ion can remaintrapped uninterruptedly for many months Finally Yb+ is the only ion with

Exp Astron

which optical clocks demonstrators have been operated in not a single but twomajor metrology laboratories

Another attractive optical clock is based on ensembles of cold neutral Sror Yb atoms trapped in an optical lattice produced by laser beams This clocktype has a lower instability than a single-ion clock thanks to the large numberof atoms (sim105) used Its development is in progress in several laboratoriesin the world (in part supported by space agencies) and has to date reached6 times 10minus17 instability at 5000 s and 1 times 10minus16 uncertainty [49 50] with potentialfor even higher performance However it is more complex and costly thanthe Yb+ ion clock since the cold atom preparation and trapping subunitsinclude more components require higher laser powers and the respectiveoperating procedures are more complex It is therefore not considered for thepresent mission designed to be developed in the very near future with limitedresources but is an interesting candidate for missions without such constraintsIts estimated physical parameters are indicated in Table 2 for comparison

42 Main instruments

421 Microwave clock

The PHARAO cold atom clock is one of the two atomic clocks of the spacemission ACES managed by ESA It is proposed here to include a secondversion of this instrument as its costperformance ratio is very attractive forthe EGE scientific objectives Its concept is very similar to ground basedatomic fountains but with the major difference of zero-g operation Atomsslowly launched in free flight cross two microwave fields tuned to the transitionbetween the two hyperfine levels of the cesium ground state The interrogationmethod based on two separate oscillating fields (Ramsey scheme) allowsthe detection of an atomic line whose width is inversely proportional to thetransit time between the two microwave cavities The resonant microwavefield at 9192631770 GHz (SI definition of the second) is synthesized startingfrom an ultrastable quartz oscillator (USO) in the FCDP phaselocked to themicrowave output of the microwave-optical local oscillator and stabilized tothe clock line using the error signal generated by the cesium resonator In thisway the intrinsic qualities of the cesium hyperfine transition both in terms ofaccuracy and frequency stability are transferred to the USO In a microgravityenvironment the velocity of atoms along the ballistic trajectories is constantand can be changed continuously over almost two orders of magnitude (5ndash500 cms) Differently from atomic fountain clocks presently operated onground very long interaction times (up to few seconds) will be possible whilekeeping reasonable the size of the instrument

The instrument consists of 4 main elements a laser source which providesthe light required to cool and detect the cesium atoms (Fig 5 left) a frequencychain which generates the 92 GHz microwave signal a cesium vacuum tubewhere the interaction between the microwave field and the atoms occurs

Exp Astron

Table 2 Resources and technology readiness level (TRL) of the EGE spacecraft

Unit Volume Mass Power TRL Comment[l] [kg] [W]

Ion clock 100 70 60Atom source 4Atom preparation 6Atom trap 4Atom interrogation 6Clock control 6

Microwave clock 200 91 110 8 TRL by end 2008Local oscillator

Stabilized laser 18 15 23 6Frequency comb 20 16 38 6

Microwave link 18 18 55 8FCDP 7 8 12 8GPS receiver 2 5 7 9CCR 2 2 0 9Payload sum 367 225 305Thermal control 14 17 30 8OBDH 20 24 38 9Star tracker 1 1 8 9Radiation shield 10 40 0 9Support structure 170 204 0 9Power incl solar arr 50 58 10 9Propulsion 21 25 20 9AOCS 36 43 82 9TMTC 8 10 15 9Satellite bus sum 330 422 203 Flexbus conceptPayload plus bus incl 20 margin 8364 7764 6096Propellant 40Launch adapter 100SC wet masspower incl Margins 9164Lattice clock 135 105 152 Alternative clock

Atom source 4Atom preparation 6Atom trap 4Atom interrogation 6Clock control 6

Also included are the resources for an optical lattice clock based on neutral Strontium atoms

(Fig 5 right) and a computer The engineering model of PHARAO is fullyassembled at CNES Toulouse and undergoes functional and performancetests until the end of 2008 The operating temperature range is 10ndash33 andnon-operating temperatures are minus45 +60 The cesium tube consists in aUHV chamber pumped by getters and a 3 ls ion pump a cesium reservoira microwave cavity coils to provide a uniform magnetic field and three layersof magnetic shields Vacuum windows with fiber-optics collimators enable thetransmission of the cooling and detection light onto the cesium atoms

In autonomous mode PHARAO uses the USO as interrogation oscillatorand will provide a clock signal with fractional frequency stability below1 times 10minus13 (τs) minus12 and inaccuracy near 1 times 10minus16 When using the microwave-

Exp Astron

Fig 5 Left The PHARAO laser source engineering model with cover removed Dimensionsare 530 times 350 times 150 mm3 and the mass is 200 kg Ten polarization maintaining optical fibers(yellow) guide the laser beams to the cesium tube All diode lasers are mounted on Peltier coolersfor temperature regulation within 2 mK (courtesy EADS SODERN) Right The PHARAOengineering model cesium tube without the two external magnetic shields The volume is 990 times336 times 444 mm3 and the total mass is 44 kg

optical local oscillator PHARAO will provide a reduced frequency instabilityof 3 times 10minus14 (τs) minus12 This corresponds to 1 times 10minus15 at 1000 s and 3 times 10minus16

at 104 s

422 Ion optical clock

The 171Yb+ ion clock shown in Figs 6 and 7 provides an atomic frequencyreference based on a single 171Yb+ ion confined within an RF electric end-captrap (similar to a Paul trap) within an ultra high vacuum (UHV) chamber andlaser cooled to sim1 mK close to the Doppler cooling limit As a result the ionrsquosfirst order Doppler motion is completely removed and the ion experienceslittle collisional perturbation from its environment Under these conditionsthe narrow linewidth 2S12 (F = 0 mF = 0) minus 2 D32 (F = 2 mF = 0) quadrupole

Fig 6 Left Simplified level scheme of 171Yb+ showing relevant cooling (blue) auxiliary (red)quadrupole clock (green) and octupole clock (dark blue) transitions δνnat is the natural linewidthof the clock transition Middle close-up of Paul-type RF trap (PTB Braunschweig) the ion istrapped in the hole (diameter 14 mm) Right Complete ion trap with vacuum flange height 10 cm(NPL Teddington)

Exp Astron

935 nm D32-state repumper laser

638 nm F72-state repumper laser

739 nm extended cavity diode laser

AOM

PMT

Ion trap UHV chamber

vacuum pumps

Fibre-free space interface

Mu-metal shielding

EC

T

B

Magnetic field

Temperature

Ion fluorescence Processor

Clock laser frequency step and control

4355 nm

Frequengy doubling II

3695 nm cooling wave

Low-drift cavity

Low-drift cavity 171Yb+ 435 nm ion clock unit

876 nm clock laser wave from MOLO

Frequengy doubling

Fig 7 Simplified schematic of the single-ion optical clock unit AOM acousto-optic frequencyshifter PMT photomultiplier tube EC endcap T temperature stabilization system B magneticfield

clock transition at 4355 nm can be interrogated by spectrally narrow and stableclock laser light A diagram of the simplified energy level scheme for 171Yb+showing the relevant cooling auxiliary and clock transitions is given in Fig 6All these wavelengths are provided by diode lasers operated in fundamentalor frequency doubled mode The clock laser light is obtained from the 871 nmwave provided by the MOLO unit (see Section 423) after frequency-doublingin single pass by a KNbO3 nonlinear-optical crystal to 4355 nm The clocktransition spectral profile is observed by stepping the clock laser frequency bymeans of an acousto-optic frequency shifter (AOM in Fig 7) and recordingthe statistics of the quantum jumps in cooling laser induced fluorescence asa function of this frequency Under clock operation the AOM repeatedlysteps back and forward between the half-intensity points on the transitionprofile monitoring the signal imbalance between these points Any detectedimbalance is servo-corrected to zero by feedback to the AOM

In detail the optical clock architecture will comprise (Fig 7)

bull an RF end-cap trap for ionising and confining a single 171Yb+ ion withinan ultra-high vacuum chamber pumped by a small ion pump and non-evaporable getter pump The trap is driven by means of an ac voltage ofa few hundred volts at a drive frequency of sim10 MHz A small oven withseveral mg of 171Yb isotope is heated to provide a low flux of Yb atomsfrom which single ions can be ionised by electron bombardment from ahot wire filament within the trap potential well Low voltage dc is applied

Exp Astron

to additional electrodes to position the ion precisely at trap potentialcentre in order to minimise the ionrsquos micromotion at the drive frequencyThis needs automatic periodic monitoring and correction especially in theperiod after re-loading an ion into the trap External magnetic field coilsin three orthogonal axes allow the nulling of external fields and settingof a fixed field of sim1 μT The trap and coils are surrounded by mu-metalshielding to minimise external field changes during the orbit A level oftemperature control of 1 K of the region surrounding the trap is requiredfor maintaining the blackbody-induced frequency shift uncertainty withtemperature at the 10minus17 level

bull a laser platform to provide frequency doubling of an extended cavity739 nm diode laser for Doppler cooling of the ion on the 2S12(F = 1) minus2 P12 (F = 0) dipole transition at 3695 nm Light from an amplified diodelaser device frequency-doubled in a periodically poled LiTaO3 crystal insingle-pass will be used Typical laser powers used for driving the coolingtransition below saturation are about 2 μW for a beam waist in the trapof about 50 μm By modulation of the injection current of the 739 nmdiode laser at 147 GHz a sideband is generated that excites the 2S12 (F =0) rarr2 P12 (F = 1) ldquorepumperrdquo transition in order to avoid optical pump-ing between the ground hyperfine states

bull The platform also houses the auxiliary lasers at 935 and 638 nm The935 nm diode provides repumping of the ion from the 2 D32(F = 1)metastable level after occasional branching decays to this level during thecooling sequence The 638 nm diode allows fast recovery of the ion fromthe very-long-lived 2F72 metastable state after very occasional collisionaldecay to that state Currently extended cavity lasers are forseen butdistributed-feedback lasers may be available in the near future

bull a high NA lens imaging system and photomultiplier detection system torecord the statistics of 369 nm fluorescence quantum jumps as a functionof clock laser frequency step

bull a fibre system to deliver the various cooling auxiliary and clock light fromsource to trap making use of achromatic doublets where necessary at thefibre-free space interface for launching into the trap

bull a monitoring and control processor which provides the primary coolingand clock laser pulse magnetic field and detection sequencing to observeand lock to the ion clock transition frequency The processor also monitorsfrequency and amplitude data necessary to determine normal laser andion operational conditions and initiate resetting and recovery algorithmswhere necessary and laser unit failure The actual magnetic field presenton the atomic volume is determined and actively stabilized so that aninaccuracy at the level of 1 times 10minus17 can be achieved

bull a redundancy level of at least two units for cooling clock and repumperlasers plus similar for frequency doubling crystals All redundancy unitsfor each wavelength will be fibre multiplexed as standard allowing redun-dant unit activation on determination of prior unit failure

Exp Astron

For an optimized quantum-projection-noise-limited performance of thesingle-ion clock Ramsey interrogation with a cycle time equal to the lifetimeof the metastable D32 level of 50 ms will be performed leading to an Allandeviation of about 27 times 10minus15 (τ s)minus12 This is sim45 times 10minus17 at 1 h satisfyingrequired stabilities for the science objectives Current laboratory results are astability of 1 times 10minus15 at 30 s averaging down to 3 times 10minus16 at 1000 s (PTBGermany) Also comparisons between two independent traps showed agree-ment at the level of several parts in 10minus16 over a series of eight measurementruns

The 171Yb+ ion clock offers a second option namely the 2S12ndash2 F72 octu-pole transition at 476 nm (see Table 1 and Fig 6) [48] It has an extremelylong upper state lifetime low electric quadrupole and second order Zeemansystematic shifts and therefore is a candidate for an even higher performanceclock if a clock laser system with lower instability than the above is providedApart from the clock laser it shares the laser system of the quadrupoleclock Thus with moderate additional resources the Yb+ ion apparatus couldprovide two optical clocks in a single package and specifically perform directfrequency comparisons between the quadrupole and the octupole transitionwith a large sensitivity Aα

The ion clock will operate under microprocessor control providing thecontrol algorithms for the integrated pulse sequencing for cooling repumpingclock interrogation magnetic field switching cooling fluorescence monitoringThe controller will also provide error flags for critical processes and systemreset and correction where necessary This will include ion re-loading andmicromotion reduction algorithms Automatic monitoring of laser power andspectral quality with signal re-optimisation or laser failure determination isrequired Redundant laser units will be pre-aligned and multiplexed into thefibre delivery to the trap so that redundant units can be readily activated in thecase of diode laser failure

423 Microwave-optical local oscillator (MOLO)

This oscillator provides two ultrastable frequency outputs one in the opticalregion the other in the microwave region the two frequencies being coher-ently related The MOLO is composed of three subunits the clock laser thereference cavity and the frequency comb see Fig 8

The clock laser subunit delivers the light for the excitation of the opticalclock transition It is generated by an extended-cavity diode laser at a wave-length of 871 nm Most of the output wave of this laser is sent to the ionclock The laser subunit containing two lasers for redundancy will require3 kg 3 l 15 W The frequency instability of the clock laser must be superiorto the minimum clock instability 27 times 10minus15 (τ s)minus12 for τ up to the timeconstant (sim10 s) of the servo system that locks the laser to the ionrsquos resonancesignal In order to achieve this the laser frequency is stabilized to the reference

Exp Astron

871 nm clock laser

EO

Detector and servo

AOM Femtosecond

frequency comb

To optical clock 10 GHz To FCDP

Microwave-optical local oscillator

ULE cavity

Fig 8 Microwave-optical local oscillator (MOLO) consisting of clock laser reference ULE cavityand femtosecond frequency comb (OFC) Radiofrequency inputs and outputs are not shown EOelectro-optic modulator ULE ultra-low expansion AOM frequency shifter

cavity where the PoundndashDreverndashHall method is used About 10 μW of thelaser output are used for this

The system will exhibit sim04 Hz (sim1 times 10minus15 relative) linewidth and afrequency drift lt01 Hzs The cavity consists of two high-reflectivity mirrorsoptically contacted to a cylinder of diameter 70 mm length 150 mm made ofULE (a glass exhibiting ultra low thermal expansion coefficient) with a cavityfinesse of about 300000 Residual vibrations of the satellite (arms lt 1 times 10minus6 g)should not be a limiting factor to the desired linewidth for safety marginan optimized cavity support can reduce the sensitivity to accelerations Foracoustic and thermal isolation the cavity resides inside an aluminium vacuumchamber equipped with a small (3 ls) ion-getter pump Two stages of polishedaluminium shields are implemented around the cavity which are activelytemperature stabilized so that the cavity is precisely kept at the zero crossingtemperature of the ULE thermal expansion coefficient Heat application andremoval is by thermoelectric elements between the shield and the vacuumchamber Additionally the entire vacuum chamber is temperature-stabilized

During transportation and launch the cavity will be rigidly fixed insidethe vacuum chamber by pressing on it from the sides by piezo-mechanicalactuators allowing accelerations of several g After bringing the satellite onthe orbit the actuators will be released

The optical setup for frequency stabilization will be a classic Pound-Drever-Hall scheme in reflection using an electro-optic modulator (EO) Couplingof the clock laser radiation to the cavity is via single-mode polarization main-taining optical fiber The fiber in- and out-couplers are mounted on miniaturepiezo-motor driven multiaxis translation stages Microprocessor control ofthe translation stages performs a laser-to-fiber incoupling stabilization andlaser-to-cavity mode-matching optimization (positioning inaccuracy lt10 μmalignment inaccuracy lt10 μrad)

The frequency stability of the clock laser will be limited by thermal 1f noisein the mirrors with an fractional Allan deviation slightly below 1 times 10minus15 from1 s to a few 10 s Resources including control electronics are 15 l 12 kg 8 W

Exp Astron

A femtosecond fiber laser based optical frequency comb generator (OFCFig 8) will be used for the clock frequency comparisons and to obtain aspectrally ultrapure microwave local oscillator output for the operation ofPHARAO for the microwave link and for the GNSS receiver The fem-tosecond laser is based on erbium-doped fiber pumped by fiber-coupled laserdiodes It emits 100 fs pulses at a repetition rate of 250 MHz with an averagepower of 100 mW The emission spectrum is centered around 1560 nm Spec-tral broadening to more than one octave is obtained via nonlinear interactionin a microstructure fibre so that the range from about 1050 to 2100 nm iscovered

Second harmonic generation of a portion of the broadened spectrum allowsobtaining the required light for the measurement of the clock laser frequencyat 871 nm The OFC will be referenced to the clock laser frequency (afterintroduction a controllable shift via the frequency shifter AOM) by makingthe carrier envelope offset frequency vanish so that the repetition rate fr isa high-order subharmonic of the optical clock reference frequency f1 = m1 frThe link between the optical and the microwave domain is then simply thedistribution of the repetition rate frequency or more conveniently a high(order 40) harmonic of the comb repetition rate 10 GHz

It has been demonstrated by several groups that frequency comparison andfrequency division with an OFC can be performed to very high accuracy andstability beyond the limitations imposed by the optical clocks themselves Forexample a fiber laser OFC with an instability (Allan deviation) of 6 times 10minus17

in 1 s has been demonstrated [51] OFCs based on fiber technology possessgood long-term reliability and high power efficiency The projection for a spacequalified OFC based on current industrial terrestrial technology is 16 kg mass20 l volume and 38 W power requirement

424 Frequency comparison and distribution package (FCDP)

This package distributes the ultra-stable 10 GHz reference frequency outputof the MOLO to the MWL performs frequency comparisons between MOLOand PHARAO and generates a variety of different reference signals for theindividual instruments including the 250 MHz reference clock to the OFCAll signals are coherent to the MOLO In the FCDP a USO is locked to theMOLO signal by a digital PLL A divider chain generates timing pulses (1 pps1000 pps etc) and keeps a continuous local time scale From the digital phasecomparison data the phase noise density spectrum between the USO and theMOLO is available by telemetry (TM) to assess the overall frequency purityin-flight

As the FCDP distributes the full performance reference from the MOLOdirectly to the MWL there is minimum impact of the FCDP on overall clockcomparison performance The USO-phase lock loop uses a similar designas for ACES-FCDP The digital PLL with linear phase detector allows theUSO to be slightly offset from its reference which is selectable by groundcommand This results in an accurate on-board time scale despite any offset

Exp Astron

of the MOLO reference frequency which can be synchronised to UTC Thefrequency synthesis chain re-uses modules developed for ACES-FCDP andtheir frequency inaccuracy Expected relative uncertainty (Allan deviation)of the frequency comparison between two signals provided to the FCDP is47 times 10minus17 at 1000 s 95 times 10minus18 at 10000 s

During non-availability of the MOLO ie during system start-up andcontingency conditions the USO runs as flywheel (ie it is unlocked withinstability lt15 times 10minus13 from 1 to 100 s) It will still provide useable butless accurate references to the instruments including the reference to OFCCommon view ground-to-ground time- and frequency comparison and supplyof the remaining instruments remain possible at the USO provided accuracy

Mass volume and power consumption are 8 kg 7 l 12 W Controldata rateis 10 kBday 50 kBday (FCDP alone)

425 Microwave link (MWL)

The microwave link is a two-way triple-frequency microwave link (Fig 9) Itis designed to perform accurate frequency comparison as well as time-transfer

Ku-Band Up-linkPower Tx 5 WCarrier 13475 GHzPN-Code 100 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

S-Band Down-linkPower Tx 1 WCarrier 2248 MHzPN-Code 1 MChips1pps 1 time marker sData 5 kBits

Ku-Band Down-linkPower Tx 1 WCarrier 1470333 GHzPN-Code 100 MChips1pps 1 time marker sData 5 kBits

EGE SC

S

Ku

KaKa-Band Up-linkPower Tx 5 WCarrier 345 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

Ka-Band Down-LinkPower Tx 1 WCarrier 32 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 5 kBits

GS Antenna diameterEGE terminal 1 mACES terminal 06 m

Fig 9 EGE microwave two-way link concept and characteristics See text for a description Thelink will be able to perform frequency comparisons with up to ten ground stations simultaneously(only one active comparison is shown) pps pulse per second Tx transmitter SC spacecraft

Exp Astron

between on-board and on-ground clocks Two-way operation allows for com-pensation of the effects due to spacecraft location and its motion relative toground Its two bi-directional links operate in Ka-band (32345 GHz) andKu-band (147135 GHz) The high carrier frequency of the up- and down-links allows for a noticeable reduction of the ionospheric delay A third down-link channel is in S-band (22 GHz) and is used to determine the ionospheretotal electron content MWL uses a combination of carrier phase and codephase data to measure with high precision the absolute ionospheric delayas well as ionospheric delay variations during a ground contact The useof three frequency bands allows to resolve third order ionospheric termsKey modulation parameters are Ka band Tone 250 MHz Ku band PN100 MChips S-band 1 MChips

The EGE-MWL concept is a considerably improved version of the ACES-MWL Microwave carrier generation is directly from the 10 GHz microwaveoutput available from the MOLO and having exceptionally low close-in phasenoise without intermediate frequency dividers or other complex synthesiserstages commonly used in RF transponders The high frequency carriers arekey to achieve the desired frequency comparison accuracy

The Ka-band carrier is generated by tripling the FCDP-generated signalThe Ku-band is subsequently generated by division by two using a verystable regenerative divider Finally the assigned radio frequency is obtainedby adding a low frequency signal using an image reject mixer A mixedanalog-digital receiver design will enhance its performance by use of a flexibledigital signal processing and digital control loops The number of simultaneousground contacts is increased from four (as in the ACES mission) to ten or moreusing the digital concept while the number of physical receiver channels perband is reduced from four to two one for operation to ground and one for in-flight calibration The concept allows adaptation of receiver loop parametersto optimise for the different portions of a highly elliptic orbit while reducingsystematic uncertainties due to high Doppler and Doppler rate Real-timedata exchange to ground terminals is by additional data modulation on all RFsignals The spacecraft carries two different small antennae for perigee and forapogee with gains of 8 dBi 20 dBi beamwidths 100 (full earth coverage) 14respectively power is 1 W for each

The MWL end-to-end performance between the attached clocks includingspace- and ground terminals is predicted to be a relative Allan deviation of1 times 10minus16 at 1000 s (perigee) and 1 times 10minus17 at 10000 s (apogee) Thesevalues are based on conservative implementation losses and other instrumentallosses assumed as 6 dB at apogee and 10 dB at perigee including additionallosses due to atmosphere An attitude knowledge of 1 is required

The MWL link will also support precise orbitographyndashcomplementary tothe GNSS receiver Generation of ranging data is simultaneous to two-waytime transfer and is available under all weather conditions Orbit determina-tion accuracy is expected to be better than 1 cm well matching EGE clockrequirements The value is expected based on demonstrated results fromother missions (METEOR ERS-2 TOPEXPOSEIDON) Laser ranging to

Exp Astron

the CCR array is used a few times during the mission to calibrate the MWLranging

In case of non-availability of the ultrapure signal from MOLO an additional10 GHz output is available from the FCDP to support ground-to-ground clockcomparison in common-view mode even after the nominal mission periodwhen the neither the clocks nor MOLO may be available any longer Basedon the PRARE experimentrsquos 12 years of operation this can be assumed forthe FCDPMWL combination as well The potential extended operational pe-riod makes the proposed configuration particularly attractive to metrologicallaboratories that rely on availability of long-lasting data in support of thegeneration of UTC

The resources of MWL are as follows mass 18 kg volume 18 l power55 W TC 2 kBday TM 120 MBday

426 Auxiliary units

The star camera assembly on the EGE spacecraft will provide real-timeattitude data primarily for the orientation of the spacecraft (with accuracy of1 in real-time) and the proper reduction of the GNSSSLR data

A GNSS receiver is used for the precise orbit determination time andfrequency transfer coarse positioning (lt1 m) for real time use by the Attitudeand Orbit Control as well as the back-up for time tagging of all payloaddata For the time and frequency transfer the receiver is nominally using theonboard frequency reference The receiver assembly includes three specialantennas By the year 2015 more than 100 GNSS satellites are expected tobe available for precise orbit determination permitting orbit inaccuracy of 1ndash2 cm RMS for altitudes up to approx 10000 km At apogee a combinationof GNSS signals and MWL ranging will be used with expected inaccuracyof 10 cm

5 Basic spacecraft specifications

The mission foresees a single spacecraft with a mission duration limited toapprox 3 years for cost reasons Attitude control along three axes withless than 1 error on a timescale of 1000 s is required The control can beperformed with cold gas thrusters magnetotorquers or reaction wheels Theselection will be done on the basis of the vibrations induced by these systems(desired level arms lt 10minus6 g)

Cold gas thrusters or monopropellant thrusters will be required to maintainthe orbit parameters during the mission by small orbit correction manoeuvresand will also guarantee a proper tuning of the orbit parameters after the releasefrom the launcher The ground segment (telemetry control and data transfer)could be via non-ESA stations user-operated to save costs

Further resources are indicated in Table 2

Exp Astron

The satellite bus design drivers are nadir-pointing microwave antennasthermal stability for the payload instruments no strong magnetic sources lowmicrovibrations harsh radiation environment due to crossing of the radiationbelts once per orbit power demand by science payload (with the possibility toreduce power demand by stand-by of some components during specific missionphases eg eclipse) mass limit due to required high altitude with low costlaunch

The satellite bus design can be based on established platforms such asFLEXBUS or PRIMA used on several past missions Their design philosophydiffers one having body-mounted solar arrays the other wings The latter canprovide larger amounts of power The vibration level induced by the panelswhen they are fixed and the duration of solar array drive periods per orbitwill have to be considered in the selection The FLEXBUS concept was usedfor a mission already requiring a low level of residual accelerations and theplatform is optimised for this parameter

For most of the satellite components one can consider using compo-nents and electronic units from navigation or communication satellites whichtypically cruise at altitudes between 15000 and 25000 km altitude LISAPathfinder which will cross the radiation belts during spiralling to theLagrange point will face similar problems concerning radiation as EGE as wellas XMM which is in a Molniya-type orbit with similar radiation characteristics

The instruments will therefore be located in the satellite structure such asto realise ionizing radiation shielding by the spacecraft structure and non-sensitive equipment to the maximum extent possible the need for additionalradiation shielding (including trapped proton radiation) for especially sensitiveparts of the instruments must be determined by ground tests It should bementioned that an orbit with higher perigee and less radiation dose is also anoption as it leads to a tolerable reduction of the science goals

The required plusmn3C temperature stability at the interface to the payloadenclosure is feasible The thermal control will employ the usual means (anenclosure with high thermal inertia heaters and a suitable radiator) Thelocation of the radiator will need a special study given the non-standard orbit

Magnetic field present during the mission can affect the properties of theonboard clock Using SPENVIS the magnetic field has been calculated alongthe spacecraft orbit the field intensity presents maxima around 02 G anda minimum value of the order of 10minus3 G As discussed above this is not alimitation to the instrument performance

6 Discussion and conclusion

The accuracy of the measurements depends crucially on the stability andaccuracy of the clocks While the performance of ground clocks (even thoseeventually deployed in new ground stations dedicated to EGE) will likelyimprove significantly from the present level the space clock design will beldquofrozenrdquo approx 10 years before data acquisition Thus their performance

Exp Astron

level will be significantly below the future ground station clock performanceAs specification we have chosen a performance level that has been demon-strated with laboratory breadboards recently We see no fundamental reasonthat this laboratory performance cannot also be achieved in a space ion opticalclock The experience with the PHARAO cold atom clock has shown thatan engineering model of a laser-based cold atom clock has a performancesimilar to that of a laboratory prototype The perturbation of the ion clock fre-quency due to magnetic fields along the orbit can be modelled and suppressedwith shielding and the residual effects measured and compensated for Themoderate temperature stability (plusmn 3C) to be provided in the experimentalbay can be achieved with thermal shields Accurate temperature controlof particular units is done by active stabilization of the units themselves

One important aspect is radiation which may affect all optical and trapcomponents as well as the lasers and reference cavity The radiation envi-ronment on the EGE orbit is significantly harsher than the near-earth ISSorbit on which ACES will operate The radiation effects must therefore becharacterized comprehensively with appropriate tests and then the radiationshield for the EGE instruments designed

In summary the Einstein Gravity Explorer mission (EGE) is a satellitemission aimed at a precise measurement of the properties of space-time usingatomic clocks By testing one of the most fundamental predictions of EinsteinrsquosTheory of General Relativity the gravitational redshift and thereby searchingfor quantum effects in gravity it explores one of the most important frontiers infundamental physics The EGE mission reaches beyond fundamental physicsproviding an important spin-off in geophysics Furthermore the technologicalaspects of the mission are ground-breaking for future time-keeping and ref-erence frame systems in space and thus are also expected to be of wide useto science and technology in general The payload of the mission has beenconceived as a compromise between cost and performance needing to fit thecost envelope of an ESA medium-class mission1

Acknowledgement We thank S Cesare for helpful information

References

1 Will CM The confrontation between general relativity and experiment Living Rev Relativ9 3 (2006)

2 Ni W-T Empirical foundations of the relativistic gravity Int J Mod Phys D 14 901 (2005)doi101142S0218271805007139

1Finally we point out that a spacecraft with a significantly reduced and thus lower-cost payload (noclocks only MOLO FCDP MWL) on a circular orbit of appropriate height could be used for along-duration earth science mission dedicated specifically to the measurement of the geopotential

Exp Astron

3 Mission profile

31 Orbit

The orbit must satisfy the following requirements

Primary 1 Large gravitational potential difference between apogee andperigee

2 sufficient contact time with at least two ground stations atperigee

Secondary Simultaneous visibility of satellite from distant (inter-continental) ground stations for several hours

To satisfy the primary objective the orbit must be highly elliptic The perigeeshould be low A very low perigee means a short contact time which decreasesthe accuracy of the frequency measurement A compromise is a perigeealtitude of sim2500 km For a sufficiently high apogee this yields a U of 60of the theoretical maximum between the earth surface and infinity

The baseline scenario is an orbit of approx 12 h period with the followingproperties

Apogee Altitude 37856 km (minus 10 deg latitude one over Atlantic andPacific resp)

Perigee Altitude 2500 km (+ 10 deg latitude one over Guatemala andMalaysia resp)

Inclination 634 argument of perigee 170RAAN 25 True Anomaly 220Apogee drift rate 70 kmday perigee drift rate sim 14 kmday

True anomaly and right ascension of the ascending node (RAAN) wereoptimized to minimize orbital precession Figure 3 shows the ground track andthe ground coverage at one of the perigees Common-view contact time forstations in Europe is about 6 h while between a station in Europe and in theUSA it is about 35 h Frequency comparison windows between a given groundstation and EGE repeat every 24 h

A final selection of the orbit requires an in-depth study of the costs of thepropellant and maneuvers as well as of the setting up of new ground stations

32 Ground segment

The mission requires two separate ground segments the science groundsegment and the standard satellite control segment which supports telemetryand telecommand (TMTC)

Exp Astron

Fig 3 Baseline EGE orbit One (out of two) apogee and one (out of two) perigee ground pointsare shown The ground coverage at one of the perigees is shown with circles at elevation anglesteps of 10 starting from 10

321 Science ground segment

As described above one part of the mission is to perform frequency com-parisons between ground-to-space and ground-to-ground (common-view andnon-common view) The use of a two-way microwave link (MWL) systemhas the advantage that a precise orbit or the precise relative velocity of thesatellite must not be known [4] The orbit determination requirements fora two-way system are significantly relaxed compared to the requirements tomodel the relativistic effects on the clocks The frequency comparison resultsare available both on ground and on the satellite

For ground-to-satellite clock comparisons five ground stations arefavourable two capable of viewing perigee three for viewing apogee Thisallows for the classic three-clock configuration where the performance of eachclock can be obtained individually The perigee stations will be observing thesatellite for sim25 min every perigee passage for the duration of the missionThe apogee stations will observe for sim3 h every apogee passage Of coursea larger number of stations as well as longer observation time are usefulfor robustness redundancy and orbit determination The stations could beinstalled on a temporary basis for the duration of the mission

The stations must have access to optical clocks of performance at least ashigh as todayrsquos state-of-the-art ground clocks but preferably higher Theywill include an optical frequency comb by means of which the optical clockfrequency is transformed to a radio frequency for comparison with the EGEclocks High-accuracy fiber-optic link technology already existing today allowsdistances of sim100 km between the ground station and the correspondingoptical clock allowing a certain freedom in the choice of the clock locationWith further increase of the fiber-optic link range currently being investigated

Exp Astron

the area over which the ground station can be positioned relative to the opticalclock will increase This option is of importance for the implementation of ageophysics measurement campaign for which a large number of clock locationsis of interest

The above ground stations will be complemented by some 15 additionalground terminals provided by the world-wide laboratories equipped with highperformance primary clocks at the time of the mission These will participate inorbit determination to 1 cm accuracy along the full orbit which can be achievedwith non-optical atomic clocks The EGE MWL is fully compatible to plannedACES ground terminals which might reduce the number of required stations

With a larger number of transportable optical clock ground stations de-ployed and moved across the continents a geophysics campaign could beimplemented to map the gravitational potential

Ground station equipment MWL uses dedicated small transportable termi-nals which can be easily installed at any location of interest Each terminalconsists of a steerable dish antenna of 1 m diameter with pointing accuracy of01 using a transmitted power of 5 W per band The high directivity antenna(beam aperture Ka-band 07 Ku-band 14 S-band 20) avoids multi-pathfrom nearby objects Antenna noise temperatures are Ka-band 450 K Ku-band 340 K Operation is at elevations above 10 and measurements areperformed above 20 For ranging purposes terminal location is determined byGPS surveying Each ground station computes its own ionospheric correctionsbased on a triple-band receiver The terminal has a built-in delay monitorwhich allows calibrated ranging during several months once its bias has beendetermined by laser ranging Its M and C interface is via local area networkto either a userrsquos computer or directly to the science network control centreby remote MampC The stationrsquos operation is fully automatic and unattendedSchedule and orbital data are received via LAN once per day Signal acquisi-tion is fully autonomous Data transmission is after each pass The terminal hasan uplink telecommand capability of 1 kBits and a downlink data capacity ofup to 10 kBitss For the local user there is a visual interface providing quick-look data from real-time comparison to the spacecraft clocks to verify theirgood health The final full performance product is calculated at the sciencedata centre

Local metrological data (temperature pressure humidity) are recorded tocorrect for tropospheric delay of the ranging data The main advantage ofMWL terminals compared to laser based systems is their all-weather capabil-ity although microwaves suffer higher propagation errors in the ionosphereand troposphere

The EGE ground terminal will profit from the ERS-2 mission PRARE(precise range and range rate equipment) [45] experiments whose terminalshad similar design and have been operated under extreme climatic conditionsincl Antarctica and high temperature regions from existing designs fromthe space segment (mission ACES) and no further special developments aredeemed necessary Cost per EGE master station is modest

Exp Astron

Satellite tracking via laser ranging Laser ranging is alternative to MWL rang-ing and also serves for absolute calibration of the latter For this purpose anon-board corner cube reflector (CCR) is provided Ranging will be done withthe International Laser Ranging Service Ideally the satellite will be rangedat every perigee passage and every apogee passage simultaneously with thefrequency comparison procedures Todayrsquos ranging precision is sufficient tosatisfy the science goals For apogee observations in the northern hemispherea large number of laser ranging stations is available The number of stationscapable of ranging at or near perigee in the southern hemisphere is six andthus is sufficient even considering unfavourable weather conditions at someof these The MWL ground stations are able to range even under thoseconditions

322 Control ground segment

Average data transfer rate is estimated at 300 MBday (approx 40 kBs onaverage) Data will be stored on board (2 GB capacity required) and will betransferred to a receiving station on the Earth once a day

XPLC

incl Data StorageIF to TMTC Link

HK amp TMTC

to ground

Opt Clock

AOM

MOLO

PHARAO

FCDP

10 GHz

100 MHz

250 MHz

GNSS Rx

MWLTampF Link

uplink Data

UTC(k)

TampF Space vs Ground

TM from GS

Tim

e S

cale

Orbit GNSS Time UTC

TampF Space vs Ground

MampC Time Scale

MampC

MampC

MampC

MampC

Mamp

C

PHARAO vs Opt Clock

PHARAO vs Opt Clock

10 GHz amp Epoch

10 MHz

amp Epoch

Epoch

10 MHz

amp Epoch

Opt Link

Fig 4 Schematic of the payload showing subunits and their signals The results of the frequencycomparison between the two on-board clocks (optical clock and microwave clock PHARAO) areobtained from a combination of the frequency values applied to the frequency shifters AOM (seeFigs 7 and 8) and those output from the frequency control and distribution package (FCDP)as indicated by the arrows ldquoPHARAO vs Opt Clockrdquo The satellite clock to ground clockcomparison results are output from the MWL XPLC is the on-board computer and is involvedin generation of signals for instrument control The optical link shown dashed is a possible add-onproviding an alternative to the MWL GS ground stations MampC monitoring and control HKhousekeeping TampF time and frequency

Exp Astron

4 Main payload instruments

41 Overview

The satellite payload consists of a pair of cold-atom clocks a microwave-optical local oscillator (MOLO) an on-board frequency distribution unit(FCDP) a satellite-to-ground clock comparison unit (MWL) and the onboardcomputercontrol system (XPLC) see Fig 4 Auxiliary units are the posi-tion determination GPSGalileo receiver (GNSS Rx) attitude monitor (startracker not shown) and the corner cube reflector array (CCR not shown)

The science goals require clocks of very high frequency stability on thetimescales of half the orbital period (6 h) as well as perigee passage (sim1000 s)see Fig 2 Only clocks based on cold atoms are suited for the science goalsTable 1 reports values for a subset of atomic clocks that have been investigatednot including clocks that require cryogenic operation or are too complex foruse in an medium-class mission The choice of a clock pair rather than a singleone fulfils several requirements it addresses more science goals improves thevalidity of some science results and it enables calibration accuracy evaluationand redundancy The validity of the science results is enhanced since theclocks differ not only in the employed atoms but also in terms of havingdifferent sensitivity to perturbations by electric and magnetic fields electronicserrors and misalignments A comparison of the results arising from each clockallows for an understanding of measurement errors Both the clocksrsquo and thefrequency comparison unitrsquos performances will be determined after launchand regularly verified during the course of the mission The only reliable

Table 1 Properties of different clock candidates for EGE

Clock Clock Fractional σ = Fractional Sensitivity Figure of Sensitivityatom frequency inaccuracy instability to α Aα merit |Aα| to meq

at 1000 s σ [1016] QCD

Sr+ 445 THz 3 times 10minus17 (p) 5 times 10minus17 (p) 042 08 (p) ndashYb+ (1) 688 THz 3 times 10minus16 (d) 3 times 10minus16 (d) 088 029 (d) ndash

3 times 10minus17 (p) 1 (p)Yb+ (2) 642 THz lt 1 times 10minus17 (p) 1 times 10minus17 (p) minus53 53 (p) ndashSr 435 THz 1 times 10minus16 (d) 1 times 10minus16 (d) 009 009 (d) ndash

lt 1 times 10minus17 (p) 09 (p)Yb 519 THz 7 times 10minus16 (d) 1 times 10minus16 (d) 031 03 (d) ndash

lt 1 times 10minus17 (p) 1 times 10minus17 (p) 31 (p)Cs 91 GHz 2 times 10minus16 (d) 1 times 10minus15 (d) 275 027 (d) 1Yb+ 126 GHz 1 times 10minus15 (p) 1 times 10minus15 (p) 34 034 (p) 1

A relevant figure of merit is the ratio of |AX | and fractional frequency instability σ on the perigeevisibility timescale The Cs clock is the only high performance clock with adequate sensitivity tothe particle masses me mq The indicated Cs clock performance was achieved using a microwavelocal oscillator based on a cryogenic oscillator On EGE the latter is to be replaced by an opticallocal oscillator-based device (MOLO) with similar short-term performance but without the needto operate at cryogenic temperature The clock instability at 10000 s is typically similar to theindicated inaccuracy(p) projected (d) demonstrated (1) quadropole (2) octupole

Exp Astron

way to achieve this is by having at least two clocks on board The on-boardclock-clock comparison provides performance evaluation independent of theground-satellite comparison unit Once the satellite clock performances areestablished the link performance can be established as well

The clocks are similar in basic structure but differ in the detailed implemen-tation They consist of

a source of atomsa preparation subunit which cools the atoms to near-standstill (millikelvintemperature or below)a trapping subunit that confines the atom(s) to a tiny volumean interrogation subunita clock control electronics package

The clocks share a common central unit the optical-microwave local oscil-lator which serves as a flywheel oscillator with outstanding stability on shorttimescales (01ndash10 s) for both clocks It consists of a laser stabilized to an ultra-low loss optical resonator made of ultra-low-expansion glass A microwavesignal is derived from the laser by phase-locking a frequency comb to the laserwave and using a multiple of the repetition rate

While the subunits of any optical clock have a similar technological basis(optics electrooptics acoustooptics thermal control vacuum systems) theydiffer in their specifications Depending on the choice of the atom species de-livery confinement and interrogation methods clocks of different perfor-mance result A number of laboratory optical clock demonstrators have beendeveloped worldwide This ensures availability of the know-how for the indus-trial implementation of the flight models

The first clock is a microwave clock based on the interrogation of slowlymoving ensembles of cold Cs atoms It is essentially based on the PHARAOclock [5 6] developed to engineering model level by CNES in the frame-work of the ESA ISS mission ACES In EGE this clock is upgraded usingthe microwave-optical local oscillator This improves significantly the clockperformance The specification is a relative instability (Allan deviation) of3 times 10minus14 (τ s)minus12 ie at the integration time τ = 1000 s it drops to 1 times 10minus15and to 3 times 10minus16 at 10000 s While this specification is at least a factor 3 lessstringent than for the optical clocks the significantly lower cost of obtaining aflight model and the sensitivity of its frequency to the particle masses makes ita suitable choice

The optical clock baseline instrument is a single-ion optical clock Amongvarious suitable ion species (Hg+ Al+ Sr+ Yb+) [44ndash47] the Yb+ ion hasbeen selected Its breadboard performance is established at the instability levelof 3 times 10minus16 over 1000 s it exhibits a large sensitivity coefficient Aα hassmall sensitivity to magnetic fields and exhibits significant potential for furtherimprovements (use of the octupole transition with sim6 times larger Aα andlower overall instability [48] see below) A trapped Yb+ ion is also a veryrobust system it has been demonstrated that a single Yb+ ion can remaintrapped uninterruptedly for many months Finally Yb+ is the only ion with

Exp Astron

which optical clocks demonstrators have been operated in not a single but twomajor metrology laboratories

Another attractive optical clock is based on ensembles of cold neutral Sror Yb atoms trapped in an optical lattice produced by laser beams This clocktype has a lower instability than a single-ion clock thanks to the large numberof atoms (sim105) used Its development is in progress in several laboratoriesin the world (in part supported by space agencies) and has to date reached6 times 10minus17 instability at 5000 s and 1 times 10minus16 uncertainty [49 50] with potentialfor even higher performance However it is more complex and costly thanthe Yb+ ion clock since the cold atom preparation and trapping subunitsinclude more components require higher laser powers and the respectiveoperating procedures are more complex It is therefore not considered for thepresent mission designed to be developed in the very near future with limitedresources but is an interesting candidate for missions without such constraintsIts estimated physical parameters are indicated in Table 2 for comparison

42 Main instruments

421 Microwave clock

The PHARAO cold atom clock is one of the two atomic clocks of the spacemission ACES managed by ESA It is proposed here to include a secondversion of this instrument as its costperformance ratio is very attractive forthe EGE scientific objectives Its concept is very similar to ground basedatomic fountains but with the major difference of zero-g operation Atomsslowly launched in free flight cross two microwave fields tuned to the transitionbetween the two hyperfine levels of the cesium ground state The interrogationmethod based on two separate oscillating fields (Ramsey scheme) allowsthe detection of an atomic line whose width is inversely proportional to thetransit time between the two microwave cavities The resonant microwavefield at 9192631770 GHz (SI definition of the second) is synthesized startingfrom an ultrastable quartz oscillator (USO) in the FCDP phaselocked to themicrowave output of the microwave-optical local oscillator and stabilized tothe clock line using the error signal generated by the cesium resonator In thisway the intrinsic qualities of the cesium hyperfine transition both in terms ofaccuracy and frequency stability are transferred to the USO In a microgravityenvironment the velocity of atoms along the ballistic trajectories is constantand can be changed continuously over almost two orders of magnitude (5ndash500 cms) Differently from atomic fountain clocks presently operated onground very long interaction times (up to few seconds) will be possible whilekeeping reasonable the size of the instrument

The instrument consists of 4 main elements a laser source which providesthe light required to cool and detect the cesium atoms (Fig 5 left) a frequencychain which generates the 92 GHz microwave signal a cesium vacuum tubewhere the interaction between the microwave field and the atoms occurs

Exp Astron

Table 2 Resources and technology readiness level (TRL) of the EGE spacecraft

Unit Volume Mass Power TRL Comment[l] [kg] [W]

Ion clock 100 70 60Atom source 4Atom preparation 6Atom trap 4Atom interrogation 6Clock control 6

Microwave clock 200 91 110 8 TRL by end 2008Local oscillator

Stabilized laser 18 15 23 6Frequency comb 20 16 38 6

Microwave link 18 18 55 8FCDP 7 8 12 8GPS receiver 2 5 7 9CCR 2 2 0 9Payload sum 367 225 305Thermal control 14 17 30 8OBDH 20 24 38 9Star tracker 1 1 8 9Radiation shield 10 40 0 9Support structure 170 204 0 9Power incl solar arr 50 58 10 9Propulsion 21 25 20 9AOCS 36 43 82 9TMTC 8 10 15 9Satellite bus sum 330 422 203 Flexbus conceptPayload plus bus incl 20 margin 8364 7764 6096Propellant 40Launch adapter 100SC wet masspower incl Margins 9164Lattice clock 135 105 152 Alternative clock

Atom source 4Atom preparation 6Atom trap 4Atom interrogation 6Clock control 6

Also included are the resources for an optical lattice clock based on neutral Strontium atoms

(Fig 5 right) and a computer The engineering model of PHARAO is fullyassembled at CNES Toulouse and undergoes functional and performancetests until the end of 2008 The operating temperature range is 10ndash33 andnon-operating temperatures are minus45 +60 The cesium tube consists in aUHV chamber pumped by getters and a 3 ls ion pump a cesium reservoira microwave cavity coils to provide a uniform magnetic field and three layersof magnetic shields Vacuum windows with fiber-optics collimators enable thetransmission of the cooling and detection light onto the cesium atoms

In autonomous mode PHARAO uses the USO as interrogation oscillatorand will provide a clock signal with fractional frequency stability below1 times 10minus13 (τs) minus12 and inaccuracy near 1 times 10minus16 When using the microwave-

Exp Astron

Fig 5 Left The PHARAO laser source engineering model with cover removed Dimensionsare 530 times 350 times 150 mm3 and the mass is 200 kg Ten polarization maintaining optical fibers(yellow) guide the laser beams to the cesium tube All diode lasers are mounted on Peltier coolersfor temperature regulation within 2 mK (courtesy EADS SODERN) Right The PHARAOengineering model cesium tube without the two external magnetic shields The volume is 990 times336 times 444 mm3 and the total mass is 44 kg

optical local oscillator PHARAO will provide a reduced frequency instabilityof 3 times 10minus14 (τs) minus12 This corresponds to 1 times 10minus15 at 1000 s and 3 times 10minus16

at 104 s

422 Ion optical clock

The 171Yb+ ion clock shown in Figs 6 and 7 provides an atomic frequencyreference based on a single 171Yb+ ion confined within an RF electric end-captrap (similar to a Paul trap) within an ultra high vacuum (UHV) chamber andlaser cooled to sim1 mK close to the Doppler cooling limit As a result the ionrsquosfirst order Doppler motion is completely removed and the ion experienceslittle collisional perturbation from its environment Under these conditionsthe narrow linewidth 2S12 (F = 0 mF = 0) minus 2 D32 (F = 2 mF = 0) quadrupole

Fig 6 Left Simplified level scheme of 171Yb+ showing relevant cooling (blue) auxiliary (red)quadrupole clock (green) and octupole clock (dark blue) transitions δνnat is the natural linewidthof the clock transition Middle close-up of Paul-type RF trap (PTB Braunschweig) the ion istrapped in the hole (diameter 14 mm) Right Complete ion trap with vacuum flange height 10 cm(NPL Teddington)

Exp Astron

935 nm D32-state repumper laser

638 nm F72-state repumper laser

739 nm extended cavity diode laser

AOM

PMT

Ion trap UHV chamber

vacuum pumps

Fibre-free space interface

Mu-metal shielding

EC

T

B

Magnetic field

Temperature

Ion fluorescence Processor

Clock laser frequency step and control

4355 nm

Frequengy doubling II

3695 nm cooling wave

Low-drift cavity

Low-drift cavity 171Yb+ 435 nm ion clock unit

876 nm clock laser wave from MOLO

Frequengy doubling

Fig 7 Simplified schematic of the single-ion optical clock unit AOM acousto-optic frequencyshifter PMT photomultiplier tube EC endcap T temperature stabilization system B magneticfield

clock transition at 4355 nm can be interrogated by spectrally narrow and stableclock laser light A diagram of the simplified energy level scheme for 171Yb+showing the relevant cooling auxiliary and clock transitions is given in Fig 6All these wavelengths are provided by diode lasers operated in fundamentalor frequency doubled mode The clock laser light is obtained from the 871 nmwave provided by the MOLO unit (see Section 423) after frequency-doublingin single pass by a KNbO3 nonlinear-optical crystal to 4355 nm The clocktransition spectral profile is observed by stepping the clock laser frequency bymeans of an acousto-optic frequency shifter (AOM in Fig 7) and recordingthe statistics of the quantum jumps in cooling laser induced fluorescence asa function of this frequency Under clock operation the AOM repeatedlysteps back and forward between the half-intensity points on the transitionprofile monitoring the signal imbalance between these points Any detectedimbalance is servo-corrected to zero by feedback to the AOM

In detail the optical clock architecture will comprise (Fig 7)

bull an RF end-cap trap for ionising and confining a single 171Yb+ ion withinan ultra-high vacuum chamber pumped by a small ion pump and non-evaporable getter pump The trap is driven by means of an ac voltage ofa few hundred volts at a drive frequency of sim10 MHz A small oven withseveral mg of 171Yb isotope is heated to provide a low flux of Yb atomsfrom which single ions can be ionised by electron bombardment from ahot wire filament within the trap potential well Low voltage dc is applied

Exp Astron

to additional electrodes to position the ion precisely at trap potentialcentre in order to minimise the ionrsquos micromotion at the drive frequencyThis needs automatic periodic monitoring and correction especially in theperiod after re-loading an ion into the trap External magnetic field coilsin three orthogonal axes allow the nulling of external fields and settingof a fixed field of sim1 μT The trap and coils are surrounded by mu-metalshielding to minimise external field changes during the orbit A level oftemperature control of 1 K of the region surrounding the trap is requiredfor maintaining the blackbody-induced frequency shift uncertainty withtemperature at the 10minus17 level

bull a laser platform to provide frequency doubling of an extended cavity739 nm diode laser for Doppler cooling of the ion on the 2S12(F = 1) minus2 P12 (F = 0) dipole transition at 3695 nm Light from an amplified diodelaser device frequency-doubled in a periodically poled LiTaO3 crystal insingle-pass will be used Typical laser powers used for driving the coolingtransition below saturation are about 2 μW for a beam waist in the trapof about 50 μm By modulation of the injection current of the 739 nmdiode laser at 147 GHz a sideband is generated that excites the 2S12 (F =0) rarr2 P12 (F = 1) ldquorepumperrdquo transition in order to avoid optical pump-ing between the ground hyperfine states

bull The platform also houses the auxiliary lasers at 935 and 638 nm The935 nm diode provides repumping of the ion from the 2 D32(F = 1)metastable level after occasional branching decays to this level during thecooling sequence The 638 nm diode allows fast recovery of the ion fromthe very-long-lived 2F72 metastable state after very occasional collisionaldecay to that state Currently extended cavity lasers are forseen butdistributed-feedback lasers may be available in the near future

bull a high NA lens imaging system and photomultiplier detection system torecord the statistics of 369 nm fluorescence quantum jumps as a functionof clock laser frequency step

bull a fibre system to deliver the various cooling auxiliary and clock light fromsource to trap making use of achromatic doublets where necessary at thefibre-free space interface for launching into the trap

bull a monitoring and control processor which provides the primary coolingand clock laser pulse magnetic field and detection sequencing to observeand lock to the ion clock transition frequency The processor also monitorsfrequency and amplitude data necessary to determine normal laser andion operational conditions and initiate resetting and recovery algorithmswhere necessary and laser unit failure The actual magnetic field presenton the atomic volume is determined and actively stabilized so that aninaccuracy at the level of 1 times 10minus17 can be achieved

bull a redundancy level of at least two units for cooling clock and repumperlasers plus similar for frequency doubling crystals All redundancy unitsfor each wavelength will be fibre multiplexed as standard allowing redun-dant unit activation on determination of prior unit failure

Exp Astron

For an optimized quantum-projection-noise-limited performance of thesingle-ion clock Ramsey interrogation with a cycle time equal to the lifetimeof the metastable D32 level of 50 ms will be performed leading to an Allandeviation of about 27 times 10minus15 (τ s)minus12 This is sim45 times 10minus17 at 1 h satisfyingrequired stabilities for the science objectives Current laboratory results are astability of 1 times 10minus15 at 30 s averaging down to 3 times 10minus16 at 1000 s (PTBGermany) Also comparisons between two independent traps showed agree-ment at the level of several parts in 10minus16 over a series of eight measurementruns

The 171Yb+ ion clock offers a second option namely the 2S12ndash2 F72 octu-pole transition at 476 nm (see Table 1 and Fig 6) [48] It has an extremelylong upper state lifetime low electric quadrupole and second order Zeemansystematic shifts and therefore is a candidate for an even higher performanceclock if a clock laser system with lower instability than the above is providedApart from the clock laser it shares the laser system of the quadrupoleclock Thus with moderate additional resources the Yb+ ion apparatus couldprovide two optical clocks in a single package and specifically perform directfrequency comparisons between the quadrupole and the octupole transitionwith a large sensitivity Aα

The ion clock will operate under microprocessor control providing thecontrol algorithms for the integrated pulse sequencing for cooling repumpingclock interrogation magnetic field switching cooling fluorescence monitoringThe controller will also provide error flags for critical processes and systemreset and correction where necessary This will include ion re-loading andmicromotion reduction algorithms Automatic monitoring of laser power andspectral quality with signal re-optimisation or laser failure determination isrequired Redundant laser units will be pre-aligned and multiplexed into thefibre delivery to the trap so that redundant units can be readily activated in thecase of diode laser failure

423 Microwave-optical local oscillator (MOLO)

This oscillator provides two ultrastable frequency outputs one in the opticalregion the other in the microwave region the two frequencies being coher-ently related The MOLO is composed of three subunits the clock laser thereference cavity and the frequency comb see Fig 8

The clock laser subunit delivers the light for the excitation of the opticalclock transition It is generated by an extended-cavity diode laser at a wave-length of 871 nm Most of the output wave of this laser is sent to the ionclock The laser subunit containing two lasers for redundancy will require3 kg 3 l 15 W The frequency instability of the clock laser must be superiorto the minimum clock instability 27 times 10minus15 (τ s)minus12 for τ up to the timeconstant (sim10 s) of the servo system that locks the laser to the ionrsquos resonancesignal In order to achieve this the laser frequency is stabilized to the reference

Exp Astron

871 nm clock laser

EO

Detector and servo

AOM Femtosecond

frequency comb

To optical clock 10 GHz To FCDP

Microwave-optical local oscillator

ULE cavity

Fig 8 Microwave-optical local oscillator (MOLO) consisting of clock laser reference ULE cavityand femtosecond frequency comb (OFC) Radiofrequency inputs and outputs are not shown EOelectro-optic modulator ULE ultra-low expansion AOM frequency shifter

cavity where the PoundndashDreverndashHall method is used About 10 μW of thelaser output are used for this

The system will exhibit sim04 Hz (sim1 times 10minus15 relative) linewidth and afrequency drift lt01 Hzs The cavity consists of two high-reflectivity mirrorsoptically contacted to a cylinder of diameter 70 mm length 150 mm made ofULE (a glass exhibiting ultra low thermal expansion coefficient) with a cavityfinesse of about 300000 Residual vibrations of the satellite (arms lt 1 times 10minus6 g)should not be a limiting factor to the desired linewidth for safety marginan optimized cavity support can reduce the sensitivity to accelerations Foracoustic and thermal isolation the cavity resides inside an aluminium vacuumchamber equipped with a small (3 ls) ion-getter pump Two stages of polishedaluminium shields are implemented around the cavity which are activelytemperature stabilized so that the cavity is precisely kept at the zero crossingtemperature of the ULE thermal expansion coefficient Heat application andremoval is by thermoelectric elements between the shield and the vacuumchamber Additionally the entire vacuum chamber is temperature-stabilized

During transportation and launch the cavity will be rigidly fixed insidethe vacuum chamber by pressing on it from the sides by piezo-mechanicalactuators allowing accelerations of several g After bringing the satellite onthe orbit the actuators will be released

The optical setup for frequency stabilization will be a classic Pound-Drever-Hall scheme in reflection using an electro-optic modulator (EO) Couplingof the clock laser radiation to the cavity is via single-mode polarization main-taining optical fiber The fiber in- and out-couplers are mounted on miniaturepiezo-motor driven multiaxis translation stages Microprocessor control ofthe translation stages performs a laser-to-fiber incoupling stabilization andlaser-to-cavity mode-matching optimization (positioning inaccuracy lt10 μmalignment inaccuracy lt10 μrad)

The frequency stability of the clock laser will be limited by thermal 1f noisein the mirrors with an fractional Allan deviation slightly below 1 times 10minus15 from1 s to a few 10 s Resources including control electronics are 15 l 12 kg 8 W

Exp Astron

A femtosecond fiber laser based optical frequency comb generator (OFCFig 8) will be used for the clock frequency comparisons and to obtain aspectrally ultrapure microwave local oscillator output for the operation ofPHARAO for the microwave link and for the GNSS receiver The fem-tosecond laser is based on erbium-doped fiber pumped by fiber-coupled laserdiodes It emits 100 fs pulses at a repetition rate of 250 MHz with an averagepower of 100 mW The emission spectrum is centered around 1560 nm Spec-tral broadening to more than one octave is obtained via nonlinear interactionin a microstructure fibre so that the range from about 1050 to 2100 nm iscovered

Second harmonic generation of a portion of the broadened spectrum allowsobtaining the required light for the measurement of the clock laser frequencyat 871 nm The OFC will be referenced to the clock laser frequency (afterintroduction a controllable shift via the frequency shifter AOM) by makingthe carrier envelope offset frequency vanish so that the repetition rate fr isa high-order subharmonic of the optical clock reference frequency f1 = m1 frThe link between the optical and the microwave domain is then simply thedistribution of the repetition rate frequency or more conveniently a high(order 40) harmonic of the comb repetition rate 10 GHz

It has been demonstrated by several groups that frequency comparison andfrequency division with an OFC can be performed to very high accuracy andstability beyond the limitations imposed by the optical clocks themselves Forexample a fiber laser OFC with an instability (Allan deviation) of 6 times 10minus17

in 1 s has been demonstrated [51] OFCs based on fiber technology possessgood long-term reliability and high power efficiency The projection for a spacequalified OFC based on current industrial terrestrial technology is 16 kg mass20 l volume and 38 W power requirement

424 Frequency comparison and distribution package (FCDP)

This package distributes the ultra-stable 10 GHz reference frequency outputof the MOLO to the MWL performs frequency comparisons between MOLOand PHARAO and generates a variety of different reference signals for theindividual instruments including the 250 MHz reference clock to the OFCAll signals are coherent to the MOLO In the FCDP a USO is locked to theMOLO signal by a digital PLL A divider chain generates timing pulses (1 pps1000 pps etc) and keeps a continuous local time scale From the digital phasecomparison data the phase noise density spectrum between the USO and theMOLO is available by telemetry (TM) to assess the overall frequency purityin-flight

As the FCDP distributes the full performance reference from the MOLOdirectly to the MWL there is minimum impact of the FCDP on overall clockcomparison performance The USO-phase lock loop uses a similar designas for ACES-FCDP The digital PLL with linear phase detector allows theUSO to be slightly offset from its reference which is selectable by groundcommand This results in an accurate on-board time scale despite any offset

Exp Astron

of the MOLO reference frequency which can be synchronised to UTC Thefrequency synthesis chain re-uses modules developed for ACES-FCDP andtheir frequency inaccuracy Expected relative uncertainty (Allan deviation)of the frequency comparison between two signals provided to the FCDP is47 times 10minus17 at 1000 s 95 times 10minus18 at 10000 s

During non-availability of the MOLO ie during system start-up andcontingency conditions the USO runs as flywheel (ie it is unlocked withinstability lt15 times 10minus13 from 1 to 100 s) It will still provide useable butless accurate references to the instruments including the reference to OFCCommon view ground-to-ground time- and frequency comparison and supplyof the remaining instruments remain possible at the USO provided accuracy

Mass volume and power consumption are 8 kg 7 l 12 W Controldata rateis 10 kBday 50 kBday (FCDP alone)

425 Microwave link (MWL)

The microwave link is a two-way triple-frequency microwave link (Fig 9) Itis designed to perform accurate frequency comparison as well as time-transfer

Ku-Band Up-linkPower Tx 5 WCarrier 13475 GHzPN-Code 100 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

S-Band Down-linkPower Tx 1 WCarrier 2248 MHzPN-Code 1 MChips1pps 1 time marker sData 5 kBits

Ku-Band Down-linkPower Tx 1 WCarrier 1470333 GHzPN-Code 100 MChips1pps 1 time marker sData 5 kBits

EGE SC

S

Ku

KaKa-Band Up-linkPower Tx 5 WCarrier 345 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

Ka-Band Down-LinkPower Tx 1 WCarrier 32 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 5 kBits

GS Antenna diameterEGE terminal 1 mACES terminal 06 m

Fig 9 EGE microwave two-way link concept and characteristics See text for a description Thelink will be able to perform frequency comparisons with up to ten ground stations simultaneously(only one active comparison is shown) pps pulse per second Tx transmitter SC spacecraft

Exp Astron

between on-board and on-ground clocks Two-way operation allows for com-pensation of the effects due to spacecraft location and its motion relative toground Its two bi-directional links operate in Ka-band (32345 GHz) andKu-band (147135 GHz) The high carrier frequency of the up- and down-links allows for a noticeable reduction of the ionospheric delay A third down-link channel is in S-band (22 GHz) and is used to determine the ionospheretotal electron content MWL uses a combination of carrier phase and codephase data to measure with high precision the absolute ionospheric delayas well as ionospheric delay variations during a ground contact The useof three frequency bands allows to resolve third order ionospheric termsKey modulation parameters are Ka band Tone 250 MHz Ku band PN100 MChips S-band 1 MChips

The EGE-MWL concept is a considerably improved version of the ACES-MWL Microwave carrier generation is directly from the 10 GHz microwaveoutput available from the MOLO and having exceptionally low close-in phasenoise without intermediate frequency dividers or other complex synthesiserstages commonly used in RF transponders The high frequency carriers arekey to achieve the desired frequency comparison accuracy

The Ka-band carrier is generated by tripling the FCDP-generated signalThe Ku-band is subsequently generated by division by two using a verystable regenerative divider Finally the assigned radio frequency is obtainedby adding a low frequency signal using an image reject mixer A mixedanalog-digital receiver design will enhance its performance by use of a flexibledigital signal processing and digital control loops The number of simultaneousground contacts is increased from four (as in the ACES mission) to ten or moreusing the digital concept while the number of physical receiver channels perband is reduced from four to two one for operation to ground and one for in-flight calibration The concept allows adaptation of receiver loop parametersto optimise for the different portions of a highly elliptic orbit while reducingsystematic uncertainties due to high Doppler and Doppler rate Real-timedata exchange to ground terminals is by additional data modulation on all RFsignals The spacecraft carries two different small antennae for perigee and forapogee with gains of 8 dBi 20 dBi beamwidths 100 (full earth coverage) 14respectively power is 1 W for each

The MWL end-to-end performance between the attached clocks includingspace- and ground terminals is predicted to be a relative Allan deviation of1 times 10minus16 at 1000 s (perigee) and 1 times 10minus17 at 10000 s (apogee) Thesevalues are based on conservative implementation losses and other instrumentallosses assumed as 6 dB at apogee and 10 dB at perigee including additionallosses due to atmosphere An attitude knowledge of 1 is required

The MWL link will also support precise orbitographyndashcomplementary tothe GNSS receiver Generation of ranging data is simultaneous to two-waytime transfer and is available under all weather conditions Orbit determina-tion accuracy is expected to be better than 1 cm well matching EGE clockrequirements The value is expected based on demonstrated results fromother missions (METEOR ERS-2 TOPEXPOSEIDON) Laser ranging to

Exp Astron

the CCR array is used a few times during the mission to calibrate the MWLranging

In case of non-availability of the ultrapure signal from MOLO an additional10 GHz output is available from the FCDP to support ground-to-ground clockcomparison in common-view mode even after the nominal mission periodwhen the neither the clocks nor MOLO may be available any longer Basedon the PRARE experimentrsquos 12 years of operation this can be assumed forthe FCDPMWL combination as well The potential extended operational pe-riod makes the proposed configuration particularly attractive to metrologicallaboratories that rely on availability of long-lasting data in support of thegeneration of UTC

The resources of MWL are as follows mass 18 kg volume 18 l power55 W TC 2 kBday TM 120 MBday

426 Auxiliary units

The star camera assembly on the EGE spacecraft will provide real-timeattitude data primarily for the orientation of the spacecraft (with accuracy of1 in real-time) and the proper reduction of the GNSSSLR data

A GNSS receiver is used for the precise orbit determination time andfrequency transfer coarse positioning (lt1 m) for real time use by the Attitudeand Orbit Control as well as the back-up for time tagging of all payloaddata For the time and frequency transfer the receiver is nominally using theonboard frequency reference The receiver assembly includes three specialantennas By the year 2015 more than 100 GNSS satellites are expected tobe available for precise orbit determination permitting orbit inaccuracy of 1ndash2 cm RMS for altitudes up to approx 10000 km At apogee a combinationof GNSS signals and MWL ranging will be used with expected inaccuracyof 10 cm

5 Basic spacecraft specifications

The mission foresees a single spacecraft with a mission duration limited toapprox 3 years for cost reasons Attitude control along three axes withless than 1 error on a timescale of 1000 s is required The control can beperformed with cold gas thrusters magnetotorquers or reaction wheels Theselection will be done on the basis of the vibrations induced by these systems(desired level arms lt 10minus6 g)

Cold gas thrusters or monopropellant thrusters will be required to maintainthe orbit parameters during the mission by small orbit correction manoeuvresand will also guarantee a proper tuning of the orbit parameters after the releasefrom the launcher The ground segment (telemetry control and data transfer)could be via non-ESA stations user-operated to save costs

Further resources are indicated in Table 2

Exp Astron

The satellite bus design drivers are nadir-pointing microwave antennasthermal stability for the payload instruments no strong magnetic sources lowmicrovibrations harsh radiation environment due to crossing of the radiationbelts once per orbit power demand by science payload (with the possibility toreduce power demand by stand-by of some components during specific missionphases eg eclipse) mass limit due to required high altitude with low costlaunch

The satellite bus design can be based on established platforms such asFLEXBUS or PRIMA used on several past missions Their design philosophydiffers one having body-mounted solar arrays the other wings The latter canprovide larger amounts of power The vibration level induced by the panelswhen they are fixed and the duration of solar array drive periods per orbitwill have to be considered in the selection The FLEXBUS concept was usedfor a mission already requiring a low level of residual accelerations and theplatform is optimised for this parameter

For most of the satellite components one can consider using compo-nents and electronic units from navigation or communication satellites whichtypically cruise at altitudes between 15000 and 25000 km altitude LISAPathfinder which will cross the radiation belts during spiralling to theLagrange point will face similar problems concerning radiation as EGE as wellas XMM which is in a Molniya-type orbit with similar radiation characteristics

The instruments will therefore be located in the satellite structure such asto realise ionizing radiation shielding by the spacecraft structure and non-sensitive equipment to the maximum extent possible the need for additionalradiation shielding (including trapped proton radiation) for especially sensitiveparts of the instruments must be determined by ground tests It should bementioned that an orbit with higher perigee and less radiation dose is also anoption as it leads to a tolerable reduction of the science goals

The required plusmn3C temperature stability at the interface to the payloadenclosure is feasible The thermal control will employ the usual means (anenclosure with high thermal inertia heaters and a suitable radiator) Thelocation of the radiator will need a special study given the non-standard orbit

Magnetic field present during the mission can affect the properties of theonboard clock Using SPENVIS the magnetic field has been calculated alongthe spacecraft orbit the field intensity presents maxima around 02 G anda minimum value of the order of 10minus3 G As discussed above this is not alimitation to the instrument performance

6 Discussion and conclusion

The accuracy of the measurements depends crucially on the stability andaccuracy of the clocks While the performance of ground clocks (even thoseeventually deployed in new ground stations dedicated to EGE) will likelyimprove significantly from the present level the space clock design will beldquofrozenrdquo approx 10 years before data acquisition Thus their performance

Exp Astron

level will be significantly below the future ground station clock performanceAs specification we have chosen a performance level that has been demon-strated with laboratory breadboards recently We see no fundamental reasonthat this laboratory performance cannot also be achieved in a space ion opticalclock The experience with the PHARAO cold atom clock has shown thatan engineering model of a laser-based cold atom clock has a performancesimilar to that of a laboratory prototype The perturbation of the ion clock fre-quency due to magnetic fields along the orbit can be modelled and suppressedwith shielding and the residual effects measured and compensated for Themoderate temperature stability (plusmn 3C) to be provided in the experimentalbay can be achieved with thermal shields Accurate temperature controlof particular units is done by active stabilization of the units themselves

One important aspect is radiation which may affect all optical and trapcomponents as well as the lasers and reference cavity The radiation envi-ronment on the EGE orbit is significantly harsher than the near-earth ISSorbit on which ACES will operate The radiation effects must therefore becharacterized comprehensively with appropriate tests and then the radiationshield for the EGE instruments designed

In summary the Einstein Gravity Explorer mission (EGE) is a satellitemission aimed at a precise measurement of the properties of space-time usingatomic clocks By testing one of the most fundamental predictions of EinsteinrsquosTheory of General Relativity the gravitational redshift and thereby searchingfor quantum effects in gravity it explores one of the most important frontiers infundamental physics The EGE mission reaches beyond fundamental physicsproviding an important spin-off in geophysics Furthermore the technologicalaspects of the mission are ground-breaking for future time-keeping and ref-erence frame systems in space and thus are also expected to be of wide useto science and technology in general The payload of the mission has beenconceived as a compromise between cost and performance needing to fit thecost envelope of an ESA medium-class mission1

Acknowledgement We thank S Cesare for helpful information

References

1 Will CM The confrontation between general relativity and experiment Living Rev Relativ9 3 (2006)

2 Ni W-T Empirical foundations of the relativistic gravity Int J Mod Phys D 14 901 (2005)doi101142S0218271805007139

1Finally we point out that a spacecraft with a significantly reduced and thus lower-cost payload (noclocks only MOLO FCDP MWL) on a circular orbit of appropriate height could be used for along-duration earth science mission dedicated specifically to the measurement of the geopotential

Exp Astron

Fig 3 Baseline EGE orbit One (out of two) apogee and one (out of two) perigee ground pointsare shown The ground coverage at one of the perigees is shown with circles at elevation anglesteps of 10 starting from 10

321 Science ground segment

As described above one part of the mission is to perform frequency com-parisons between ground-to-space and ground-to-ground (common-view andnon-common view) The use of a two-way microwave link (MWL) systemhas the advantage that a precise orbit or the precise relative velocity of thesatellite must not be known [4] The orbit determination requirements fora two-way system are significantly relaxed compared to the requirements tomodel the relativistic effects on the clocks The frequency comparison resultsare available both on ground and on the satellite

For ground-to-satellite clock comparisons five ground stations arefavourable two capable of viewing perigee three for viewing apogee Thisallows for the classic three-clock configuration where the performance of eachclock can be obtained individually The perigee stations will be observing thesatellite for sim25 min every perigee passage for the duration of the missionThe apogee stations will observe for sim3 h every apogee passage Of coursea larger number of stations as well as longer observation time are usefulfor robustness redundancy and orbit determination The stations could beinstalled on a temporary basis for the duration of the mission

The stations must have access to optical clocks of performance at least ashigh as todayrsquos state-of-the-art ground clocks but preferably higher Theywill include an optical frequency comb by means of which the optical clockfrequency is transformed to a radio frequency for comparison with the EGEclocks High-accuracy fiber-optic link technology already existing today allowsdistances of sim100 km between the ground station and the correspondingoptical clock allowing a certain freedom in the choice of the clock locationWith further increase of the fiber-optic link range currently being investigated

Exp Astron

the area over which the ground station can be positioned relative to the opticalclock will increase This option is of importance for the implementation of ageophysics measurement campaign for which a large number of clock locationsis of interest

The above ground stations will be complemented by some 15 additionalground terminals provided by the world-wide laboratories equipped with highperformance primary clocks at the time of the mission These will participate inorbit determination to 1 cm accuracy along the full orbit which can be achievedwith non-optical atomic clocks The EGE MWL is fully compatible to plannedACES ground terminals which might reduce the number of required stations

With a larger number of transportable optical clock ground stations de-ployed and moved across the continents a geophysics campaign could beimplemented to map the gravitational potential

Ground station equipment MWL uses dedicated small transportable termi-nals which can be easily installed at any location of interest Each terminalconsists of a steerable dish antenna of 1 m diameter with pointing accuracy of01 using a transmitted power of 5 W per band The high directivity antenna(beam aperture Ka-band 07 Ku-band 14 S-band 20) avoids multi-pathfrom nearby objects Antenna noise temperatures are Ka-band 450 K Ku-band 340 K Operation is at elevations above 10 and measurements areperformed above 20 For ranging purposes terminal location is determined byGPS surveying Each ground station computes its own ionospheric correctionsbased on a triple-band receiver The terminal has a built-in delay monitorwhich allows calibrated ranging during several months once its bias has beendetermined by laser ranging Its M and C interface is via local area networkto either a userrsquos computer or directly to the science network control centreby remote MampC The stationrsquos operation is fully automatic and unattendedSchedule and orbital data are received via LAN once per day Signal acquisi-tion is fully autonomous Data transmission is after each pass The terminal hasan uplink telecommand capability of 1 kBits and a downlink data capacity ofup to 10 kBitss For the local user there is a visual interface providing quick-look data from real-time comparison to the spacecraft clocks to verify theirgood health The final full performance product is calculated at the sciencedata centre

Local metrological data (temperature pressure humidity) are recorded tocorrect for tropospheric delay of the ranging data The main advantage ofMWL terminals compared to laser based systems is their all-weather capabil-ity although microwaves suffer higher propagation errors in the ionosphereand troposphere

The EGE ground terminal will profit from the ERS-2 mission PRARE(precise range and range rate equipment) [45] experiments whose terminalshad similar design and have been operated under extreme climatic conditionsincl Antarctica and high temperature regions from existing designs fromthe space segment (mission ACES) and no further special developments aredeemed necessary Cost per EGE master station is modest

Exp Astron

Satellite tracking via laser ranging Laser ranging is alternative to MWL rang-ing and also serves for absolute calibration of the latter For this purpose anon-board corner cube reflector (CCR) is provided Ranging will be done withthe International Laser Ranging Service Ideally the satellite will be rangedat every perigee passage and every apogee passage simultaneously with thefrequency comparison procedures Todayrsquos ranging precision is sufficient tosatisfy the science goals For apogee observations in the northern hemispherea large number of laser ranging stations is available The number of stationscapable of ranging at or near perigee in the southern hemisphere is six andthus is sufficient even considering unfavourable weather conditions at someof these The MWL ground stations are able to range even under thoseconditions

322 Control ground segment

Average data transfer rate is estimated at 300 MBday (approx 40 kBs onaverage) Data will be stored on board (2 GB capacity required) and will betransferred to a receiving station on the Earth once a day

XPLC

incl Data StorageIF to TMTC Link

HK amp TMTC

to ground

Opt Clock

AOM

MOLO

PHARAO

FCDP

10 GHz

100 MHz

250 MHz

GNSS Rx

MWLTampF Link

uplink Data

UTC(k)

TampF Space vs Ground

TM from GS

Tim

e S

cale

Orbit GNSS Time UTC

TampF Space vs Ground

MampC Time Scale

MampC

MampC

MampC

MampC

Mamp

C

PHARAO vs Opt Clock

PHARAO vs Opt Clock

10 GHz amp Epoch

10 MHz

amp Epoch

Epoch

10 MHz

amp Epoch

Opt Link

Fig 4 Schematic of the payload showing subunits and their signals The results of the frequencycomparison between the two on-board clocks (optical clock and microwave clock PHARAO) areobtained from a combination of the frequency values applied to the frequency shifters AOM (seeFigs 7 and 8) and those output from the frequency control and distribution package (FCDP)as indicated by the arrows ldquoPHARAO vs Opt Clockrdquo The satellite clock to ground clockcomparison results are output from the MWL XPLC is the on-board computer and is involvedin generation of signals for instrument control The optical link shown dashed is a possible add-onproviding an alternative to the MWL GS ground stations MampC monitoring and control HKhousekeeping TampF time and frequency

Exp Astron

4 Main payload instruments

41 Overview

The satellite payload consists of a pair of cold-atom clocks a microwave-optical local oscillator (MOLO) an on-board frequency distribution unit(FCDP) a satellite-to-ground clock comparison unit (MWL) and the onboardcomputercontrol system (XPLC) see Fig 4 Auxiliary units are the posi-tion determination GPSGalileo receiver (GNSS Rx) attitude monitor (startracker not shown) and the corner cube reflector array (CCR not shown)

The science goals require clocks of very high frequency stability on thetimescales of half the orbital period (6 h) as well as perigee passage (sim1000 s)see Fig 2 Only clocks based on cold atoms are suited for the science goalsTable 1 reports values for a subset of atomic clocks that have been investigatednot including clocks that require cryogenic operation or are too complex foruse in an medium-class mission The choice of a clock pair rather than a singleone fulfils several requirements it addresses more science goals improves thevalidity of some science results and it enables calibration accuracy evaluationand redundancy The validity of the science results is enhanced since theclocks differ not only in the employed atoms but also in terms of havingdifferent sensitivity to perturbations by electric and magnetic fields electronicserrors and misalignments A comparison of the results arising from each clockallows for an understanding of measurement errors Both the clocksrsquo and thefrequency comparison unitrsquos performances will be determined after launchand regularly verified during the course of the mission The only reliable

Table 1 Properties of different clock candidates for EGE

Clock Clock Fractional σ = Fractional Sensitivity Figure of Sensitivityatom frequency inaccuracy instability to α Aα merit |Aα| to meq

at 1000 s σ [1016] QCD

Sr+ 445 THz 3 times 10minus17 (p) 5 times 10minus17 (p) 042 08 (p) ndashYb+ (1) 688 THz 3 times 10minus16 (d) 3 times 10minus16 (d) 088 029 (d) ndash

3 times 10minus17 (p) 1 (p)Yb+ (2) 642 THz lt 1 times 10minus17 (p) 1 times 10minus17 (p) minus53 53 (p) ndashSr 435 THz 1 times 10minus16 (d) 1 times 10minus16 (d) 009 009 (d) ndash

lt 1 times 10minus17 (p) 09 (p)Yb 519 THz 7 times 10minus16 (d) 1 times 10minus16 (d) 031 03 (d) ndash

lt 1 times 10minus17 (p) 1 times 10minus17 (p) 31 (p)Cs 91 GHz 2 times 10minus16 (d) 1 times 10minus15 (d) 275 027 (d) 1Yb+ 126 GHz 1 times 10minus15 (p) 1 times 10minus15 (p) 34 034 (p) 1

A relevant figure of merit is the ratio of |AX | and fractional frequency instability σ on the perigeevisibility timescale The Cs clock is the only high performance clock with adequate sensitivity tothe particle masses me mq The indicated Cs clock performance was achieved using a microwavelocal oscillator based on a cryogenic oscillator On EGE the latter is to be replaced by an opticallocal oscillator-based device (MOLO) with similar short-term performance but without the needto operate at cryogenic temperature The clock instability at 10000 s is typically similar to theindicated inaccuracy(p) projected (d) demonstrated (1) quadropole (2) octupole

Exp Astron

way to achieve this is by having at least two clocks on board The on-boardclock-clock comparison provides performance evaluation independent of theground-satellite comparison unit Once the satellite clock performances areestablished the link performance can be established as well

The clocks are similar in basic structure but differ in the detailed implemen-tation They consist of

a source of atomsa preparation subunit which cools the atoms to near-standstill (millikelvintemperature or below)a trapping subunit that confines the atom(s) to a tiny volumean interrogation subunita clock control electronics package

The clocks share a common central unit the optical-microwave local oscil-lator which serves as a flywheel oscillator with outstanding stability on shorttimescales (01ndash10 s) for both clocks It consists of a laser stabilized to an ultra-low loss optical resonator made of ultra-low-expansion glass A microwavesignal is derived from the laser by phase-locking a frequency comb to the laserwave and using a multiple of the repetition rate

While the subunits of any optical clock have a similar technological basis(optics electrooptics acoustooptics thermal control vacuum systems) theydiffer in their specifications Depending on the choice of the atom species de-livery confinement and interrogation methods clocks of different perfor-mance result A number of laboratory optical clock demonstrators have beendeveloped worldwide This ensures availability of the know-how for the indus-trial implementation of the flight models

The first clock is a microwave clock based on the interrogation of slowlymoving ensembles of cold Cs atoms It is essentially based on the PHARAOclock [5 6] developed to engineering model level by CNES in the frame-work of the ESA ISS mission ACES In EGE this clock is upgraded usingthe microwave-optical local oscillator This improves significantly the clockperformance The specification is a relative instability (Allan deviation) of3 times 10minus14 (τ s)minus12 ie at the integration time τ = 1000 s it drops to 1 times 10minus15and to 3 times 10minus16 at 10000 s While this specification is at least a factor 3 lessstringent than for the optical clocks the significantly lower cost of obtaining aflight model and the sensitivity of its frequency to the particle masses makes ita suitable choice

The optical clock baseline instrument is a single-ion optical clock Amongvarious suitable ion species (Hg+ Al+ Sr+ Yb+) [44ndash47] the Yb+ ion hasbeen selected Its breadboard performance is established at the instability levelof 3 times 10minus16 over 1000 s it exhibits a large sensitivity coefficient Aα hassmall sensitivity to magnetic fields and exhibits significant potential for furtherimprovements (use of the octupole transition with sim6 times larger Aα andlower overall instability [48] see below) A trapped Yb+ ion is also a veryrobust system it has been demonstrated that a single Yb+ ion can remaintrapped uninterruptedly for many months Finally Yb+ is the only ion with

Exp Astron

which optical clocks demonstrators have been operated in not a single but twomajor metrology laboratories

Another attractive optical clock is based on ensembles of cold neutral Sror Yb atoms trapped in an optical lattice produced by laser beams This clocktype has a lower instability than a single-ion clock thanks to the large numberof atoms (sim105) used Its development is in progress in several laboratoriesin the world (in part supported by space agencies) and has to date reached6 times 10minus17 instability at 5000 s and 1 times 10minus16 uncertainty [49 50] with potentialfor even higher performance However it is more complex and costly thanthe Yb+ ion clock since the cold atom preparation and trapping subunitsinclude more components require higher laser powers and the respectiveoperating procedures are more complex It is therefore not considered for thepresent mission designed to be developed in the very near future with limitedresources but is an interesting candidate for missions without such constraintsIts estimated physical parameters are indicated in Table 2 for comparison

42 Main instruments

421 Microwave clock

The PHARAO cold atom clock is one of the two atomic clocks of the spacemission ACES managed by ESA It is proposed here to include a secondversion of this instrument as its costperformance ratio is very attractive forthe EGE scientific objectives Its concept is very similar to ground basedatomic fountains but with the major difference of zero-g operation Atomsslowly launched in free flight cross two microwave fields tuned to the transitionbetween the two hyperfine levels of the cesium ground state The interrogationmethod based on two separate oscillating fields (Ramsey scheme) allowsthe detection of an atomic line whose width is inversely proportional to thetransit time between the two microwave cavities The resonant microwavefield at 9192631770 GHz (SI definition of the second) is synthesized startingfrom an ultrastable quartz oscillator (USO) in the FCDP phaselocked to themicrowave output of the microwave-optical local oscillator and stabilized tothe clock line using the error signal generated by the cesium resonator In thisway the intrinsic qualities of the cesium hyperfine transition both in terms ofaccuracy and frequency stability are transferred to the USO In a microgravityenvironment the velocity of atoms along the ballistic trajectories is constantand can be changed continuously over almost two orders of magnitude (5ndash500 cms) Differently from atomic fountain clocks presently operated onground very long interaction times (up to few seconds) will be possible whilekeeping reasonable the size of the instrument

The instrument consists of 4 main elements a laser source which providesthe light required to cool and detect the cesium atoms (Fig 5 left) a frequencychain which generates the 92 GHz microwave signal a cesium vacuum tubewhere the interaction between the microwave field and the atoms occurs

Exp Astron

Table 2 Resources and technology readiness level (TRL) of the EGE spacecraft

Unit Volume Mass Power TRL Comment[l] [kg] [W]

Ion clock 100 70 60Atom source 4Atom preparation 6Atom trap 4Atom interrogation 6Clock control 6

Microwave clock 200 91 110 8 TRL by end 2008Local oscillator

Stabilized laser 18 15 23 6Frequency comb 20 16 38 6

Microwave link 18 18 55 8FCDP 7 8 12 8GPS receiver 2 5 7 9CCR 2 2 0 9Payload sum 367 225 305Thermal control 14 17 30 8OBDH 20 24 38 9Star tracker 1 1 8 9Radiation shield 10 40 0 9Support structure 170 204 0 9Power incl solar arr 50 58 10 9Propulsion 21 25 20 9AOCS 36 43 82 9TMTC 8 10 15 9Satellite bus sum 330 422 203 Flexbus conceptPayload plus bus incl 20 margin 8364 7764 6096Propellant 40Launch adapter 100SC wet masspower incl Margins 9164Lattice clock 135 105 152 Alternative clock

Atom source 4Atom preparation 6Atom trap 4Atom interrogation 6Clock control 6

Also included are the resources for an optical lattice clock based on neutral Strontium atoms

(Fig 5 right) and a computer The engineering model of PHARAO is fullyassembled at CNES Toulouse and undergoes functional and performancetests until the end of 2008 The operating temperature range is 10ndash33 andnon-operating temperatures are minus45 +60 The cesium tube consists in aUHV chamber pumped by getters and a 3 ls ion pump a cesium reservoira microwave cavity coils to provide a uniform magnetic field and three layersof magnetic shields Vacuum windows with fiber-optics collimators enable thetransmission of the cooling and detection light onto the cesium atoms

In autonomous mode PHARAO uses the USO as interrogation oscillatorand will provide a clock signal with fractional frequency stability below1 times 10minus13 (τs) minus12 and inaccuracy near 1 times 10minus16 When using the microwave-

Exp Astron

Fig 5 Left The PHARAO laser source engineering model with cover removed Dimensionsare 530 times 350 times 150 mm3 and the mass is 200 kg Ten polarization maintaining optical fibers(yellow) guide the laser beams to the cesium tube All diode lasers are mounted on Peltier coolersfor temperature regulation within 2 mK (courtesy EADS SODERN) Right The PHARAOengineering model cesium tube without the two external magnetic shields The volume is 990 times336 times 444 mm3 and the total mass is 44 kg

optical local oscillator PHARAO will provide a reduced frequency instabilityof 3 times 10minus14 (τs) minus12 This corresponds to 1 times 10minus15 at 1000 s and 3 times 10minus16

at 104 s

422 Ion optical clock

The 171Yb+ ion clock shown in Figs 6 and 7 provides an atomic frequencyreference based on a single 171Yb+ ion confined within an RF electric end-captrap (similar to a Paul trap) within an ultra high vacuum (UHV) chamber andlaser cooled to sim1 mK close to the Doppler cooling limit As a result the ionrsquosfirst order Doppler motion is completely removed and the ion experienceslittle collisional perturbation from its environment Under these conditionsthe narrow linewidth 2S12 (F = 0 mF = 0) minus 2 D32 (F = 2 mF = 0) quadrupole

Fig 6 Left Simplified level scheme of 171Yb+ showing relevant cooling (blue) auxiliary (red)quadrupole clock (green) and octupole clock (dark blue) transitions δνnat is the natural linewidthof the clock transition Middle close-up of Paul-type RF trap (PTB Braunschweig) the ion istrapped in the hole (diameter 14 mm) Right Complete ion trap with vacuum flange height 10 cm(NPL Teddington)

Exp Astron

935 nm D32-state repumper laser

638 nm F72-state repumper laser

739 nm extended cavity diode laser

AOM

PMT

Ion trap UHV chamber

vacuum pumps

Fibre-free space interface

Mu-metal shielding

EC

T

B

Magnetic field

Temperature

Ion fluorescence Processor

Clock laser frequency step and control

4355 nm

Frequengy doubling II

3695 nm cooling wave

Low-drift cavity

Low-drift cavity 171Yb+ 435 nm ion clock unit

876 nm clock laser wave from MOLO

Frequengy doubling

Fig 7 Simplified schematic of the single-ion optical clock unit AOM acousto-optic frequencyshifter PMT photomultiplier tube EC endcap T temperature stabilization system B magneticfield

clock transition at 4355 nm can be interrogated by spectrally narrow and stableclock laser light A diagram of the simplified energy level scheme for 171Yb+showing the relevant cooling auxiliary and clock transitions is given in Fig 6All these wavelengths are provided by diode lasers operated in fundamentalor frequency doubled mode The clock laser light is obtained from the 871 nmwave provided by the MOLO unit (see Section 423) after frequency-doublingin single pass by a KNbO3 nonlinear-optical crystal to 4355 nm The clocktransition spectral profile is observed by stepping the clock laser frequency bymeans of an acousto-optic frequency shifter (AOM in Fig 7) and recordingthe statistics of the quantum jumps in cooling laser induced fluorescence asa function of this frequency Under clock operation the AOM repeatedlysteps back and forward between the half-intensity points on the transitionprofile monitoring the signal imbalance between these points Any detectedimbalance is servo-corrected to zero by feedback to the AOM

In detail the optical clock architecture will comprise (Fig 7)

bull an RF end-cap trap for ionising and confining a single 171Yb+ ion withinan ultra-high vacuum chamber pumped by a small ion pump and non-evaporable getter pump The trap is driven by means of an ac voltage ofa few hundred volts at a drive frequency of sim10 MHz A small oven withseveral mg of 171Yb isotope is heated to provide a low flux of Yb atomsfrom which single ions can be ionised by electron bombardment from ahot wire filament within the trap potential well Low voltage dc is applied

Exp Astron

to additional electrodes to position the ion precisely at trap potentialcentre in order to minimise the ionrsquos micromotion at the drive frequencyThis needs automatic periodic monitoring and correction especially in theperiod after re-loading an ion into the trap External magnetic field coilsin three orthogonal axes allow the nulling of external fields and settingof a fixed field of sim1 μT The trap and coils are surrounded by mu-metalshielding to minimise external field changes during the orbit A level oftemperature control of 1 K of the region surrounding the trap is requiredfor maintaining the blackbody-induced frequency shift uncertainty withtemperature at the 10minus17 level

bull a laser platform to provide frequency doubling of an extended cavity739 nm diode laser for Doppler cooling of the ion on the 2S12(F = 1) minus2 P12 (F = 0) dipole transition at 3695 nm Light from an amplified diodelaser device frequency-doubled in a periodically poled LiTaO3 crystal insingle-pass will be used Typical laser powers used for driving the coolingtransition below saturation are about 2 μW for a beam waist in the trapof about 50 μm By modulation of the injection current of the 739 nmdiode laser at 147 GHz a sideband is generated that excites the 2S12 (F =0) rarr2 P12 (F = 1) ldquorepumperrdquo transition in order to avoid optical pump-ing between the ground hyperfine states

bull The platform also houses the auxiliary lasers at 935 and 638 nm The935 nm diode provides repumping of the ion from the 2 D32(F = 1)metastable level after occasional branching decays to this level during thecooling sequence The 638 nm diode allows fast recovery of the ion fromthe very-long-lived 2F72 metastable state after very occasional collisionaldecay to that state Currently extended cavity lasers are forseen butdistributed-feedback lasers may be available in the near future

bull a high NA lens imaging system and photomultiplier detection system torecord the statistics of 369 nm fluorescence quantum jumps as a functionof clock laser frequency step

bull a fibre system to deliver the various cooling auxiliary and clock light fromsource to trap making use of achromatic doublets where necessary at thefibre-free space interface for launching into the trap

bull a monitoring and control processor which provides the primary coolingand clock laser pulse magnetic field and detection sequencing to observeand lock to the ion clock transition frequency The processor also monitorsfrequency and amplitude data necessary to determine normal laser andion operational conditions and initiate resetting and recovery algorithmswhere necessary and laser unit failure The actual magnetic field presenton the atomic volume is determined and actively stabilized so that aninaccuracy at the level of 1 times 10minus17 can be achieved

bull a redundancy level of at least two units for cooling clock and repumperlasers plus similar for frequency doubling crystals All redundancy unitsfor each wavelength will be fibre multiplexed as standard allowing redun-dant unit activation on determination of prior unit failure

Exp Astron

For an optimized quantum-projection-noise-limited performance of thesingle-ion clock Ramsey interrogation with a cycle time equal to the lifetimeof the metastable D32 level of 50 ms will be performed leading to an Allandeviation of about 27 times 10minus15 (τ s)minus12 This is sim45 times 10minus17 at 1 h satisfyingrequired stabilities for the science objectives Current laboratory results are astability of 1 times 10minus15 at 30 s averaging down to 3 times 10minus16 at 1000 s (PTBGermany) Also comparisons between two independent traps showed agree-ment at the level of several parts in 10minus16 over a series of eight measurementruns

The 171Yb+ ion clock offers a second option namely the 2S12ndash2 F72 octu-pole transition at 476 nm (see Table 1 and Fig 6) [48] It has an extremelylong upper state lifetime low electric quadrupole and second order Zeemansystematic shifts and therefore is a candidate for an even higher performanceclock if a clock laser system with lower instability than the above is providedApart from the clock laser it shares the laser system of the quadrupoleclock Thus with moderate additional resources the Yb+ ion apparatus couldprovide two optical clocks in a single package and specifically perform directfrequency comparisons between the quadrupole and the octupole transitionwith a large sensitivity Aα

The ion clock will operate under microprocessor control providing thecontrol algorithms for the integrated pulse sequencing for cooling repumpingclock interrogation magnetic field switching cooling fluorescence monitoringThe controller will also provide error flags for critical processes and systemreset and correction where necessary This will include ion re-loading andmicromotion reduction algorithms Automatic monitoring of laser power andspectral quality with signal re-optimisation or laser failure determination isrequired Redundant laser units will be pre-aligned and multiplexed into thefibre delivery to the trap so that redundant units can be readily activated in thecase of diode laser failure

423 Microwave-optical local oscillator (MOLO)

This oscillator provides two ultrastable frequency outputs one in the opticalregion the other in the microwave region the two frequencies being coher-ently related The MOLO is composed of three subunits the clock laser thereference cavity and the frequency comb see Fig 8

The clock laser subunit delivers the light for the excitation of the opticalclock transition It is generated by an extended-cavity diode laser at a wave-length of 871 nm Most of the output wave of this laser is sent to the ionclock The laser subunit containing two lasers for redundancy will require3 kg 3 l 15 W The frequency instability of the clock laser must be superiorto the minimum clock instability 27 times 10minus15 (τ s)minus12 for τ up to the timeconstant (sim10 s) of the servo system that locks the laser to the ionrsquos resonancesignal In order to achieve this the laser frequency is stabilized to the reference

Exp Astron

871 nm clock laser

EO

Detector and servo

AOM Femtosecond

frequency comb

To optical clock 10 GHz To FCDP

Microwave-optical local oscillator

ULE cavity

Fig 8 Microwave-optical local oscillator (MOLO) consisting of clock laser reference ULE cavityand femtosecond frequency comb (OFC) Radiofrequency inputs and outputs are not shown EOelectro-optic modulator ULE ultra-low expansion AOM frequency shifter

cavity where the PoundndashDreverndashHall method is used About 10 μW of thelaser output are used for this

The system will exhibit sim04 Hz (sim1 times 10minus15 relative) linewidth and afrequency drift lt01 Hzs The cavity consists of two high-reflectivity mirrorsoptically contacted to a cylinder of diameter 70 mm length 150 mm made ofULE (a glass exhibiting ultra low thermal expansion coefficient) with a cavityfinesse of about 300000 Residual vibrations of the satellite (arms lt 1 times 10minus6 g)should not be a limiting factor to the desired linewidth for safety marginan optimized cavity support can reduce the sensitivity to accelerations Foracoustic and thermal isolation the cavity resides inside an aluminium vacuumchamber equipped with a small (3 ls) ion-getter pump Two stages of polishedaluminium shields are implemented around the cavity which are activelytemperature stabilized so that the cavity is precisely kept at the zero crossingtemperature of the ULE thermal expansion coefficient Heat application andremoval is by thermoelectric elements between the shield and the vacuumchamber Additionally the entire vacuum chamber is temperature-stabilized

During transportation and launch the cavity will be rigidly fixed insidethe vacuum chamber by pressing on it from the sides by piezo-mechanicalactuators allowing accelerations of several g After bringing the satellite onthe orbit the actuators will be released

The optical setup for frequency stabilization will be a classic Pound-Drever-Hall scheme in reflection using an electro-optic modulator (EO) Couplingof the clock laser radiation to the cavity is via single-mode polarization main-taining optical fiber The fiber in- and out-couplers are mounted on miniaturepiezo-motor driven multiaxis translation stages Microprocessor control ofthe translation stages performs a laser-to-fiber incoupling stabilization andlaser-to-cavity mode-matching optimization (positioning inaccuracy lt10 μmalignment inaccuracy lt10 μrad)

The frequency stability of the clock laser will be limited by thermal 1f noisein the mirrors with an fractional Allan deviation slightly below 1 times 10minus15 from1 s to a few 10 s Resources including control electronics are 15 l 12 kg 8 W

Exp Astron

A femtosecond fiber laser based optical frequency comb generator (OFCFig 8) will be used for the clock frequency comparisons and to obtain aspectrally ultrapure microwave local oscillator output for the operation ofPHARAO for the microwave link and for the GNSS receiver The fem-tosecond laser is based on erbium-doped fiber pumped by fiber-coupled laserdiodes It emits 100 fs pulses at a repetition rate of 250 MHz with an averagepower of 100 mW The emission spectrum is centered around 1560 nm Spec-tral broadening to more than one octave is obtained via nonlinear interactionin a microstructure fibre so that the range from about 1050 to 2100 nm iscovered

Second harmonic generation of a portion of the broadened spectrum allowsobtaining the required light for the measurement of the clock laser frequencyat 871 nm The OFC will be referenced to the clock laser frequency (afterintroduction a controllable shift via the frequency shifter AOM) by makingthe carrier envelope offset frequency vanish so that the repetition rate fr isa high-order subharmonic of the optical clock reference frequency f1 = m1 frThe link between the optical and the microwave domain is then simply thedistribution of the repetition rate frequency or more conveniently a high(order 40) harmonic of the comb repetition rate 10 GHz

It has been demonstrated by several groups that frequency comparison andfrequency division with an OFC can be performed to very high accuracy andstability beyond the limitations imposed by the optical clocks themselves Forexample a fiber laser OFC with an instability (Allan deviation) of 6 times 10minus17

in 1 s has been demonstrated [51] OFCs based on fiber technology possessgood long-term reliability and high power efficiency The projection for a spacequalified OFC based on current industrial terrestrial technology is 16 kg mass20 l volume and 38 W power requirement

424 Frequency comparison and distribution package (FCDP)

This package distributes the ultra-stable 10 GHz reference frequency outputof the MOLO to the MWL performs frequency comparisons between MOLOand PHARAO and generates a variety of different reference signals for theindividual instruments including the 250 MHz reference clock to the OFCAll signals are coherent to the MOLO In the FCDP a USO is locked to theMOLO signal by a digital PLL A divider chain generates timing pulses (1 pps1000 pps etc) and keeps a continuous local time scale From the digital phasecomparison data the phase noise density spectrum between the USO and theMOLO is available by telemetry (TM) to assess the overall frequency purityin-flight

As the FCDP distributes the full performance reference from the MOLOdirectly to the MWL there is minimum impact of the FCDP on overall clockcomparison performance The USO-phase lock loop uses a similar designas for ACES-FCDP The digital PLL with linear phase detector allows theUSO to be slightly offset from its reference which is selectable by groundcommand This results in an accurate on-board time scale despite any offset

Exp Astron

of the MOLO reference frequency which can be synchronised to UTC Thefrequency synthesis chain re-uses modules developed for ACES-FCDP andtheir frequency inaccuracy Expected relative uncertainty (Allan deviation)of the frequency comparison between two signals provided to the FCDP is47 times 10minus17 at 1000 s 95 times 10minus18 at 10000 s

During non-availability of the MOLO ie during system start-up andcontingency conditions the USO runs as flywheel (ie it is unlocked withinstability lt15 times 10minus13 from 1 to 100 s) It will still provide useable butless accurate references to the instruments including the reference to OFCCommon view ground-to-ground time- and frequency comparison and supplyof the remaining instruments remain possible at the USO provided accuracy

Mass volume and power consumption are 8 kg 7 l 12 W Controldata rateis 10 kBday 50 kBday (FCDP alone)

425 Microwave link (MWL)

The microwave link is a two-way triple-frequency microwave link (Fig 9) Itis designed to perform accurate frequency comparison as well as time-transfer

Ku-Band Up-linkPower Tx 5 WCarrier 13475 GHzPN-Code 100 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

S-Band Down-linkPower Tx 1 WCarrier 2248 MHzPN-Code 1 MChips1pps 1 time marker sData 5 kBits

Ku-Band Down-linkPower Tx 1 WCarrier 1470333 GHzPN-Code 100 MChips1pps 1 time marker sData 5 kBits

EGE SC

S

Ku

KaKa-Band Up-linkPower Tx 5 WCarrier 345 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

Ka-Band Down-LinkPower Tx 1 WCarrier 32 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 5 kBits

GS Antenna diameterEGE terminal 1 mACES terminal 06 m

Fig 9 EGE microwave two-way link concept and characteristics See text for a description Thelink will be able to perform frequency comparisons with up to ten ground stations simultaneously(only one active comparison is shown) pps pulse per second Tx transmitter SC spacecraft

Exp Astron

between on-board and on-ground clocks Two-way operation allows for com-pensation of the effects due to spacecraft location and its motion relative toground Its two bi-directional links operate in Ka-band (32345 GHz) andKu-band (147135 GHz) The high carrier frequency of the up- and down-links allows for a noticeable reduction of the ionospheric delay A third down-link channel is in S-band (22 GHz) and is used to determine the ionospheretotal electron content MWL uses a combination of carrier phase and codephase data to measure with high precision the absolute ionospheric delayas well as ionospheric delay variations during a ground contact The useof three frequency bands allows to resolve third order ionospheric termsKey modulation parameters are Ka band Tone 250 MHz Ku band PN100 MChips S-band 1 MChips

The EGE-MWL concept is a considerably improved version of the ACES-MWL Microwave carrier generation is directly from the 10 GHz microwaveoutput available from the MOLO and having exceptionally low close-in phasenoise without intermediate frequency dividers or other complex synthesiserstages commonly used in RF transponders The high frequency carriers arekey to achieve the desired frequency comparison accuracy

The Ka-band carrier is generated by tripling the FCDP-generated signalThe Ku-band is subsequently generated by division by two using a verystable regenerative divider Finally the assigned radio frequency is obtainedby adding a low frequency signal using an image reject mixer A mixedanalog-digital receiver design will enhance its performance by use of a flexibledigital signal processing and digital control loops The number of simultaneousground contacts is increased from four (as in the ACES mission) to ten or moreusing the digital concept while the number of physical receiver channels perband is reduced from four to two one for operation to ground and one for in-flight calibration The concept allows adaptation of receiver loop parametersto optimise for the different portions of a highly elliptic orbit while reducingsystematic uncertainties due to high Doppler and Doppler rate Real-timedata exchange to ground terminals is by additional data modulation on all RFsignals The spacecraft carries two different small antennae for perigee and forapogee with gains of 8 dBi 20 dBi beamwidths 100 (full earth coverage) 14respectively power is 1 W for each

The MWL end-to-end performance between the attached clocks includingspace- and ground terminals is predicted to be a relative Allan deviation of1 times 10minus16 at 1000 s (perigee) and 1 times 10minus17 at 10000 s (apogee) Thesevalues are based on conservative implementation losses and other instrumentallosses assumed as 6 dB at apogee and 10 dB at perigee including additionallosses due to atmosphere An attitude knowledge of 1 is required

The MWL link will also support precise orbitographyndashcomplementary tothe GNSS receiver Generation of ranging data is simultaneous to two-waytime transfer and is available under all weather conditions Orbit determina-tion accuracy is expected to be better than 1 cm well matching EGE clockrequirements The value is expected based on demonstrated results fromother missions (METEOR ERS-2 TOPEXPOSEIDON) Laser ranging to

Exp Astron

the CCR array is used a few times during the mission to calibrate the MWLranging

In case of non-availability of the ultrapure signal from MOLO an additional10 GHz output is available from the FCDP to support ground-to-ground clockcomparison in common-view mode even after the nominal mission periodwhen the neither the clocks nor MOLO may be available any longer Basedon the PRARE experimentrsquos 12 years of operation this can be assumed forthe FCDPMWL combination as well The potential extended operational pe-riod makes the proposed configuration particularly attractive to metrologicallaboratories that rely on availability of long-lasting data in support of thegeneration of UTC

The resources of MWL are as follows mass 18 kg volume 18 l power55 W TC 2 kBday TM 120 MBday

426 Auxiliary units

The star camera assembly on the EGE spacecraft will provide real-timeattitude data primarily for the orientation of the spacecraft (with accuracy of1 in real-time) and the proper reduction of the GNSSSLR data

A GNSS receiver is used for the precise orbit determination time andfrequency transfer coarse positioning (lt1 m) for real time use by the Attitudeand Orbit Control as well as the back-up for time tagging of all payloaddata For the time and frequency transfer the receiver is nominally using theonboard frequency reference The receiver assembly includes three specialantennas By the year 2015 more than 100 GNSS satellites are expected tobe available for precise orbit determination permitting orbit inaccuracy of 1ndash2 cm RMS for altitudes up to approx 10000 km At apogee a combinationof GNSS signals and MWL ranging will be used with expected inaccuracyof 10 cm

5 Basic spacecraft specifications

The mission foresees a single spacecraft with a mission duration limited toapprox 3 years for cost reasons Attitude control along three axes withless than 1 error on a timescale of 1000 s is required The control can beperformed with cold gas thrusters magnetotorquers or reaction wheels Theselection will be done on the basis of the vibrations induced by these systems(desired level arms lt 10minus6 g)

Cold gas thrusters or monopropellant thrusters will be required to maintainthe orbit parameters during the mission by small orbit correction manoeuvresand will also guarantee a proper tuning of the orbit parameters after the releasefrom the launcher The ground segment (telemetry control and data transfer)could be via non-ESA stations user-operated to save costs

Further resources are indicated in Table 2

Exp Astron

The satellite bus design drivers are nadir-pointing microwave antennasthermal stability for the payload instruments no strong magnetic sources lowmicrovibrations harsh radiation environment due to crossing of the radiationbelts once per orbit power demand by science payload (with the possibility toreduce power demand by stand-by of some components during specific missionphases eg eclipse) mass limit due to required high altitude with low costlaunch

The satellite bus design can be based on established platforms such asFLEXBUS or PRIMA used on several past missions Their design philosophydiffers one having body-mounted solar arrays the other wings The latter canprovide larger amounts of power The vibration level induced by the panelswhen they are fixed and the duration of solar array drive periods per orbitwill have to be considered in the selection The FLEXBUS concept was usedfor a mission already requiring a low level of residual accelerations and theplatform is optimised for this parameter

For most of the satellite components one can consider using compo-nents and electronic units from navigation or communication satellites whichtypically cruise at altitudes between 15000 and 25000 km altitude LISAPathfinder which will cross the radiation belts during spiralling to theLagrange point will face similar problems concerning radiation as EGE as wellas XMM which is in a Molniya-type orbit with similar radiation characteristics

The instruments will therefore be located in the satellite structure such asto realise ionizing radiation shielding by the spacecraft structure and non-sensitive equipment to the maximum extent possible the need for additionalradiation shielding (including trapped proton radiation) for especially sensitiveparts of the instruments must be determined by ground tests It should bementioned that an orbit with higher perigee and less radiation dose is also anoption as it leads to a tolerable reduction of the science goals

The required plusmn3C temperature stability at the interface to the payloadenclosure is feasible The thermal control will employ the usual means (anenclosure with high thermal inertia heaters and a suitable radiator) Thelocation of the radiator will need a special study given the non-standard orbit

Magnetic field present during the mission can affect the properties of theonboard clock Using SPENVIS the magnetic field has been calculated alongthe spacecraft orbit the field intensity presents maxima around 02 G anda minimum value of the order of 10minus3 G As discussed above this is not alimitation to the instrument performance

6 Discussion and conclusion

The accuracy of the measurements depends crucially on the stability andaccuracy of the clocks While the performance of ground clocks (even thoseeventually deployed in new ground stations dedicated to EGE) will likelyimprove significantly from the present level the space clock design will beldquofrozenrdquo approx 10 years before data acquisition Thus their performance

Exp Astron

level will be significantly below the future ground station clock performanceAs specification we have chosen a performance level that has been demon-strated with laboratory breadboards recently We see no fundamental reasonthat this laboratory performance cannot also be achieved in a space ion opticalclock The experience with the PHARAO cold atom clock has shown thatan engineering model of a laser-based cold atom clock has a performancesimilar to that of a laboratory prototype The perturbation of the ion clock fre-quency due to magnetic fields along the orbit can be modelled and suppressedwith shielding and the residual effects measured and compensated for Themoderate temperature stability (plusmn 3C) to be provided in the experimentalbay can be achieved with thermal shields Accurate temperature controlof particular units is done by active stabilization of the units themselves

One important aspect is radiation which may affect all optical and trapcomponents as well as the lasers and reference cavity The radiation envi-ronment on the EGE orbit is significantly harsher than the near-earth ISSorbit on which ACES will operate The radiation effects must therefore becharacterized comprehensively with appropriate tests and then the radiationshield for the EGE instruments designed

In summary the Einstein Gravity Explorer mission (EGE) is a satellitemission aimed at a precise measurement of the properties of space-time usingatomic clocks By testing one of the most fundamental predictions of EinsteinrsquosTheory of General Relativity the gravitational redshift and thereby searchingfor quantum effects in gravity it explores one of the most important frontiers infundamental physics The EGE mission reaches beyond fundamental physicsproviding an important spin-off in geophysics Furthermore the technologicalaspects of the mission are ground-breaking for future time-keeping and ref-erence frame systems in space and thus are also expected to be of wide useto science and technology in general The payload of the mission has beenconceived as a compromise between cost and performance needing to fit thecost envelope of an ESA medium-class mission1

Acknowledgement We thank S Cesare for helpful information

References

1 Will CM The confrontation between general relativity and experiment Living Rev Relativ9 3 (2006)

2 Ni W-T Empirical foundations of the relativistic gravity Int J Mod Phys D 14 901 (2005)doi101142S0218271805007139

1Finally we point out that a spacecraft with a significantly reduced and thus lower-cost payload (noclocks only MOLO FCDP MWL) on a circular orbit of appropriate height could be used for along-duration earth science mission dedicated specifically to the measurement of the geopotential

Exp Astron

the area over which the ground station can be positioned relative to the opticalclock will increase This option is of importance for the implementation of ageophysics measurement campaign for which a large number of clock locationsis of interest

The above ground stations will be complemented by some 15 additionalground terminals provided by the world-wide laboratories equipped with highperformance primary clocks at the time of the mission These will participate inorbit determination to 1 cm accuracy along the full orbit which can be achievedwith non-optical atomic clocks The EGE MWL is fully compatible to plannedACES ground terminals which might reduce the number of required stations

With a larger number of transportable optical clock ground stations de-ployed and moved across the continents a geophysics campaign could beimplemented to map the gravitational potential

Ground station equipment MWL uses dedicated small transportable termi-nals which can be easily installed at any location of interest Each terminalconsists of a steerable dish antenna of 1 m diameter with pointing accuracy of01 using a transmitted power of 5 W per band The high directivity antenna(beam aperture Ka-band 07 Ku-band 14 S-band 20) avoids multi-pathfrom nearby objects Antenna noise temperatures are Ka-band 450 K Ku-band 340 K Operation is at elevations above 10 and measurements areperformed above 20 For ranging purposes terminal location is determined byGPS surveying Each ground station computes its own ionospheric correctionsbased on a triple-band receiver The terminal has a built-in delay monitorwhich allows calibrated ranging during several months once its bias has beendetermined by laser ranging Its M and C interface is via local area networkto either a userrsquos computer or directly to the science network control centreby remote MampC The stationrsquos operation is fully automatic and unattendedSchedule and orbital data are received via LAN once per day Signal acquisi-tion is fully autonomous Data transmission is after each pass The terminal hasan uplink telecommand capability of 1 kBits and a downlink data capacity ofup to 10 kBitss For the local user there is a visual interface providing quick-look data from real-time comparison to the spacecraft clocks to verify theirgood health The final full performance product is calculated at the sciencedata centre

Local metrological data (temperature pressure humidity) are recorded tocorrect for tropospheric delay of the ranging data The main advantage ofMWL terminals compared to laser based systems is their all-weather capabil-ity although microwaves suffer higher propagation errors in the ionosphereand troposphere

The EGE ground terminal will profit from the ERS-2 mission PRARE(precise range and range rate equipment) [45] experiments whose terminalshad similar design and have been operated under extreme climatic conditionsincl Antarctica and high temperature regions from existing designs fromthe space segment (mission ACES) and no further special developments aredeemed necessary Cost per EGE master station is modest

Exp Astron

Satellite tracking via laser ranging Laser ranging is alternative to MWL rang-ing and also serves for absolute calibration of the latter For this purpose anon-board corner cube reflector (CCR) is provided Ranging will be done withthe International Laser Ranging Service Ideally the satellite will be rangedat every perigee passage and every apogee passage simultaneously with thefrequency comparison procedures Todayrsquos ranging precision is sufficient tosatisfy the science goals For apogee observations in the northern hemispherea large number of laser ranging stations is available The number of stationscapable of ranging at or near perigee in the southern hemisphere is six andthus is sufficient even considering unfavourable weather conditions at someof these The MWL ground stations are able to range even under thoseconditions

322 Control ground segment

Average data transfer rate is estimated at 300 MBday (approx 40 kBs onaverage) Data will be stored on board (2 GB capacity required) and will betransferred to a receiving station on the Earth once a day

XPLC

incl Data StorageIF to TMTC Link

HK amp TMTC

to ground

Opt Clock

AOM

MOLO

PHARAO

FCDP

10 GHz

100 MHz

250 MHz

GNSS Rx

MWLTampF Link

uplink Data

UTC(k)

TampF Space vs Ground

TM from GS

Tim

e S

cale

Orbit GNSS Time UTC

TampF Space vs Ground

MampC Time Scale

MampC

MampC

MampC

MampC

Mamp

C

PHARAO vs Opt Clock

PHARAO vs Opt Clock

10 GHz amp Epoch

10 MHz

amp Epoch

Epoch

10 MHz

amp Epoch

Opt Link

Fig 4 Schematic of the payload showing subunits and their signals The results of the frequencycomparison between the two on-board clocks (optical clock and microwave clock PHARAO) areobtained from a combination of the frequency values applied to the frequency shifters AOM (seeFigs 7 and 8) and those output from the frequency control and distribution package (FCDP)as indicated by the arrows ldquoPHARAO vs Opt Clockrdquo The satellite clock to ground clockcomparison results are output from the MWL XPLC is the on-board computer and is involvedin generation of signals for instrument control The optical link shown dashed is a possible add-onproviding an alternative to the MWL GS ground stations MampC monitoring and control HKhousekeeping TampF time and frequency

Exp Astron

4 Main payload instruments

41 Overview

The satellite payload consists of a pair of cold-atom clocks a microwave-optical local oscillator (MOLO) an on-board frequency distribution unit(FCDP) a satellite-to-ground clock comparison unit (MWL) and the onboardcomputercontrol system (XPLC) see Fig 4 Auxiliary units are the posi-tion determination GPSGalileo receiver (GNSS Rx) attitude monitor (startracker not shown) and the corner cube reflector array (CCR not shown)

The science goals require clocks of very high frequency stability on thetimescales of half the orbital period (6 h) as well as perigee passage (sim1000 s)see Fig 2 Only clocks based on cold atoms are suited for the science goalsTable 1 reports values for a subset of atomic clocks that have been investigatednot including clocks that require cryogenic operation or are too complex foruse in an medium-class mission The choice of a clock pair rather than a singleone fulfils several requirements it addresses more science goals improves thevalidity of some science results and it enables calibration accuracy evaluationand redundancy The validity of the science results is enhanced since theclocks differ not only in the employed atoms but also in terms of havingdifferent sensitivity to perturbations by electric and magnetic fields electronicserrors and misalignments A comparison of the results arising from each clockallows for an understanding of measurement errors Both the clocksrsquo and thefrequency comparison unitrsquos performances will be determined after launchand regularly verified during the course of the mission The only reliable

Table 1 Properties of different clock candidates for EGE

Clock Clock Fractional σ = Fractional Sensitivity Figure of Sensitivityatom frequency inaccuracy instability to α Aα merit |Aα| to meq

at 1000 s σ [1016] QCD

Sr+ 445 THz 3 times 10minus17 (p) 5 times 10minus17 (p) 042 08 (p) ndashYb+ (1) 688 THz 3 times 10minus16 (d) 3 times 10minus16 (d) 088 029 (d) ndash

3 times 10minus17 (p) 1 (p)Yb+ (2) 642 THz lt 1 times 10minus17 (p) 1 times 10minus17 (p) minus53 53 (p) ndashSr 435 THz 1 times 10minus16 (d) 1 times 10minus16 (d) 009 009 (d) ndash

lt 1 times 10minus17 (p) 09 (p)Yb 519 THz 7 times 10minus16 (d) 1 times 10minus16 (d) 031 03 (d) ndash

lt 1 times 10minus17 (p) 1 times 10minus17 (p) 31 (p)Cs 91 GHz 2 times 10minus16 (d) 1 times 10minus15 (d) 275 027 (d) 1Yb+ 126 GHz 1 times 10minus15 (p) 1 times 10minus15 (p) 34 034 (p) 1

A relevant figure of merit is the ratio of |AX | and fractional frequency instability σ on the perigeevisibility timescale The Cs clock is the only high performance clock with adequate sensitivity tothe particle masses me mq The indicated Cs clock performance was achieved using a microwavelocal oscillator based on a cryogenic oscillator On EGE the latter is to be replaced by an opticallocal oscillator-based device (MOLO) with similar short-term performance but without the needto operate at cryogenic temperature The clock instability at 10000 s is typically similar to theindicated inaccuracy(p) projected (d) demonstrated (1) quadropole (2) octupole

Exp Astron

way to achieve this is by having at least two clocks on board The on-boardclock-clock comparison provides performance evaluation independent of theground-satellite comparison unit Once the satellite clock performances areestablished the link performance can be established as well

The clocks are similar in basic structure but differ in the detailed implemen-tation They consist of

a source of atomsa preparation subunit which cools the atoms to near-standstill (millikelvintemperature or below)a trapping subunit that confines the atom(s) to a tiny volumean interrogation subunita clock control electronics package

The clocks share a common central unit the optical-microwave local oscil-lator which serves as a flywheel oscillator with outstanding stability on shorttimescales (01ndash10 s) for both clocks It consists of a laser stabilized to an ultra-low loss optical resonator made of ultra-low-expansion glass A microwavesignal is derived from the laser by phase-locking a frequency comb to the laserwave and using a multiple of the repetition rate

While the subunits of any optical clock have a similar technological basis(optics electrooptics acoustooptics thermal control vacuum systems) theydiffer in their specifications Depending on the choice of the atom species de-livery confinement and interrogation methods clocks of different perfor-mance result A number of laboratory optical clock demonstrators have beendeveloped worldwide This ensures availability of the know-how for the indus-trial implementation of the flight models

The first clock is a microwave clock based on the interrogation of slowlymoving ensembles of cold Cs atoms It is essentially based on the PHARAOclock [5 6] developed to engineering model level by CNES in the frame-work of the ESA ISS mission ACES In EGE this clock is upgraded usingthe microwave-optical local oscillator This improves significantly the clockperformance The specification is a relative instability (Allan deviation) of3 times 10minus14 (τ s)minus12 ie at the integration time τ = 1000 s it drops to 1 times 10minus15and to 3 times 10minus16 at 10000 s While this specification is at least a factor 3 lessstringent than for the optical clocks the significantly lower cost of obtaining aflight model and the sensitivity of its frequency to the particle masses makes ita suitable choice

The optical clock baseline instrument is a single-ion optical clock Amongvarious suitable ion species (Hg+ Al+ Sr+ Yb+) [44ndash47] the Yb+ ion hasbeen selected Its breadboard performance is established at the instability levelof 3 times 10minus16 over 1000 s it exhibits a large sensitivity coefficient Aα hassmall sensitivity to magnetic fields and exhibits significant potential for furtherimprovements (use of the octupole transition with sim6 times larger Aα andlower overall instability [48] see below) A trapped Yb+ ion is also a veryrobust system it has been demonstrated that a single Yb+ ion can remaintrapped uninterruptedly for many months Finally Yb+ is the only ion with

Exp Astron

which optical clocks demonstrators have been operated in not a single but twomajor metrology laboratories

Another attractive optical clock is based on ensembles of cold neutral Sror Yb atoms trapped in an optical lattice produced by laser beams This clocktype has a lower instability than a single-ion clock thanks to the large numberof atoms (sim105) used Its development is in progress in several laboratoriesin the world (in part supported by space agencies) and has to date reached6 times 10minus17 instability at 5000 s and 1 times 10minus16 uncertainty [49 50] with potentialfor even higher performance However it is more complex and costly thanthe Yb+ ion clock since the cold atom preparation and trapping subunitsinclude more components require higher laser powers and the respectiveoperating procedures are more complex It is therefore not considered for thepresent mission designed to be developed in the very near future with limitedresources but is an interesting candidate for missions without such constraintsIts estimated physical parameters are indicated in Table 2 for comparison

42 Main instruments

421 Microwave clock

The PHARAO cold atom clock is one of the two atomic clocks of the spacemission ACES managed by ESA It is proposed here to include a secondversion of this instrument as its costperformance ratio is very attractive forthe EGE scientific objectives Its concept is very similar to ground basedatomic fountains but with the major difference of zero-g operation Atomsslowly launched in free flight cross two microwave fields tuned to the transitionbetween the two hyperfine levels of the cesium ground state The interrogationmethod based on two separate oscillating fields (Ramsey scheme) allowsthe detection of an atomic line whose width is inversely proportional to thetransit time between the two microwave cavities The resonant microwavefield at 9192631770 GHz (SI definition of the second) is synthesized startingfrom an ultrastable quartz oscillator (USO) in the FCDP phaselocked to themicrowave output of the microwave-optical local oscillator and stabilized tothe clock line using the error signal generated by the cesium resonator In thisway the intrinsic qualities of the cesium hyperfine transition both in terms ofaccuracy and frequency stability are transferred to the USO In a microgravityenvironment the velocity of atoms along the ballistic trajectories is constantand can be changed continuously over almost two orders of magnitude (5ndash500 cms) Differently from atomic fountain clocks presently operated onground very long interaction times (up to few seconds) will be possible whilekeeping reasonable the size of the instrument

The instrument consists of 4 main elements a laser source which providesthe light required to cool and detect the cesium atoms (Fig 5 left) a frequencychain which generates the 92 GHz microwave signal a cesium vacuum tubewhere the interaction between the microwave field and the atoms occurs

Exp Astron

Table 2 Resources and technology readiness level (TRL) of the EGE spacecraft

Unit Volume Mass Power TRL Comment[l] [kg] [W]

Ion clock 100 70 60Atom source 4Atom preparation 6Atom trap 4Atom interrogation 6Clock control 6

Microwave clock 200 91 110 8 TRL by end 2008Local oscillator

Stabilized laser 18 15 23 6Frequency comb 20 16 38 6

Microwave link 18 18 55 8FCDP 7 8 12 8GPS receiver 2 5 7 9CCR 2 2 0 9Payload sum 367 225 305Thermal control 14 17 30 8OBDH 20 24 38 9Star tracker 1 1 8 9Radiation shield 10 40 0 9Support structure 170 204 0 9Power incl solar arr 50 58 10 9Propulsion 21 25 20 9AOCS 36 43 82 9TMTC 8 10 15 9Satellite bus sum 330 422 203 Flexbus conceptPayload plus bus incl 20 margin 8364 7764 6096Propellant 40Launch adapter 100SC wet masspower incl Margins 9164Lattice clock 135 105 152 Alternative clock

Atom source 4Atom preparation 6Atom trap 4Atom interrogation 6Clock control 6

Also included are the resources for an optical lattice clock based on neutral Strontium atoms

(Fig 5 right) and a computer The engineering model of PHARAO is fullyassembled at CNES Toulouse and undergoes functional and performancetests until the end of 2008 The operating temperature range is 10ndash33 andnon-operating temperatures are minus45 +60 The cesium tube consists in aUHV chamber pumped by getters and a 3 ls ion pump a cesium reservoira microwave cavity coils to provide a uniform magnetic field and three layersof magnetic shields Vacuum windows with fiber-optics collimators enable thetransmission of the cooling and detection light onto the cesium atoms

In autonomous mode PHARAO uses the USO as interrogation oscillatorand will provide a clock signal with fractional frequency stability below1 times 10minus13 (τs) minus12 and inaccuracy near 1 times 10minus16 When using the microwave-

Exp Astron

Fig 5 Left The PHARAO laser source engineering model with cover removed Dimensionsare 530 times 350 times 150 mm3 and the mass is 200 kg Ten polarization maintaining optical fibers(yellow) guide the laser beams to the cesium tube All diode lasers are mounted on Peltier coolersfor temperature regulation within 2 mK (courtesy EADS SODERN) Right The PHARAOengineering model cesium tube without the two external magnetic shields The volume is 990 times336 times 444 mm3 and the total mass is 44 kg

optical local oscillator PHARAO will provide a reduced frequency instabilityof 3 times 10minus14 (τs) minus12 This corresponds to 1 times 10minus15 at 1000 s and 3 times 10minus16

at 104 s

422 Ion optical clock

The 171Yb+ ion clock shown in Figs 6 and 7 provides an atomic frequencyreference based on a single 171Yb+ ion confined within an RF electric end-captrap (similar to a Paul trap) within an ultra high vacuum (UHV) chamber andlaser cooled to sim1 mK close to the Doppler cooling limit As a result the ionrsquosfirst order Doppler motion is completely removed and the ion experienceslittle collisional perturbation from its environment Under these conditionsthe narrow linewidth 2S12 (F = 0 mF = 0) minus 2 D32 (F = 2 mF = 0) quadrupole

Fig 6 Left Simplified level scheme of 171Yb+ showing relevant cooling (blue) auxiliary (red)quadrupole clock (green) and octupole clock (dark blue) transitions δνnat is the natural linewidthof the clock transition Middle close-up of Paul-type RF trap (PTB Braunschweig) the ion istrapped in the hole (diameter 14 mm) Right Complete ion trap with vacuum flange height 10 cm(NPL Teddington)

Exp Astron

935 nm D32-state repumper laser

638 nm F72-state repumper laser

739 nm extended cavity diode laser

AOM

PMT

Ion trap UHV chamber

vacuum pumps

Fibre-free space interface

Mu-metal shielding

EC

T

B

Magnetic field

Temperature

Ion fluorescence Processor

Clock laser frequency step and control

4355 nm

Frequengy doubling II

3695 nm cooling wave

Low-drift cavity

Low-drift cavity 171Yb+ 435 nm ion clock unit

876 nm clock laser wave from MOLO

Frequengy doubling

Fig 7 Simplified schematic of the single-ion optical clock unit AOM acousto-optic frequencyshifter PMT photomultiplier tube EC endcap T temperature stabilization system B magneticfield

clock transition at 4355 nm can be interrogated by spectrally narrow and stableclock laser light A diagram of the simplified energy level scheme for 171Yb+showing the relevant cooling auxiliary and clock transitions is given in Fig 6All these wavelengths are provided by diode lasers operated in fundamentalor frequency doubled mode The clock laser light is obtained from the 871 nmwave provided by the MOLO unit (see Section 423) after frequency-doublingin single pass by a KNbO3 nonlinear-optical crystal to 4355 nm The clocktransition spectral profile is observed by stepping the clock laser frequency bymeans of an acousto-optic frequency shifter (AOM in Fig 7) and recordingthe statistics of the quantum jumps in cooling laser induced fluorescence asa function of this frequency Under clock operation the AOM repeatedlysteps back and forward between the half-intensity points on the transitionprofile monitoring the signal imbalance between these points Any detectedimbalance is servo-corrected to zero by feedback to the AOM

In detail the optical clock architecture will comprise (Fig 7)

bull an RF end-cap trap for ionising and confining a single 171Yb+ ion withinan ultra-high vacuum chamber pumped by a small ion pump and non-evaporable getter pump The trap is driven by means of an ac voltage ofa few hundred volts at a drive frequency of sim10 MHz A small oven withseveral mg of 171Yb isotope is heated to provide a low flux of Yb atomsfrom which single ions can be ionised by electron bombardment from ahot wire filament within the trap potential well Low voltage dc is applied

Exp Astron

to additional electrodes to position the ion precisely at trap potentialcentre in order to minimise the ionrsquos micromotion at the drive frequencyThis needs automatic periodic monitoring and correction especially in theperiod after re-loading an ion into the trap External magnetic field coilsin three orthogonal axes allow the nulling of external fields and settingof a fixed field of sim1 μT The trap and coils are surrounded by mu-metalshielding to minimise external field changes during the orbit A level oftemperature control of 1 K of the region surrounding the trap is requiredfor maintaining the blackbody-induced frequency shift uncertainty withtemperature at the 10minus17 level

bull a laser platform to provide frequency doubling of an extended cavity739 nm diode laser for Doppler cooling of the ion on the 2S12(F = 1) minus2 P12 (F = 0) dipole transition at 3695 nm Light from an amplified diodelaser device frequency-doubled in a periodically poled LiTaO3 crystal insingle-pass will be used Typical laser powers used for driving the coolingtransition below saturation are about 2 μW for a beam waist in the trapof about 50 μm By modulation of the injection current of the 739 nmdiode laser at 147 GHz a sideband is generated that excites the 2S12 (F =0) rarr2 P12 (F = 1) ldquorepumperrdquo transition in order to avoid optical pump-ing between the ground hyperfine states

bull The platform also houses the auxiliary lasers at 935 and 638 nm The935 nm diode provides repumping of the ion from the 2 D32(F = 1)metastable level after occasional branching decays to this level during thecooling sequence The 638 nm diode allows fast recovery of the ion fromthe very-long-lived 2F72 metastable state after very occasional collisionaldecay to that state Currently extended cavity lasers are forseen butdistributed-feedback lasers may be available in the near future

bull a high NA lens imaging system and photomultiplier detection system torecord the statistics of 369 nm fluorescence quantum jumps as a functionof clock laser frequency step

bull a fibre system to deliver the various cooling auxiliary and clock light fromsource to trap making use of achromatic doublets where necessary at thefibre-free space interface for launching into the trap

bull a monitoring and control processor which provides the primary coolingand clock laser pulse magnetic field and detection sequencing to observeand lock to the ion clock transition frequency The processor also monitorsfrequency and amplitude data necessary to determine normal laser andion operational conditions and initiate resetting and recovery algorithmswhere necessary and laser unit failure The actual magnetic field presenton the atomic volume is determined and actively stabilized so that aninaccuracy at the level of 1 times 10minus17 can be achieved

bull a redundancy level of at least two units for cooling clock and repumperlasers plus similar for frequency doubling crystals All redundancy unitsfor each wavelength will be fibre multiplexed as standard allowing redun-dant unit activation on determination of prior unit failure

Exp Astron

For an optimized quantum-projection-noise-limited performance of thesingle-ion clock Ramsey interrogation with a cycle time equal to the lifetimeof the metastable D32 level of 50 ms will be performed leading to an Allandeviation of about 27 times 10minus15 (τ s)minus12 This is sim45 times 10minus17 at 1 h satisfyingrequired stabilities for the science objectives Current laboratory results are astability of 1 times 10minus15 at 30 s averaging down to 3 times 10minus16 at 1000 s (PTBGermany) Also comparisons between two independent traps showed agree-ment at the level of several parts in 10minus16 over a series of eight measurementruns

The 171Yb+ ion clock offers a second option namely the 2S12ndash2 F72 octu-pole transition at 476 nm (see Table 1 and Fig 6) [48] It has an extremelylong upper state lifetime low electric quadrupole and second order Zeemansystematic shifts and therefore is a candidate for an even higher performanceclock if a clock laser system with lower instability than the above is providedApart from the clock laser it shares the laser system of the quadrupoleclock Thus with moderate additional resources the Yb+ ion apparatus couldprovide two optical clocks in a single package and specifically perform directfrequency comparisons between the quadrupole and the octupole transitionwith a large sensitivity Aα

The ion clock will operate under microprocessor control providing thecontrol algorithms for the integrated pulse sequencing for cooling repumpingclock interrogation magnetic field switching cooling fluorescence monitoringThe controller will also provide error flags for critical processes and systemreset and correction where necessary This will include ion re-loading andmicromotion reduction algorithms Automatic monitoring of laser power andspectral quality with signal re-optimisation or laser failure determination isrequired Redundant laser units will be pre-aligned and multiplexed into thefibre delivery to the trap so that redundant units can be readily activated in thecase of diode laser failure

423 Microwave-optical local oscillator (MOLO)

This oscillator provides two ultrastable frequency outputs one in the opticalregion the other in the microwave region the two frequencies being coher-ently related The MOLO is composed of three subunits the clock laser thereference cavity and the frequency comb see Fig 8

The clock laser subunit delivers the light for the excitation of the opticalclock transition It is generated by an extended-cavity diode laser at a wave-length of 871 nm Most of the output wave of this laser is sent to the ionclock The laser subunit containing two lasers for redundancy will require3 kg 3 l 15 W The frequency instability of the clock laser must be superiorto the minimum clock instability 27 times 10minus15 (τ s)minus12 for τ up to the timeconstant (sim10 s) of the servo system that locks the laser to the ionrsquos resonancesignal In order to achieve this the laser frequency is stabilized to the reference

Exp Astron

871 nm clock laser

EO

Detector and servo

AOM Femtosecond

frequency comb

To optical clock 10 GHz To FCDP

Microwave-optical local oscillator

ULE cavity

Fig 8 Microwave-optical local oscillator (MOLO) consisting of clock laser reference ULE cavityand femtosecond frequency comb (OFC) Radiofrequency inputs and outputs are not shown EOelectro-optic modulator ULE ultra-low expansion AOM frequency shifter

cavity where the PoundndashDreverndashHall method is used About 10 μW of thelaser output are used for this

The system will exhibit sim04 Hz (sim1 times 10minus15 relative) linewidth and afrequency drift lt01 Hzs The cavity consists of two high-reflectivity mirrorsoptically contacted to a cylinder of diameter 70 mm length 150 mm made ofULE (a glass exhibiting ultra low thermal expansion coefficient) with a cavityfinesse of about 300000 Residual vibrations of the satellite (arms lt 1 times 10minus6 g)should not be a limiting factor to the desired linewidth for safety marginan optimized cavity support can reduce the sensitivity to accelerations Foracoustic and thermal isolation the cavity resides inside an aluminium vacuumchamber equipped with a small (3 ls) ion-getter pump Two stages of polishedaluminium shields are implemented around the cavity which are activelytemperature stabilized so that the cavity is precisely kept at the zero crossingtemperature of the ULE thermal expansion coefficient Heat application andremoval is by thermoelectric elements between the shield and the vacuumchamber Additionally the entire vacuum chamber is temperature-stabilized

During transportation and launch the cavity will be rigidly fixed insidethe vacuum chamber by pressing on it from the sides by piezo-mechanicalactuators allowing accelerations of several g After bringing the satellite onthe orbit the actuators will be released

The optical setup for frequency stabilization will be a classic Pound-Drever-Hall scheme in reflection using an electro-optic modulator (EO) Couplingof the clock laser radiation to the cavity is via single-mode polarization main-taining optical fiber The fiber in- and out-couplers are mounted on miniaturepiezo-motor driven multiaxis translation stages Microprocessor control ofthe translation stages performs a laser-to-fiber incoupling stabilization andlaser-to-cavity mode-matching optimization (positioning inaccuracy lt10 μmalignment inaccuracy lt10 μrad)

The frequency stability of the clock laser will be limited by thermal 1f noisein the mirrors with an fractional Allan deviation slightly below 1 times 10minus15 from1 s to a few 10 s Resources including control electronics are 15 l 12 kg 8 W

Exp Astron

A femtosecond fiber laser based optical frequency comb generator (OFCFig 8) will be used for the clock frequency comparisons and to obtain aspectrally ultrapure microwave local oscillator output for the operation ofPHARAO for the microwave link and for the GNSS receiver The fem-tosecond laser is based on erbium-doped fiber pumped by fiber-coupled laserdiodes It emits 100 fs pulses at a repetition rate of 250 MHz with an averagepower of 100 mW The emission spectrum is centered around 1560 nm Spec-tral broadening to more than one octave is obtained via nonlinear interactionin a microstructure fibre so that the range from about 1050 to 2100 nm iscovered

Second harmonic generation of a portion of the broadened spectrum allowsobtaining the required light for the measurement of the clock laser frequencyat 871 nm The OFC will be referenced to the clock laser frequency (afterintroduction a controllable shift via the frequency shifter AOM) by makingthe carrier envelope offset frequency vanish so that the repetition rate fr isa high-order subharmonic of the optical clock reference frequency f1 = m1 frThe link between the optical and the microwave domain is then simply thedistribution of the repetition rate frequency or more conveniently a high(order 40) harmonic of the comb repetition rate 10 GHz

It has been demonstrated by several groups that frequency comparison andfrequency division with an OFC can be performed to very high accuracy andstability beyond the limitations imposed by the optical clocks themselves Forexample a fiber laser OFC with an instability (Allan deviation) of 6 times 10minus17

in 1 s has been demonstrated [51] OFCs based on fiber technology possessgood long-term reliability and high power efficiency The projection for a spacequalified OFC based on current industrial terrestrial technology is 16 kg mass20 l volume and 38 W power requirement

424 Frequency comparison and distribution package (FCDP)

This package distributes the ultra-stable 10 GHz reference frequency outputof the MOLO to the MWL performs frequency comparisons between MOLOand PHARAO and generates a variety of different reference signals for theindividual instruments including the 250 MHz reference clock to the OFCAll signals are coherent to the MOLO In the FCDP a USO is locked to theMOLO signal by a digital PLL A divider chain generates timing pulses (1 pps1000 pps etc) and keeps a continuous local time scale From the digital phasecomparison data the phase noise density spectrum between the USO and theMOLO is available by telemetry (TM) to assess the overall frequency purityin-flight

As the FCDP distributes the full performance reference from the MOLOdirectly to the MWL there is minimum impact of the FCDP on overall clockcomparison performance The USO-phase lock loop uses a similar designas for ACES-FCDP The digital PLL with linear phase detector allows theUSO to be slightly offset from its reference which is selectable by groundcommand This results in an accurate on-board time scale despite any offset

Exp Astron

of the MOLO reference frequency which can be synchronised to UTC Thefrequency synthesis chain re-uses modules developed for ACES-FCDP andtheir frequency inaccuracy Expected relative uncertainty (Allan deviation)of the frequency comparison between two signals provided to the FCDP is47 times 10minus17 at 1000 s 95 times 10minus18 at 10000 s

During non-availability of the MOLO ie during system start-up andcontingency conditions the USO runs as flywheel (ie it is unlocked withinstability lt15 times 10minus13 from 1 to 100 s) It will still provide useable butless accurate references to the instruments including the reference to OFCCommon view ground-to-ground time- and frequency comparison and supplyof the remaining instruments remain possible at the USO provided accuracy

Mass volume and power consumption are 8 kg 7 l 12 W Controldata rateis 10 kBday 50 kBday (FCDP alone)

425 Microwave link (MWL)

The microwave link is a two-way triple-frequency microwave link (Fig 9) Itis designed to perform accurate frequency comparison as well as time-transfer

Ku-Band Up-linkPower Tx 5 WCarrier 13475 GHzPN-Code 100 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

S-Band Down-linkPower Tx 1 WCarrier 2248 MHzPN-Code 1 MChips1pps 1 time marker sData 5 kBits

Ku-Band Down-linkPower Tx 1 WCarrier 1470333 GHzPN-Code 100 MChips1pps 1 time marker sData 5 kBits

EGE SC

S

Ku

KaKa-Band Up-linkPower Tx 5 WCarrier 345 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

Ka-Band Down-LinkPower Tx 1 WCarrier 32 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 5 kBits

GS Antenna diameterEGE terminal 1 mACES terminal 06 m

Fig 9 EGE microwave two-way link concept and characteristics See text for a description Thelink will be able to perform frequency comparisons with up to ten ground stations simultaneously(only one active comparison is shown) pps pulse per second Tx transmitter SC spacecraft

Exp Astron

between on-board and on-ground clocks Two-way operation allows for com-pensation of the effects due to spacecraft location and its motion relative toground Its two bi-directional links operate in Ka-band (32345 GHz) andKu-band (147135 GHz) The high carrier frequency of the up- and down-links allows for a noticeable reduction of the ionospheric delay A third down-link channel is in S-band (22 GHz) and is used to determine the ionospheretotal electron content MWL uses a combination of carrier phase and codephase data to measure with high precision the absolute ionospheric delayas well as ionospheric delay variations during a ground contact The useof three frequency bands allows to resolve third order ionospheric termsKey modulation parameters are Ka band Tone 250 MHz Ku band PN100 MChips S-band 1 MChips

The EGE-MWL concept is a considerably improved version of the ACES-MWL Microwave carrier generation is directly from the 10 GHz microwaveoutput available from the MOLO and having exceptionally low close-in phasenoise without intermediate frequency dividers or other complex synthesiserstages commonly used in RF transponders The high frequency carriers arekey to achieve the desired frequency comparison accuracy

The Ka-band carrier is generated by tripling the FCDP-generated signalThe Ku-band is subsequently generated by division by two using a verystable regenerative divider Finally the assigned radio frequency is obtainedby adding a low frequency signal using an image reject mixer A mixedanalog-digital receiver design will enhance its performance by use of a flexibledigital signal processing and digital control loops The number of simultaneousground contacts is increased from four (as in the ACES mission) to ten or moreusing the digital concept while the number of physical receiver channels perband is reduced from four to two one for operation to ground and one for in-flight calibration The concept allows adaptation of receiver loop parametersto optimise for the different portions of a highly elliptic orbit while reducingsystematic uncertainties due to high Doppler and Doppler rate Real-timedata exchange to ground terminals is by additional data modulation on all RFsignals The spacecraft carries two different small antennae for perigee and forapogee with gains of 8 dBi 20 dBi beamwidths 100 (full earth coverage) 14respectively power is 1 W for each

The MWL end-to-end performance between the attached clocks includingspace- and ground terminals is predicted to be a relative Allan deviation of1 times 10minus16 at 1000 s (perigee) and 1 times 10minus17 at 10000 s (apogee) Thesevalues are based on conservative implementation losses and other instrumentallosses assumed as 6 dB at apogee and 10 dB at perigee including additionallosses due to atmosphere An attitude knowledge of 1 is required

The MWL link will also support precise orbitographyndashcomplementary tothe GNSS receiver Generation of ranging data is simultaneous to two-waytime transfer and is available under all weather conditions Orbit determina-tion accuracy is expected to be better than 1 cm well matching EGE clockrequirements The value is expected based on demonstrated results fromother missions (METEOR ERS-2 TOPEXPOSEIDON) Laser ranging to

Exp Astron

the CCR array is used a few times during the mission to calibrate the MWLranging

In case of non-availability of the ultrapure signal from MOLO an additional10 GHz output is available from the FCDP to support ground-to-ground clockcomparison in common-view mode even after the nominal mission periodwhen the neither the clocks nor MOLO may be available any longer Basedon the PRARE experimentrsquos 12 years of operation this can be assumed forthe FCDPMWL combination as well The potential extended operational pe-riod makes the proposed configuration particularly attractive to metrologicallaboratories that rely on availability of long-lasting data in support of thegeneration of UTC

The resources of MWL are as follows mass 18 kg volume 18 l power55 W TC 2 kBday TM 120 MBday

426 Auxiliary units

The star camera assembly on the EGE spacecraft will provide real-timeattitude data primarily for the orientation of the spacecraft (with accuracy of1 in real-time) and the proper reduction of the GNSSSLR data

A GNSS receiver is used for the precise orbit determination time andfrequency transfer coarse positioning (lt1 m) for real time use by the Attitudeand Orbit Control as well as the back-up for time tagging of all payloaddata For the time and frequency transfer the receiver is nominally using theonboard frequency reference The receiver assembly includes three specialantennas By the year 2015 more than 100 GNSS satellites are expected tobe available for precise orbit determination permitting orbit inaccuracy of 1ndash2 cm RMS for altitudes up to approx 10000 km At apogee a combinationof GNSS signals and MWL ranging will be used with expected inaccuracyof 10 cm

5 Basic spacecraft specifications

The mission foresees a single spacecraft with a mission duration limited toapprox 3 years for cost reasons Attitude control along three axes withless than 1 error on a timescale of 1000 s is required The control can beperformed with cold gas thrusters magnetotorquers or reaction wheels Theselection will be done on the basis of the vibrations induced by these systems(desired level arms lt 10minus6 g)

Cold gas thrusters or monopropellant thrusters will be required to maintainthe orbit parameters during the mission by small orbit correction manoeuvresand will also guarantee a proper tuning of the orbit parameters after the releasefrom the launcher The ground segment (telemetry control and data transfer)could be via non-ESA stations user-operated to save costs

Further resources are indicated in Table 2

Exp Astron

The satellite bus design drivers are nadir-pointing microwave antennasthermal stability for the payload instruments no strong magnetic sources lowmicrovibrations harsh radiation environment due to crossing of the radiationbelts once per orbit power demand by science payload (with the possibility toreduce power demand by stand-by of some components during specific missionphases eg eclipse) mass limit due to required high altitude with low costlaunch

The satellite bus design can be based on established platforms such asFLEXBUS or PRIMA used on several past missions Their design philosophydiffers one having body-mounted solar arrays the other wings The latter canprovide larger amounts of power The vibration level induced by the panelswhen they are fixed and the duration of solar array drive periods per orbitwill have to be considered in the selection The FLEXBUS concept was usedfor a mission already requiring a low level of residual accelerations and theplatform is optimised for this parameter

For most of the satellite components one can consider using compo-nents and electronic units from navigation or communication satellites whichtypically cruise at altitudes between 15000 and 25000 km altitude LISAPathfinder which will cross the radiation belts during spiralling to theLagrange point will face similar problems concerning radiation as EGE as wellas XMM which is in a Molniya-type orbit with similar radiation characteristics

The instruments will therefore be located in the satellite structure such asto realise ionizing radiation shielding by the spacecraft structure and non-sensitive equipment to the maximum extent possible the need for additionalradiation shielding (including trapped proton radiation) for especially sensitiveparts of the instruments must be determined by ground tests It should bementioned that an orbit with higher perigee and less radiation dose is also anoption as it leads to a tolerable reduction of the science goals

The required plusmn3C temperature stability at the interface to the payloadenclosure is feasible The thermal control will employ the usual means (anenclosure with high thermal inertia heaters and a suitable radiator) Thelocation of the radiator will need a special study given the non-standard orbit

Magnetic field present during the mission can affect the properties of theonboard clock Using SPENVIS the magnetic field has been calculated alongthe spacecraft orbit the field intensity presents maxima around 02 G anda minimum value of the order of 10minus3 G As discussed above this is not alimitation to the instrument performance

6 Discussion and conclusion

The accuracy of the measurements depends crucially on the stability andaccuracy of the clocks While the performance of ground clocks (even thoseeventually deployed in new ground stations dedicated to EGE) will likelyimprove significantly from the present level the space clock design will beldquofrozenrdquo approx 10 years before data acquisition Thus their performance

Exp Astron

level will be significantly below the future ground station clock performanceAs specification we have chosen a performance level that has been demon-strated with laboratory breadboards recently We see no fundamental reasonthat this laboratory performance cannot also be achieved in a space ion opticalclock The experience with the PHARAO cold atom clock has shown thatan engineering model of a laser-based cold atom clock has a performancesimilar to that of a laboratory prototype The perturbation of the ion clock fre-quency due to magnetic fields along the orbit can be modelled and suppressedwith shielding and the residual effects measured and compensated for Themoderate temperature stability (plusmn 3C) to be provided in the experimentalbay can be achieved with thermal shields Accurate temperature controlof particular units is done by active stabilization of the units themselves

One important aspect is radiation which may affect all optical and trapcomponents as well as the lasers and reference cavity The radiation envi-ronment on the EGE orbit is significantly harsher than the near-earth ISSorbit on which ACES will operate The radiation effects must therefore becharacterized comprehensively with appropriate tests and then the radiationshield for the EGE instruments designed

In summary the Einstein Gravity Explorer mission (EGE) is a satellitemission aimed at a precise measurement of the properties of space-time usingatomic clocks By testing one of the most fundamental predictions of EinsteinrsquosTheory of General Relativity the gravitational redshift and thereby searchingfor quantum effects in gravity it explores one of the most important frontiers infundamental physics The EGE mission reaches beyond fundamental physicsproviding an important spin-off in geophysics Furthermore the technologicalaspects of the mission are ground-breaking for future time-keeping and ref-erence frame systems in space and thus are also expected to be of wide useto science and technology in general The payload of the mission has beenconceived as a compromise between cost and performance needing to fit thecost envelope of an ESA medium-class mission1

Acknowledgement We thank S Cesare for helpful information

References

1 Will CM The confrontation between general relativity and experiment Living Rev Relativ9 3 (2006)

2 Ni W-T Empirical foundations of the relativistic gravity Int J Mod Phys D 14 901 (2005)doi101142S0218271805007139

1Finally we point out that a spacecraft with a significantly reduced and thus lower-cost payload (noclocks only MOLO FCDP MWL) on a circular orbit of appropriate height could be used for along-duration earth science mission dedicated specifically to the measurement of the geopotential

Exp Astron

Satellite tracking via laser ranging Laser ranging is alternative to MWL rang-ing and also serves for absolute calibration of the latter For this purpose anon-board corner cube reflector (CCR) is provided Ranging will be done withthe International Laser Ranging Service Ideally the satellite will be rangedat every perigee passage and every apogee passage simultaneously with thefrequency comparison procedures Todayrsquos ranging precision is sufficient tosatisfy the science goals For apogee observations in the northern hemispherea large number of laser ranging stations is available The number of stationscapable of ranging at or near perigee in the southern hemisphere is six andthus is sufficient even considering unfavourable weather conditions at someof these The MWL ground stations are able to range even under thoseconditions

322 Control ground segment

Average data transfer rate is estimated at 300 MBday (approx 40 kBs onaverage) Data will be stored on board (2 GB capacity required) and will betransferred to a receiving station on the Earth once a day

XPLC

incl Data StorageIF to TMTC Link

HK amp TMTC

to ground

Opt Clock

AOM

MOLO

PHARAO

FCDP

10 GHz

100 MHz

250 MHz

GNSS Rx

MWLTampF Link

uplink Data

UTC(k)

TampF Space vs Ground

TM from GS

Tim

e S

cale

Orbit GNSS Time UTC

TampF Space vs Ground

MampC Time Scale

MampC

MampC

MampC

MampC

Mamp

C

PHARAO vs Opt Clock

PHARAO vs Opt Clock

10 GHz amp Epoch

10 MHz

amp Epoch

Epoch

10 MHz

amp Epoch

Opt Link

Fig 4 Schematic of the payload showing subunits and their signals The results of the frequencycomparison between the two on-board clocks (optical clock and microwave clock PHARAO) areobtained from a combination of the frequency values applied to the frequency shifters AOM (seeFigs 7 and 8) and those output from the frequency control and distribution package (FCDP)as indicated by the arrows ldquoPHARAO vs Opt Clockrdquo The satellite clock to ground clockcomparison results are output from the MWL XPLC is the on-board computer and is involvedin generation of signals for instrument control The optical link shown dashed is a possible add-onproviding an alternative to the MWL GS ground stations MampC monitoring and control HKhousekeeping TampF time and frequency

Exp Astron

4 Main payload instruments

41 Overview

The satellite payload consists of a pair of cold-atom clocks a microwave-optical local oscillator (MOLO) an on-board frequency distribution unit(FCDP) a satellite-to-ground clock comparison unit (MWL) and the onboardcomputercontrol system (XPLC) see Fig 4 Auxiliary units are the posi-tion determination GPSGalileo receiver (GNSS Rx) attitude monitor (startracker not shown) and the corner cube reflector array (CCR not shown)

The science goals require clocks of very high frequency stability on thetimescales of half the orbital period (6 h) as well as perigee passage (sim1000 s)see Fig 2 Only clocks based on cold atoms are suited for the science goalsTable 1 reports values for a subset of atomic clocks that have been investigatednot including clocks that require cryogenic operation or are too complex foruse in an medium-class mission The choice of a clock pair rather than a singleone fulfils several requirements it addresses more science goals improves thevalidity of some science results and it enables calibration accuracy evaluationand redundancy The validity of the science results is enhanced since theclocks differ not only in the employed atoms but also in terms of havingdifferent sensitivity to perturbations by electric and magnetic fields electronicserrors and misalignments A comparison of the results arising from each clockallows for an understanding of measurement errors Both the clocksrsquo and thefrequency comparison unitrsquos performances will be determined after launchand regularly verified during the course of the mission The only reliable

Table 1 Properties of different clock candidates for EGE

Clock Clock Fractional σ = Fractional Sensitivity Figure of Sensitivityatom frequency inaccuracy instability to α Aα merit |Aα| to meq

at 1000 s σ [1016] QCD

Sr+ 445 THz 3 times 10minus17 (p) 5 times 10minus17 (p) 042 08 (p) ndashYb+ (1) 688 THz 3 times 10minus16 (d) 3 times 10minus16 (d) 088 029 (d) ndash

3 times 10minus17 (p) 1 (p)Yb+ (2) 642 THz lt 1 times 10minus17 (p) 1 times 10minus17 (p) minus53 53 (p) ndashSr 435 THz 1 times 10minus16 (d) 1 times 10minus16 (d) 009 009 (d) ndash

lt 1 times 10minus17 (p) 09 (p)Yb 519 THz 7 times 10minus16 (d) 1 times 10minus16 (d) 031 03 (d) ndash

lt 1 times 10minus17 (p) 1 times 10minus17 (p) 31 (p)Cs 91 GHz 2 times 10minus16 (d) 1 times 10minus15 (d) 275 027 (d) 1Yb+ 126 GHz 1 times 10minus15 (p) 1 times 10minus15 (p) 34 034 (p) 1

A relevant figure of merit is the ratio of |AX | and fractional frequency instability σ on the perigeevisibility timescale The Cs clock is the only high performance clock with adequate sensitivity tothe particle masses me mq The indicated Cs clock performance was achieved using a microwavelocal oscillator based on a cryogenic oscillator On EGE the latter is to be replaced by an opticallocal oscillator-based device (MOLO) with similar short-term performance but without the needto operate at cryogenic temperature The clock instability at 10000 s is typically similar to theindicated inaccuracy(p) projected (d) demonstrated (1) quadropole (2) octupole

Exp Astron

way to achieve this is by having at least two clocks on board The on-boardclock-clock comparison provides performance evaluation independent of theground-satellite comparison unit Once the satellite clock performances areestablished the link performance can be established as well

The clocks are similar in basic structure but differ in the detailed implemen-tation They consist of

a source of atomsa preparation subunit which cools the atoms to near-standstill (millikelvintemperature or below)a trapping subunit that confines the atom(s) to a tiny volumean interrogation subunita clock control electronics package

The clocks share a common central unit the optical-microwave local oscil-lator which serves as a flywheel oscillator with outstanding stability on shorttimescales (01ndash10 s) for both clocks It consists of a laser stabilized to an ultra-low loss optical resonator made of ultra-low-expansion glass A microwavesignal is derived from the laser by phase-locking a frequency comb to the laserwave and using a multiple of the repetition rate

While the subunits of any optical clock have a similar technological basis(optics electrooptics acoustooptics thermal control vacuum systems) theydiffer in their specifications Depending on the choice of the atom species de-livery confinement and interrogation methods clocks of different perfor-mance result A number of laboratory optical clock demonstrators have beendeveloped worldwide This ensures availability of the know-how for the indus-trial implementation of the flight models

The first clock is a microwave clock based on the interrogation of slowlymoving ensembles of cold Cs atoms It is essentially based on the PHARAOclock [5 6] developed to engineering model level by CNES in the frame-work of the ESA ISS mission ACES In EGE this clock is upgraded usingthe microwave-optical local oscillator This improves significantly the clockperformance The specification is a relative instability (Allan deviation) of3 times 10minus14 (τ s)minus12 ie at the integration time τ = 1000 s it drops to 1 times 10minus15and to 3 times 10minus16 at 10000 s While this specification is at least a factor 3 lessstringent than for the optical clocks the significantly lower cost of obtaining aflight model and the sensitivity of its frequency to the particle masses makes ita suitable choice

The optical clock baseline instrument is a single-ion optical clock Amongvarious suitable ion species (Hg+ Al+ Sr+ Yb+) [44ndash47] the Yb+ ion hasbeen selected Its breadboard performance is established at the instability levelof 3 times 10minus16 over 1000 s it exhibits a large sensitivity coefficient Aα hassmall sensitivity to magnetic fields and exhibits significant potential for furtherimprovements (use of the octupole transition with sim6 times larger Aα andlower overall instability [48] see below) A trapped Yb+ ion is also a veryrobust system it has been demonstrated that a single Yb+ ion can remaintrapped uninterruptedly for many months Finally Yb+ is the only ion with

Exp Astron

which optical clocks demonstrators have been operated in not a single but twomajor metrology laboratories

Another attractive optical clock is based on ensembles of cold neutral Sror Yb atoms trapped in an optical lattice produced by laser beams This clocktype has a lower instability than a single-ion clock thanks to the large numberof atoms (sim105) used Its development is in progress in several laboratoriesin the world (in part supported by space agencies) and has to date reached6 times 10minus17 instability at 5000 s and 1 times 10minus16 uncertainty [49 50] with potentialfor even higher performance However it is more complex and costly thanthe Yb+ ion clock since the cold atom preparation and trapping subunitsinclude more components require higher laser powers and the respectiveoperating procedures are more complex It is therefore not considered for thepresent mission designed to be developed in the very near future with limitedresources but is an interesting candidate for missions without such constraintsIts estimated physical parameters are indicated in Table 2 for comparison

42 Main instruments

421 Microwave clock

The PHARAO cold atom clock is one of the two atomic clocks of the spacemission ACES managed by ESA It is proposed here to include a secondversion of this instrument as its costperformance ratio is very attractive forthe EGE scientific objectives Its concept is very similar to ground basedatomic fountains but with the major difference of zero-g operation Atomsslowly launched in free flight cross two microwave fields tuned to the transitionbetween the two hyperfine levels of the cesium ground state The interrogationmethod based on two separate oscillating fields (Ramsey scheme) allowsthe detection of an atomic line whose width is inversely proportional to thetransit time between the two microwave cavities The resonant microwavefield at 9192631770 GHz (SI definition of the second) is synthesized startingfrom an ultrastable quartz oscillator (USO) in the FCDP phaselocked to themicrowave output of the microwave-optical local oscillator and stabilized tothe clock line using the error signal generated by the cesium resonator In thisway the intrinsic qualities of the cesium hyperfine transition both in terms ofaccuracy and frequency stability are transferred to the USO In a microgravityenvironment the velocity of atoms along the ballistic trajectories is constantand can be changed continuously over almost two orders of magnitude (5ndash500 cms) Differently from atomic fountain clocks presently operated onground very long interaction times (up to few seconds) will be possible whilekeeping reasonable the size of the instrument

The instrument consists of 4 main elements a laser source which providesthe light required to cool and detect the cesium atoms (Fig 5 left) a frequencychain which generates the 92 GHz microwave signal a cesium vacuum tubewhere the interaction between the microwave field and the atoms occurs

Exp Astron

Table 2 Resources and technology readiness level (TRL) of the EGE spacecraft

Unit Volume Mass Power TRL Comment[l] [kg] [W]

Ion clock 100 70 60Atom source 4Atom preparation 6Atom trap 4Atom interrogation 6Clock control 6

Microwave clock 200 91 110 8 TRL by end 2008Local oscillator

Stabilized laser 18 15 23 6Frequency comb 20 16 38 6

Microwave link 18 18 55 8FCDP 7 8 12 8GPS receiver 2 5 7 9CCR 2 2 0 9Payload sum 367 225 305Thermal control 14 17 30 8OBDH 20 24 38 9Star tracker 1 1 8 9Radiation shield 10 40 0 9Support structure 170 204 0 9Power incl solar arr 50 58 10 9Propulsion 21 25 20 9AOCS 36 43 82 9TMTC 8 10 15 9Satellite bus sum 330 422 203 Flexbus conceptPayload plus bus incl 20 margin 8364 7764 6096Propellant 40Launch adapter 100SC wet masspower incl Margins 9164Lattice clock 135 105 152 Alternative clock

Atom source 4Atom preparation 6Atom trap 4Atom interrogation 6Clock control 6

Also included are the resources for an optical lattice clock based on neutral Strontium atoms

(Fig 5 right) and a computer The engineering model of PHARAO is fullyassembled at CNES Toulouse and undergoes functional and performancetests until the end of 2008 The operating temperature range is 10ndash33 andnon-operating temperatures are minus45 +60 The cesium tube consists in aUHV chamber pumped by getters and a 3 ls ion pump a cesium reservoira microwave cavity coils to provide a uniform magnetic field and three layersof magnetic shields Vacuum windows with fiber-optics collimators enable thetransmission of the cooling and detection light onto the cesium atoms

In autonomous mode PHARAO uses the USO as interrogation oscillatorand will provide a clock signal with fractional frequency stability below1 times 10minus13 (τs) minus12 and inaccuracy near 1 times 10minus16 When using the microwave-

Exp Astron

Fig 5 Left The PHARAO laser source engineering model with cover removed Dimensionsare 530 times 350 times 150 mm3 and the mass is 200 kg Ten polarization maintaining optical fibers(yellow) guide the laser beams to the cesium tube All diode lasers are mounted on Peltier coolersfor temperature regulation within 2 mK (courtesy EADS SODERN) Right The PHARAOengineering model cesium tube without the two external magnetic shields The volume is 990 times336 times 444 mm3 and the total mass is 44 kg

optical local oscillator PHARAO will provide a reduced frequency instabilityof 3 times 10minus14 (τs) minus12 This corresponds to 1 times 10minus15 at 1000 s and 3 times 10minus16

at 104 s

422 Ion optical clock

The 171Yb+ ion clock shown in Figs 6 and 7 provides an atomic frequencyreference based on a single 171Yb+ ion confined within an RF electric end-captrap (similar to a Paul trap) within an ultra high vacuum (UHV) chamber andlaser cooled to sim1 mK close to the Doppler cooling limit As a result the ionrsquosfirst order Doppler motion is completely removed and the ion experienceslittle collisional perturbation from its environment Under these conditionsthe narrow linewidth 2S12 (F = 0 mF = 0) minus 2 D32 (F = 2 mF = 0) quadrupole

Fig 6 Left Simplified level scheme of 171Yb+ showing relevant cooling (blue) auxiliary (red)quadrupole clock (green) and octupole clock (dark blue) transitions δνnat is the natural linewidthof the clock transition Middle close-up of Paul-type RF trap (PTB Braunschweig) the ion istrapped in the hole (diameter 14 mm) Right Complete ion trap with vacuum flange height 10 cm(NPL Teddington)

Exp Astron

935 nm D32-state repumper laser

638 nm F72-state repumper laser

739 nm extended cavity diode laser

AOM

PMT

Ion trap UHV chamber

vacuum pumps

Fibre-free space interface

Mu-metal shielding

EC

T

B

Magnetic field

Temperature

Ion fluorescence Processor

Clock laser frequency step and control

4355 nm

Frequengy doubling II

3695 nm cooling wave

Low-drift cavity

Low-drift cavity 171Yb+ 435 nm ion clock unit

876 nm clock laser wave from MOLO

Frequengy doubling

Fig 7 Simplified schematic of the single-ion optical clock unit AOM acousto-optic frequencyshifter PMT photomultiplier tube EC endcap T temperature stabilization system B magneticfield

clock transition at 4355 nm can be interrogated by spectrally narrow and stableclock laser light A diagram of the simplified energy level scheme for 171Yb+showing the relevant cooling auxiliary and clock transitions is given in Fig 6All these wavelengths are provided by diode lasers operated in fundamentalor frequency doubled mode The clock laser light is obtained from the 871 nmwave provided by the MOLO unit (see Section 423) after frequency-doublingin single pass by a KNbO3 nonlinear-optical crystal to 4355 nm The clocktransition spectral profile is observed by stepping the clock laser frequency bymeans of an acousto-optic frequency shifter (AOM in Fig 7) and recordingthe statistics of the quantum jumps in cooling laser induced fluorescence asa function of this frequency Under clock operation the AOM repeatedlysteps back and forward between the half-intensity points on the transitionprofile monitoring the signal imbalance between these points Any detectedimbalance is servo-corrected to zero by feedback to the AOM

In detail the optical clock architecture will comprise (Fig 7)

bull an RF end-cap trap for ionising and confining a single 171Yb+ ion withinan ultra-high vacuum chamber pumped by a small ion pump and non-evaporable getter pump The trap is driven by means of an ac voltage ofa few hundred volts at a drive frequency of sim10 MHz A small oven withseveral mg of 171Yb isotope is heated to provide a low flux of Yb atomsfrom which single ions can be ionised by electron bombardment from ahot wire filament within the trap potential well Low voltage dc is applied

Exp Astron

to additional electrodes to position the ion precisely at trap potentialcentre in order to minimise the ionrsquos micromotion at the drive frequencyThis needs automatic periodic monitoring and correction especially in theperiod after re-loading an ion into the trap External magnetic field coilsin three orthogonal axes allow the nulling of external fields and settingof a fixed field of sim1 μT The trap and coils are surrounded by mu-metalshielding to minimise external field changes during the orbit A level oftemperature control of 1 K of the region surrounding the trap is requiredfor maintaining the blackbody-induced frequency shift uncertainty withtemperature at the 10minus17 level

bull a laser platform to provide frequency doubling of an extended cavity739 nm diode laser for Doppler cooling of the ion on the 2S12(F = 1) minus2 P12 (F = 0) dipole transition at 3695 nm Light from an amplified diodelaser device frequency-doubled in a periodically poled LiTaO3 crystal insingle-pass will be used Typical laser powers used for driving the coolingtransition below saturation are about 2 μW for a beam waist in the trapof about 50 μm By modulation of the injection current of the 739 nmdiode laser at 147 GHz a sideband is generated that excites the 2S12 (F =0) rarr2 P12 (F = 1) ldquorepumperrdquo transition in order to avoid optical pump-ing between the ground hyperfine states

bull The platform also houses the auxiliary lasers at 935 and 638 nm The935 nm diode provides repumping of the ion from the 2 D32(F = 1)metastable level after occasional branching decays to this level during thecooling sequence The 638 nm diode allows fast recovery of the ion fromthe very-long-lived 2F72 metastable state after very occasional collisionaldecay to that state Currently extended cavity lasers are forseen butdistributed-feedback lasers may be available in the near future

bull a high NA lens imaging system and photomultiplier detection system torecord the statistics of 369 nm fluorescence quantum jumps as a functionof clock laser frequency step

bull a fibre system to deliver the various cooling auxiliary and clock light fromsource to trap making use of achromatic doublets where necessary at thefibre-free space interface for launching into the trap

bull a monitoring and control processor which provides the primary coolingand clock laser pulse magnetic field and detection sequencing to observeand lock to the ion clock transition frequency The processor also monitorsfrequency and amplitude data necessary to determine normal laser andion operational conditions and initiate resetting and recovery algorithmswhere necessary and laser unit failure The actual magnetic field presenton the atomic volume is determined and actively stabilized so that aninaccuracy at the level of 1 times 10minus17 can be achieved

bull a redundancy level of at least two units for cooling clock and repumperlasers plus similar for frequency doubling crystals All redundancy unitsfor each wavelength will be fibre multiplexed as standard allowing redun-dant unit activation on determination of prior unit failure

Exp Astron

For an optimized quantum-projection-noise-limited performance of thesingle-ion clock Ramsey interrogation with a cycle time equal to the lifetimeof the metastable D32 level of 50 ms will be performed leading to an Allandeviation of about 27 times 10minus15 (τ s)minus12 This is sim45 times 10minus17 at 1 h satisfyingrequired stabilities for the science objectives Current laboratory results are astability of 1 times 10minus15 at 30 s averaging down to 3 times 10minus16 at 1000 s (PTBGermany) Also comparisons between two independent traps showed agree-ment at the level of several parts in 10minus16 over a series of eight measurementruns

The 171Yb+ ion clock offers a second option namely the 2S12ndash2 F72 octu-pole transition at 476 nm (see Table 1 and Fig 6) [48] It has an extremelylong upper state lifetime low electric quadrupole and second order Zeemansystematic shifts and therefore is a candidate for an even higher performanceclock if a clock laser system with lower instability than the above is providedApart from the clock laser it shares the laser system of the quadrupoleclock Thus with moderate additional resources the Yb+ ion apparatus couldprovide two optical clocks in a single package and specifically perform directfrequency comparisons between the quadrupole and the octupole transitionwith a large sensitivity Aα

The ion clock will operate under microprocessor control providing thecontrol algorithms for the integrated pulse sequencing for cooling repumpingclock interrogation magnetic field switching cooling fluorescence monitoringThe controller will also provide error flags for critical processes and systemreset and correction where necessary This will include ion re-loading andmicromotion reduction algorithms Automatic monitoring of laser power andspectral quality with signal re-optimisation or laser failure determination isrequired Redundant laser units will be pre-aligned and multiplexed into thefibre delivery to the trap so that redundant units can be readily activated in thecase of diode laser failure

423 Microwave-optical local oscillator (MOLO)

This oscillator provides two ultrastable frequency outputs one in the opticalregion the other in the microwave region the two frequencies being coher-ently related The MOLO is composed of three subunits the clock laser thereference cavity and the frequency comb see Fig 8

The clock laser subunit delivers the light for the excitation of the opticalclock transition It is generated by an extended-cavity diode laser at a wave-length of 871 nm Most of the output wave of this laser is sent to the ionclock The laser subunit containing two lasers for redundancy will require3 kg 3 l 15 W The frequency instability of the clock laser must be superiorto the minimum clock instability 27 times 10minus15 (τ s)minus12 for τ up to the timeconstant (sim10 s) of the servo system that locks the laser to the ionrsquos resonancesignal In order to achieve this the laser frequency is stabilized to the reference

Exp Astron

871 nm clock laser

EO

Detector and servo

AOM Femtosecond

frequency comb

To optical clock 10 GHz To FCDP

Microwave-optical local oscillator

ULE cavity

Fig 8 Microwave-optical local oscillator (MOLO) consisting of clock laser reference ULE cavityand femtosecond frequency comb (OFC) Radiofrequency inputs and outputs are not shown EOelectro-optic modulator ULE ultra-low expansion AOM frequency shifter

cavity where the PoundndashDreverndashHall method is used About 10 μW of thelaser output are used for this

The system will exhibit sim04 Hz (sim1 times 10minus15 relative) linewidth and afrequency drift lt01 Hzs The cavity consists of two high-reflectivity mirrorsoptically contacted to a cylinder of diameter 70 mm length 150 mm made ofULE (a glass exhibiting ultra low thermal expansion coefficient) with a cavityfinesse of about 300000 Residual vibrations of the satellite (arms lt 1 times 10minus6 g)should not be a limiting factor to the desired linewidth for safety marginan optimized cavity support can reduce the sensitivity to accelerations Foracoustic and thermal isolation the cavity resides inside an aluminium vacuumchamber equipped with a small (3 ls) ion-getter pump Two stages of polishedaluminium shields are implemented around the cavity which are activelytemperature stabilized so that the cavity is precisely kept at the zero crossingtemperature of the ULE thermal expansion coefficient Heat application andremoval is by thermoelectric elements between the shield and the vacuumchamber Additionally the entire vacuum chamber is temperature-stabilized

During transportation and launch the cavity will be rigidly fixed insidethe vacuum chamber by pressing on it from the sides by piezo-mechanicalactuators allowing accelerations of several g After bringing the satellite onthe orbit the actuators will be released

The optical setup for frequency stabilization will be a classic Pound-Drever-Hall scheme in reflection using an electro-optic modulator (EO) Couplingof the clock laser radiation to the cavity is via single-mode polarization main-taining optical fiber The fiber in- and out-couplers are mounted on miniaturepiezo-motor driven multiaxis translation stages Microprocessor control ofthe translation stages performs a laser-to-fiber incoupling stabilization andlaser-to-cavity mode-matching optimization (positioning inaccuracy lt10 μmalignment inaccuracy lt10 μrad)

The frequency stability of the clock laser will be limited by thermal 1f noisein the mirrors with an fractional Allan deviation slightly below 1 times 10minus15 from1 s to a few 10 s Resources including control electronics are 15 l 12 kg 8 W

Exp Astron

A femtosecond fiber laser based optical frequency comb generator (OFCFig 8) will be used for the clock frequency comparisons and to obtain aspectrally ultrapure microwave local oscillator output for the operation ofPHARAO for the microwave link and for the GNSS receiver The fem-tosecond laser is based on erbium-doped fiber pumped by fiber-coupled laserdiodes It emits 100 fs pulses at a repetition rate of 250 MHz with an averagepower of 100 mW The emission spectrum is centered around 1560 nm Spec-tral broadening to more than one octave is obtained via nonlinear interactionin a microstructure fibre so that the range from about 1050 to 2100 nm iscovered

Second harmonic generation of a portion of the broadened spectrum allowsobtaining the required light for the measurement of the clock laser frequencyat 871 nm The OFC will be referenced to the clock laser frequency (afterintroduction a controllable shift via the frequency shifter AOM) by makingthe carrier envelope offset frequency vanish so that the repetition rate fr isa high-order subharmonic of the optical clock reference frequency f1 = m1 frThe link between the optical and the microwave domain is then simply thedistribution of the repetition rate frequency or more conveniently a high(order 40) harmonic of the comb repetition rate 10 GHz

It has been demonstrated by several groups that frequency comparison andfrequency division with an OFC can be performed to very high accuracy andstability beyond the limitations imposed by the optical clocks themselves Forexample a fiber laser OFC with an instability (Allan deviation) of 6 times 10minus17

in 1 s has been demonstrated [51] OFCs based on fiber technology possessgood long-term reliability and high power efficiency The projection for a spacequalified OFC based on current industrial terrestrial technology is 16 kg mass20 l volume and 38 W power requirement

424 Frequency comparison and distribution package (FCDP)

This package distributes the ultra-stable 10 GHz reference frequency outputof the MOLO to the MWL performs frequency comparisons between MOLOand PHARAO and generates a variety of different reference signals for theindividual instruments including the 250 MHz reference clock to the OFCAll signals are coherent to the MOLO In the FCDP a USO is locked to theMOLO signal by a digital PLL A divider chain generates timing pulses (1 pps1000 pps etc) and keeps a continuous local time scale From the digital phasecomparison data the phase noise density spectrum between the USO and theMOLO is available by telemetry (TM) to assess the overall frequency purityin-flight

As the FCDP distributes the full performance reference from the MOLOdirectly to the MWL there is minimum impact of the FCDP on overall clockcomparison performance The USO-phase lock loop uses a similar designas for ACES-FCDP The digital PLL with linear phase detector allows theUSO to be slightly offset from its reference which is selectable by groundcommand This results in an accurate on-board time scale despite any offset

Exp Astron

of the MOLO reference frequency which can be synchronised to UTC Thefrequency synthesis chain re-uses modules developed for ACES-FCDP andtheir frequency inaccuracy Expected relative uncertainty (Allan deviation)of the frequency comparison between two signals provided to the FCDP is47 times 10minus17 at 1000 s 95 times 10minus18 at 10000 s

During non-availability of the MOLO ie during system start-up andcontingency conditions the USO runs as flywheel (ie it is unlocked withinstability lt15 times 10minus13 from 1 to 100 s) It will still provide useable butless accurate references to the instruments including the reference to OFCCommon view ground-to-ground time- and frequency comparison and supplyof the remaining instruments remain possible at the USO provided accuracy

Mass volume and power consumption are 8 kg 7 l 12 W Controldata rateis 10 kBday 50 kBday (FCDP alone)

425 Microwave link (MWL)

The microwave link is a two-way triple-frequency microwave link (Fig 9) Itis designed to perform accurate frequency comparison as well as time-transfer

Ku-Band Up-linkPower Tx 5 WCarrier 13475 GHzPN-Code 100 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

S-Band Down-linkPower Tx 1 WCarrier 2248 MHzPN-Code 1 MChips1pps 1 time marker sData 5 kBits

Ku-Band Down-linkPower Tx 1 WCarrier 1470333 GHzPN-Code 100 MChips1pps 1 time marker sData 5 kBits

EGE SC

S

Ku

KaKa-Band Up-linkPower Tx 5 WCarrier 345 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

Ka-Band Down-LinkPower Tx 1 WCarrier 32 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 5 kBits

GS Antenna diameterEGE terminal 1 mACES terminal 06 m

Fig 9 EGE microwave two-way link concept and characteristics See text for a description Thelink will be able to perform frequency comparisons with up to ten ground stations simultaneously(only one active comparison is shown) pps pulse per second Tx transmitter SC spacecraft

Exp Astron

between on-board and on-ground clocks Two-way operation allows for com-pensation of the effects due to spacecraft location and its motion relative toground Its two bi-directional links operate in Ka-band (32345 GHz) andKu-band (147135 GHz) The high carrier frequency of the up- and down-links allows for a noticeable reduction of the ionospheric delay A third down-link channel is in S-band (22 GHz) and is used to determine the ionospheretotal electron content MWL uses a combination of carrier phase and codephase data to measure with high precision the absolute ionospheric delayas well as ionospheric delay variations during a ground contact The useof three frequency bands allows to resolve third order ionospheric termsKey modulation parameters are Ka band Tone 250 MHz Ku band PN100 MChips S-band 1 MChips

The EGE-MWL concept is a considerably improved version of the ACES-MWL Microwave carrier generation is directly from the 10 GHz microwaveoutput available from the MOLO and having exceptionally low close-in phasenoise without intermediate frequency dividers or other complex synthesiserstages commonly used in RF transponders The high frequency carriers arekey to achieve the desired frequency comparison accuracy

The Ka-band carrier is generated by tripling the FCDP-generated signalThe Ku-band is subsequently generated by division by two using a verystable regenerative divider Finally the assigned radio frequency is obtainedby adding a low frequency signal using an image reject mixer A mixedanalog-digital receiver design will enhance its performance by use of a flexibledigital signal processing and digital control loops The number of simultaneousground contacts is increased from four (as in the ACES mission) to ten or moreusing the digital concept while the number of physical receiver channels perband is reduced from four to two one for operation to ground and one for in-flight calibration The concept allows adaptation of receiver loop parametersto optimise for the different portions of a highly elliptic orbit while reducingsystematic uncertainties due to high Doppler and Doppler rate Real-timedata exchange to ground terminals is by additional data modulation on all RFsignals The spacecraft carries two different small antennae for perigee and forapogee with gains of 8 dBi 20 dBi beamwidths 100 (full earth coverage) 14respectively power is 1 W for each

The MWL end-to-end performance between the attached clocks includingspace- and ground terminals is predicted to be a relative Allan deviation of1 times 10minus16 at 1000 s (perigee) and 1 times 10minus17 at 10000 s (apogee) Thesevalues are based on conservative implementation losses and other instrumentallosses assumed as 6 dB at apogee and 10 dB at perigee including additionallosses due to atmosphere An attitude knowledge of 1 is required

The MWL link will also support precise orbitographyndashcomplementary tothe GNSS receiver Generation of ranging data is simultaneous to two-waytime transfer and is available under all weather conditions Orbit determina-tion accuracy is expected to be better than 1 cm well matching EGE clockrequirements The value is expected based on demonstrated results fromother missions (METEOR ERS-2 TOPEXPOSEIDON) Laser ranging to

Exp Astron

the CCR array is used a few times during the mission to calibrate the MWLranging

In case of non-availability of the ultrapure signal from MOLO an additional10 GHz output is available from the FCDP to support ground-to-ground clockcomparison in common-view mode even after the nominal mission periodwhen the neither the clocks nor MOLO may be available any longer Basedon the PRARE experimentrsquos 12 years of operation this can be assumed forthe FCDPMWL combination as well The potential extended operational pe-riod makes the proposed configuration particularly attractive to metrologicallaboratories that rely on availability of long-lasting data in support of thegeneration of UTC

The resources of MWL are as follows mass 18 kg volume 18 l power55 W TC 2 kBday TM 120 MBday

426 Auxiliary units

The star camera assembly on the EGE spacecraft will provide real-timeattitude data primarily for the orientation of the spacecraft (with accuracy of1 in real-time) and the proper reduction of the GNSSSLR data

A GNSS receiver is used for the precise orbit determination time andfrequency transfer coarse positioning (lt1 m) for real time use by the Attitudeand Orbit Control as well as the back-up for time tagging of all payloaddata For the time and frequency transfer the receiver is nominally using theonboard frequency reference The receiver assembly includes three specialantennas By the year 2015 more than 100 GNSS satellites are expected tobe available for precise orbit determination permitting orbit inaccuracy of 1ndash2 cm RMS for altitudes up to approx 10000 km At apogee a combinationof GNSS signals and MWL ranging will be used with expected inaccuracyof 10 cm

5 Basic spacecraft specifications

The mission foresees a single spacecraft with a mission duration limited toapprox 3 years for cost reasons Attitude control along three axes withless than 1 error on a timescale of 1000 s is required The control can beperformed with cold gas thrusters magnetotorquers or reaction wheels Theselection will be done on the basis of the vibrations induced by these systems(desired level arms lt 10minus6 g)

Cold gas thrusters or monopropellant thrusters will be required to maintainthe orbit parameters during the mission by small orbit correction manoeuvresand will also guarantee a proper tuning of the orbit parameters after the releasefrom the launcher The ground segment (telemetry control and data transfer)could be via non-ESA stations user-operated to save costs

Further resources are indicated in Table 2

Exp Astron

The satellite bus design drivers are nadir-pointing microwave antennasthermal stability for the payload instruments no strong magnetic sources lowmicrovibrations harsh radiation environment due to crossing of the radiationbelts once per orbit power demand by science payload (with the possibility toreduce power demand by stand-by of some components during specific missionphases eg eclipse) mass limit due to required high altitude with low costlaunch

The satellite bus design can be based on established platforms such asFLEXBUS or PRIMA used on several past missions Their design philosophydiffers one having body-mounted solar arrays the other wings The latter canprovide larger amounts of power The vibration level induced by the panelswhen they are fixed and the duration of solar array drive periods per orbitwill have to be considered in the selection The FLEXBUS concept was usedfor a mission already requiring a low level of residual accelerations and theplatform is optimised for this parameter

For most of the satellite components one can consider using compo-nents and electronic units from navigation or communication satellites whichtypically cruise at altitudes between 15000 and 25000 km altitude LISAPathfinder which will cross the radiation belts during spiralling to theLagrange point will face similar problems concerning radiation as EGE as wellas XMM which is in a Molniya-type orbit with similar radiation characteristics

The instruments will therefore be located in the satellite structure such asto realise ionizing radiation shielding by the spacecraft structure and non-sensitive equipment to the maximum extent possible the need for additionalradiation shielding (including trapped proton radiation) for especially sensitiveparts of the instruments must be determined by ground tests It should bementioned that an orbit with higher perigee and less radiation dose is also anoption as it leads to a tolerable reduction of the science goals

The required plusmn3C temperature stability at the interface to the payloadenclosure is feasible The thermal control will employ the usual means (anenclosure with high thermal inertia heaters and a suitable radiator) Thelocation of the radiator will need a special study given the non-standard orbit

Magnetic field present during the mission can affect the properties of theonboard clock Using SPENVIS the magnetic field has been calculated alongthe spacecraft orbit the field intensity presents maxima around 02 G anda minimum value of the order of 10minus3 G As discussed above this is not alimitation to the instrument performance

6 Discussion and conclusion

The accuracy of the measurements depends crucially on the stability andaccuracy of the clocks While the performance of ground clocks (even thoseeventually deployed in new ground stations dedicated to EGE) will likelyimprove significantly from the present level the space clock design will beldquofrozenrdquo approx 10 years before data acquisition Thus their performance

Exp Astron

level will be significantly below the future ground station clock performanceAs specification we have chosen a performance level that has been demon-strated with laboratory breadboards recently We see no fundamental reasonthat this laboratory performance cannot also be achieved in a space ion opticalclock The experience with the PHARAO cold atom clock has shown thatan engineering model of a laser-based cold atom clock has a performancesimilar to that of a laboratory prototype The perturbation of the ion clock fre-quency due to magnetic fields along the orbit can be modelled and suppressedwith shielding and the residual effects measured and compensated for Themoderate temperature stability (plusmn 3C) to be provided in the experimentalbay can be achieved with thermal shields Accurate temperature controlof particular units is done by active stabilization of the units themselves

One important aspect is radiation which may affect all optical and trapcomponents as well as the lasers and reference cavity The radiation envi-ronment on the EGE orbit is significantly harsher than the near-earth ISSorbit on which ACES will operate The radiation effects must therefore becharacterized comprehensively with appropriate tests and then the radiationshield for the EGE instruments designed

In summary the Einstein Gravity Explorer mission (EGE) is a satellitemission aimed at a precise measurement of the properties of space-time usingatomic clocks By testing one of the most fundamental predictions of EinsteinrsquosTheory of General Relativity the gravitational redshift and thereby searchingfor quantum effects in gravity it explores one of the most important frontiers infundamental physics The EGE mission reaches beyond fundamental physicsproviding an important spin-off in geophysics Furthermore the technologicalaspects of the mission are ground-breaking for future time-keeping and ref-erence frame systems in space and thus are also expected to be of wide useto science and technology in general The payload of the mission has beenconceived as a compromise between cost and performance needing to fit thecost envelope of an ESA medium-class mission1

Acknowledgement We thank S Cesare for helpful information

References

1 Will CM The confrontation between general relativity and experiment Living Rev Relativ9 3 (2006)

2 Ni W-T Empirical foundations of the relativistic gravity Int J Mod Phys D 14 901 (2005)doi101142S0218271805007139

1Finally we point out that a spacecraft with a significantly reduced and thus lower-cost payload (noclocks only MOLO FCDP MWL) on a circular orbit of appropriate height could be used for along-duration earth science mission dedicated specifically to the measurement of the geopotential

Exp Astron

4 Main payload instruments

41 Overview

The satellite payload consists of a pair of cold-atom clocks a microwave-optical local oscillator (MOLO) an on-board frequency distribution unit(FCDP) a satellite-to-ground clock comparison unit (MWL) and the onboardcomputercontrol system (XPLC) see Fig 4 Auxiliary units are the posi-tion determination GPSGalileo receiver (GNSS Rx) attitude monitor (startracker not shown) and the corner cube reflector array (CCR not shown)

The science goals require clocks of very high frequency stability on thetimescales of half the orbital period (6 h) as well as perigee passage (sim1000 s)see Fig 2 Only clocks based on cold atoms are suited for the science goalsTable 1 reports values for a subset of atomic clocks that have been investigatednot including clocks that require cryogenic operation or are too complex foruse in an medium-class mission The choice of a clock pair rather than a singleone fulfils several requirements it addresses more science goals improves thevalidity of some science results and it enables calibration accuracy evaluationand redundancy The validity of the science results is enhanced since theclocks differ not only in the employed atoms but also in terms of havingdifferent sensitivity to perturbations by electric and magnetic fields electronicserrors and misalignments A comparison of the results arising from each clockallows for an understanding of measurement errors Both the clocksrsquo and thefrequency comparison unitrsquos performances will be determined after launchand regularly verified during the course of the mission The only reliable

Table 1 Properties of different clock candidates for EGE

Clock Clock Fractional σ = Fractional Sensitivity Figure of Sensitivityatom frequency inaccuracy instability to α Aα merit |Aα| to meq

at 1000 s σ [1016] QCD

Sr+ 445 THz 3 times 10minus17 (p) 5 times 10minus17 (p) 042 08 (p) ndashYb+ (1) 688 THz 3 times 10minus16 (d) 3 times 10minus16 (d) 088 029 (d) ndash

3 times 10minus17 (p) 1 (p)Yb+ (2) 642 THz lt 1 times 10minus17 (p) 1 times 10minus17 (p) minus53 53 (p) ndashSr 435 THz 1 times 10minus16 (d) 1 times 10minus16 (d) 009 009 (d) ndash

lt 1 times 10minus17 (p) 09 (p)Yb 519 THz 7 times 10minus16 (d) 1 times 10minus16 (d) 031 03 (d) ndash

lt 1 times 10minus17 (p) 1 times 10minus17 (p) 31 (p)Cs 91 GHz 2 times 10minus16 (d) 1 times 10minus15 (d) 275 027 (d) 1Yb+ 126 GHz 1 times 10minus15 (p) 1 times 10minus15 (p) 34 034 (p) 1

A relevant figure of merit is the ratio of |AX | and fractional frequency instability σ on the perigeevisibility timescale The Cs clock is the only high performance clock with adequate sensitivity tothe particle masses me mq The indicated Cs clock performance was achieved using a microwavelocal oscillator based on a cryogenic oscillator On EGE the latter is to be replaced by an opticallocal oscillator-based device (MOLO) with similar short-term performance but without the needto operate at cryogenic temperature The clock instability at 10000 s is typically similar to theindicated inaccuracy(p) projected (d) demonstrated (1) quadropole (2) octupole

Exp Astron

way to achieve this is by having at least two clocks on board The on-boardclock-clock comparison provides performance evaluation independent of theground-satellite comparison unit Once the satellite clock performances areestablished the link performance can be established as well

The clocks are similar in basic structure but differ in the detailed implemen-tation They consist of

a source of atomsa preparation subunit which cools the atoms to near-standstill (millikelvintemperature or below)a trapping subunit that confines the atom(s) to a tiny volumean interrogation subunita clock control electronics package

The clocks share a common central unit the optical-microwave local oscil-lator which serves as a flywheel oscillator with outstanding stability on shorttimescales (01ndash10 s) for both clocks It consists of a laser stabilized to an ultra-low loss optical resonator made of ultra-low-expansion glass A microwavesignal is derived from the laser by phase-locking a frequency comb to the laserwave and using a multiple of the repetition rate

While the subunits of any optical clock have a similar technological basis(optics electrooptics acoustooptics thermal control vacuum systems) theydiffer in their specifications Depending on the choice of the atom species de-livery confinement and interrogation methods clocks of different perfor-mance result A number of laboratory optical clock demonstrators have beendeveloped worldwide This ensures availability of the know-how for the indus-trial implementation of the flight models

The first clock is a microwave clock based on the interrogation of slowlymoving ensembles of cold Cs atoms It is essentially based on the PHARAOclock [5 6] developed to engineering model level by CNES in the frame-work of the ESA ISS mission ACES In EGE this clock is upgraded usingthe microwave-optical local oscillator This improves significantly the clockperformance The specification is a relative instability (Allan deviation) of3 times 10minus14 (τ s)minus12 ie at the integration time τ = 1000 s it drops to 1 times 10minus15and to 3 times 10minus16 at 10000 s While this specification is at least a factor 3 lessstringent than for the optical clocks the significantly lower cost of obtaining aflight model and the sensitivity of its frequency to the particle masses makes ita suitable choice

The optical clock baseline instrument is a single-ion optical clock Amongvarious suitable ion species (Hg+ Al+ Sr+ Yb+) [44ndash47] the Yb+ ion hasbeen selected Its breadboard performance is established at the instability levelof 3 times 10minus16 over 1000 s it exhibits a large sensitivity coefficient Aα hassmall sensitivity to magnetic fields and exhibits significant potential for furtherimprovements (use of the octupole transition with sim6 times larger Aα andlower overall instability [48] see below) A trapped Yb+ ion is also a veryrobust system it has been demonstrated that a single Yb+ ion can remaintrapped uninterruptedly for many months Finally Yb+ is the only ion with

Exp Astron

which optical clocks demonstrators have been operated in not a single but twomajor metrology laboratories

Another attractive optical clock is based on ensembles of cold neutral Sror Yb atoms trapped in an optical lattice produced by laser beams This clocktype has a lower instability than a single-ion clock thanks to the large numberof atoms (sim105) used Its development is in progress in several laboratoriesin the world (in part supported by space agencies) and has to date reached6 times 10minus17 instability at 5000 s and 1 times 10minus16 uncertainty [49 50] with potentialfor even higher performance However it is more complex and costly thanthe Yb+ ion clock since the cold atom preparation and trapping subunitsinclude more components require higher laser powers and the respectiveoperating procedures are more complex It is therefore not considered for thepresent mission designed to be developed in the very near future with limitedresources but is an interesting candidate for missions without such constraintsIts estimated physical parameters are indicated in Table 2 for comparison

42 Main instruments

421 Microwave clock

The PHARAO cold atom clock is one of the two atomic clocks of the spacemission ACES managed by ESA It is proposed here to include a secondversion of this instrument as its costperformance ratio is very attractive forthe EGE scientific objectives Its concept is very similar to ground basedatomic fountains but with the major difference of zero-g operation Atomsslowly launched in free flight cross two microwave fields tuned to the transitionbetween the two hyperfine levels of the cesium ground state The interrogationmethod based on two separate oscillating fields (Ramsey scheme) allowsthe detection of an atomic line whose width is inversely proportional to thetransit time between the two microwave cavities The resonant microwavefield at 9192631770 GHz (SI definition of the second) is synthesized startingfrom an ultrastable quartz oscillator (USO) in the FCDP phaselocked to themicrowave output of the microwave-optical local oscillator and stabilized tothe clock line using the error signal generated by the cesium resonator In thisway the intrinsic qualities of the cesium hyperfine transition both in terms ofaccuracy and frequency stability are transferred to the USO In a microgravityenvironment the velocity of atoms along the ballistic trajectories is constantand can be changed continuously over almost two orders of magnitude (5ndash500 cms) Differently from atomic fountain clocks presently operated onground very long interaction times (up to few seconds) will be possible whilekeeping reasonable the size of the instrument

The instrument consists of 4 main elements a laser source which providesthe light required to cool and detect the cesium atoms (Fig 5 left) a frequencychain which generates the 92 GHz microwave signal a cesium vacuum tubewhere the interaction between the microwave field and the atoms occurs

Exp Astron

Table 2 Resources and technology readiness level (TRL) of the EGE spacecraft

Unit Volume Mass Power TRL Comment[l] [kg] [W]

Ion clock 100 70 60Atom source 4Atom preparation 6Atom trap 4Atom interrogation 6Clock control 6

Microwave clock 200 91 110 8 TRL by end 2008Local oscillator

Stabilized laser 18 15 23 6Frequency comb 20 16 38 6

Microwave link 18 18 55 8FCDP 7 8 12 8GPS receiver 2 5 7 9CCR 2 2 0 9Payload sum 367 225 305Thermal control 14 17 30 8OBDH 20 24 38 9Star tracker 1 1 8 9Radiation shield 10 40 0 9Support structure 170 204 0 9Power incl solar arr 50 58 10 9Propulsion 21 25 20 9AOCS 36 43 82 9TMTC 8 10 15 9Satellite bus sum 330 422 203 Flexbus conceptPayload plus bus incl 20 margin 8364 7764 6096Propellant 40Launch adapter 100SC wet masspower incl Margins 9164Lattice clock 135 105 152 Alternative clock

Atom source 4Atom preparation 6Atom trap 4Atom interrogation 6Clock control 6

Also included are the resources for an optical lattice clock based on neutral Strontium atoms

(Fig 5 right) and a computer The engineering model of PHARAO is fullyassembled at CNES Toulouse and undergoes functional and performancetests until the end of 2008 The operating temperature range is 10ndash33 andnon-operating temperatures are minus45 +60 The cesium tube consists in aUHV chamber pumped by getters and a 3 ls ion pump a cesium reservoira microwave cavity coils to provide a uniform magnetic field and three layersof magnetic shields Vacuum windows with fiber-optics collimators enable thetransmission of the cooling and detection light onto the cesium atoms

In autonomous mode PHARAO uses the USO as interrogation oscillatorand will provide a clock signal with fractional frequency stability below1 times 10minus13 (τs) minus12 and inaccuracy near 1 times 10minus16 When using the microwave-

Exp Astron

Fig 5 Left The PHARAO laser source engineering model with cover removed Dimensionsare 530 times 350 times 150 mm3 and the mass is 200 kg Ten polarization maintaining optical fibers(yellow) guide the laser beams to the cesium tube All diode lasers are mounted on Peltier coolersfor temperature regulation within 2 mK (courtesy EADS SODERN) Right The PHARAOengineering model cesium tube without the two external magnetic shields The volume is 990 times336 times 444 mm3 and the total mass is 44 kg

optical local oscillator PHARAO will provide a reduced frequency instabilityof 3 times 10minus14 (τs) minus12 This corresponds to 1 times 10minus15 at 1000 s and 3 times 10minus16

at 104 s

422 Ion optical clock

The 171Yb+ ion clock shown in Figs 6 and 7 provides an atomic frequencyreference based on a single 171Yb+ ion confined within an RF electric end-captrap (similar to a Paul trap) within an ultra high vacuum (UHV) chamber andlaser cooled to sim1 mK close to the Doppler cooling limit As a result the ionrsquosfirst order Doppler motion is completely removed and the ion experienceslittle collisional perturbation from its environment Under these conditionsthe narrow linewidth 2S12 (F = 0 mF = 0) minus 2 D32 (F = 2 mF = 0) quadrupole

Fig 6 Left Simplified level scheme of 171Yb+ showing relevant cooling (blue) auxiliary (red)quadrupole clock (green) and octupole clock (dark blue) transitions δνnat is the natural linewidthof the clock transition Middle close-up of Paul-type RF trap (PTB Braunschweig) the ion istrapped in the hole (diameter 14 mm) Right Complete ion trap with vacuum flange height 10 cm(NPL Teddington)

Exp Astron

935 nm D32-state repumper laser

638 nm F72-state repumper laser

739 nm extended cavity diode laser

AOM

PMT

Ion trap UHV chamber

vacuum pumps

Fibre-free space interface

Mu-metal shielding

EC

T

B

Magnetic field

Temperature

Ion fluorescence Processor

Clock laser frequency step and control

4355 nm

Frequengy doubling II

3695 nm cooling wave

Low-drift cavity

Low-drift cavity 171Yb+ 435 nm ion clock unit

876 nm clock laser wave from MOLO

Frequengy doubling

Fig 7 Simplified schematic of the single-ion optical clock unit AOM acousto-optic frequencyshifter PMT photomultiplier tube EC endcap T temperature stabilization system B magneticfield

clock transition at 4355 nm can be interrogated by spectrally narrow and stableclock laser light A diagram of the simplified energy level scheme for 171Yb+showing the relevant cooling auxiliary and clock transitions is given in Fig 6All these wavelengths are provided by diode lasers operated in fundamentalor frequency doubled mode The clock laser light is obtained from the 871 nmwave provided by the MOLO unit (see Section 423) after frequency-doublingin single pass by a KNbO3 nonlinear-optical crystal to 4355 nm The clocktransition spectral profile is observed by stepping the clock laser frequency bymeans of an acousto-optic frequency shifter (AOM in Fig 7) and recordingthe statistics of the quantum jumps in cooling laser induced fluorescence asa function of this frequency Under clock operation the AOM repeatedlysteps back and forward between the half-intensity points on the transitionprofile monitoring the signal imbalance between these points Any detectedimbalance is servo-corrected to zero by feedback to the AOM

In detail the optical clock architecture will comprise (Fig 7)

bull an RF end-cap trap for ionising and confining a single 171Yb+ ion withinan ultra-high vacuum chamber pumped by a small ion pump and non-evaporable getter pump The trap is driven by means of an ac voltage ofa few hundred volts at a drive frequency of sim10 MHz A small oven withseveral mg of 171Yb isotope is heated to provide a low flux of Yb atomsfrom which single ions can be ionised by electron bombardment from ahot wire filament within the trap potential well Low voltage dc is applied

Exp Astron

to additional electrodes to position the ion precisely at trap potentialcentre in order to minimise the ionrsquos micromotion at the drive frequencyThis needs automatic periodic monitoring and correction especially in theperiod after re-loading an ion into the trap External magnetic field coilsin three orthogonal axes allow the nulling of external fields and settingof a fixed field of sim1 μT The trap and coils are surrounded by mu-metalshielding to minimise external field changes during the orbit A level oftemperature control of 1 K of the region surrounding the trap is requiredfor maintaining the blackbody-induced frequency shift uncertainty withtemperature at the 10minus17 level

bull a laser platform to provide frequency doubling of an extended cavity739 nm diode laser for Doppler cooling of the ion on the 2S12(F = 1) minus2 P12 (F = 0) dipole transition at 3695 nm Light from an amplified diodelaser device frequency-doubled in a periodically poled LiTaO3 crystal insingle-pass will be used Typical laser powers used for driving the coolingtransition below saturation are about 2 μW for a beam waist in the trapof about 50 μm By modulation of the injection current of the 739 nmdiode laser at 147 GHz a sideband is generated that excites the 2S12 (F =0) rarr2 P12 (F = 1) ldquorepumperrdquo transition in order to avoid optical pump-ing between the ground hyperfine states

bull The platform also houses the auxiliary lasers at 935 and 638 nm The935 nm diode provides repumping of the ion from the 2 D32(F = 1)metastable level after occasional branching decays to this level during thecooling sequence The 638 nm diode allows fast recovery of the ion fromthe very-long-lived 2F72 metastable state after very occasional collisionaldecay to that state Currently extended cavity lasers are forseen butdistributed-feedback lasers may be available in the near future

bull a high NA lens imaging system and photomultiplier detection system torecord the statistics of 369 nm fluorescence quantum jumps as a functionof clock laser frequency step

bull a fibre system to deliver the various cooling auxiliary and clock light fromsource to trap making use of achromatic doublets where necessary at thefibre-free space interface for launching into the trap

bull a monitoring and control processor which provides the primary coolingand clock laser pulse magnetic field and detection sequencing to observeand lock to the ion clock transition frequency The processor also monitorsfrequency and amplitude data necessary to determine normal laser andion operational conditions and initiate resetting and recovery algorithmswhere necessary and laser unit failure The actual magnetic field presenton the atomic volume is determined and actively stabilized so that aninaccuracy at the level of 1 times 10minus17 can be achieved

bull a redundancy level of at least two units for cooling clock and repumperlasers plus similar for frequency doubling crystals All redundancy unitsfor each wavelength will be fibre multiplexed as standard allowing redun-dant unit activation on determination of prior unit failure

Exp Astron

For an optimized quantum-projection-noise-limited performance of thesingle-ion clock Ramsey interrogation with a cycle time equal to the lifetimeof the metastable D32 level of 50 ms will be performed leading to an Allandeviation of about 27 times 10minus15 (τ s)minus12 This is sim45 times 10minus17 at 1 h satisfyingrequired stabilities for the science objectives Current laboratory results are astability of 1 times 10minus15 at 30 s averaging down to 3 times 10minus16 at 1000 s (PTBGermany) Also comparisons between two independent traps showed agree-ment at the level of several parts in 10minus16 over a series of eight measurementruns

The 171Yb+ ion clock offers a second option namely the 2S12ndash2 F72 octu-pole transition at 476 nm (see Table 1 and Fig 6) [48] It has an extremelylong upper state lifetime low electric quadrupole and second order Zeemansystematic shifts and therefore is a candidate for an even higher performanceclock if a clock laser system with lower instability than the above is providedApart from the clock laser it shares the laser system of the quadrupoleclock Thus with moderate additional resources the Yb+ ion apparatus couldprovide two optical clocks in a single package and specifically perform directfrequency comparisons between the quadrupole and the octupole transitionwith a large sensitivity Aα

The ion clock will operate under microprocessor control providing thecontrol algorithms for the integrated pulse sequencing for cooling repumpingclock interrogation magnetic field switching cooling fluorescence monitoringThe controller will also provide error flags for critical processes and systemreset and correction where necessary This will include ion re-loading andmicromotion reduction algorithms Automatic monitoring of laser power andspectral quality with signal re-optimisation or laser failure determination isrequired Redundant laser units will be pre-aligned and multiplexed into thefibre delivery to the trap so that redundant units can be readily activated in thecase of diode laser failure

423 Microwave-optical local oscillator (MOLO)

This oscillator provides two ultrastable frequency outputs one in the opticalregion the other in the microwave region the two frequencies being coher-ently related The MOLO is composed of three subunits the clock laser thereference cavity and the frequency comb see Fig 8

The clock laser subunit delivers the light for the excitation of the opticalclock transition It is generated by an extended-cavity diode laser at a wave-length of 871 nm Most of the output wave of this laser is sent to the ionclock The laser subunit containing two lasers for redundancy will require3 kg 3 l 15 W The frequency instability of the clock laser must be superiorto the minimum clock instability 27 times 10minus15 (τ s)minus12 for τ up to the timeconstant (sim10 s) of the servo system that locks the laser to the ionrsquos resonancesignal In order to achieve this the laser frequency is stabilized to the reference

Exp Astron

871 nm clock laser

EO

Detector and servo

AOM Femtosecond

frequency comb

To optical clock 10 GHz To FCDP

Microwave-optical local oscillator

ULE cavity

Fig 8 Microwave-optical local oscillator (MOLO) consisting of clock laser reference ULE cavityand femtosecond frequency comb (OFC) Radiofrequency inputs and outputs are not shown EOelectro-optic modulator ULE ultra-low expansion AOM frequency shifter

cavity where the PoundndashDreverndashHall method is used About 10 μW of thelaser output are used for this

The system will exhibit sim04 Hz (sim1 times 10minus15 relative) linewidth and afrequency drift lt01 Hzs The cavity consists of two high-reflectivity mirrorsoptically contacted to a cylinder of diameter 70 mm length 150 mm made ofULE (a glass exhibiting ultra low thermal expansion coefficient) with a cavityfinesse of about 300000 Residual vibrations of the satellite (arms lt 1 times 10minus6 g)should not be a limiting factor to the desired linewidth for safety marginan optimized cavity support can reduce the sensitivity to accelerations Foracoustic and thermal isolation the cavity resides inside an aluminium vacuumchamber equipped with a small (3 ls) ion-getter pump Two stages of polishedaluminium shields are implemented around the cavity which are activelytemperature stabilized so that the cavity is precisely kept at the zero crossingtemperature of the ULE thermal expansion coefficient Heat application andremoval is by thermoelectric elements between the shield and the vacuumchamber Additionally the entire vacuum chamber is temperature-stabilized

During transportation and launch the cavity will be rigidly fixed insidethe vacuum chamber by pressing on it from the sides by piezo-mechanicalactuators allowing accelerations of several g After bringing the satellite onthe orbit the actuators will be released

The optical setup for frequency stabilization will be a classic Pound-Drever-Hall scheme in reflection using an electro-optic modulator (EO) Couplingof the clock laser radiation to the cavity is via single-mode polarization main-taining optical fiber The fiber in- and out-couplers are mounted on miniaturepiezo-motor driven multiaxis translation stages Microprocessor control ofthe translation stages performs a laser-to-fiber incoupling stabilization andlaser-to-cavity mode-matching optimization (positioning inaccuracy lt10 μmalignment inaccuracy lt10 μrad)

The frequency stability of the clock laser will be limited by thermal 1f noisein the mirrors with an fractional Allan deviation slightly below 1 times 10minus15 from1 s to a few 10 s Resources including control electronics are 15 l 12 kg 8 W

Exp Astron

A femtosecond fiber laser based optical frequency comb generator (OFCFig 8) will be used for the clock frequency comparisons and to obtain aspectrally ultrapure microwave local oscillator output for the operation ofPHARAO for the microwave link and for the GNSS receiver The fem-tosecond laser is based on erbium-doped fiber pumped by fiber-coupled laserdiodes It emits 100 fs pulses at a repetition rate of 250 MHz with an averagepower of 100 mW The emission spectrum is centered around 1560 nm Spec-tral broadening to more than one octave is obtained via nonlinear interactionin a microstructure fibre so that the range from about 1050 to 2100 nm iscovered

Second harmonic generation of a portion of the broadened spectrum allowsobtaining the required light for the measurement of the clock laser frequencyat 871 nm The OFC will be referenced to the clock laser frequency (afterintroduction a controllable shift via the frequency shifter AOM) by makingthe carrier envelope offset frequency vanish so that the repetition rate fr isa high-order subharmonic of the optical clock reference frequency f1 = m1 frThe link between the optical and the microwave domain is then simply thedistribution of the repetition rate frequency or more conveniently a high(order 40) harmonic of the comb repetition rate 10 GHz

It has been demonstrated by several groups that frequency comparison andfrequency division with an OFC can be performed to very high accuracy andstability beyond the limitations imposed by the optical clocks themselves Forexample a fiber laser OFC with an instability (Allan deviation) of 6 times 10minus17

in 1 s has been demonstrated [51] OFCs based on fiber technology possessgood long-term reliability and high power efficiency The projection for a spacequalified OFC based on current industrial terrestrial technology is 16 kg mass20 l volume and 38 W power requirement

424 Frequency comparison and distribution package (FCDP)

This package distributes the ultra-stable 10 GHz reference frequency outputof the MOLO to the MWL performs frequency comparisons between MOLOand PHARAO and generates a variety of different reference signals for theindividual instruments including the 250 MHz reference clock to the OFCAll signals are coherent to the MOLO In the FCDP a USO is locked to theMOLO signal by a digital PLL A divider chain generates timing pulses (1 pps1000 pps etc) and keeps a continuous local time scale From the digital phasecomparison data the phase noise density spectrum between the USO and theMOLO is available by telemetry (TM) to assess the overall frequency purityin-flight

As the FCDP distributes the full performance reference from the MOLOdirectly to the MWL there is minimum impact of the FCDP on overall clockcomparison performance The USO-phase lock loop uses a similar designas for ACES-FCDP The digital PLL with linear phase detector allows theUSO to be slightly offset from its reference which is selectable by groundcommand This results in an accurate on-board time scale despite any offset

Exp Astron

of the MOLO reference frequency which can be synchronised to UTC Thefrequency synthesis chain re-uses modules developed for ACES-FCDP andtheir frequency inaccuracy Expected relative uncertainty (Allan deviation)of the frequency comparison between two signals provided to the FCDP is47 times 10minus17 at 1000 s 95 times 10minus18 at 10000 s

During non-availability of the MOLO ie during system start-up andcontingency conditions the USO runs as flywheel (ie it is unlocked withinstability lt15 times 10minus13 from 1 to 100 s) It will still provide useable butless accurate references to the instruments including the reference to OFCCommon view ground-to-ground time- and frequency comparison and supplyof the remaining instruments remain possible at the USO provided accuracy

Mass volume and power consumption are 8 kg 7 l 12 W Controldata rateis 10 kBday 50 kBday (FCDP alone)

425 Microwave link (MWL)

The microwave link is a two-way triple-frequency microwave link (Fig 9) Itis designed to perform accurate frequency comparison as well as time-transfer

Ku-Band Up-linkPower Tx 5 WCarrier 13475 GHzPN-Code 100 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

S-Band Down-linkPower Tx 1 WCarrier 2248 MHzPN-Code 1 MChips1pps 1 time marker sData 5 kBits

Ku-Band Down-linkPower Tx 1 WCarrier 1470333 GHzPN-Code 100 MChips1pps 1 time marker sData 5 kBits

EGE SC

S

Ku

KaKa-Band Up-linkPower Tx 5 WCarrier 345 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

Ka-Band Down-LinkPower Tx 1 WCarrier 32 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 5 kBits

GS Antenna diameterEGE terminal 1 mACES terminal 06 m

Fig 9 EGE microwave two-way link concept and characteristics See text for a description Thelink will be able to perform frequency comparisons with up to ten ground stations simultaneously(only one active comparison is shown) pps pulse per second Tx transmitter SC spacecraft

Exp Astron

between on-board and on-ground clocks Two-way operation allows for com-pensation of the effects due to spacecraft location and its motion relative toground Its two bi-directional links operate in Ka-band (32345 GHz) andKu-band (147135 GHz) The high carrier frequency of the up- and down-links allows for a noticeable reduction of the ionospheric delay A third down-link channel is in S-band (22 GHz) and is used to determine the ionospheretotal electron content MWL uses a combination of carrier phase and codephase data to measure with high precision the absolute ionospheric delayas well as ionospheric delay variations during a ground contact The useof three frequency bands allows to resolve third order ionospheric termsKey modulation parameters are Ka band Tone 250 MHz Ku band PN100 MChips S-band 1 MChips

The EGE-MWL concept is a considerably improved version of the ACES-MWL Microwave carrier generation is directly from the 10 GHz microwaveoutput available from the MOLO and having exceptionally low close-in phasenoise without intermediate frequency dividers or other complex synthesiserstages commonly used in RF transponders The high frequency carriers arekey to achieve the desired frequency comparison accuracy

The Ka-band carrier is generated by tripling the FCDP-generated signalThe Ku-band is subsequently generated by division by two using a verystable regenerative divider Finally the assigned radio frequency is obtainedby adding a low frequency signal using an image reject mixer A mixedanalog-digital receiver design will enhance its performance by use of a flexibledigital signal processing and digital control loops The number of simultaneousground contacts is increased from four (as in the ACES mission) to ten or moreusing the digital concept while the number of physical receiver channels perband is reduced from four to two one for operation to ground and one for in-flight calibration The concept allows adaptation of receiver loop parametersto optimise for the different portions of a highly elliptic orbit while reducingsystematic uncertainties due to high Doppler and Doppler rate Real-timedata exchange to ground terminals is by additional data modulation on all RFsignals The spacecraft carries two different small antennae for perigee and forapogee with gains of 8 dBi 20 dBi beamwidths 100 (full earth coverage) 14respectively power is 1 W for each

The MWL end-to-end performance between the attached clocks includingspace- and ground terminals is predicted to be a relative Allan deviation of1 times 10minus16 at 1000 s (perigee) and 1 times 10minus17 at 10000 s (apogee) Thesevalues are based on conservative implementation losses and other instrumentallosses assumed as 6 dB at apogee and 10 dB at perigee including additionallosses due to atmosphere An attitude knowledge of 1 is required

The MWL link will also support precise orbitographyndashcomplementary tothe GNSS receiver Generation of ranging data is simultaneous to two-waytime transfer and is available under all weather conditions Orbit determina-tion accuracy is expected to be better than 1 cm well matching EGE clockrequirements The value is expected based on demonstrated results fromother missions (METEOR ERS-2 TOPEXPOSEIDON) Laser ranging to

Exp Astron

the CCR array is used a few times during the mission to calibrate the MWLranging

In case of non-availability of the ultrapure signal from MOLO an additional10 GHz output is available from the FCDP to support ground-to-ground clockcomparison in common-view mode even after the nominal mission periodwhen the neither the clocks nor MOLO may be available any longer Basedon the PRARE experimentrsquos 12 years of operation this can be assumed forthe FCDPMWL combination as well The potential extended operational pe-riod makes the proposed configuration particularly attractive to metrologicallaboratories that rely on availability of long-lasting data in support of thegeneration of UTC

The resources of MWL are as follows mass 18 kg volume 18 l power55 W TC 2 kBday TM 120 MBday

426 Auxiliary units

The star camera assembly on the EGE spacecraft will provide real-timeattitude data primarily for the orientation of the spacecraft (with accuracy of1 in real-time) and the proper reduction of the GNSSSLR data

A GNSS receiver is used for the precise orbit determination time andfrequency transfer coarse positioning (lt1 m) for real time use by the Attitudeand Orbit Control as well as the back-up for time tagging of all payloaddata For the time and frequency transfer the receiver is nominally using theonboard frequency reference The receiver assembly includes three specialantennas By the year 2015 more than 100 GNSS satellites are expected tobe available for precise orbit determination permitting orbit inaccuracy of 1ndash2 cm RMS for altitudes up to approx 10000 km At apogee a combinationof GNSS signals and MWL ranging will be used with expected inaccuracyof 10 cm

5 Basic spacecraft specifications

The mission foresees a single spacecraft with a mission duration limited toapprox 3 years for cost reasons Attitude control along three axes withless than 1 error on a timescale of 1000 s is required The control can beperformed with cold gas thrusters magnetotorquers or reaction wheels Theselection will be done on the basis of the vibrations induced by these systems(desired level arms lt 10minus6 g)

Cold gas thrusters or monopropellant thrusters will be required to maintainthe orbit parameters during the mission by small orbit correction manoeuvresand will also guarantee a proper tuning of the orbit parameters after the releasefrom the launcher The ground segment (telemetry control and data transfer)could be via non-ESA stations user-operated to save costs

Further resources are indicated in Table 2

Exp Astron

The satellite bus design drivers are nadir-pointing microwave antennasthermal stability for the payload instruments no strong magnetic sources lowmicrovibrations harsh radiation environment due to crossing of the radiationbelts once per orbit power demand by science payload (with the possibility toreduce power demand by stand-by of some components during specific missionphases eg eclipse) mass limit due to required high altitude with low costlaunch

The satellite bus design can be based on established platforms such asFLEXBUS or PRIMA used on several past missions Their design philosophydiffers one having body-mounted solar arrays the other wings The latter canprovide larger amounts of power The vibration level induced by the panelswhen they are fixed and the duration of solar array drive periods per orbitwill have to be considered in the selection The FLEXBUS concept was usedfor a mission already requiring a low level of residual accelerations and theplatform is optimised for this parameter

For most of the satellite components one can consider using compo-nents and electronic units from navigation or communication satellites whichtypically cruise at altitudes between 15000 and 25000 km altitude LISAPathfinder which will cross the radiation belts during spiralling to theLagrange point will face similar problems concerning radiation as EGE as wellas XMM which is in a Molniya-type orbit with similar radiation characteristics

The instruments will therefore be located in the satellite structure such asto realise ionizing radiation shielding by the spacecraft structure and non-sensitive equipment to the maximum extent possible the need for additionalradiation shielding (including trapped proton radiation) for especially sensitiveparts of the instruments must be determined by ground tests It should bementioned that an orbit with higher perigee and less radiation dose is also anoption as it leads to a tolerable reduction of the science goals

The required plusmn3C temperature stability at the interface to the payloadenclosure is feasible The thermal control will employ the usual means (anenclosure with high thermal inertia heaters and a suitable radiator) Thelocation of the radiator will need a special study given the non-standard orbit

Magnetic field present during the mission can affect the properties of theonboard clock Using SPENVIS the magnetic field has been calculated alongthe spacecraft orbit the field intensity presents maxima around 02 G anda minimum value of the order of 10minus3 G As discussed above this is not alimitation to the instrument performance

6 Discussion and conclusion

The accuracy of the measurements depends crucially on the stability andaccuracy of the clocks While the performance of ground clocks (even thoseeventually deployed in new ground stations dedicated to EGE) will likelyimprove significantly from the present level the space clock design will beldquofrozenrdquo approx 10 years before data acquisition Thus their performance

Exp Astron

level will be significantly below the future ground station clock performanceAs specification we have chosen a performance level that has been demon-strated with laboratory breadboards recently We see no fundamental reasonthat this laboratory performance cannot also be achieved in a space ion opticalclock The experience with the PHARAO cold atom clock has shown thatan engineering model of a laser-based cold atom clock has a performancesimilar to that of a laboratory prototype The perturbation of the ion clock fre-quency due to magnetic fields along the orbit can be modelled and suppressedwith shielding and the residual effects measured and compensated for Themoderate temperature stability (plusmn 3C) to be provided in the experimentalbay can be achieved with thermal shields Accurate temperature controlof particular units is done by active stabilization of the units themselves

One important aspect is radiation which may affect all optical and trapcomponents as well as the lasers and reference cavity The radiation envi-ronment on the EGE orbit is significantly harsher than the near-earth ISSorbit on which ACES will operate The radiation effects must therefore becharacterized comprehensively with appropriate tests and then the radiationshield for the EGE instruments designed

In summary the Einstein Gravity Explorer mission (EGE) is a satellitemission aimed at a precise measurement of the properties of space-time usingatomic clocks By testing one of the most fundamental predictions of EinsteinrsquosTheory of General Relativity the gravitational redshift and thereby searchingfor quantum effects in gravity it explores one of the most important frontiers infundamental physics The EGE mission reaches beyond fundamental physicsproviding an important spin-off in geophysics Furthermore the technologicalaspects of the mission are ground-breaking for future time-keeping and ref-erence frame systems in space and thus are also expected to be of wide useto science and technology in general The payload of the mission has beenconceived as a compromise between cost and performance needing to fit thecost envelope of an ESA medium-class mission1

Acknowledgement We thank S Cesare for helpful information

References

1 Will CM The confrontation between general relativity and experiment Living Rev Relativ9 3 (2006)

2 Ni W-T Empirical foundations of the relativistic gravity Int J Mod Phys D 14 901 (2005)doi101142S0218271805007139

1Finally we point out that a spacecraft with a significantly reduced and thus lower-cost payload (noclocks only MOLO FCDP MWL) on a circular orbit of appropriate height could be used for along-duration earth science mission dedicated specifically to the measurement of the geopotential

Exp Astron

way to achieve this is by having at least two clocks on board The on-boardclock-clock comparison provides performance evaluation independent of theground-satellite comparison unit Once the satellite clock performances areestablished the link performance can be established as well

The clocks are similar in basic structure but differ in the detailed implemen-tation They consist of

a source of atomsa preparation subunit which cools the atoms to near-standstill (millikelvintemperature or below)a trapping subunit that confines the atom(s) to a tiny volumean interrogation subunita clock control electronics package

The clocks share a common central unit the optical-microwave local oscil-lator which serves as a flywheel oscillator with outstanding stability on shorttimescales (01ndash10 s) for both clocks It consists of a laser stabilized to an ultra-low loss optical resonator made of ultra-low-expansion glass A microwavesignal is derived from the laser by phase-locking a frequency comb to the laserwave and using a multiple of the repetition rate

While the subunits of any optical clock have a similar technological basis(optics electrooptics acoustooptics thermal control vacuum systems) theydiffer in their specifications Depending on the choice of the atom species de-livery confinement and interrogation methods clocks of different perfor-mance result A number of laboratory optical clock demonstrators have beendeveloped worldwide This ensures availability of the know-how for the indus-trial implementation of the flight models

The first clock is a microwave clock based on the interrogation of slowlymoving ensembles of cold Cs atoms It is essentially based on the PHARAOclock [5 6] developed to engineering model level by CNES in the frame-work of the ESA ISS mission ACES In EGE this clock is upgraded usingthe microwave-optical local oscillator This improves significantly the clockperformance The specification is a relative instability (Allan deviation) of3 times 10minus14 (τ s)minus12 ie at the integration time τ = 1000 s it drops to 1 times 10minus15and to 3 times 10minus16 at 10000 s While this specification is at least a factor 3 lessstringent than for the optical clocks the significantly lower cost of obtaining aflight model and the sensitivity of its frequency to the particle masses makes ita suitable choice

The optical clock baseline instrument is a single-ion optical clock Amongvarious suitable ion species (Hg+ Al+ Sr+ Yb+) [44ndash47] the Yb+ ion hasbeen selected Its breadboard performance is established at the instability levelof 3 times 10minus16 over 1000 s it exhibits a large sensitivity coefficient Aα hassmall sensitivity to magnetic fields and exhibits significant potential for furtherimprovements (use of the octupole transition with sim6 times larger Aα andlower overall instability [48] see below) A trapped Yb+ ion is also a veryrobust system it has been demonstrated that a single Yb+ ion can remaintrapped uninterruptedly for many months Finally Yb+ is the only ion with

Exp Astron

which optical clocks demonstrators have been operated in not a single but twomajor metrology laboratories

Another attractive optical clock is based on ensembles of cold neutral Sror Yb atoms trapped in an optical lattice produced by laser beams This clocktype has a lower instability than a single-ion clock thanks to the large numberof atoms (sim105) used Its development is in progress in several laboratoriesin the world (in part supported by space agencies) and has to date reached6 times 10minus17 instability at 5000 s and 1 times 10minus16 uncertainty [49 50] with potentialfor even higher performance However it is more complex and costly thanthe Yb+ ion clock since the cold atom preparation and trapping subunitsinclude more components require higher laser powers and the respectiveoperating procedures are more complex It is therefore not considered for thepresent mission designed to be developed in the very near future with limitedresources but is an interesting candidate for missions without such constraintsIts estimated physical parameters are indicated in Table 2 for comparison

42 Main instruments

421 Microwave clock

The PHARAO cold atom clock is one of the two atomic clocks of the spacemission ACES managed by ESA It is proposed here to include a secondversion of this instrument as its costperformance ratio is very attractive forthe EGE scientific objectives Its concept is very similar to ground basedatomic fountains but with the major difference of zero-g operation Atomsslowly launched in free flight cross two microwave fields tuned to the transitionbetween the two hyperfine levels of the cesium ground state The interrogationmethod based on two separate oscillating fields (Ramsey scheme) allowsthe detection of an atomic line whose width is inversely proportional to thetransit time between the two microwave cavities The resonant microwavefield at 9192631770 GHz (SI definition of the second) is synthesized startingfrom an ultrastable quartz oscillator (USO) in the FCDP phaselocked to themicrowave output of the microwave-optical local oscillator and stabilized tothe clock line using the error signal generated by the cesium resonator In thisway the intrinsic qualities of the cesium hyperfine transition both in terms ofaccuracy and frequency stability are transferred to the USO In a microgravityenvironment the velocity of atoms along the ballistic trajectories is constantand can be changed continuously over almost two orders of magnitude (5ndash500 cms) Differently from atomic fountain clocks presently operated onground very long interaction times (up to few seconds) will be possible whilekeeping reasonable the size of the instrument

The instrument consists of 4 main elements a laser source which providesthe light required to cool and detect the cesium atoms (Fig 5 left) a frequencychain which generates the 92 GHz microwave signal a cesium vacuum tubewhere the interaction between the microwave field and the atoms occurs

Exp Astron

Table 2 Resources and technology readiness level (TRL) of the EGE spacecraft

Unit Volume Mass Power TRL Comment[l] [kg] [W]

Ion clock 100 70 60Atom source 4Atom preparation 6Atom trap 4Atom interrogation 6Clock control 6

Microwave clock 200 91 110 8 TRL by end 2008Local oscillator

Stabilized laser 18 15 23 6Frequency comb 20 16 38 6

Microwave link 18 18 55 8FCDP 7 8 12 8GPS receiver 2 5 7 9CCR 2 2 0 9Payload sum 367 225 305Thermal control 14 17 30 8OBDH 20 24 38 9Star tracker 1 1 8 9Radiation shield 10 40 0 9Support structure 170 204 0 9Power incl solar arr 50 58 10 9Propulsion 21 25 20 9AOCS 36 43 82 9TMTC 8 10 15 9Satellite bus sum 330 422 203 Flexbus conceptPayload plus bus incl 20 margin 8364 7764 6096Propellant 40Launch adapter 100SC wet masspower incl Margins 9164Lattice clock 135 105 152 Alternative clock

Atom source 4Atom preparation 6Atom trap 4Atom interrogation 6Clock control 6

Also included are the resources for an optical lattice clock based on neutral Strontium atoms

(Fig 5 right) and a computer The engineering model of PHARAO is fullyassembled at CNES Toulouse and undergoes functional and performancetests until the end of 2008 The operating temperature range is 10ndash33 andnon-operating temperatures are minus45 +60 The cesium tube consists in aUHV chamber pumped by getters and a 3 ls ion pump a cesium reservoira microwave cavity coils to provide a uniform magnetic field and three layersof magnetic shields Vacuum windows with fiber-optics collimators enable thetransmission of the cooling and detection light onto the cesium atoms

In autonomous mode PHARAO uses the USO as interrogation oscillatorand will provide a clock signal with fractional frequency stability below1 times 10minus13 (τs) minus12 and inaccuracy near 1 times 10minus16 When using the microwave-

Exp Astron

Fig 5 Left The PHARAO laser source engineering model with cover removed Dimensionsare 530 times 350 times 150 mm3 and the mass is 200 kg Ten polarization maintaining optical fibers(yellow) guide the laser beams to the cesium tube All diode lasers are mounted on Peltier coolersfor temperature regulation within 2 mK (courtesy EADS SODERN) Right The PHARAOengineering model cesium tube without the two external magnetic shields The volume is 990 times336 times 444 mm3 and the total mass is 44 kg

optical local oscillator PHARAO will provide a reduced frequency instabilityof 3 times 10minus14 (τs) minus12 This corresponds to 1 times 10minus15 at 1000 s and 3 times 10minus16

at 104 s

422 Ion optical clock

The 171Yb+ ion clock shown in Figs 6 and 7 provides an atomic frequencyreference based on a single 171Yb+ ion confined within an RF electric end-captrap (similar to a Paul trap) within an ultra high vacuum (UHV) chamber andlaser cooled to sim1 mK close to the Doppler cooling limit As a result the ionrsquosfirst order Doppler motion is completely removed and the ion experienceslittle collisional perturbation from its environment Under these conditionsthe narrow linewidth 2S12 (F = 0 mF = 0) minus 2 D32 (F = 2 mF = 0) quadrupole

Fig 6 Left Simplified level scheme of 171Yb+ showing relevant cooling (blue) auxiliary (red)quadrupole clock (green) and octupole clock (dark blue) transitions δνnat is the natural linewidthof the clock transition Middle close-up of Paul-type RF trap (PTB Braunschweig) the ion istrapped in the hole (diameter 14 mm) Right Complete ion trap with vacuum flange height 10 cm(NPL Teddington)

Exp Astron

935 nm D32-state repumper laser

638 nm F72-state repumper laser

739 nm extended cavity diode laser

AOM

PMT

Ion trap UHV chamber

vacuum pumps

Fibre-free space interface

Mu-metal shielding

EC

T

B

Magnetic field

Temperature

Ion fluorescence Processor

Clock laser frequency step and control

4355 nm

Frequengy doubling II

3695 nm cooling wave

Low-drift cavity

Low-drift cavity 171Yb+ 435 nm ion clock unit

876 nm clock laser wave from MOLO

Frequengy doubling

Fig 7 Simplified schematic of the single-ion optical clock unit AOM acousto-optic frequencyshifter PMT photomultiplier tube EC endcap T temperature stabilization system B magneticfield

clock transition at 4355 nm can be interrogated by spectrally narrow and stableclock laser light A diagram of the simplified energy level scheme for 171Yb+showing the relevant cooling auxiliary and clock transitions is given in Fig 6All these wavelengths are provided by diode lasers operated in fundamentalor frequency doubled mode The clock laser light is obtained from the 871 nmwave provided by the MOLO unit (see Section 423) after frequency-doublingin single pass by a KNbO3 nonlinear-optical crystal to 4355 nm The clocktransition spectral profile is observed by stepping the clock laser frequency bymeans of an acousto-optic frequency shifter (AOM in Fig 7) and recordingthe statistics of the quantum jumps in cooling laser induced fluorescence asa function of this frequency Under clock operation the AOM repeatedlysteps back and forward between the half-intensity points on the transitionprofile monitoring the signal imbalance between these points Any detectedimbalance is servo-corrected to zero by feedback to the AOM

In detail the optical clock architecture will comprise (Fig 7)

bull an RF end-cap trap for ionising and confining a single 171Yb+ ion withinan ultra-high vacuum chamber pumped by a small ion pump and non-evaporable getter pump The trap is driven by means of an ac voltage ofa few hundred volts at a drive frequency of sim10 MHz A small oven withseveral mg of 171Yb isotope is heated to provide a low flux of Yb atomsfrom which single ions can be ionised by electron bombardment from ahot wire filament within the trap potential well Low voltage dc is applied

Exp Astron

to additional electrodes to position the ion precisely at trap potentialcentre in order to minimise the ionrsquos micromotion at the drive frequencyThis needs automatic periodic monitoring and correction especially in theperiod after re-loading an ion into the trap External magnetic field coilsin three orthogonal axes allow the nulling of external fields and settingof a fixed field of sim1 μT The trap and coils are surrounded by mu-metalshielding to minimise external field changes during the orbit A level oftemperature control of 1 K of the region surrounding the trap is requiredfor maintaining the blackbody-induced frequency shift uncertainty withtemperature at the 10minus17 level

bull a laser platform to provide frequency doubling of an extended cavity739 nm diode laser for Doppler cooling of the ion on the 2S12(F = 1) minus2 P12 (F = 0) dipole transition at 3695 nm Light from an amplified diodelaser device frequency-doubled in a periodically poled LiTaO3 crystal insingle-pass will be used Typical laser powers used for driving the coolingtransition below saturation are about 2 μW for a beam waist in the trapof about 50 μm By modulation of the injection current of the 739 nmdiode laser at 147 GHz a sideband is generated that excites the 2S12 (F =0) rarr2 P12 (F = 1) ldquorepumperrdquo transition in order to avoid optical pump-ing between the ground hyperfine states

bull The platform also houses the auxiliary lasers at 935 and 638 nm The935 nm diode provides repumping of the ion from the 2 D32(F = 1)metastable level after occasional branching decays to this level during thecooling sequence The 638 nm diode allows fast recovery of the ion fromthe very-long-lived 2F72 metastable state after very occasional collisionaldecay to that state Currently extended cavity lasers are forseen butdistributed-feedback lasers may be available in the near future

bull a high NA lens imaging system and photomultiplier detection system torecord the statistics of 369 nm fluorescence quantum jumps as a functionof clock laser frequency step

bull a fibre system to deliver the various cooling auxiliary and clock light fromsource to trap making use of achromatic doublets where necessary at thefibre-free space interface for launching into the trap

bull a monitoring and control processor which provides the primary coolingand clock laser pulse magnetic field and detection sequencing to observeand lock to the ion clock transition frequency The processor also monitorsfrequency and amplitude data necessary to determine normal laser andion operational conditions and initiate resetting and recovery algorithmswhere necessary and laser unit failure The actual magnetic field presenton the atomic volume is determined and actively stabilized so that aninaccuracy at the level of 1 times 10minus17 can be achieved

bull a redundancy level of at least two units for cooling clock and repumperlasers plus similar for frequency doubling crystals All redundancy unitsfor each wavelength will be fibre multiplexed as standard allowing redun-dant unit activation on determination of prior unit failure

Exp Astron

For an optimized quantum-projection-noise-limited performance of thesingle-ion clock Ramsey interrogation with a cycle time equal to the lifetimeof the metastable D32 level of 50 ms will be performed leading to an Allandeviation of about 27 times 10minus15 (τ s)minus12 This is sim45 times 10minus17 at 1 h satisfyingrequired stabilities for the science objectives Current laboratory results are astability of 1 times 10minus15 at 30 s averaging down to 3 times 10minus16 at 1000 s (PTBGermany) Also comparisons between two independent traps showed agree-ment at the level of several parts in 10minus16 over a series of eight measurementruns

The 171Yb+ ion clock offers a second option namely the 2S12ndash2 F72 octu-pole transition at 476 nm (see Table 1 and Fig 6) [48] It has an extremelylong upper state lifetime low electric quadrupole and second order Zeemansystematic shifts and therefore is a candidate for an even higher performanceclock if a clock laser system with lower instability than the above is providedApart from the clock laser it shares the laser system of the quadrupoleclock Thus with moderate additional resources the Yb+ ion apparatus couldprovide two optical clocks in a single package and specifically perform directfrequency comparisons between the quadrupole and the octupole transitionwith a large sensitivity Aα

The ion clock will operate under microprocessor control providing thecontrol algorithms for the integrated pulse sequencing for cooling repumpingclock interrogation magnetic field switching cooling fluorescence monitoringThe controller will also provide error flags for critical processes and systemreset and correction where necessary This will include ion re-loading andmicromotion reduction algorithms Automatic monitoring of laser power andspectral quality with signal re-optimisation or laser failure determination isrequired Redundant laser units will be pre-aligned and multiplexed into thefibre delivery to the trap so that redundant units can be readily activated in thecase of diode laser failure

423 Microwave-optical local oscillator (MOLO)

This oscillator provides two ultrastable frequency outputs one in the opticalregion the other in the microwave region the two frequencies being coher-ently related The MOLO is composed of three subunits the clock laser thereference cavity and the frequency comb see Fig 8

The clock laser subunit delivers the light for the excitation of the opticalclock transition It is generated by an extended-cavity diode laser at a wave-length of 871 nm Most of the output wave of this laser is sent to the ionclock The laser subunit containing two lasers for redundancy will require3 kg 3 l 15 W The frequency instability of the clock laser must be superiorto the minimum clock instability 27 times 10minus15 (τ s)minus12 for τ up to the timeconstant (sim10 s) of the servo system that locks the laser to the ionrsquos resonancesignal In order to achieve this the laser frequency is stabilized to the reference

Exp Astron

871 nm clock laser

EO

Detector and servo

AOM Femtosecond

frequency comb

To optical clock 10 GHz To FCDP

Microwave-optical local oscillator

ULE cavity

Fig 8 Microwave-optical local oscillator (MOLO) consisting of clock laser reference ULE cavityand femtosecond frequency comb (OFC) Radiofrequency inputs and outputs are not shown EOelectro-optic modulator ULE ultra-low expansion AOM frequency shifter

cavity where the PoundndashDreverndashHall method is used About 10 μW of thelaser output are used for this

The system will exhibit sim04 Hz (sim1 times 10minus15 relative) linewidth and afrequency drift lt01 Hzs The cavity consists of two high-reflectivity mirrorsoptically contacted to a cylinder of diameter 70 mm length 150 mm made ofULE (a glass exhibiting ultra low thermal expansion coefficient) with a cavityfinesse of about 300000 Residual vibrations of the satellite (arms lt 1 times 10minus6 g)should not be a limiting factor to the desired linewidth for safety marginan optimized cavity support can reduce the sensitivity to accelerations Foracoustic and thermal isolation the cavity resides inside an aluminium vacuumchamber equipped with a small (3 ls) ion-getter pump Two stages of polishedaluminium shields are implemented around the cavity which are activelytemperature stabilized so that the cavity is precisely kept at the zero crossingtemperature of the ULE thermal expansion coefficient Heat application andremoval is by thermoelectric elements between the shield and the vacuumchamber Additionally the entire vacuum chamber is temperature-stabilized

During transportation and launch the cavity will be rigidly fixed insidethe vacuum chamber by pressing on it from the sides by piezo-mechanicalactuators allowing accelerations of several g After bringing the satellite onthe orbit the actuators will be released

The optical setup for frequency stabilization will be a classic Pound-Drever-Hall scheme in reflection using an electro-optic modulator (EO) Couplingof the clock laser radiation to the cavity is via single-mode polarization main-taining optical fiber The fiber in- and out-couplers are mounted on miniaturepiezo-motor driven multiaxis translation stages Microprocessor control ofthe translation stages performs a laser-to-fiber incoupling stabilization andlaser-to-cavity mode-matching optimization (positioning inaccuracy lt10 μmalignment inaccuracy lt10 μrad)

The frequency stability of the clock laser will be limited by thermal 1f noisein the mirrors with an fractional Allan deviation slightly below 1 times 10minus15 from1 s to a few 10 s Resources including control electronics are 15 l 12 kg 8 W

Exp Astron

A femtosecond fiber laser based optical frequency comb generator (OFCFig 8) will be used for the clock frequency comparisons and to obtain aspectrally ultrapure microwave local oscillator output for the operation ofPHARAO for the microwave link and for the GNSS receiver The fem-tosecond laser is based on erbium-doped fiber pumped by fiber-coupled laserdiodes It emits 100 fs pulses at a repetition rate of 250 MHz with an averagepower of 100 mW The emission spectrum is centered around 1560 nm Spec-tral broadening to more than one octave is obtained via nonlinear interactionin a microstructure fibre so that the range from about 1050 to 2100 nm iscovered

Second harmonic generation of a portion of the broadened spectrum allowsobtaining the required light for the measurement of the clock laser frequencyat 871 nm The OFC will be referenced to the clock laser frequency (afterintroduction a controllable shift via the frequency shifter AOM) by makingthe carrier envelope offset frequency vanish so that the repetition rate fr isa high-order subharmonic of the optical clock reference frequency f1 = m1 frThe link between the optical and the microwave domain is then simply thedistribution of the repetition rate frequency or more conveniently a high(order 40) harmonic of the comb repetition rate 10 GHz

It has been demonstrated by several groups that frequency comparison andfrequency division with an OFC can be performed to very high accuracy andstability beyond the limitations imposed by the optical clocks themselves Forexample a fiber laser OFC with an instability (Allan deviation) of 6 times 10minus17

in 1 s has been demonstrated [51] OFCs based on fiber technology possessgood long-term reliability and high power efficiency The projection for a spacequalified OFC based on current industrial terrestrial technology is 16 kg mass20 l volume and 38 W power requirement

424 Frequency comparison and distribution package (FCDP)

This package distributes the ultra-stable 10 GHz reference frequency outputof the MOLO to the MWL performs frequency comparisons between MOLOand PHARAO and generates a variety of different reference signals for theindividual instruments including the 250 MHz reference clock to the OFCAll signals are coherent to the MOLO In the FCDP a USO is locked to theMOLO signal by a digital PLL A divider chain generates timing pulses (1 pps1000 pps etc) and keeps a continuous local time scale From the digital phasecomparison data the phase noise density spectrum between the USO and theMOLO is available by telemetry (TM) to assess the overall frequency purityin-flight

As the FCDP distributes the full performance reference from the MOLOdirectly to the MWL there is minimum impact of the FCDP on overall clockcomparison performance The USO-phase lock loop uses a similar designas for ACES-FCDP The digital PLL with linear phase detector allows theUSO to be slightly offset from its reference which is selectable by groundcommand This results in an accurate on-board time scale despite any offset

Exp Astron

of the MOLO reference frequency which can be synchronised to UTC Thefrequency synthesis chain re-uses modules developed for ACES-FCDP andtheir frequency inaccuracy Expected relative uncertainty (Allan deviation)of the frequency comparison between two signals provided to the FCDP is47 times 10minus17 at 1000 s 95 times 10minus18 at 10000 s

During non-availability of the MOLO ie during system start-up andcontingency conditions the USO runs as flywheel (ie it is unlocked withinstability lt15 times 10minus13 from 1 to 100 s) It will still provide useable butless accurate references to the instruments including the reference to OFCCommon view ground-to-ground time- and frequency comparison and supplyof the remaining instruments remain possible at the USO provided accuracy

Mass volume and power consumption are 8 kg 7 l 12 W Controldata rateis 10 kBday 50 kBday (FCDP alone)

425 Microwave link (MWL)

The microwave link is a two-way triple-frequency microwave link (Fig 9) Itis designed to perform accurate frequency comparison as well as time-transfer

Ku-Band Up-linkPower Tx 5 WCarrier 13475 GHzPN-Code 100 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

S-Band Down-linkPower Tx 1 WCarrier 2248 MHzPN-Code 1 MChips1pps 1 time marker sData 5 kBits

Ku-Band Down-linkPower Tx 1 WCarrier 1470333 GHzPN-Code 100 MChips1pps 1 time marker sData 5 kBits

EGE SC

S

Ku

KaKa-Band Up-linkPower Tx 5 WCarrier 345 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

Ka-Band Down-LinkPower Tx 1 WCarrier 32 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 5 kBits

GS Antenna diameterEGE terminal 1 mACES terminal 06 m

Fig 9 EGE microwave two-way link concept and characteristics See text for a description Thelink will be able to perform frequency comparisons with up to ten ground stations simultaneously(only one active comparison is shown) pps pulse per second Tx transmitter SC spacecraft

Exp Astron

between on-board and on-ground clocks Two-way operation allows for com-pensation of the effects due to spacecraft location and its motion relative toground Its two bi-directional links operate in Ka-band (32345 GHz) andKu-band (147135 GHz) The high carrier frequency of the up- and down-links allows for a noticeable reduction of the ionospheric delay A third down-link channel is in S-band (22 GHz) and is used to determine the ionospheretotal electron content MWL uses a combination of carrier phase and codephase data to measure with high precision the absolute ionospheric delayas well as ionospheric delay variations during a ground contact The useof three frequency bands allows to resolve third order ionospheric termsKey modulation parameters are Ka band Tone 250 MHz Ku band PN100 MChips S-band 1 MChips

The EGE-MWL concept is a considerably improved version of the ACES-MWL Microwave carrier generation is directly from the 10 GHz microwaveoutput available from the MOLO and having exceptionally low close-in phasenoise without intermediate frequency dividers or other complex synthesiserstages commonly used in RF transponders The high frequency carriers arekey to achieve the desired frequency comparison accuracy

The Ka-band carrier is generated by tripling the FCDP-generated signalThe Ku-band is subsequently generated by division by two using a verystable regenerative divider Finally the assigned radio frequency is obtainedby adding a low frequency signal using an image reject mixer A mixedanalog-digital receiver design will enhance its performance by use of a flexibledigital signal processing and digital control loops The number of simultaneousground contacts is increased from four (as in the ACES mission) to ten or moreusing the digital concept while the number of physical receiver channels perband is reduced from four to two one for operation to ground and one for in-flight calibration The concept allows adaptation of receiver loop parametersto optimise for the different portions of a highly elliptic orbit while reducingsystematic uncertainties due to high Doppler and Doppler rate Real-timedata exchange to ground terminals is by additional data modulation on all RFsignals The spacecraft carries two different small antennae for perigee and forapogee with gains of 8 dBi 20 dBi beamwidths 100 (full earth coverage) 14respectively power is 1 W for each

The MWL end-to-end performance between the attached clocks includingspace- and ground terminals is predicted to be a relative Allan deviation of1 times 10minus16 at 1000 s (perigee) and 1 times 10minus17 at 10000 s (apogee) Thesevalues are based on conservative implementation losses and other instrumentallosses assumed as 6 dB at apogee and 10 dB at perigee including additionallosses due to atmosphere An attitude knowledge of 1 is required

The MWL link will also support precise orbitographyndashcomplementary tothe GNSS receiver Generation of ranging data is simultaneous to two-waytime transfer and is available under all weather conditions Orbit determina-tion accuracy is expected to be better than 1 cm well matching EGE clockrequirements The value is expected based on demonstrated results fromother missions (METEOR ERS-2 TOPEXPOSEIDON) Laser ranging to

Exp Astron

the CCR array is used a few times during the mission to calibrate the MWLranging

In case of non-availability of the ultrapure signal from MOLO an additional10 GHz output is available from the FCDP to support ground-to-ground clockcomparison in common-view mode even after the nominal mission periodwhen the neither the clocks nor MOLO may be available any longer Basedon the PRARE experimentrsquos 12 years of operation this can be assumed forthe FCDPMWL combination as well The potential extended operational pe-riod makes the proposed configuration particularly attractive to metrologicallaboratories that rely on availability of long-lasting data in support of thegeneration of UTC

The resources of MWL are as follows mass 18 kg volume 18 l power55 W TC 2 kBday TM 120 MBday

426 Auxiliary units

The star camera assembly on the EGE spacecraft will provide real-timeattitude data primarily for the orientation of the spacecraft (with accuracy of1 in real-time) and the proper reduction of the GNSSSLR data

A GNSS receiver is used for the precise orbit determination time andfrequency transfer coarse positioning (lt1 m) for real time use by the Attitudeand Orbit Control as well as the back-up for time tagging of all payloaddata For the time and frequency transfer the receiver is nominally using theonboard frequency reference The receiver assembly includes three specialantennas By the year 2015 more than 100 GNSS satellites are expected tobe available for precise orbit determination permitting orbit inaccuracy of 1ndash2 cm RMS for altitudes up to approx 10000 km At apogee a combinationof GNSS signals and MWL ranging will be used with expected inaccuracyof 10 cm

5 Basic spacecraft specifications

The mission foresees a single spacecraft with a mission duration limited toapprox 3 years for cost reasons Attitude control along three axes withless than 1 error on a timescale of 1000 s is required The control can beperformed with cold gas thrusters magnetotorquers or reaction wheels Theselection will be done on the basis of the vibrations induced by these systems(desired level arms lt 10minus6 g)

Cold gas thrusters or monopropellant thrusters will be required to maintainthe orbit parameters during the mission by small orbit correction manoeuvresand will also guarantee a proper tuning of the orbit parameters after the releasefrom the launcher The ground segment (telemetry control and data transfer)could be via non-ESA stations user-operated to save costs

Further resources are indicated in Table 2

Exp Astron

The satellite bus design drivers are nadir-pointing microwave antennasthermal stability for the payload instruments no strong magnetic sources lowmicrovibrations harsh radiation environment due to crossing of the radiationbelts once per orbit power demand by science payload (with the possibility toreduce power demand by stand-by of some components during specific missionphases eg eclipse) mass limit due to required high altitude with low costlaunch

The satellite bus design can be based on established platforms such asFLEXBUS or PRIMA used on several past missions Their design philosophydiffers one having body-mounted solar arrays the other wings The latter canprovide larger amounts of power The vibration level induced by the panelswhen they are fixed and the duration of solar array drive periods per orbitwill have to be considered in the selection The FLEXBUS concept was usedfor a mission already requiring a low level of residual accelerations and theplatform is optimised for this parameter

For most of the satellite components one can consider using compo-nents and electronic units from navigation or communication satellites whichtypically cruise at altitudes between 15000 and 25000 km altitude LISAPathfinder which will cross the radiation belts during spiralling to theLagrange point will face similar problems concerning radiation as EGE as wellas XMM which is in a Molniya-type orbit with similar radiation characteristics

The instruments will therefore be located in the satellite structure such asto realise ionizing radiation shielding by the spacecraft structure and non-sensitive equipment to the maximum extent possible the need for additionalradiation shielding (including trapped proton radiation) for especially sensitiveparts of the instruments must be determined by ground tests It should bementioned that an orbit with higher perigee and less radiation dose is also anoption as it leads to a tolerable reduction of the science goals

The required plusmn3C temperature stability at the interface to the payloadenclosure is feasible The thermal control will employ the usual means (anenclosure with high thermal inertia heaters and a suitable radiator) Thelocation of the radiator will need a special study given the non-standard orbit

Magnetic field present during the mission can affect the properties of theonboard clock Using SPENVIS the magnetic field has been calculated alongthe spacecraft orbit the field intensity presents maxima around 02 G anda minimum value of the order of 10minus3 G As discussed above this is not alimitation to the instrument performance

6 Discussion and conclusion

The accuracy of the measurements depends crucially on the stability andaccuracy of the clocks While the performance of ground clocks (even thoseeventually deployed in new ground stations dedicated to EGE) will likelyimprove significantly from the present level the space clock design will beldquofrozenrdquo approx 10 years before data acquisition Thus their performance

Exp Astron

level will be significantly below the future ground station clock performanceAs specification we have chosen a performance level that has been demon-strated with laboratory breadboards recently We see no fundamental reasonthat this laboratory performance cannot also be achieved in a space ion opticalclock The experience with the PHARAO cold atom clock has shown thatan engineering model of a laser-based cold atom clock has a performancesimilar to that of a laboratory prototype The perturbation of the ion clock fre-quency due to magnetic fields along the orbit can be modelled and suppressedwith shielding and the residual effects measured and compensated for Themoderate temperature stability (plusmn 3C) to be provided in the experimentalbay can be achieved with thermal shields Accurate temperature controlof particular units is done by active stabilization of the units themselves

One important aspect is radiation which may affect all optical and trapcomponents as well as the lasers and reference cavity The radiation envi-ronment on the EGE orbit is significantly harsher than the near-earth ISSorbit on which ACES will operate The radiation effects must therefore becharacterized comprehensively with appropriate tests and then the radiationshield for the EGE instruments designed

In summary the Einstein Gravity Explorer mission (EGE) is a satellitemission aimed at a precise measurement of the properties of space-time usingatomic clocks By testing one of the most fundamental predictions of EinsteinrsquosTheory of General Relativity the gravitational redshift and thereby searchingfor quantum effects in gravity it explores one of the most important frontiers infundamental physics The EGE mission reaches beyond fundamental physicsproviding an important spin-off in geophysics Furthermore the technologicalaspects of the mission are ground-breaking for future time-keeping and ref-erence frame systems in space and thus are also expected to be of wide useto science and technology in general The payload of the mission has beenconceived as a compromise between cost and performance needing to fit thecost envelope of an ESA medium-class mission1

Acknowledgement We thank S Cesare for helpful information

References

1 Will CM The confrontation between general relativity and experiment Living Rev Relativ9 3 (2006)

2 Ni W-T Empirical foundations of the relativistic gravity Int J Mod Phys D 14 901 (2005)doi101142S0218271805007139

1Finally we point out that a spacecraft with a significantly reduced and thus lower-cost payload (noclocks only MOLO FCDP MWL) on a circular orbit of appropriate height could be used for along-duration earth science mission dedicated specifically to the measurement of the geopotential

Exp Astron

which optical clocks demonstrators have been operated in not a single but twomajor metrology laboratories

Another attractive optical clock is based on ensembles of cold neutral Sror Yb atoms trapped in an optical lattice produced by laser beams This clocktype has a lower instability than a single-ion clock thanks to the large numberof atoms (sim105) used Its development is in progress in several laboratoriesin the world (in part supported by space agencies) and has to date reached6 times 10minus17 instability at 5000 s and 1 times 10minus16 uncertainty [49 50] with potentialfor even higher performance However it is more complex and costly thanthe Yb+ ion clock since the cold atom preparation and trapping subunitsinclude more components require higher laser powers and the respectiveoperating procedures are more complex It is therefore not considered for thepresent mission designed to be developed in the very near future with limitedresources but is an interesting candidate for missions without such constraintsIts estimated physical parameters are indicated in Table 2 for comparison

42 Main instruments

421 Microwave clock

The PHARAO cold atom clock is one of the two atomic clocks of the spacemission ACES managed by ESA It is proposed here to include a secondversion of this instrument as its costperformance ratio is very attractive forthe EGE scientific objectives Its concept is very similar to ground basedatomic fountains but with the major difference of zero-g operation Atomsslowly launched in free flight cross two microwave fields tuned to the transitionbetween the two hyperfine levels of the cesium ground state The interrogationmethod based on two separate oscillating fields (Ramsey scheme) allowsthe detection of an atomic line whose width is inversely proportional to thetransit time between the two microwave cavities The resonant microwavefield at 9192631770 GHz (SI definition of the second) is synthesized startingfrom an ultrastable quartz oscillator (USO) in the FCDP phaselocked to themicrowave output of the microwave-optical local oscillator and stabilized tothe clock line using the error signal generated by the cesium resonator In thisway the intrinsic qualities of the cesium hyperfine transition both in terms ofaccuracy and frequency stability are transferred to the USO In a microgravityenvironment the velocity of atoms along the ballistic trajectories is constantand can be changed continuously over almost two orders of magnitude (5ndash500 cms) Differently from atomic fountain clocks presently operated onground very long interaction times (up to few seconds) will be possible whilekeeping reasonable the size of the instrument

The instrument consists of 4 main elements a laser source which providesthe light required to cool and detect the cesium atoms (Fig 5 left) a frequencychain which generates the 92 GHz microwave signal a cesium vacuum tubewhere the interaction between the microwave field and the atoms occurs

Exp Astron

Table 2 Resources and technology readiness level (TRL) of the EGE spacecraft

Unit Volume Mass Power TRL Comment[l] [kg] [W]

Ion clock 100 70 60Atom source 4Atom preparation 6Atom trap 4Atom interrogation 6Clock control 6

Microwave clock 200 91 110 8 TRL by end 2008Local oscillator

Stabilized laser 18 15 23 6Frequency comb 20 16 38 6

Microwave link 18 18 55 8FCDP 7 8 12 8GPS receiver 2 5 7 9CCR 2 2 0 9Payload sum 367 225 305Thermal control 14 17 30 8OBDH 20 24 38 9Star tracker 1 1 8 9Radiation shield 10 40 0 9Support structure 170 204 0 9Power incl solar arr 50 58 10 9Propulsion 21 25 20 9AOCS 36 43 82 9TMTC 8 10 15 9Satellite bus sum 330 422 203 Flexbus conceptPayload plus bus incl 20 margin 8364 7764 6096Propellant 40Launch adapter 100SC wet masspower incl Margins 9164Lattice clock 135 105 152 Alternative clock

Atom source 4Atom preparation 6Atom trap 4Atom interrogation 6Clock control 6

Also included are the resources for an optical lattice clock based on neutral Strontium atoms

(Fig 5 right) and a computer The engineering model of PHARAO is fullyassembled at CNES Toulouse and undergoes functional and performancetests until the end of 2008 The operating temperature range is 10ndash33 andnon-operating temperatures are minus45 +60 The cesium tube consists in aUHV chamber pumped by getters and a 3 ls ion pump a cesium reservoira microwave cavity coils to provide a uniform magnetic field and three layersof magnetic shields Vacuum windows with fiber-optics collimators enable thetransmission of the cooling and detection light onto the cesium atoms

In autonomous mode PHARAO uses the USO as interrogation oscillatorand will provide a clock signal with fractional frequency stability below1 times 10minus13 (τs) minus12 and inaccuracy near 1 times 10minus16 When using the microwave-

Exp Astron

Fig 5 Left The PHARAO laser source engineering model with cover removed Dimensionsare 530 times 350 times 150 mm3 and the mass is 200 kg Ten polarization maintaining optical fibers(yellow) guide the laser beams to the cesium tube All diode lasers are mounted on Peltier coolersfor temperature regulation within 2 mK (courtesy EADS SODERN) Right The PHARAOengineering model cesium tube without the two external magnetic shields The volume is 990 times336 times 444 mm3 and the total mass is 44 kg

optical local oscillator PHARAO will provide a reduced frequency instabilityof 3 times 10minus14 (τs) minus12 This corresponds to 1 times 10minus15 at 1000 s and 3 times 10minus16

at 104 s

422 Ion optical clock

The 171Yb+ ion clock shown in Figs 6 and 7 provides an atomic frequencyreference based on a single 171Yb+ ion confined within an RF electric end-captrap (similar to a Paul trap) within an ultra high vacuum (UHV) chamber andlaser cooled to sim1 mK close to the Doppler cooling limit As a result the ionrsquosfirst order Doppler motion is completely removed and the ion experienceslittle collisional perturbation from its environment Under these conditionsthe narrow linewidth 2S12 (F = 0 mF = 0) minus 2 D32 (F = 2 mF = 0) quadrupole

Fig 6 Left Simplified level scheme of 171Yb+ showing relevant cooling (blue) auxiliary (red)quadrupole clock (green) and octupole clock (dark blue) transitions δνnat is the natural linewidthof the clock transition Middle close-up of Paul-type RF trap (PTB Braunschweig) the ion istrapped in the hole (diameter 14 mm) Right Complete ion trap with vacuum flange height 10 cm(NPL Teddington)

Exp Astron

935 nm D32-state repumper laser

638 nm F72-state repumper laser

739 nm extended cavity diode laser

AOM

PMT

Ion trap UHV chamber

vacuum pumps

Fibre-free space interface

Mu-metal shielding

EC

T

B

Magnetic field

Temperature

Ion fluorescence Processor

Clock laser frequency step and control

4355 nm

Frequengy doubling II

3695 nm cooling wave

Low-drift cavity

Low-drift cavity 171Yb+ 435 nm ion clock unit

876 nm clock laser wave from MOLO

Frequengy doubling

Fig 7 Simplified schematic of the single-ion optical clock unit AOM acousto-optic frequencyshifter PMT photomultiplier tube EC endcap T temperature stabilization system B magneticfield

clock transition at 4355 nm can be interrogated by spectrally narrow and stableclock laser light A diagram of the simplified energy level scheme for 171Yb+showing the relevant cooling auxiliary and clock transitions is given in Fig 6All these wavelengths are provided by diode lasers operated in fundamentalor frequency doubled mode The clock laser light is obtained from the 871 nmwave provided by the MOLO unit (see Section 423) after frequency-doublingin single pass by a KNbO3 nonlinear-optical crystal to 4355 nm The clocktransition spectral profile is observed by stepping the clock laser frequency bymeans of an acousto-optic frequency shifter (AOM in Fig 7) and recordingthe statistics of the quantum jumps in cooling laser induced fluorescence asa function of this frequency Under clock operation the AOM repeatedlysteps back and forward between the half-intensity points on the transitionprofile monitoring the signal imbalance between these points Any detectedimbalance is servo-corrected to zero by feedback to the AOM

In detail the optical clock architecture will comprise (Fig 7)

bull an RF end-cap trap for ionising and confining a single 171Yb+ ion withinan ultra-high vacuum chamber pumped by a small ion pump and non-evaporable getter pump The trap is driven by means of an ac voltage ofa few hundred volts at a drive frequency of sim10 MHz A small oven withseveral mg of 171Yb isotope is heated to provide a low flux of Yb atomsfrom which single ions can be ionised by electron bombardment from ahot wire filament within the trap potential well Low voltage dc is applied

Exp Astron

to additional electrodes to position the ion precisely at trap potentialcentre in order to minimise the ionrsquos micromotion at the drive frequencyThis needs automatic periodic monitoring and correction especially in theperiod after re-loading an ion into the trap External magnetic field coilsin three orthogonal axes allow the nulling of external fields and settingof a fixed field of sim1 μT The trap and coils are surrounded by mu-metalshielding to minimise external field changes during the orbit A level oftemperature control of 1 K of the region surrounding the trap is requiredfor maintaining the blackbody-induced frequency shift uncertainty withtemperature at the 10minus17 level

bull a laser platform to provide frequency doubling of an extended cavity739 nm diode laser for Doppler cooling of the ion on the 2S12(F = 1) minus2 P12 (F = 0) dipole transition at 3695 nm Light from an amplified diodelaser device frequency-doubled in a periodically poled LiTaO3 crystal insingle-pass will be used Typical laser powers used for driving the coolingtransition below saturation are about 2 μW for a beam waist in the trapof about 50 μm By modulation of the injection current of the 739 nmdiode laser at 147 GHz a sideband is generated that excites the 2S12 (F =0) rarr2 P12 (F = 1) ldquorepumperrdquo transition in order to avoid optical pump-ing between the ground hyperfine states

bull The platform also houses the auxiliary lasers at 935 and 638 nm The935 nm diode provides repumping of the ion from the 2 D32(F = 1)metastable level after occasional branching decays to this level during thecooling sequence The 638 nm diode allows fast recovery of the ion fromthe very-long-lived 2F72 metastable state after very occasional collisionaldecay to that state Currently extended cavity lasers are forseen butdistributed-feedback lasers may be available in the near future

bull a high NA lens imaging system and photomultiplier detection system torecord the statistics of 369 nm fluorescence quantum jumps as a functionof clock laser frequency step

bull a fibre system to deliver the various cooling auxiliary and clock light fromsource to trap making use of achromatic doublets where necessary at thefibre-free space interface for launching into the trap

bull a monitoring and control processor which provides the primary coolingand clock laser pulse magnetic field and detection sequencing to observeand lock to the ion clock transition frequency The processor also monitorsfrequency and amplitude data necessary to determine normal laser andion operational conditions and initiate resetting and recovery algorithmswhere necessary and laser unit failure The actual magnetic field presenton the atomic volume is determined and actively stabilized so that aninaccuracy at the level of 1 times 10minus17 can be achieved

bull a redundancy level of at least two units for cooling clock and repumperlasers plus similar for frequency doubling crystals All redundancy unitsfor each wavelength will be fibre multiplexed as standard allowing redun-dant unit activation on determination of prior unit failure

Exp Astron

For an optimized quantum-projection-noise-limited performance of thesingle-ion clock Ramsey interrogation with a cycle time equal to the lifetimeof the metastable D32 level of 50 ms will be performed leading to an Allandeviation of about 27 times 10minus15 (τ s)minus12 This is sim45 times 10minus17 at 1 h satisfyingrequired stabilities for the science objectives Current laboratory results are astability of 1 times 10minus15 at 30 s averaging down to 3 times 10minus16 at 1000 s (PTBGermany) Also comparisons between two independent traps showed agree-ment at the level of several parts in 10minus16 over a series of eight measurementruns

The 171Yb+ ion clock offers a second option namely the 2S12ndash2 F72 octu-pole transition at 476 nm (see Table 1 and Fig 6) [48] It has an extremelylong upper state lifetime low electric quadrupole and second order Zeemansystematic shifts and therefore is a candidate for an even higher performanceclock if a clock laser system with lower instability than the above is providedApart from the clock laser it shares the laser system of the quadrupoleclock Thus with moderate additional resources the Yb+ ion apparatus couldprovide two optical clocks in a single package and specifically perform directfrequency comparisons between the quadrupole and the octupole transitionwith a large sensitivity Aα

The ion clock will operate under microprocessor control providing thecontrol algorithms for the integrated pulse sequencing for cooling repumpingclock interrogation magnetic field switching cooling fluorescence monitoringThe controller will also provide error flags for critical processes and systemreset and correction where necessary This will include ion re-loading andmicromotion reduction algorithms Automatic monitoring of laser power andspectral quality with signal re-optimisation or laser failure determination isrequired Redundant laser units will be pre-aligned and multiplexed into thefibre delivery to the trap so that redundant units can be readily activated in thecase of diode laser failure

423 Microwave-optical local oscillator (MOLO)

This oscillator provides two ultrastable frequency outputs one in the opticalregion the other in the microwave region the two frequencies being coher-ently related The MOLO is composed of three subunits the clock laser thereference cavity and the frequency comb see Fig 8

The clock laser subunit delivers the light for the excitation of the opticalclock transition It is generated by an extended-cavity diode laser at a wave-length of 871 nm Most of the output wave of this laser is sent to the ionclock The laser subunit containing two lasers for redundancy will require3 kg 3 l 15 W The frequency instability of the clock laser must be superiorto the minimum clock instability 27 times 10minus15 (τ s)minus12 for τ up to the timeconstant (sim10 s) of the servo system that locks the laser to the ionrsquos resonancesignal In order to achieve this the laser frequency is stabilized to the reference

Exp Astron

871 nm clock laser

EO

Detector and servo

AOM Femtosecond

frequency comb

To optical clock 10 GHz To FCDP

Microwave-optical local oscillator

ULE cavity

Fig 8 Microwave-optical local oscillator (MOLO) consisting of clock laser reference ULE cavityand femtosecond frequency comb (OFC) Radiofrequency inputs and outputs are not shown EOelectro-optic modulator ULE ultra-low expansion AOM frequency shifter

cavity where the PoundndashDreverndashHall method is used About 10 μW of thelaser output are used for this

The system will exhibit sim04 Hz (sim1 times 10minus15 relative) linewidth and afrequency drift lt01 Hzs The cavity consists of two high-reflectivity mirrorsoptically contacted to a cylinder of diameter 70 mm length 150 mm made ofULE (a glass exhibiting ultra low thermal expansion coefficient) with a cavityfinesse of about 300000 Residual vibrations of the satellite (arms lt 1 times 10minus6 g)should not be a limiting factor to the desired linewidth for safety marginan optimized cavity support can reduce the sensitivity to accelerations Foracoustic and thermal isolation the cavity resides inside an aluminium vacuumchamber equipped with a small (3 ls) ion-getter pump Two stages of polishedaluminium shields are implemented around the cavity which are activelytemperature stabilized so that the cavity is precisely kept at the zero crossingtemperature of the ULE thermal expansion coefficient Heat application andremoval is by thermoelectric elements between the shield and the vacuumchamber Additionally the entire vacuum chamber is temperature-stabilized

During transportation and launch the cavity will be rigidly fixed insidethe vacuum chamber by pressing on it from the sides by piezo-mechanicalactuators allowing accelerations of several g After bringing the satellite onthe orbit the actuators will be released

The optical setup for frequency stabilization will be a classic Pound-Drever-Hall scheme in reflection using an electro-optic modulator (EO) Couplingof the clock laser radiation to the cavity is via single-mode polarization main-taining optical fiber The fiber in- and out-couplers are mounted on miniaturepiezo-motor driven multiaxis translation stages Microprocessor control ofthe translation stages performs a laser-to-fiber incoupling stabilization andlaser-to-cavity mode-matching optimization (positioning inaccuracy lt10 μmalignment inaccuracy lt10 μrad)

The frequency stability of the clock laser will be limited by thermal 1f noisein the mirrors with an fractional Allan deviation slightly below 1 times 10minus15 from1 s to a few 10 s Resources including control electronics are 15 l 12 kg 8 W

Exp Astron

A femtosecond fiber laser based optical frequency comb generator (OFCFig 8) will be used for the clock frequency comparisons and to obtain aspectrally ultrapure microwave local oscillator output for the operation ofPHARAO for the microwave link and for the GNSS receiver The fem-tosecond laser is based on erbium-doped fiber pumped by fiber-coupled laserdiodes It emits 100 fs pulses at a repetition rate of 250 MHz with an averagepower of 100 mW The emission spectrum is centered around 1560 nm Spec-tral broadening to more than one octave is obtained via nonlinear interactionin a microstructure fibre so that the range from about 1050 to 2100 nm iscovered

Second harmonic generation of a portion of the broadened spectrum allowsobtaining the required light for the measurement of the clock laser frequencyat 871 nm The OFC will be referenced to the clock laser frequency (afterintroduction a controllable shift via the frequency shifter AOM) by makingthe carrier envelope offset frequency vanish so that the repetition rate fr isa high-order subharmonic of the optical clock reference frequency f1 = m1 frThe link between the optical and the microwave domain is then simply thedistribution of the repetition rate frequency or more conveniently a high(order 40) harmonic of the comb repetition rate 10 GHz

It has been demonstrated by several groups that frequency comparison andfrequency division with an OFC can be performed to very high accuracy andstability beyond the limitations imposed by the optical clocks themselves Forexample a fiber laser OFC with an instability (Allan deviation) of 6 times 10minus17

in 1 s has been demonstrated [51] OFCs based on fiber technology possessgood long-term reliability and high power efficiency The projection for a spacequalified OFC based on current industrial terrestrial technology is 16 kg mass20 l volume and 38 W power requirement

424 Frequency comparison and distribution package (FCDP)

This package distributes the ultra-stable 10 GHz reference frequency outputof the MOLO to the MWL performs frequency comparisons between MOLOand PHARAO and generates a variety of different reference signals for theindividual instruments including the 250 MHz reference clock to the OFCAll signals are coherent to the MOLO In the FCDP a USO is locked to theMOLO signal by a digital PLL A divider chain generates timing pulses (1 pps1000 pps etc) and keeps a continuous local time scale From the digital phasecomparison data the phase noise density spectrum between the USO and theMOLO is available by telemetry (TM) to assess the overall frequency purityin-flight

As the FCDP distributes the full performance reference from the MOLOdirectly to the MWL there is minimum impact of the FCDP on overall clockcomparison performance The USO-phase lock loop uses a similar designas for ACES-FCDP The digital PLL with linear phase detector allows theUSO to be slightly offset from its reference which is selectable by groundcommand This results in an accurate on-board time scale despite any offset

Exp Astron

of the MOLO reference frequency which can be synchronised to UTC Thefrequency synthesis chain re-uses modules developed for ACES-FCDP andtheir frequency inaccuracy Expected relative uncertainty (Allan deviation)of the frequency comparison between two signals provided to the FCDP is47 times 10minus17 at 1000 s 95 times 10minus18 at 10000 s

During non-availability of the MOLO ie during system start-up andcontingency conditions the USO runs as flywheel (ie it is unlocked withinstability lt15 times 10minus13 from 1 to 100 s) It will still provide useable butless accurate references to the instruments including the reference to OFCCommon view ground-to-ground time- and frequency comparison and supplyof the remaining instruments remain possible at the USO provided accuracy

Mass volume and power consumption are 8 kg 7 l 12 W Controldata rateis 10 kBday 50 kBday (FCDP alone)

425 Microwave link (MWL)

The microwave link is a two-way triple-frequency microwave link (Fig 9) Itis designed to perform accurate frequency comparison as well as time-transfer

Ku-Band Up-linkPower Tx 5 WCarrier 13475 GHzPN-Code 100 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

S-Band Down-linkPower Tx 1 WCarrier 2248 MHzPN-Code 1 MChips1pps 1 time marker sData 5 kBits

Ku-Band Down-linkPower Tx 1 WCarrier 1470333 GHzPN-Code 100 MChips1pps 1 time marker sData 5 kBits

EGE SC

S

Ku

KaKa-Band Up-linkPower Tx 5 WCarrier 345 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

Ka-Band Down-LinkPower Tx 1 WCarrier 32 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 5 kBits

GS Antenna diameterEGE terminal 1 mACES terminal 06 m

Fig 9 EGE microwave two-way link concept and characteristics See text for a description Thelink will be able to perform frequency comparisons with up to ten ground stations simultaneously(only one active comparison is shown) pps pulse per second Tx transmitter SC spacecraft

Exp Astron

between on-board and on-ground clocks Two-way operation allows for com-pensation of the effects due to spacecraft location and its motion relative toground Its two bi-directional links operate in Ka-band (32345 GHz) andKu-band (147135 GHz) The high carrier frequency of the up- and down-links allows for a noticeable reduction of the ionospheric delay A third down-link channel is in S-band (22 GHz) and is used to determine the ionospheretotal electron content MWL uses a combination of carrier phase and codephase data to measure with high precision the absolute ionospheric delayas well as ionospheric delay variations during a ground contact The useof three frequency bands allows to resolve third order ionospheric termsKey modulation parameters are Ka band Tone 250 MHz Ku band PN100 MChips S-band 1 MChips

The EGE-MWL concept is a considerably improved version of the ACES-MWL Microwave carrier generation is directly from the 10 GHz microwaveoutput available from the MOLO and having exceptionally low close-in phasenoise without intermediate frequency dividers or other complex synthesiserstages commonly used in RF transponders The high frequency carriers arekey to achieve the desired frequency comparison accuracy

The Ka-band carrier is generated by tripling the FCDP-generated signalThe Ku-band is subsequently generated by division by two using a verystable regenerative divider Finally the assigned radio frequency is obtainedby adding a low frequency signal using an image reject mixer A mixedanalog-digital receiver design will enhance its performance by use of a flexibledigital signal processing and digital control loops The number of simultaneousground contacts is increased from four (as in the ACES mission) to ten or moreusing the digital concept while the number of physical receiver channels perband is reduced from four to two one for operation to ground and one for in-flight calibration The concept allows adaptation of receiver loop parametersto optimise for the different portions of a highly elliptic orbit while reducingsystematic uncertainties due to high Doppler and Doppler rate Real-timedata exchange to ground terminals is by additional data modulation on all RFsignals The spacecraft carries two different small antennae for perigee and forapogee with gains of 8 dBi 20 dBi beamwidths 100 (full earth coverage) 14respectively power is 1 W for each

The MWL end-to-end performance between the attached clocks includingspace- and ground terminals is predicted to be a relative Allan deviation of1 times 10minus16 at 1000 s (perigee) and 1 times 10minus17 at 10000 s (apogee) Thesevalues are based on conservative implementation losses and other instrumentallosses assumed as 6 dB at apogee and 10 dB at perigee including additionallosses due to atmosphere An attitude knowledge of 1 is required

The MWL link will also support precise orbitographyndashcomplementary tothe GNSS receiver Generation of ranging data is simultaneous to two-waytime transfer and is available under all weather conditions Orbit determina-tion accuracy is expected to be better than 1 cm well matching EGE clockrequirements The value is expected based on demonstrated results fromother missions (METEOR ERS-2 TOPEXPOSEIDON) Laser ranging to

Exp Astron

the CCR array is used a few times during the mission to calibrate the MWLranging

In case of non-availability of the ultrapure signal from MOLO an additional10 GHz output is available from the FCDP to support ground-to-ground clockcomparison in common-view mode even after the nominal mission periodwhen the neither the clocks nor MOLO may be available any longer Basedon the PRARE experimentrsquos 12 years of operation this can be assumed forthe FCDPMWL combination as well The potential extended operational pe-riod makes the proposed configuration particularly attractive to metrologicallaboratories that rely on availability of long-lasting data in support of thegeneration of UTC

The resources of MWL are as follows mass 18 kg volume 18 l power55 W TC 2 kBday TM 120 MBday

426 Auxiliary units

The star camera assembly on the EGE spacecraft will provide real-timeattitude data primarily for the orientation of the spacecraft (with accuracy of1 in real-time) and the proper reduction of the GNSSSLR data

A GNSS receiver is used for the precise orbit determination time andfrequency transfer coarse positioning (lt1 m) for real time use by the Attitudeand Orbit Control as well as the back-up for time tagging of all payloaddata For the time and frequency transfer the receiver is nominally using theonboard frequency reference The receiver assembly includes three specialantennas By the year 2015 more than 100 GNSS satellites are expected tobe available for precise orbit determination permitting orbit inaccuracy of 1ndash2 cm RMS for altitudes up to approx 10000 km At apogee a combinationof GNSS signals and MWL ranging will be used with expected inaccuracyof 10 cm

5 Basic spacecraft specifications

The mission foresees a single spacecraft with a mission duration limited toapprox 3 years for cost reasons Attitude control along three axes withless than 1 error on a timescale of 1000 s is required The control can beperformed with cold gas thrusters magnetotorquers or reaction wheels Theselection will be done on the basis of the vibrations induced by these systems(desired level arms lt 10minus6 g)

Cold gas thrusters or monopropellant thrusters will be required to maintainthe orbit parameters during the mission by small orbit correction manoeuvresand will also guarantee a proper tuning of the orbit parameters after the releasefrom the launcher The ground segment (telemetry control and data transfer)could be via non-ESA stations user-operated to save costs

Further resources are indicated in Table 2

Exp Astron

The satellite bus design drivers are nadir-pointing microwave antennasthermal stability for the payload instruments no strong magnetic sources lowmicrovibrations harsh radiation environment due to crossing of the radiationbelts once per orbit power demand by science payload (with the possibility toreduce power demand by stand-by of some components during specific missionphases eg eclipse) mass limit due to required high altitude with low costlaunch

The satellite bus design can be based on established platforms such asFLEXBUS or PRIMA used on several past missions Their design philosophydiffers one having body-mounted solar arrays the other wings The latter canprovide larger amounts of power The vibration level induced by the panelswhen they are fixed and the duration of solar array drive periods per orbitwill have to be considered in the selection The FLEXBUS concept was usedfor a mission already requiring a low level of residual accelerations and theplatform is optimised for this parameter

For most of the satellite components one can consider using compo-nents and electronic units from navigation or communication satellites whichtypically cruise at altitudes between 15000 and 25000 km altitude LISAPathfinder which will cross the radiation belts during spiralling to theLagrange point will face similar problems concerning radiation as EGE as wellas XMM which is in a Molniya-type orbit with similar radiation characteristics

The instruments will therefore be located in the satellite structure such asto realise ionizing radiation shielding by the spacecraft structure and non-sensitive equipment to the maximum extent possible the need for additionalradiation shielding (including trapped proton radiation) for especially sensitiveparts of the instruments must be determined by ground tests It should bementioned that an orbit with higher perigee and less radiation dose is also anoption as it leads to a tolerable reduction of the science goals

The required plusmn3C temperature stability at the interface to the payloadenclosure is feasible The thermal control will employ the usual means (anenclosure with high thermal inertia heaters and a suitable radiator) Thelocation of the radiator will need a special study given the non-standard orbit

Magnetic field present during the mission can affect the properties of theonboard clock Using SPENVIS the magnetic field has been calculated alongthe spacecraft orbit the field intensity presents maxima around 02 G anda minimum value of the order of 10minus3 G As discussed above this is not alimitation to the instrument performance

6 Discussion and conclusion

The accuracy of the measurements depends crucially on the stability andaccuracy of the clocks While the performance of ground clocks (even thoseeventually deployed in new ground stations dedicated to EGE) will likelyimprove significantly from the present level the space clock design will beldquofrozenrdquo approx 10 years before data acquisition Thus their performance

Exp Astron

level will be significantly below the future ground station clock performanceAs specification we have chosen a performance level that has been demon-strated with laboratory breadboards recently We see no fundamental reasonthat this laboratory performance cannot also be achieved in a space ion opticalclock The experience with the PHARAO cold atom clock has shown thatan engineering model of a laser-based cold atom clock has a performancesimilar to that of a laboratory prototype The perturbation of the ion clock fre-quency due to magnetic fields along the orbit can be modelled and suppressedwith shielding and the residual effects measured and compensated for Themoderate temperature stability (plusmn 3C) to be provided in the experimentalbay can be achieved with thermal shields Accurate temperature controlof particular units is done by active stabilization of the units themselves

One important aspect is radiation which may affect all optical and trapcomponents as well as the lasers and reference cavity The radiation envi-ronment on the EGE orbit is significantly harsher than the near-earth ISSorbit on which ACES will operate The radiation effects must therefore becharacterized comprehensively with appropriate tests and then the radiationshield for the EGE instruments designed

In summary the Einstein Gravity Explorer mission (EGE) is a satellitemission aimed at a precise measurement of the properties of space-time usingatomic clocks By testing one of the most fundamental predictions of EinsteinrsquosTheory of General Relativity the gravitational redshift and thereby searchingfor quantum effects in gravity it explores one of the most important frontiers infundamental physics The EGE mission reaches beyond fundamental physicsproviding an important spin-off in geophysics Furthermore the technologicalaspects of the mission are ground-breaking for future time-keeping and ref-erence frame systems in space and thus are also expected to be of wide useto science and technology in general The payload of the mission has beenconceived as a compromise between cost and performance needing to fit thecost envelope of an ESA medium-class mission1

Acknowledgement We thank S Cesare for helpful information

References

1 Will CM The confrontation between general relativity and experiment Living Rev Relativ9 3 (2006)

2 Ni W-T Empirical foundations of the relativistic gravity Int J Mod Phys D 14 901 (2005)doi101142S0218271805007139

1Finally we point out that a spacecraft with a significantly reduced and thus lower-cost payload (noclocks only MOLO FCDP MWL) on a circular orbit of appropriate height could be used for along-duration earth science mission dedicated specifically to the measurement of the geopotential

Exp Astron

Table 2 Resources and technology readiness level (TRL) of the EGE spacecraft

Unit Volume Mass Power TRL Comment[l] [kg] [W]

Ion clock 100 70 60Atom source 4Atom preparation 6Atom trap 4Atom interrogation 6Clock control 6

Microwave clock 200 91 110 8 TRL by end 2008Local oscillator

Stabilized laser 18 15 23 6Frequency comb 20 16 38 6

Microwave link 18 18 55 8FCDP 7 8 12 8GPS receiver 2 5 7 9CCR 2 2 0 9Payload sum 367 225 305Thermal control 14 17 30 8OBDH 20 24 38 9Star tracker 1 1 8 9Radiation shield 10 40 0 9Support structure 170 204 0 9Power incl solar arr 50 58 10 9Propulsion 21 25 20 9AOCS 36 43 82 9TMTC 8 10 15 9Satellite bus sum 330 422 203 Flexbus conceptPayload plus bus incl 20 margin 8364 7764 6096Propellant 40Launch adapter 100SC wet masspower incl Margins 9164Lattice clock 135 105 152 Alternative clock

Atom source 4Atom preparation 6Atom trap 4Atom interrogation 6Clock control 6

Also included are the resources for an optical lattice clock based on neutral Strontium atoms

(Fig 5 right) and a computer The engineering model of PHARAO is fullyassembled at CNES Toulouse and undergoes functional and performancetests until the end of 2008 The operating temperature range is 10ndash33 andnon-operating temperatures are minus45 +60 The cesium tube consists in aUHV chamber pumped by getters and a 3 ls ion pump a cesium reservoira microwave cavity coils to provide a uniform magnetic field and three layersof magnetic shields Vacuum windows with fiber-optics collimators enable thetransmission of the cooling and detection light onto the cesium atoms

In autonomous mode PHARAO uses the USO as interrogation oscillatorand will provide a clock signal with fractional frequency stability below1 times 10minus13 (τs) minus12 and inaccuracy near 1 times 10minus16 When using the microwave-

Exp Astron

Fig 5 Left The PHARAO laser source engineering model with cover removed Dimensionsare 530 times 350 times 150 mm3 and the mass is 200 kg Ten polarization maintaining optical fibers(yellow) guide the laser beams to the cesium tube All diode lasers are mounted on Peltier coolersfor temperature regulation within 2 mK (courtesy EADS SODERN) Right The PHARAOengineering model cesium tube without the two external magnetic shields The volume is 990 times336 times 444 mm3 and the total mass is 44 kg

optical local oscillator PHARAO will provide a reduced frequency instabilityof 3 times 10minus14 (τs) minus12 This corresponds to 1 times 10minus15 at 1000 s and 3 times 10minus16

at 104 s

422 Ion optical clock

The 171Yb+ ion clock shown in Figs 6 and 7 provides an atomic frequencyreference based on a single 171Yb+ ion confined within an RF electric end-captrap (similar to a Paul trap) within an ultra high vacuum (UHV) chamber andlaser cooled to sim1 mK close to the Doppler cooling limit As a result the ionrsquosfirst order Doppler motion is completely removed and the ion experienceslittle collisional perturbation from its environment Under these conditionsthe narrow linewidth 2S12 (F = 0 mF = 0) minus 2 D32 (F = 2 mF = 0) quadrupole

Fig 6 Left Simplified level scheme of 171Yb+ showing relevant cooling (blue) auxiliary (red)quadrupole clock (green) and octupole clock (dark blue) transitions δνnat is the natural linewidthof the clock transition Middle close-up of Paul-type RF trap (PTB Braunschweig) the ion istrapped in the hole (diameter 14 mm) Right Complete ion trap with vacuum flange height 10 cm(NPL Teddington)

Exp Astron

935 nm D32-state repumper laser

638 nm F72-state repumper laser

739 nm extended cavity diode laser

AOM

PMT

Ion trap UHV chamber

vacuum pumps

Fibre-free space interface

Mu-metal shielding

EC

T

B

Magnetic field

Temperature

Ion fluorescence Processor

Clock laser frequency step and control

4355 nm

Frequengy doubling II

3695 nm cooling wave

Low-drift cavity

Low-drift cavity 171Yb+ 435 nm ion clock unit

876 nm clock laser wave from MOLO

Frequengy doubling

Fig 7 Simplified schematic of the single-ion optical clock unit AOM acousto-optic frequencyshifter PMT photomultiplier tube EC endcap T temperature stabilization system B magneticfield

clock transition at 4355 nm can be interrogated by spectrally narrow and stableclock laser light A diagram of the simplified energy level scheme for 171Yb+showing the relevant cooling auxiliary and clock transitions is given in Fig 6All these wavelengths are provided by diode lasers operated in fundamentalor frequency doubled mode The clock laser light is obtained from the 871 nmwave provided by the MOLO unit (see Section 423) after frequency-doublingin single pass by a KNbO3 nonlinear-optical crystal to 4355 nm The clocktransition spectral profile is observed by stepping the clock laser frequency bymeans of an acousto-optic frequency shifter (AOM in Fig 7) and recordingthe statistics of the quantum jumps in cooling laser induced fluorescence asa function of this frequency Under clock operation the AOM repeatedlysteps back and forward between the half-intensity points on the transitionprofile monitoring the signal imbalance between these points Any detectedimbalance is servo-corrected to zero by feedback to the AOM

In detail the optical clock architecture will comprise (Fig 7)

bull an RF end-cap trap for ionising and confining a single 171Yb+ ion withinan ultra-high vacuum chamber pumped by a small ion pump and non-evaporable getter pump The trap is driven by means of an ac voltage ofa few hundred volts at a drive frequency of sim10 MHz A small oven withseveral mg of 171Yb isotope is heated to provide a low flux of Yb atomsfrom which single ions can be ionised by electron bombardment from ahot wire filament within the trap potential well Low voltage dc is applied

Exp Astron

to additional electrodes to position the ion precisely at trap potentialcentre in order to minimise the ionrsquos micromotion at the drive frequencyThis needs automatic periodic monitoring and correction especially in theperiod after re-loading an ion into the trap External magnetic field coilsin three orthogonal axes allow the nulling of external fields and settingof a fixed field of sim1 μT The trap and coils are surrounded by mu-metalshielding to minimise external field changes during the orbit A level oftemperature control of 1 K of the region surrounding the trap is requiredfor maintaining the blackbody-induced frequency shift uncertainty withtemperature at the 10minus17 level

bull a laser platform to provide frequency doubling of an extended cavity739 nm diode laser for Doppler cooling of the ion on the 2S12(F = 1) minus2 P12 (F = 0) dipole transition at 3695 nm Light from an amplified diodelaser device frequency-doubled in a periodically poled LiTaO3 crystal insingle-pass will be used Typical laser powers used for driving the coolingtransition below saturation are about 2 μW for a beam waist in the trapof about 50 μm By modulation of the injection current of the 739 nmdiode laser at 147 GHz a sideband is generated that excites the 2S12 (F =0) rarr2 P12 (F = 1) ldquorepumperrdquo transition in order to avoid optical pump-ing between the ground hyperfine states

bull The platform also houses the auxiliary lasers at 935 and 638 nm The935 nm diode provides repumping of the ion from the 2 D32(F = 1)metastable level after occasional branching decays to this level during thecooling sequence The 638 nm diode allows fast recovery of the ion fromthe very-long-lived 2F72 metastable state after very occasional collisionaldecay to that state Currently extended cavity lasers are forseen butdistributed-feedback lasers may be available in the near future

bull a high NA lens imaging system and photomultiplier detection system torecord the statistics of 369 nm fluorescence quantum jumps as a functionof clock laser frequency step

bull a fibre system to deliver the various cooling auxiliary and clock light fromsource to trap making use of achromatic doublets where necessary at thefibre-free space interface for launching into the trap

bull a monitoring and control processor which provides the primary coolingand clock laser pulse magnetic field and detection sequencing to observeand lock to the ion clock transition frequency The processor also monitorsfrequency and amplitude data necessary to determine normal laser andion operational conditions and initiate resetting and recovery algorithmswhere necessary and laser unit failure The actual magnetic field presenton the atomic volume is determined and actively stabilized so that aninaccuracy at the level of 1 times 10minus17 can be achieved

bull a redundancy level of at least two units for cooling clock and repumperlasers plus similar for frequency doubling crystals All redundancy unitsfor each wavelength will be fibre multiplexed as standard allowing redun-dant unit activation on determination of prior unit failure

Exp Astron

For an optimized quantum-projection-noise-limited performance of thesingle-ion clock Ramsey interrogation with a cycle time equal to the lifetimeof the metastable D32 level of 50 ms will be performed leading to an Allandeviation of about 27 times 10minus15 (τ s)minus12 This is sim45 times 10minus17 at 1 h satisfyingrequired stabilities for the science objectives Current laboratory results are astability of 1 times 10minus15 at 30 s averaging down to 3 times 10minus16 at 1000 s (PTBGermany) Also comparisons between two independent traps showed agree-ment at the level of several parts in 10minus16 over a series of eight measurementruns

The 171Yb+ ion clock offers a second option namely the 2S12ndash2 F72 octu-pole transition at 476 nm (see Table 1 and Fig 6) [48] It has an extremelylong upper state lifetime low electric quadrupole and second order Zeemansystematic shifts and therefore is a candidate for an even higher performanceclock if a clock laser system with lower instability than the above is providedApart from the clock laser it shares the laser system of the quadrupoleclock Thus with moderate additional resources the Yb+ ion apparatus couldprovide two optical clocks in a single package and specifically perform directfrequency comparisons between the quadrupole and the octupole transitionwith a large sensitivity Aα

The ion clock will operate under microprocessor control providing thecontrol algorithms for the integrated pulse sequencing for cooling repumpingclock interrogation magnetic field switching cooling fluorescence monitoringThe controller will also provide error flags for critical processes and systemreset and correction where necessary This will include ion re-loading andmicromotion reduction algorithms Automatic monitoring of laser power andspectral quality with signal re-optimisation or laser failure determination isrequired Redundant laser units will be pre-aligned and multiplexed into thefibre delivery to the trap so that redundant units can be readily activated in thecase of diode laser failure

423 Microwave-optical local oscillator (MOLO)

This oscillator provides two ultrastable frequency outputs one in the opticalregion the other in the microwave region the two frequencies being coher-ently related The MOLO is composed of three subunits the clock laser thereference cavity and the frequency comb see Fig 8

The clock laser subunit delivers the light for the excitation of the opticalclock transition It is generated by an extended-cavity diode laser at a wave-length of 871 nm Most of the output wave of this laser is sent to the ionclock The laser subunit containing two lasers for redundancy will require3 kg 3 l 15 W The frequency instability of the clock laser must be superiorto the minimum clock instability 27 times 10minus15 (τ s)minus12 for τ up to the timeconstant (sim10 s) of the servo system that locks the laser to the ionrsquos resonancesignal In order to achieve this the laser frequency is stabilized to the reference

Exp Astron

871 nm clock laser

EO

Detector and servo

AOM Femtosecond

frequency comb

To optical clock 10 GHz To FCDP

Microwave-optical local oscillator

ULE cavity

Fig 8 Microwave-optical local oscillator (MOLO) consisting of clock laser reference ULE cavityand femtosecond frequency comb (OFC) Radiofrequency inputs and outputs are not shown EOelectro-optic modulator ULE ultra-low expansion AOM frequency shifter

cavity where the PoundndashDreverndashHall method is used About 10 μW of thelaser output are used for this

The system will exhibit sim04 Hz (sim1 times 10minus15 relative) linewidth and afrequency drift lt01 Hzs The cavity consists of two high-reflectivity mirrorsoptically contacted to a cylinder of diameter 70 mm length 150 mm made ofULE (a glass exhibiting ultra low thermal expansion coefficient) with a cavityfinesse of about 300000 Residual vibrations of the satellite (arms lt 1 times 10minus6 g)should not be a limiting factor to the desired linewidth for safety marginan optimized cavity support can reduce the sensitivity to accelerations Foracoustic and thermal isolation the cavity resides inside an aluminium vacuumchamber equipped with a small (3 ls) ion-getter pump Two stages of polishedaluminium shields are implemented around the cavity which are activelytemperature stabilized so that the cavity is precisely kept at the zero crossingtemperature of the ULE thermal expansion coefficient Heat application andremoval is by thermoelectric elements between the shield and the vacuumchamber Additionally the entire vacuum chamber is temperature-stabilized

During transportation and launch the cavity will be rigidly fixed insidethe vacuum chamber by pressing on it from the sides by piezo-mechanicalactuators allowing accelerations of several g After bringing the satellite onthe orbit the actuators will be released

The optical setup for frequency stabilization will be a classic Pound-Drever-Hall scheme in reflection using an electro-optic modulator (EO) Couplingof the clock laser radiation to the cavity is via single-mode polarization main-taining optical fiber The fiber in- and out-couplers are mounted on miniaturepiezo-motor driven multiaxis translation stages Microprocessor control ofthe translation stages performs a laser-to-fiber incoupling stabilization andlaser-to-cavity mode-matching optimization (positioning inaccuracy lt10 μmalignment inaccuracy lt10 μrad)

The frequency stability of the clock laser will be limited by thermal 1f noisein the mirrors with an fractional Allan deviation slightly below 1 times 10minus15 from1 s to a few 10 s Resources including control electronics are 15 l 12 kg 8 W

Exp Astron

A femtosecond fiber laser based optical frequency comb generator (OFCFig 8) will be used for the clock frequency comparisons and to obtain aspectrally ultrapure microwave local oscillator output for the operation ofPHARAO for the microwave link and for the GNSS receiver The fem-tosecond laser is based on erbium-doped fiber pumped by fiber-coupled laserdiodes It emits 100 fs pulses at a repetition rate of 250 MHz with an averagepower of 100 mW The emission spectrum is centered around 1560 nm Spec-tral broadening to more than one octave is obtained via nonlinear interactionin a microstructure fibre so that the range from about 1050 to 2100 nm iscovered

Second harmonic generation of a portion of the broadened spectrum allowsobtaining the required light for the measurement of the clock laser frequencyat 871 nm The OFC will be referenced to the clock laser frequency (afterintroduction a controllable shift via the frequency shifter AOM) by makingthe carrier envelope offset frequency vanish so that the repetition rate fr isa high-order subharmonic of the optical clock reference frequency f1 = m1 frThe link between the optical and the microwave domain is then simply thedistribution of the repetition rate frequency or more conveniently a high(order 40) harmonic of the comb repetition rate 10 GHz

It has been demonstrated by several groups that frequency comparison andfrequency division with an OFC can be performed to very high accuracy andstability beyond the limitations imposed by the optical clocks themselves Forexample a fiber laser OFC with an instability (Allan deviation) of 6 times 10minus17

in 1 s has been demonstrated [51] OFCs based on fiber technology possessgood long-term reliability and high power efficiency The projection for a spacequalified OFC based on current industrial terrestrial technology is 16 kg mass20 l volume and 38 W power requirement

424 Frequency comparison and distribution package (FCDP)

This package distributes the ultra-stable 10 GHz reference frequency outputof the MOLO to the MWL performs frequency comparisons between MOLOand PHARAO and generates a variety of different reference signals for theindividual instruments including the 250 MHz reference clock to the OFCAll signals are coherent to the MOLO In the FCDP a USO is locked to theMOLO signal by a digital PLL A divider chain generates timing pulses (1 pps1000 pps etc) and keeps a continuous local time scale From the digital phasecomparison data the phase noise density spectrum between the USO and theMOLO is available by telemetry (TM) to assess the overall frequency purityin-flight

As the FCDP distributes the full performance reference from the MOLOdirectly to the MWL there is minimum impact of the FCDP on overall clockcomparison performance The USO-phase lock loop uses a similar designas for ACES-FCDP The digital PLL with linear phase detector allows theUSO to be slightly offset from its reference which is selectable by groundcommand This results in an accurate on-board time scale despite any offset

Exp Astron

of the MOLO reference frequency which can be synchronised to UTC Thefrequency synthesis chain re-uses modules developed for ACES-FCDP andtheir frequency inaccuracy Expected relative uncertainty (Allan deviation)of the frequency comparison between two signals provided to the FCDP is47 times 10minus17 at 1000 s 95 times 10minus18 at 10000 s

During non-availability of the MOLO ie during system start-up andcontingency conditions the USO runs as flywheel (ie it is unlocked withinstability lt15 times 10minus13 from 1 to 100 s) It will still provide useable butless accurate references to the instruments including the reference to OFCCommon view ground-to-ground time- and frequency comparison and supplyof the remaining instruments remain possible at the USO provided accuracy

Mass volume and power consumption are 8 kg 7 l 12 W Controldata rateis 10 kBday 50 kBday (FCDP alone)

425 Microwave link (MWL)

The microwave link is a two-way triple-frequency microwave link (Fig 9) Itis designed to perform accurate frequency comparison as well as time-transfer

Ku-Band Up-linkPower Tx 5 WCarrier 13475 GHzPN-Code 100 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

S-Band Down-linkPower Tx 1 WCarrier 2248 MHzPN-Code 1 MChips1pps 1 time marker sData 5 kBits

Ku-Band Down-linkPower Tx 1 WCarrier 1470333 GHzPN-Code 100 MChips1pps 1 time marker sData 5 kBits

EGE SC

S

Ku

KaKa-Band Up-linkPower Tx 5 WCarrier 345 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

Ka-Band Down-LinkPower Tx 1 WCarrier 32 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 5 kBits

GS Antenna diameterEGE terminal 1 mACES terminal 06 m

Fig 9 EGE microwave two-way link concept and characteristics See text for a description Thelink will be able to perform frequency comparisons with up to ten ground stations simultaneously(only one active comparison is shown) pps pulse per second Tx transmitter SC spacecraft

Exp Astron

between on-board and on-ground clocks Two-way operation allows for com-pensation of the effects due to spacecraft location and its motion relative toground Its two bi-directional links operate in Ka-band (32345 GHz) andKu-band (147135 GHz) The high carrier frequency of the up- and down-links allows for a noticeable reduction of the ionospheric delay A third down-link channel is in S-band (22 GHz) and is used to determine the ionospheretotal electron content MWL uses a combination of carrier phase and codephase data to measure with high precision the absolute ionospheric delayas well as ionospheric delay variations during a ground contact The useof three frequency bands allows to resolve third order ionospheric termsKey modulation parameters are Ka band Tone 250 MHz Ku band PN100 MChips S-band 1 MChips

The EGE-MWL concept is a considerably improved version of the ACES-MWL Microwave carrier generation is directly from the 10 GHz microwaveoutput available from the MOLO and having exceptionally low close-in phasenoise without intermediate frequency dividers or other complex synthesiserstages commonly used in RF transponders The high frequency carriers arekey to achieve the desired frequency comparison accuracy

The Ka-band carrier is generated by tripling the FCDP-generated signalThe Ku-band is subsequently generated by division by two using a verystable regenerative divider Finally the assigned radio frequency is obtainedby adding a low frequency signal using an image reject mixer A mixedanalog-digital receiver design will enhance its performance by use of a flexibledigital signal processing and digital control loops The number of simultaneousground contacts is increased from four (as in the ACES mission) to ten or moreusing the digital concept while the number of physical receiver channels perband is reduced from four to two one for operation to ground and one for in-flight calibration The concept allows adaptation of receiver loop parametersto optimise for the different portions of a highly elliptic orbit while reducingsystematic uncertainties due to high Doppler and Doppler rate Real-timedata exchange to ground terminals is by additional data modulation on all RFsignals The spacecraft carries two different small antennae for perigee and forapogee with gains of 8 dBi 20 dBi beamwidths 100 (full earth coverage) 14respectively power is 1 W for each

The MWL end-to-end performance between the attached clocks includingspace- and ground terminals is predicted to be a relative Allan deviation of1 times 10minus16 at 1000 s (perigee) and 1 times 10minus17 at 10000 s (apogee) Thesevalues are based on conservative implementation losses and other instrumentallosses assumed as 6 dB at apogee and 10 dB at perigee including additionallosses due to atmosphere An attitude knowledge of 1 is required

The MWL link will also support precise orbitographyndashcomplementary tothe GNSS receiver Generation of ranging data is simultaneous to two-waytime transfer and is available under all weather conditions Orbit determina-tion accuracy is expected to be better than 1 cm well matching EGE clockrequirements The value is expected based on demonstrated results fromother missions (METEOR ERS-2 TOPEXPOSEIDON) Laser ranging to

Exp Astron

the CCR array is used a few times during the mission to calibrate the MWLranging

In case of non-availability of the ultrapure signal from MOLO an additional10 GHz output is available from the FCDP to support ground-to-ground clockcomparison in common-view mode even after the nominal mission periodwhen the neither the clocks nor MOLO may be available any longer Basedon the PRARE experimentrsquos 12 years of operation this can be assumed forthe FCDPMWL combination as well The potential extended operational pe-riod makes the proposed configuration particularly attractive to metrologicallaboratories that rely on availability of long-lasting data in support of thegeneration of UTC

The resources of MWL are as follows mass 18 kg volume 18 l power55 W TC 2 kBday TM 120 MBday

426 Auxiliary units

The star camera assembly on the EGE spacecraft will provide real-timeattitude data primarily for the orientation of the spacecraft (with accuracy of1 in real-time) and the proper reduction of the GNSSSLR data

A GNSS receiver is used for the precise orbit determination time andfrequency transfer coarse positioning (lt1 m) for real time use by the Attitudeand Orbit Control as well as the back-up for time tagging of all payloaddata For the time and frequency transfer the receiver is nominally using theonboard frequency reference The receiver assembly includes three specialantennas By the year 2015 more than 100 GNSS satellites are expected tobe available for precise orbit determination permitting orbit inaccuracy of 1ndash2 cm RMS for altitudes up to approx 10000 km At apogee a combinationof GNSS signals and MWL ranging will be used with expected inaccuracyof 10 cm

5 Basic spacecraft specifications

The mission foresees a single spacecraft with a mission duration limited toapprox 3 years for cost reasons Attitude control along three axes withless than 1 error on a timescale of 1000 s is required The control can beperformed with cold gas thrusters magnetotorquers or reaction wheels Theselection will be done on the basis of the vibrations induced by these systems(desired level arms lt 10minus6 g)

Cold gas thrusters or monopropellant thrusters will be required to maintainthe orbit parameters during the mission by small orbit correction manoeuvresand will also guarantee a proper tuning of the orbit parameters after the releasefrom the launcher The ground segment (telemetry control and data transfer)could be via non-ESA stations user-operated to save costs

Further resources are indicated in Table 2

Exp Astron

The satellite bus design drivers are nadir-pointing microwave antennasthermal stability for the payload instruments no strong magnetic sources lowmicrovibrations harsh radiation environment due to crossing of the radiationbelts once per orbit power demand by science payload (with the possibility toreduce power demand by stand-by of some components during specific missionphases eg eclipse) mass limit due to required high altitude with low costlaunch

The satellite bus design can be based on established platforms such asFLEXBUS or PRIMA used on several past missions Their design philosophydiffers one having body-mounted solar arrays the other wings The latter canprovide larger amounts of power The vibration level induced by the panelswhen they are fixed and the duration of solar array drive periods per orbitwill have to be considered in the selection The FLEXBUS concept was usedfor a mission already requiring a low level of residual accelerations and theplatform is optimised for this parameter

For most of the satellite components one can consider using compo-nents and electronic units from navigation or communication satellites whichtypically cruise at altitudes between 15000 and 25000 km altitude LISAPathfinder which will cross the radiation belts during spiralling to theLagrange point will face similar problems concerning radiation as EGE as wellas XMM which is in a Molniya-type orbit with similar radiation characteristics

The instruments will therefore be located in the satellite structure such asto realise ionizing radiation shielding by the spacecraft structure and non-sensitive equipment to the maximum extent possible the need for additionalradiation shielding (including trapped proton radiation) for especially sensitiveparts of the instruments must be determined by ground tests It should bementioned that an orbit with higher perigee and less radiation dose is also anoption as it leads to a tolerable reduction of the science goals

The required plusmn3C temperature stability at the interface to the payloadenclosure is feasible The thermal control will employ the usual means (anenclosure with high thermal inertia heaters and a suitable radiator) Thelocation of the radiator will need a special study given the non-standard orbit

Magnetic field present during the mission can affect the properties of theonboard clock Using SPENVIS the magnetic field has been calculated alongthe spacecraft orbit the field intensity presents maxima around 02 G anda minimum value of the order of 10minus3 G As discussed above this is not alimitation to the instrument performance

6 Discussion and conclusion

The accuracy of the measurements depends crucially on the stability andaccuracy of the clocks While the performance of ground clocks (even thoseeventually deployed in new ground stations dedicated to EGE) will likelyimprove significantly from the present level the space clock design will beldquofrozenrdquo approx 10 years before data acquisition Thus their performance

Exp Astron

level will be significantly below the future ground station clock performanceAs specification we have chosen a performance level that has been demon-strated with laboratory breadboards recently We see no fundamental reasonthat this laboratory performance cannot also be achieved in a space ion opticalclock The experience with the PHARAO cold atom clock has shown thatan engineering model of a laser-based cold atom clock has a performancesimilar to that of a laboratory prototype The perturbation of the ion clock fre-quency due to magnetic fields along the orbit can be modelled and suppressedwith shielding and the residual effects measured and compensated for Themoderate temperature stability (plusmn 3C) to be provided in the experimentalbay can be achieved with thermal shields Accurate temperature controlof particular units is done by active stabilization of the units themselves

One important aspect is radiation which may affect all optical and trapcomponents as well as the lasers and reference cavity The radiation envi-ronment on the EGE orbit is significantly harsher than the near-earth ISSorbit on which ACES will operate The radiation effects must therefore becharacterized comprehensively with appropriate tests and then the radiationshield for the EGE instruments designed

In summary the Einstein Gravity Explorer mission (EGE) is a satellitemission aimed at a precise measurement of the properties of space-time usingatomic clocks By testing one of the most fundamental predictions of EinsteinrsquosTheory of General Relativity the gravitational redshift and thereby searchingfor quantum effects in gravity it explores one of the most important frontiers infundamental physics The EGE mission reaches beyond fundamental physicsproviding an important spin-off in geophysics Furthermore the technologicalaspects of the mission are ground-breaking for future time-keeping and ref-erence frame systems in space and thus are also expected to be of wide useto science and technology in general The payload of the mission has beenconceived as a compromise between cost and performance needing to fit thecost envelope of an ESA medium-class mission1

Acknowledgement We thank S Cesare for helpful information

References

1 Will CM The confrontation between general relativity and experiment Living Rev Relativ9 3 (2006)

2 Ni W-T Empirical foundations of the relativistic gravity Int J Mod Phys D 14 901 (2005)doi101142S0218271805007139

1Finally we point out that a spacecraft with a significantly reduced and thus lower-cost payload (noclocks only MOLO FCDP MWL) on a circular orbit of appropriate height could be used for along-duration earth science mission dedicated specifically to the measurement of the geopotential

Exp Astron

Fig 5 Left The PHARAO laser source engineering model with cover removed Dimensionsare 530 times 350 times 150 mm3 and the mass is 200 kg Ten polarization maintaining optical fibers(yellow) guide the laser beams to the cesium tube All diode lasers are mounted on Peltier coolersfor temperature regulation within 2 mK (courtesy EADS SODERN) Right The PHARAOengineering model cesium tube without the two external magnetic shields The volume is 990 times336 times 444 mm3 and the total mass is 44 kg

optical local oscillator PHARAO will provide a reduced frequency instabilityof 3 times 10minus14 (τs) minus12 This corresponds to 1 times 10minus15 at 1000 s and 3 times 10minus16

at 104 s

422 Ion optical clock

The 171Yb+ ion clock shown in Figs 6 and 7 provides an atomic frequencyreference based on a single 171Yb+ ion confined within an RF electric end-captrap (similar to a Paul trap) within an ultra high vacuum (UHV) chamber andlaser cooled to sim1 mK close to the Doppler cooling limit As a result the ionrsquosfirst order Doppler motion is completely removed and the ion experienceslittle collisional perturbation from its environment Under these conditionsthe narrow linewidth 2S12 (F = 0 mF = 0) minus 2 D32 (F = 2 mF = 0) quadrupole

Fig 6 Left Simplified level scheme of 171Yb+ showing relevant cooling (blue) auxiliary (red)quadrupole clock (green) and octupole clock (dark blue) transitions δνnat is the natural linewidthof the clock transition Middle close-up of Paul-type RF trap (PTB Braunschweig) the ion istrapped in the hole (diameter 14 mm) Right Complete ion trap with vacuum flange height 10 cm(NPL Teddington)

Exp Astron

935 nm D32-state repumper laser

638 nm F72-state repumper laser

739 nm extended cavity diode laser

AOM

PMT

Ion trap UHV chamber

vacuum pumps

Fibre-free space interface

Mu-metal shielding

EC

T

B

Magnetic field

Temperature

Ion fluorescence Processor

Clock laser frequency step and control

4355 nm

Frequengy doubling II

3695 nm cooling wave

Low-drift cavity

Low-drift cavity 171Yb+ 435 nm ion clock unit

876 nm clock laser wave from MOLO

Frequengy doubling

Fig 7 Simplified schematic of the single-ion optical clock unit AOM acousto-optic frequencyshifter PMT photomultiplier tube EC endcap T temperature stabilization system B magneticfield

clock transition at 4355 nm can be interrogated by spectrally narrow and stableclock laser light A diagram of the simplified energy level scheme for 171Yb+showing the relevant cooling auxiliary and clock transitions is given in Fig 6All these wavelengths are provided by diode lasers operated in fundamentalor frequency doubled mode The clock laser light is obtained from the 871 nmwave provided by the MOLO unit (see Section 423) after frequency-doublingin single pass by a KNbO3 nonlinear-optical crystal to 4355 nm The clocktransition spectral profile is observed by stepping the clock laser frequency bymeans of an acousto-optic frequency shifter (AOM in Fig 7) and recordingthe statistics of the quantum jumps in cooling laser induced fluorescence asa function of this frequency Under clock operation the AOM repeatedlysteps back and forward between the half-intensity points on the transitionprofile monitoring the signal imbalance between these points Any detectedimbalance is servo-corrected to zero by feedback to the AOM

In detail the optical clock architecture will comprise (Fig 7)

bull an RF end-cap trap for ionising and confining a single 171Yb+ ion withinan ultra-high vacuum chamber pumped by a small ion pump and non-evaporable getter pump The trap is driven by means of an ac voltage ofa few hundred volts at a drive frequency of sim10 MHz A small oven withseveral mg of 171Yb isotope is heated to provide a low flux of Yb atomsfrom which single ions can be ionised by electron bombardment from ahot wire filament within the trap potential well Low voltage dc is applied

Exp Astron

to additional electrodes to position the ion precisely at trap potentialcentre in order to minimise the ionrsquos micromotion at the drive frequencyThis needs automatic periodic monitoring and correction especially in theperiod after re-loading an ion into the trap External magnetic field coilsin three orthogonal axes allow the nulling of external fields and settingof a fixed field of sim1 μT The trap and coils are surrounded by mu-metalshielding to minimise external field changes during the orbit A level oftemperature control of 1 K of the region surrounding the trap is requiredfor maintaining the blackbody-induced frequency shift uncertainty withtemperature at the 10minus17 level

bull a laser platform to provide frequency doubling of an extended cavity739 nm diode laser for Doppler cooling of the ion on the 2S12(F = 1) minus2 P12 (F = 0) dipole transition at 3695 nm Light from an amplified diodelaser device frequency-doubled in a periodically poled LiTaO3 crystal insingle-pass will be used Typical laser powers used for driving the coolingtransition below saturation are about 2 μW for a beam waist in the trapof about 50 μm By modulation of the injection current of the 739 nmdiode laser at 147 GHz a sideband is generated that excites the 2S12 (F =0) rarr2 P12 (F = 1) ldquorepumperrdquo transition in order to avoid optical pump-ing between the ground hyperfine states

bull The platform also houses the auxiliary lasers at 935 and 638 nm The935 nm diode provides repumping of the ion from the 2 D32(F = 1)metastable level after occasional branching decays to this level during thecooling sequence The 638 nm diode allows fast recovery of the ion fromthe very-long-lived 2F72 metastable state after very occasional collisionaldecay to that state Currently extended cavity lasers are forseen butdistributed-feedback lasers may be available in the near future

bull a high NA lens imaging system and photomultiplier detection system torecord the statistics of 369 nm fluorescence quantum jumps as a functionof clock laser frequency step

bull a fibre system to deliver the various cooling auxiliary and clock light fromsource to trap making use of achromatic doublets where necessary at thefibre-free space interface for launching into the trap

bull a monitoring and control processor which provides the primary coolingand clock laser pulse magnetic field and detection sequencing to observeand lock to the ion clock transition frequency The processor also monitorsfrequency and amplitude data necessary to determine normal laser andion operational conditions and initiate resetting and recovery algorithmswhere necessary and laser unit failure The actual magnetic field presenton the atomic volume is determined and actively stabilized so that aninaccuracy at the level of 1 times 10minus17 can be achieved

bull a redundancy level of at least two units for cooling clock and repumperlasers plus similar for frequency doubling crystals All redundancy unitsfor each wavelength will be fibre multiplexed as standard allowing redun-dant unit activation on determination of prior unit failure

Exp Astron

For an optimized quantum-projection-noise-limited performance of thesingle-ion clock Ramsey interrogation with a cycle time equal to the lifetimeof the metastable D32 level of 50 ms will be performed leading to an Allandeviation of about 27 times 10minus15 (τ s)minus12 This is sim45 times 10minus17 at 1 h satisfyingrequired stabilities for the science objectives Current laboratory results are astability of 1 times 10minus15 at 30 s averaging down to 3 times 10minus16 at 1000 s (PTBGermany) Also comparisons between two independent traps showed agree-ment at the level of several parts in 10minus16 over a series of eight measurementruns

The 171Yb+ ion clock offers a second option namely the 2S12ndash2 F72 octu-pole transition at 476 nm (see Table 1 and Fig 6) [48] It has an extremelylong upper state lifetime low electric quadrupole and second order Zeemansystematic shifts and therefore is a candidate for an even higher performanceclock if a clock laser system with lower instability than the above is providedApart from the clock laser it shares the laser system of the quadrupoleclock Thus with moderate additional resources the Yb+ ion apparatus couldprovide two optical clocks in a single package and specifically perform directfrequency comparisons between the quadrupole and the octupole transitionwith a large sensitivity Aα

The ion clock will operate under microprocessor control providing thecontrol algorithms for the integrated pulse sequencing for cooling repumpingclock interrogation magnetic field switching cooling fluorescence monitoringThe controller will also provide error flags for critical processes and systemreset and correction where necessary This will include ion re-loading andmicromotion reduction algorithms Automatic monitoring of laser power andspectral quality with signal re-optimisation or laser failure determination isrequired Redundant laser units will be pre-aligned and multiplexed into thefibre delivery to the trap so that redundant units can be readily activated in thecase of diode laser failure

423 Microwave-optical local oscillator (MOLO)

This oscillator provides two ultrastable frequency outputs one in the opticalregion the other in the microwave region the two frequencies being coher-ently related The MOLO is composed of three subunits the clock laser thereference cavity and the frequency comb see Fig 8

The clock laser subunit delivers the light for the excitation of the opticalclock transition It is generated by an extended-cavity diode laser at a wave-length of 871 nm Most of the output wave of this laser is sent to the ionclock The laser subunit containing two lasers for redundancy will require3 kg 3 l 15 W The frequency instability of the clock laser must be superiorto the minimum clock instability 27 times 10minus15 (τ s)minus12 for τ up to the timeconstant (sim10 s) of the servo system that locks the laser to the ionrsquos resonancesignal In order to achieve this the laser frequency is stabilized to the reference

Exp Astron

871 nm clock laser

EO

Detector and servo

AOM Femtosecond

frequency comb

To optical clock 10 GHz To FCDP

Microwave-optical local oscillator

ULE cavity

Fig 8 Microwave-optical local oscillator (MOLO) consisting of clock laser reference ULE cavityand femtosecond frequency comb (OFC) Radiofrequency inputs and outputs are not shown EOelectro-optic modulator ULE ultra-low expansion AOM frequency shifter

cavity where the PoundndashDreverndashHall method is used About 10 μW of thelaser output are used for this

The system will exhibit sim04 Hz (sim1 times 10minus15 relative) linewidth and afrequency drift lt01 Hzs The cavity consists of two high-reflectivity mirrorsoptically contacted to a cylinder of diameter 70 mm length 150 mm made ofULE (a glass exhibiting ultra low thermal expansion coefficient) with a cavityfinesse of about 300000 Residual vibrations of the satellite (arms lt 1 times 10minus6 g)should not be a limiting factor to the desired linewidth for safety marginan optimized cavity support can reduce the sensitivity to accelerations Foracoustic and thermal isolation the cavity resides inside an aluminium vacuumchamber equipped with a small (3 ls) ion-getter pump Two stages of polishedaluminium shields are implemented around the cavity which are activelytemperature stabilized so that the cavity is precisely kept at the zero crossingtemperature of the ULE thermal expansion coefficient Heat application andremoval is by thermoelectric elements between the shield and the vacuumchamber Additionally the entire vacuum chamber is temperature-stabilized

During transportation and launch the cavity will be rigidly fixed insidethe vacuum chamber by pressing on it from the sides by piezo-mechanicalactuators allowing accelerations of several g After bringing the satellite onthe orbit the actuators will be released

The optical setup for frequency stabilization will be a classic Pound-Drever-Hall scheme in reflection using an electro-optic modulator (EO) Couplingof the clock laser radiation to the cavity is via single-mode polarization main-taining optical fiber The fiber in- and out-couplers are mounted on miniaturepiezo-motor driven multiaxis translation stages Microprocessor control ofthe translation stages performs a laser-to-fiber incoupling stabilization andlaser-to-cavity mode-matching optimization (positioning inaccuracy lt10 μmalignment inaccuracy lt10 μrad)

The frequency stability of the clock laser will be limited by thermal 1f noisein the mirrors with an fractional Allan deviation slightly below 1 times 10minus15 from1 s to a few 10 s Resources including control electronics are 15 l 12 kg 8 W

Exp Astron

A femtosecond fiber laser based optical frequency comb generator (OFCFig 8) will be used for the clock frequency comparisons and to obtain aspectrally ultrapure microwave local oscillator output for the operation ofPHARAO for the microwave link and for the GNSS receiver The fem-tosecond laser is based on erbium-doped fiber pumped by fiber-coupled laserdiodes It emits 100 fs pulses at a repetition rate of 250 MHz with an averagepower of 100 mW The emission spectrum is centered around 1560 nm Spec-tral broadening to more than one octave is obtained via nonlinear interactionin a microstructure fibre so that the range from about 1050 to 2100 nm iscovered

Second harmonic generation of a portion of the broadened spectrum allowsobtaining the required light for the measurement of the clock laser frequencyat 871 nm The OFC will be referenced to the clock laser frequency (afterintroduction a controllable shift via the frequency shifter AOM) by makingthe carrier envelope offset frequency vanish so that the repetition rate fr isa high-order subharmonic of the optical clock reference frequency f1 = m1 frThe link between the optical and the microwave domain is then simply thedistribution of the repetition rate frequency or more conveniently a high(order 40) harmonic of the comb repetition rate 10 GHz

It has been demonstrated by several groups that frequency comparison andfrequency division with an OFC can be performed to very high accuracy andstability beyond the limitations imposed by the optical clocks themselves Forexample a fiber laser OFC with an instability (Allan deviation) of 6 times 10minus17

in 1 s has been demonstrated [51] OFCs based on fiber technology possessgood long-term reliability and high power efficiency The projection for a spacequalified OFC based on current industrial terrestrial technology is 16 kg mass20 l volume and 38 W power requirement

424 Frequency comparison and distribution package (FCDP)

This package distributes the ultra-stable 10 GHz reference frequency outputof the MOLO to the MWL performs frequency comparisons between MOLOand PHARAO and generates a variety of different reference signals for theindividual instruments including the 250 MHz reference clock to the OFCAll signals are coherent to the MOLO In the FCDP a USO is locked to theMOLO signal by a digital PLL A divider chain generates timing pulses (1 pps1000 pps etc) and keeps a continuous local time scale From the digital phasecomparison data the phase noise density spectrum between the USO and theMOLO is available by telemetry (TM) to assess the overall frequency purityin-flight

As the FCDP distributes the full performance reference from the MOLOdirectly to the MWL there is minimum impact of the FCDP on overall clockcomparison performance The USO-phase lock loop uses a similar designas for ACES-FCDP The digital PLL with linear phase detector allows theUSO to be slightly offset from its reference which is selectable by groundcommand This results in an accurate on-board time scale despite any offset

Exp Astron

of the MOLO reference frequency which can be synchronised to UTC Thefrequency synthesis chain re-uses modules developed for ACES-FCDP andtheir frequency inaccuracy Expected relative uncertainty (Allan deviation)of the frequency comparison between two signals provided to the FCDP is47 times 10minus17 at 1000 s 95 times 10minus18 at 10000 s

During non-availability of the MOLO ie during system start-up andcontingency conditions the USO runs as flywheel (ie it is unlocked withinstability lt15 times 10minus13 from 1 to 100 s) It will still provide useable butless accurate references to the instruments including the reference to OFCCommon view ground-to-ground time- and frequency comparison and supplyof the remaining instruments remain possible at the USO provided accuracy

Mass volume and power consumption are 8 kg 7 l 12 W Controldata rateis 10 kBday 50 kBday (FCDP alone)

425 Microwave link (MWL)

The microwave link is a two-way triple-frequency microwave link (Fig 9) Itis designed to perform accurate frequency comparison as well as time-transfer

Ku-Band Up-linkPower Tx 5 WCarrier 13475 GHzPN-Code 100 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

S-Band Down-linkPower Tx 1 WCarrier 2248 MHzPN-Code 1 MChips1pps 1 time marker sData 5 kBits

Ku-Band Down-linkPower Tx 1 WCarrier 1470333 GHzPN-Code 100 MChips1pps 1 time marker sData 5 kBits

EGE SC

S

Ku

KaKa-Band Up-linkPower Tx 5 WCarrier 345 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

Ka-Band Down-LinkPower Tx 1 WCarrier 32 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 5 kBits

GS Antenna diameterEGE terminal 1 mACES terminal 06 m

Fig 9 EGE microwave two-way link concept and characteristics See text for a description Thelink will be able to perform frequency comparisons with up to ten ground stations simultaneously(only one active comparison is shown) pps pulse per second Tx transmitter SC spacecraft

Exp Astron

between on-board and on-ground clocks Two-way operation allows for com-pensation of the effects due to spacecraft location and its motion relative toground Its two bi-directional links operate in Ka-band (32345 GHz) andKu-band (147135 GHz) The high carrier frequency of the up- and down-links allows for a noticeable reduction of the ionospheric delay A third down-link channel is in S-band (22 GHz) and is used to determine the ionospheretotal electron content MWL uses a combination of carrier phase and codephase data to measure with high precision the absolute ionospheric delayas well as ionospheric delay variations during a ground contact The useof three frequency bands allows to resolve third order ionospheric termsKey modulation parameters are Ka band Tone 250 MHz Ku band PN100 MChips S-band 1 MChips

The EGE-MWL concept is a considerably improved version of the ACES-MWL Microwave carrier generation is directly from the 10 GHz microwaveoutput available from the MOLO and having exceptionally low close-in phasenoise without intermediate frequency dividers or other complex synthesiserstages commonly used in RF transponders The high frequency carriers arekey to achieve the desired frequency comparison accuracy

The Ka-band carrier is generated by tripling the FCDP-generated signalThe Ku-band is subsequently generated by division by two using a verystable regenerative divider Finally the assigned radio frequency is obtainedby adding a low frequency signal using an image reject mixer A mixedanalog-digital receiver design will enhance its performance by use of a flexibledigital signal processing and digital control loops The number of simultaneousground contacts is increased from four (as in the ACES mission) to ten or moreusing the digital concept while the number of physical receiver channels perband is reduced from four to two one for operation to ground and one for in-flight calibration The concept allows adaptation of receiver loop parametersto optimise for the different portions of a highly elliptic orbit while reducingsystematic uncertainties due to high Doppler and Doppler rate Real-timedata exchange to ground terminals is by additional data modulation on all RFsignals The spacecraft carries two different small antennae for perigee and forapogee with gains of 8 dBi 20 dBi beamwidths 100 (full earth coverage) 14respectively power is 1 W for each

The MWL end-to-end performance between the attached clocks includingspace- and ground terminals is predicted to be a relative Allan deviation of1 times 10minus16 at 1000 s (perigee) and 1 times 10minus17 at 10000 s (apogee) Thesevalues are based on conservative implementation losses and other instrumentallosses assumed as 6 dB at apogee and 10 dB at perigee including additionallosses due to atmosphere An attitude knowledge of 1 is required

The MWL link will also support precise orbitographyndashcomplementary tothe GNSS receiver Generation of ranging data is simultaneous to two-waytime transfer and is available under all weather conditions Orbit determina-tion accuracy is expected to be better than 1 cm well matching EGE clockrequirements The value is expected based on demonstrated results fromother missions (METEOR ERS-2 TOPEXPOSEIDON) Laser ranging to

Exp Astron

the CCR array is used a few times during the mission to calibrate the MWLranging

In case of non-availability of the ultrapure signal from MOLO an additional10 GHz output is available from the FCDP to support ground-to-ground clockcomparison in common-view mode even after the nominal mission periodwhen the neither the clocks nor MOLO may be available any longer Basedon the PRARE experimentrsquos 12 years of operation this can be assumed forthe FCDPMWL combination as well The potential extended operational pe-riod makes the proposed configuration particularly attractive to metrologicallaboratories that rely on availability of long-lasting data in support of thegeneration of UTC

The resources of MWL are as follows mass 18 kg volume 18 l power55 W TC 2 kBday TM 120 MBday

426 Auxiliary units

The star camera assembly on the EGE spacecraft will provide real-timeattitude data primarily for the orientation of the spacecraft (with accuracy of1 in real-time) and the proper reduction of the GNSSSLR data

A GNSS receiver is used for the precise orbit determination time andfrequency transfer coarse positioning (lt1 m) for real time use by the Attitudeand Orbit Control as well as the back-up for time tagging of all payloaddata For the time and frequency transfer the receiver is nominally using theonboard frequency reference The receiver assembly includes three specialantennas By the year 2015 more than 100 GNSS satellites are expected tobe available for precise orbit determination permitting orbit inaccuracy of 1ndash2 cm RMS for altitudes up to approx 10000 km At apogee a combinationof GNSS signals and MWL ranging will be used with expected inaccuracyof 10 cm

5 Basic spacecraft specifications

The mission foresees a single spacecraft with a mission duration limited toapprox 3 years for cost reasons Attitude control along three axes withless than 1 error on a timescale of 1000 s is required The control can beperformed with cold gas thrusters magnetotorquers or reaction wheels Theselection will be done on the basis of the vibrations induced by these systems(desired level arms lt 10minus6 g)

Cold gas thrusters or monopropellant thrusters will be required to maintainthe orbit parameters during the mission by small orbit correction manoeuvresand will also guarantee a proper tuning of the orbit parameters after the releasefrom the launcher The ground segment (telemetry control and data transfer)could be via non-ESA stations user-operated to save costs

Further resources are indicated in Table 2

Exp Astron

The satellite bus design drivers are nadir-pointing microwave antennasthermal stability for the payload instruments no strong magnetic sources lowmicrovibrations harsh radiation environment due to crossing of the radiationbelts once per orbit power demand by science payload (with the possibility toreduce power demand by stand-by of some components during specific missionphases eg eclipse) mass limit due to required high altitude with low costlaunch

The satellite bus design can be based on established platforms such asFLEXBUS or PRIMA used on several past missions Their design philosophydiffers one having body-mounted solar arrays the other wings The latter canprovide larger amounts of power The vibration level induced by the panelswhen they are fixed and the duration of solar array drive periods per orbitwill have to be considered in the selection The FLEXBUS concept was usedfor a mission already requiring a low level of residual accelerations and theplatform is optimised for this parameter

For most of the satellite components one can consider using compo-nents and electronic units from navigation or communication satellites whichtypically cruise at altitudes between 15000 and 25000 km altitude LISAPathfinder which will cross the radiation belts during spiralling to theLagrange point will face similar problems concerning radiation as EGE as wellas XMM which is in a Molniya-type orbit with similar radiation characteristics

The instruments will therefore be located in the satellite structure such asto realise ionizing radiation shielding by the spacecraft structure and non-sensitive equipment to the maximum extent possible the need for additionalradiation shielding (including trapped proton radiation) for especially sensitiveparts of the instruments must be determined by ground tests It should bementioned that an orbit with higher perigee and less radiation dose is also anoption as it leads to a tolerable reduction of the science goals

The required plusmn3C temperature stability at the interface to the payloadenclosure is feasible The thermal control will employ the usual means (anenclosure with high thermal inertia heaters and a suitable radiator) Thelocation of the radiator will need a special study given the non-standard orbit

Magnetic field present during the mission can affect the properties of theonboard clock Using SPENVIS the magnetic field has been calculated alongthe spacecraft orbit the field intensity presents maxima around 02 G anda minimum value of the order of 10minus3 G As discussed above this is not alimitation to the instrument performance

6 Discussion and conclusion

The accuracy of the measurements depends crucially on the stability andaccuracy of the clocks While the performance of ground clocks (even thoseeventually deployed in new ground stations dedicated to EGE) will likelyimprove significantly from the present level the space clock design will beldquofrozenrdquo approx 10 years before data acquisition Thus their performance

Exp Astron

level will be significantly below the future ground station clock performanceAs specification we have chosen a performance level that has been demon-strated with laboratory breadboards recently We see no fundamental reasonthat this laboratory performance cannot also be achieved in a space ion opticalclock The experience with the PHARAO cold atom clock has shown thatan engineering model of a laser-based cold atom clock has a performancesimilar to that of a laboratory prototype The perturbation of the ion clock fre-quency due to magnetic fields along the orbit can be modelled and suppressedwith shielding and the residual effects measured and compensated for Themoderate temperature stability (plusmn 3C) to be provided in the experimentalbay can be achieved with thermal shields Accurate temperature controlof particular units is done by active stabilization of the units themselves

One important aspect is radiation which may affect all optical and trapcomponents as well as the lasers and reference cavity The radiation envi-ronment on the EGE orbit is significantly harsher than the near-earth ISSorbit on which ACES will operate The radiation effects must therefore becharacterized comprehensively with appropriate tests and then the radiationshield for the EGE instruments designed

In summary the Einstein Gravity Explorer mission (EGE) is a satellitemission aimed at a precise measurement of the properties of space-time usingatomic clocks By testing one of the most fundamental predictions of EinsteinrsquosTheory of General Relativity the gravitational redshift and thereby searchingfor quantum effects in gravity it explores one of the most important frontiers infundamental physics The EGE mission reaches beyond fundamental physicsproviding an important spin-off in geophysics Furthermore the technologicalaspects of the mission are ground-breaking for future time-keeping and ref-erence frame systems in space and thus are also expected to be of wide useto science and technology in general The payload of the mission has beenconceived as a compromise between cost and performance needing to fit thecost envelope of an ESA medium-class mission1

Acknowledgement We thank S Cesare for helpful information

References

1 Will CM The confrontation between general relativity and experiment Living Rev Relativ9 3 (2006)

2 Ni W-T Empirical foundations of the relativistic gravity Int J Mod Phys D 14 901 (2005)doi101142S0218271805007139

1Finally we point out that a spacecraft with a significantly reduced and thus lower-cost payload (noclocks only MOLO FCDP MWL) on a circular orbit of appropriate height could be used for along-duration earth science mission dedicated specifically to the measurement of the geopotential

Exp Astron

935 nm D32-state repumper laser

638 nm F72-state repumper laser

739 nm extended cavity diode laser

AOM

PMT

Ion trap UHV chamber

vacuum pumps

Fibre-free space interface

Mu-metal shielding

EC

T

B

Magnetic field

Temperature

Ion fluorescence Processor

Clock laser frequency step and control

4355 nm

Frequengy doubling II

3695 nm cooling wave

Low-drift cavity

Low-drift cavity 171Yb+ 435 nm ion clock unit

876 nm clock laser wave from MOLO

Frequengy doubling

Fig 7 Simplified schematic of the single-ion optical clock unit AOM acousto-optic frequencyshifter PMT photomultiplier tube EC endcap T temperature stabilization system B magneticfield

clock transition at 4355 nm can be interrogated by spectrally narrow and stableclock laser light A diagram of the simplified energy level scheme for 171Yb+showing the relevant cooling auxiliary and clock transitions is given in Fig 6All these wavelengths are provided by diode lasers operated in fundamentalor frequency doubled mode The clock laser light is obtained from the 871 nmwave provided by the MOLO unit (see Section 423) after frequency-doublingin single pass by a KNbO3 nonlinear-optical crystal to 4355 nm The clocktransition spectral profile is observed by stepping the clock laser frequency bymeans of an acousto-optic frequency shifter (AOM in Fig 7) and recordingthe statistics of the quantum jumps in cooling laser induced fluorescence asa function of this frequency Under clock operation the AOM repeatedlysteps back and forward between the half-intensity points on the transitionprofile monitoring the signal imbalance between these points Any detectedimbalance is servo-corrected to zero by feedback to the AOM

In detail the optical clock architecture will comprise (Fig 7)

bull an RF end-cap trap for ionising and confining a single 171Yb+ ion withinan ultra-high vacuum chamber pumped by a small ion pump and non-evaporable getter pump The trap is driven by means of an ac voltage ofa few hundred volts at a drive frequency of sim10 MHz A small oven withseveral mg of 171Yb isotope is heated to provide a low flux of Yb atomsfrom which single ions can be ionised by electron bombardment from ahot wire filament within the trap potential well Low voltage dc is applied

Exp Astron

to additional electrodes to position the ion precisely at trap potentialcentre in order to minimise the ionrsquos micromotion at the drive frequencyThis needs automatic periodic monitoring and correction especially in theperiod after re-loading an ion into the trap External magnetic field coilsin three orthogonal axes allow the nulling of external fields and settingof a fixed field of sim1 μT The trap and coils are surrounded by mu-metalshielding to minimise external field changes during the orbit A level oftemperature control of 1 K of the region surrounding the trap is requiredfor maintaining the blackbody-induced frequency shift uncertainty withtemperature at the 10minus17 level

bull a laser platform to provide frequency doubling of an extended cavity739 nm diode laser for Doppler cooling of the ion on the 2S12(F = 1) minus2 P12 (F = 0) dipole transition at 3695 nm Light from an amplified diodelaser device frequency-doubled in a periodically poled LiTaO3 crystal insingle-pass will be used Typical laser powers used for driving the coolingtransition below saturation are about 2 μW for a beam waist in the trapof about 50 μm By modulation of the injection current of the 739 nmdiode laser at 147 GHz a sideband is generated that excites the 2S12 (F =0) rarr2 P12 (F = 1) ldquorepumperrdquo transition in order to avoid optical pump-ing between the ground hyperfine states

bull The platform also houses the auxiliary lasers at 935 and 638 nm The935 nm diode provides repumping of the ion from the 2 D32(F = 1)metastable level after occasional branching decays to this level during thecooling sequence The 638 nm diode allows fast recovery of the ion fromthe very-long-lived 2F72 metastable state after very occasional collisionaldecay to that state Currently extended cavity lasers are forseen butdistributed-feedback lasers may be available in the near future

bull a high NA lens imaging system and photomultiplier detection system torecord the statistics of 369 nm fluorescence quantum jumps as a functionof clock laser frequency step

bull a fibre system to deliver the various cooling auxiliary and clock light fromsource to trap making use of achromatic doublets where necessary at thefibre-free space interface for launching into the trap

bull a monitoring and control processor which provides the primary coolingand clock laser pulse magnetic field and detection sequencing to observeand lock to the ion clock transition frequency The processor also monitorsfrequency and amplitude data necessary to determine normal laser andion operational conditions and initiate resetting and recovery algorithmswhere necessary and laser unit failure The actual magnetic field presenton the atomic volume is determined and actively stabilized so that aninaccuracy at the level of 1 times 10minus17 can be achieved

bull a redundancy level of at least two units for cooling clock and repumperlasers plus similar for frequency doubling crystals All redundancy unitsfor each wavelength will be fibre multiplexed as standard allowing redun-dant unit activation on determination of prior unit failure

Exp Astron

For an optimized quantum-projection-noise-limited performance of thesingle-ion clock Ramsey interrogation with a cycle time equal to the lifetimeof the metastable D32 level of 50 ms will be performed leading to an Allandeviation of about 27 times 10minus15 (τ s)minus12 This is sim45 times 10minus17 at 1 h satisfyingrequired stabilities for the science objectives Current laboratory results are astability of 1 times 10minus15 at 30 s averaging down to 3 times 10minus16 at 1000 s (PTBGermany) Also comparisons between two independent traps showed agree-ment at the level of several parts in 10minus16 over a series of eight measurementruns

The 171Yb+ ion clock offers a second option namely the 2S12ndash2 F72 octu-pole transition at 476 nm (see Table 1 and Fig 6) [48] It has an extremelylong upper state lifetime low electric quadrupole and second order Zeemansystematic shifts and therefore is a candidate for an even higher performanceclock if a clock laser system with lower instability than the above is providedApart from the clock laser it shares the laser system of the quadrupoleclock Thus with moderate additional resources the Yb+ ion apparatus couldprovide two optical clocks in a single package and specifically perform directfrequency comparisons between the quadrupole and the octupole transitionwith a large sensitivity Aα

The ion clock will operate under microprocessor control providing thecontrol algorithms for the integrated pulse sequencing for cooling repumpingclock interrogation magnetic field switching cooling fluorescence monitoringThe controller will also provide error flags for critical processes and systemreset and correction where necessary This will include ion re-loading andmicromotion reduction algorithms Automatic monitoring of laser power andspectral quality with signal re-optimisation or laser failure determination isrequired Redundant laser units will be pre-aligned and multiplexed into thefibre delivery to the trap so that redundant units can be readily activated in thecase of diode laser failure

423 Microwave-optical local oscillator (MOLO)

This oscillator provides two ultrastable frequency outputs one in the opticalregion the other in the microwave region the two frequencies being coher-ently related The MOLO is composed of three subunits the clock laser thereference cavity and the frequency comb see Fig 8

The clock laser subunit delivers the light for the excitation of the opticalclock transition It is generated by an extended-cavity diode laser at a wave-length of 871 nm Most of the output wave of this laser is sent to the ionclock The laser subunit containing two lasers for redundancy will require3 kg 3 l 15 W The frequency instability of the clock laser must be superiorto the minimum clock instability 27 times 10minus15 (τ s)minus12 for τ up to the timeconstant (sim10 s) of the servo system that locks the laser to the ionrsquos resonancesignal In order to achieve this the laser frequency is stabilized to the reference

Exp Astron

871 nm clock laser

EO

Detector and servo

AOM Femtosecond

frequency comb

To optical clock 10 GHz To FCDP

Microwave-optical local oscillator

ULE cavity

Fig 8 Microwave-optical local oscillator (MOLO) consisting of clock laser reference ULE cavityand femtosecond frequency comb (OFC) Radiofrequency inputs and outputs are not shown EOelectro-optic modulator ULE ultra-low expansion AOM frequency shifter

cavity where the PoundndashDreverndashHall method is used About 10 μW of thelaser output are used for this

The system will exhibit sim04 Hz (sim1 times 10minus15 relative) linewidth and afrequency drift lt01 Hzs The cavity consists of two high-reflectivity mirrorsoptically contacted to a cylinder of diameter 70 mm length 150 mm made ofULE (a glass exhibiting ultra low thermal expansion coefficient) with a cavityfinesse of about 300000 Residual vibrations of the satellite (arms lt 1 times 10minus6 g)should not be a limiting factor to the desired linewidth for safety marginan optimized cavity support can reduce the sensitivity to accelerations Foracoustic and thermal isolation the cavity resides inside an aluminium vacuumchamber equipped with a small (3 ls) ion-getter pump Two stages of polishedaluminium shields are implemented around the cavity which are activelytemperature stabilized so that the cavity is precisely kept at the zero crossingtemperature of the ULE thermal expansion coefficient Heat application andremoval is by thermoelectric elements between the shield and the vacuumchamber Additionally the entire vacuum chamber is temperature-stabilized

During transportation and launch the cavity will be rigidly fixed insidethe vacuum chamber by pressing on it from the sides by piezo-mechanicalactuators allowing accelerations of several g After bringing the satellite onthe orbit the actuators will be released

The optical setup for frequency stabilization will be a classic Pound-Drever-Hall scheme in reflection using an electro-optic modulator (EO) Couplingof the clock laser radiation to the cavity is via single-mode polarization main-taining optical fiber The fiber in- and out-couplers are mounted on miniaturepiezo-motor driven multiaxis translation stages Microprocessor control ofthe translation stages performs a laser-to-fiber incoupling stabilization andlaser-to-cavity mode-matching optimization (positioning inaccuracy lt10 μmalignment inaccuracy lt10 μrad)

The frequency stability of the clock laser will be limited by thermal 1f noisein the mirrors with an fractional Allan deviation slightly below 1 times 10minus15 from1 s to a few 10 s Resources including control electronics are 15 l 12 kg 8 W

Exp Astron

A femtosecond fiber laser based optical frequency comb generator (OFCFig 8) will be used for the clock frequency comparisons and to obtain aspectrally ultrapure microwave local oscillator output for the operation ofPHARAO for the microwave link and for the GNSS receiver The fem-tosecond laser is based on erbium-doped fiber pumped by fiber-coupled laserdiodes It emits 100 fs pulses at a repetition rate of 250 MHz with an averagepower of 100 mW The emission spectrum is centered around 1560 nm Spec-tral broadening to more than one octave is obtained via nonlinear interactionin a microstructure fibre so that the range from about 1050 to 2100 nm iscovered

Second harmonic generation of a portion of the broadened spectrum allowsobtaining the required light for the measurement of the clock laser frequencyat 871 nm The OFC will be referenced to the clock laser frequency (afterintroduction a controllable shift via the frequency shifter AOM) by makingthe carrier envelope offset frequency vanish so that the repetition rate fr isa high-order subharmonic of the optical clock reference frequency f1 = m1 frThe link between the optical and the microwave domain is then simply thedistribution of the repetition rate frequency or more conveniently a high(order 40) harmonic of the comb repetition rate 10 GHz

It has been demonstrated by several groups that frequency comparison andfrequency division with an OFC can be performed to very high accuracy andstability beyond the limitations imposed by the optical clocks themselves Forexample a fiber laser OFC with an instability (Allan deviation) of 6 times 10minus17

in 1 s has been demonstrated [51] OFCs based on fiber technology possessgood long-term reliability and high power efficiency The projection for a spacequalified OFC based on current industrial terrestrial technology is 16 kg mass20 l volume and 38 W power requirement

424 Frequency comparison and distribution package (FCDP)

This package distributes the ultra-stable 10 GHz reference frequency outputof the MOLO to the MWL performs frequency comparisons between MOLOand PHARAO and generates a variety of different reference signals for theindividual instruments including the 250 MHz reference clock to the OFCAll signals are coherent to the MOLO In the FCDP a USO is locked to theMOLO signal by a digital PLL A divider chain generates timing pulses (1 pps1000 pps etc) and keeps a continuous local time scale From the digital phasecomparison data the phase noise density spectrum between the USO and theMOLO is available by telemetry (TM) to assess the overall frequency purityin-flight

As the FCDP distributes the full performance reference from the MOLOdirectly to the MWL there is minimum impact of the FCDP on overall clockcomparison performance The USO-phase lock loop uses a similar designas for ACES-FCDP The digital PLL with linear phase detector allows theUSO to be slightly offset from its reference which is selectable by groundcommand This results in an accurate on-board time scale despite any offset

Exp Astron

of the MOLO reference frequency which can be synchronised to UTC Thefrequency synthesis chain re-uses modules developed for ACES-FCDP andtheir frequency inaccuracy Expected relative uncertainty (Allan deviation)of the frequency comparison between two signals provided to the FCDP is47 times 10minus17 at 1000 s 95 times 10minus18 at 10000 s

During non-availability of the MOLO ie during system start-up andcontingency conditions the USO runs as flywheel (ie it is unlocked withinstability lt15 times 10minus13 from 1 to 100 s) It will still provide useable butless accurate references to the instruments including the reference to OFCCommon view ground-to-ground time- and frequency comparison and supplyof the remaining instruments remain possible at the USO provided accuracy

Mass volume and power consumption are 8 kg 7 l 12 W Controldata rateis 10 kBday 50 kBday (FCDP alone)

425 Microwave link (MWL)

The microwave link is a two-way triple-frequency microwave link (Fig 9) Itis designed to perform accurate frequency comparison as well as time-transfer

Ku-Band Up-linkPower Tx 5 WCarrier 13475 GHzPN-Code 100 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

S-Band Down-linkPower Tx 1 WCarrier 2248 MHzPN-Code 1 MChips1pps 1 time marker sData 5 kBits

Ku-Band Down-linkPower Tx 1 WCarrier 1470333 GHzPN-Code 100 MChips1pps 1 time marker sData 5 kBits

EGE SC

S

Ku

KaKa-Band Up-linkPower Tx 5 WCarrier 345 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

Ka-Band Down-LinkPower Tx 1 WCarrier 32 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 5 kBits

GS Antenna diameterEGE terminal 1 mACES terminal 06 m

Fig 9 EGE microwave two-way link concept and characteristics See text for a description Thelink will be able to perform frequency comparisons with up to ten ground stations simultaneously(only one active comparison is shown) pps pulse per second Tx transmitter SC spacecraft

Exp Astron

between on-board and on-ground clocks Two-way operation allows for com-pensation of the effects due to spacecraft location and its motion relative toground Its two bi-directional links operate in Ka-band (32345 GHz) andKu-band (147135 GHz) The high carrier frequency of the up- and down-links allows for a noticeable reduction of the ionospheric delay A third down-link channel is in S-band (22 GHz) and is used to determine the ionospheretotal electron content MWL uses a combination of carrier phase and codephase data to measure with high precision the absolute ionospheric delayas well as ionospheric delay variations during a ground contact The useof three frequency bands allows to resolve third order ionospheric termsKey modulation parameters are Ka band Tone 250 MHz Ku band PN100 MChips S-band 1 MChips

The EGE-MWL concept is a considerably improved version of the ACES-MWL Microwave carrier generation is directly from the 10 GHz microwaveoutput available from the MOLO and having exceptionally low close-in phasenoise without intermediate frequency dividers or other complex synthesiserstages commonly used in RF transponders The high frequency carriers arekey to achieve the desired frequency comparison accuracy

The Ka-band carrier is generated by tripling the FCDP-generated signalThe Ku-band is subsequently generated by division by two using a verystable regenerative divider Finally the assigned radio frequency is obtainedby adding a low frequency signal using an image reject mixer A mixedanalog-digital receiver design will enhance its performance by use of a flexibledigital signal processing and digital control loops The number of simultaneousground contacts is increased from four (as in the ACES mission) to ten or moreusing the digital concept while the number of physical receiver channels perband is reduced from four to two one for operation to ground and one for in-flight calibration The concept allows adaptation of receiver loop parametersto optimise for the different portions of a highly elliptic orbit while reducingsystematic uncertainties due to high Doppler and Doppler rate Real-timedata exchange to ground terminals is by additional data modulation on all RFsignals The spacecraft carries two different small antennae for perigee and forapogee with gains of 8 dBi 20 dBi beamwidths 100 (full earth coverage) 14respectively power is 1 W for each

The MWL end-to-end performance between the attached clocks includingspace- and ground terminals is predicted to be a relative Allan deviation of1 times 10minus16 at 1000 s (perigee) and 1 times 10minus17 at 10000 s (apogee) Thesevalues are based on conservative implementation losses and other instrumentallosses assumed as 6 dB at apogee and 10 dB at perigee including additionallosses due to atmosphere An attitude knowledge of 1 is required

The MWL link will also support precise orbitographyndashcomplementary tothe GNSS receiver Generation of ranging data is simultaneous to two-waytime transfer and is available under all weather conditions Orbit determina-tion accuracy is expected to be better than 1 cm well matching EGE clockrequirements The value is expected based on demonstrated results fromother missions (METEOR ERS-2 TOPEXPOSEIDON) Laser ranging to

Exp Astron

the CCR array is used a few times during the mission to calibrate the MWLranging

In case of non-availability of the ultrapure signal from MOLO an additional10 GHz output is available from the FCDP to support ground-to-ground clockcomparison in common-view mode even after the nominal mission periodwhen the neither the clocks nor MOLO may be available any longer Basedon the PRARE experimentrsquos 12 years of operation this can be assumed forthe FCDPMWL combination as well The potential extended operational pe-riod makes the proposed configuration particularly attractive to metrologicallaboratories that rely on availability of long-lasting data in support of thegeneration of UTC

The resources of MWL are as follows mass 18 kg volume 18 l power55 W TC 2 kBday TM 120 MBday

426 Auxiliary units

The star camera assembly on the EGE spacecraft will provide real-timeattitude data primarily for the orientation of the spacecraft (with accuracy of1 in real-time) and the proper reduction of the GNSSSLR data

A GNSS receiver is used for the precise orbit determination time andfrequency transfer coarse positioning (lt1 m) for real time use by the Attitudeand Orbit Control as well as the back-up for time tagging of all payloaddata For the time and frequency transfer the receiver is nominally using theonboard frequency reference The receiver assembly includes three specialantennas By the year 2015 more than 100 GNSS satellites are expected tobe available for precise orbit determination permitting orbit inaccuracy of 1ndash2 cm RMS for altitudes up to approx 10000 km At apogee a combinationof GNSS signals and MWL ranging will be used with expected inaccuracyof 10 cm

5 Basic spacecraft specifications

The mission foresees a single spacecraft with a mission duration limited toapprox 3 years for cost reasons Attitude control along three axes withless than 1 error on a timescale of 1000 s is required The control can beperformed with cold gas thrusters magnetotorquers or reaction wheels Theselection will be done on the basis of the vibrations induced by these systems(desired level arms lt 10minus6 g)

Cold gas thrusters or monopropellant thrusters will be required to maintainthe orbit parameters during the mission by small orbit correction manoeuvresand will also guarantee a proper tuning of the orbit parameters after the releasefrom the launcher The ground segment (telemetry control and data transfer)could be via non-ESA stations user-operated to save costs

Further resources are indicated in Table 2

Exp Astron

The satellite bus design drivers are nadir-pointing microwave antennasthermal stability for the payload instruments no strong magnetic sources lowmicrovibrations harsh radiation environment due to crossing of the radiationbelts once per orbit power demand by science payload (with the possibility toreduce power demand by stand-by of some components during specific missionphases eg eclipse) mass limit due to required high altitude with low costlaunch

The satellite bus design can be based on established platforms such asFLEXBUS or PRIMA used on several past missions Their design philosophydiffers one having body-mounted solar arrays the other wings The latter canprovide larger amounts of power The vibration level induced by the panelswhen they are fixed and the duration of solar array drive periods per orbitwill have to be considered in the selection The FLEXBUS concept was usedfor a mission already requiring a low level of residual accelerations and theplatform is optimised for this parameter

For most of the satellite components one can consider using compo-nents and electronic units from navigation or communication satellites whichtypically cruise at altitudes between 15000 and 25000 km altitude LISAPathfinder which will cross the radiation belts during spiralling to theLagrange point will face similar problems concerning radiation as EGE as wellas XMM which is in a Molniya-type orbit with similar radiation characteristics

The instruments will therefore be located in the satellite structure such asto realise ionizing radiation shielding by the spacecraft structure and non-sensitive equipment to the maximum extent possible the need for additionalradiation shielding (including trapped proton radiation) for especially sensitiveparts of the instruments must be determined by ground tests It should bementioned that an orbit with higher perigee and less radiation dose is also anoption as it leads to a tolerable reduction of the science goals

The required plusmn3C temperature stability at the interface to the payloadenclosure is feasible The thermal control will employ the usual means (anenclosure with high thermal inertia heaters and a suitable radiator) Thelocation of the radiator will need a special study given the non-standard orbit

Magnetic field present during the mission can affect the properties of theonboard clock Using SPENVIS the magnetic field has been calculated alongthe spacecraft orbit the field intensity presents maxima around 02 G anda minimum value of the order of 10minus3 G As discussed above this is not alimitation to the instrument performance

6 Discussion and conclusion

The accuracy of the measurements depends crucially on the stability andaccuracy of the clocks While the performance of ground clocks (even thoseeventually deployed in new ground stations dedicated to EGE) will likelyimprove significantly from the present level the space clock design will beldquofrozenrdquo approx 10 years before data acquisition Thus their performance

Exp Astron

level will be significantly below the future ground station clock performanceAs specification we have chosen a performance level that has been demon-strated with laboratory breadboards recently We see no fundamental reasonthat this laboratory performance cannot also be achieved in a space ion opticalclock The experience with the PHARAO cold atom clock has shown thatan engineering model of a laser-based cold atom clock has a performancesimilar to that of a laboratory prototype The perturbation of the ion clock fre-quency due to magnetic fields along the orbit can be modelled and suppressedwith shielding and the residual effects measured and compensated for Themoderate temperature stability (plusmn 3C) to be provided in the experimentalbay can be achieved with thermal shields Accurate temperature controlof particular units is done by active stabilization of the units themselves

One important aspect is radiation which may affect all optical and trapcomponents as well as the lasers and reference cavity The radiation envi-ronment on the EGE orbit is significantly harsher than the near-earth ISSorbit on which ACES will operate The radiation effects must therefore becharacterized comprehensively with appropriate tests and then the radiationshield for the EGE instruments designed

In summary the Einstein Gravity Explorer mission (EGE) is a satellitemission aimed at a precise measurement of the properties of space-time usingatomic clocks By testing one of the most fundamental predictions of EinsteinrsquosTheory of General Relativity the gravitational redshift and thereby searchingfor quantum effects in gravity it explores one of the most important frontiers infundamental physics The EGE mission reaches beyond fundamental physicsproviding an important spin-off in geophysics Furthermore the technologicalaspects of the mission are ground-breaking for future time-keeping and ref-erence frame systems in space and thus are also expected to be of wide useto science and technology in general The payload of the mission has beenconceived as a compromise between cost and performance needing to fit thecost envelope of an ESA medium-class mission1

Acknowledgement We thank S Cesare for helpful information

References

1 Will CM The confrontation between general relativity and experiment Living Rev Relativ9 3 (2006)

2 Ni W-T Empirical foundations of the relativistic gravity Int J Mod Phys D 14 901 (2005)doi101142S0218271805007139

1Finally we point out that a spacecraft with a significantly reduced and thus lower-cost payload (noclocks only MOLO FCDP MWL) on a circular orbit of appropriate height could be used for along-duration earth science mission dedicated specifically to the measurement of the geopotential

Exp Astron

to additional electrodes to position the ion precisely at trap potentialcentre in order to minimise the ionrsquos micromotion at the drive frequencyThis needs automatic periodic monitoring and correction especially in theperiod after re-loading an ion into the trap External magnetic field coilsin three orthogonal axes allow the nulling of external fields and settingof a fixed field of sim1 μT The trap and coils are surrounded by mu-metalshielding to minimise external field changes during the orbit A level oftemperature control of 1 K of the region surrounding the trap is requiredfor maintaining the blackbody-induced frequency shift uncertainty withtemperature at the 10minus17 level

bull a laser platform to provide frequency doubling of an extended cavity739 nm diode laser for Doppler cooling of the ion on the 2S12(F = 1) minus2 P12 (F = 0) dipole transition at 3695 nm Light from an amplified diodelaser device frequency-doubled in a periodically poled LiTaO3 crystal insingle-pass will be used Typical laser powers used for driving the coolingtransition below saturation are about 2 μW for a beam waist in the trapof about 50 μm By modulation of the injection current of the 739 nmdiode laser at 147 GHz a sideband is generated that excites the 2S12 (F =0) rarr2 P12 (F = 1) ldquorepumperrdquo transition in order to avoid optical pump-ing between the ground hyperfine states

bull The platform also houses the auxiliary lasers at 935 and 638 nm The935 nm diode provides repumping of the ion from the 2 D32(F = 1)metastable level after occasional branching decays to this level during thecooling sequence The 638 nm diode allows fast recovery of the ion fromthe very-long-lived 2F72 metastable state after very occasional collisionaldecay to that state Currently extended cavity lasers are forseen butdistributed-feedback lasers may be available in the near future

bull a high NA lens imaging system and photomultiplier detection system torecord the statistics of 369 nm fluorescence quantum jumps as a functionof clock laser frequency step

bull a fibre system to deliver the various cooling auxiliary and clock light fromsource to trap making use of achromatic doublets where necessary at thefibre-free space interface for launching into the trap

bull a monitoring and control processor which provides the primary coolingand clock laser pulse magnetic field and detection sequencing to observeand lock to the ion clock transition frequency The processor also monitorsfrequency and amplitude data necessary to determine normal laser andion operational conditions and initiate resetting and recovery algorithmswhere necessary and laser unit failure The actual magnetic field presenton the atomic volume is determined and actively stabilized so that aninaccuracy at the level of 1 times 10minus17 can be achieved

bull a redundancy level of at least two units for cooling clock and repumperlasers plus similar for frequency doubling crystals All redundancy unitsfor each wavelength will be fibre multiplexed as standard allowing redun-dant unit activation on determination of prior unit failure

Exp Astron

For an optimized quantum-projection-noise-limited performance of thesingle-ion clock Ramsey interrogation with a cycle time equal to the lifetimeof the metastable D32 level of 50 ms will be performed leading to an Allandeviation of about 27 times 10minus15 (τ s)minus12 This is sim45 times 10minus17 at 1 h satisfyingrequired stabilities for the science objectives Current laboratory results are astability of 1 times 10minus15 at 30 s averaging down to 3 times 10minus16 at 1000 s (PTBGermany) Also comparisons between two independent traps showed agree-ment at the level of several parts in 10minus16 over a series of eight measurementruns

The 171Yb+ ion clock offers a second option namely the 2S12ndash2 F72 octu-pole transition at 476 nm (see Table 1 and Fig 6) [48] It has an extremelylong upper state lifetime low electric quadrupole and second order Zeemansystematic shifts and therefore is a candidate for an even higher performanceclock if a clock laser system with lower instability than the above is providedApart from the clock laser it shares the laser system of the quadrupoleclock Thus with moderate additional resources the Yb+ ion apparatus couldprovide two optical clocks in a single package and specifically perform directfrequency comparisons between the quadrupole and the octupole transitionwith a large sensitivity Aα

The ion clock will operate under microprocessor control providing thecontrol algorithms for the integrated pulse sequencing for cooling repumpingclock interrogation magnetic field switching cooling fluorescence monitoringThe controller will also provide error flags for critical processes and systemreset and correction where necessary This will include ion re-loading andmicromotion reduction algorithms Automatic monitoring of laser power andspectral quality with signal re-optimisation or laser failure determination isrequired Redundant laser units will be pre-aligned and multiplexed into thefibre delivery to the trap so that redundant units can be readily activated in thecase of diode laser failure

423 Microwave-optical local oscillator (MOLO)

This oscillator provides two ultrastable frequency outputs one in the opticalregion the other in the microwave region the two frequencies being coher-ently related The MOLO is composed of three subunits the clock laser thereference cavity and the frequency comb see Fig 8

The clock laser subunit delivers the light for the excitation of the opticalclock transition It is generated by an extended-cavity diode laser at a wave-length of 871 nm Most of the output wave of this laser is sent to the ionclock The laser subunit containing two lasers for redundancy will require3 kg 3 l 15 W The frequency instability of the clock laser must be superiorto the minimum clock instability 27 times 10minus15 (τ s)minus12 for τ up to the timeconstant (sim10 s) of the servo system that locks the laser to the ionrsquos resonancesignal In order to achieve this the laser frequency is stabilized to the reference

Exp Astron

871 nm clock laser

EO

Detector and servo

AOM Femtosecond

frequency comb

To optical clock 10 GHz To FCDP

Microwave-optical local oscillator

ULE cavity

Fig 8 Microwave-optical local oscillator (MOLO) consisting of clock laser reference ULE cavityand femtosecond frequency comb (OFC) Radiofrequency inputs and outputs are not shown EOelectro-optic modulator ULE ultra-low expansion AOM frequency shifter

cavity where the PoundndashDreverndashHall method is used About 10 μW of thelaser output are used for this

The system will exhibit sim04 Hz (sim1 times 10minus15 relative) linewidth and afrequency drift lt01 Hzs The cavity consists of two high-reflectivity mirrorsoptically contacted to a cylinder of diameter 70 mm length 150 mm made ofULE (a glass exhibiting ultra low thermal expansion coefficient) with a cavityfinesse of about 300000 Residual vibrations of the satellite (arms lt 1 times 10minus6 g)should not be a limiting factor to the desired linewidth for safety marginan optimized cavity support can reduce the sensitivity to accelerations Foracoustic and thermal isolation the cavity resides inside an aluminium vacuumchamber equipped with a small (3 ls) ion-getter pump Two stages of polishedaluminium shields are implemented around the cavity which are activelytemperature stabilized so that the cavity is precisely kept at the zero crossingtemperature of the ULE thermal expansion coefficient Heat application andremoval is by thermoelectric elements between the shield and the vacuumchamber Additionally the entire vacuum chamber is temperature-stabilized

During transportation and launch the cavity will be rigidly fixed insidethe vacuum chamber by pressing on it from the sides by piezo-mechanicalactuators allowing accelerations of several g After bringing the satellite onthe orbit the actuators will be released

The optical setup for frequency stabilization will be a classic Pound-Drever-Hall scheme in reflection using an electro-optic modulator (EO) Couplingof the clock laser radiation to the cavity is via single-mode polarization main-taining optical fiber The fiber in- and out-couplers are mounted on miniaturepiezo-motor driven multiaxis translation stages Microprocessor control ofthe translation stages performs a laser-to-fiber incoupling stabilization andlaser-to-cavity mode-matching optimization (positioning inaccuracy lt10 μmalignment inaccuracy lt10 μrad)

The frequency stability of the clock laser will be limited by thermal 1f noisein the mirrors with an fractional Allan deviation slightly below 1 times 10minus15 from1 s to a few 10 s Resources including control electronics are 15 l 12 kg 8 W

Exp Astron

A femtosecond fiber laser based optical frequency comb generator (OFCFig 8) will be used for the clock frequency comparisons and to obtain aspectrally ultrapure microwave local oscillator output for the operation ofPHARAO for the microwave link and for the GNSS receiver The fem-tosecond laser is based on erbium-doped fiber pumped by fiber-coupled laserdiodes It emits 100 fs pulses at a repetition rate of 250 MHz with an averagepower of 100 mW The emission spectrum is centered around 1560 nm Spec-tral broadening to more than one octave is obtained via nonlinear interactionin a microstructure fibre so that the range from about 1050 to 2100 nm iscovered

Second harmonic generation of a portion of the broadened spectrum allowsobtaining the required light for the measurement of the clock laser frequencyat 871 nm The OFC will be referenced to the clock laser frequency (afterintroduction a controllable shift via the frequency shifter AOM) by makingthe carrier envelope offset frequency vanish so that the repetition rate fr isa high-order subharmonic of the optical clock reference frequency f1 = m1 frThe link between the optical and the microwave domain is then simply thedistribution of the repetition rate frequency or more conveniently a high(order 40) harmonic of the comb repetition rate 10 GHz

It has been demonstrated by several groups that frequency comparison andfrequency division with an OFC can be performed to very high accuracy andstability beyond the limitations imposed by the optical clocks themselves Forexample a fiber laser OFC with an instability (Allan deviation) of 6 times 10minus17

in 1 s has been demonstrated [51] OFCs based on fiber technology possessgood long-term reliability and high power efficiency The projection for a spacequalified OFC based on current industrial terrestrial technology is 16 kg mass20 l volume and 38 W power requirement

424 Frequency comparison and distribution package (FCDP)

This package distributes the ultra-stable 10 GHz reference frequency outputof the MOLO to the MWL performs frequency comparisons between MOLOand PHARAO and generates a variety of different reference signals for theindividual instruments including the 250 MHz reference clock to the OFCAll signals are coherent to the MOLO In the FCDP a USO is locked to theMOLO signal by a digital PLL A divider chain generates timing pulses (1 pps1000 pps etc) and keeps a continuous local time scale From the digital phasecomparison data the phase noise density spectrum between the USO and theMOLO is available by telemetry (TM) to assess the overall frequency purityin-flight

As the FCDP distributes the full performance reference from the MOLOdirectly to the MWL there is minimum impact of the FCDP on overall clockcomparison performance The USO-phase lock loop uses a similar designas for ACES-FCDP The digital PLL with linear phase detector allows theUSO to be slightly offset from its reference which is selectable by groundcommand This results in an accurate on-board time scale despite any offset

Exp Astron

of the MOLO reference frequency which can be synchronised to UTC Thefrequency synthesis chain re-uses modules developed for ACES-FCDP andtheir frequency inaccuracy Expected relative uncertainty (Allan deviation)of the frequency comparison between two signals provided to the FCDP is47 times 10minus17 at 1000 s 95 times 10minus18 at 10000 s

During non-availability of the MOLO ie during system start-up andcontingency conditions the USO runs as flywheel (ie it is unlocked withinstability lt15 times 10minus13 from 1 to 100 s) It will still provide useable butless accurate references to the instruments including the reference to OFCCommon view ground-to-ground time- and frequency comparison and supplyof the remaining instruments remain possible at the USO provided accuracy

Mass volume and power consumption are 8 kg 7 l 12 W Controldata rateis 10 kBday 50 kBday (FCDP alone)

425 Microwave link (MWL)

The microwave link is a two-way triple-frequency microwave link (Fig 9) Itis designed to perform accurate frequency comparison as well as time-transfer

Ku-Band Up-linkPower Tx 5 WCarrier 13475 GHzPN-Code 100 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

S-Band Down-linkPower Tx 1 WCarrier 2248 MHzPN-Code 1 MChips1pps 1 time marker sData 5 kBits

Ku-Band Down-linkPower Tx 1 WCarrier 1470333 GHzPN-Code 100 MChips1pps 1 time marker sData 5 kBits

EGE SC

S

Ku

KaKa-Band Up-linkPower Tx 5 WCarrier 345 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

Ka-Band Down-LinkPower Tx 1 WCarrier 32 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 5 kBits

GS Antenna diameterEGE terminal 1 mACES terminal 06 m

Fig 9 EGE microwave two-way link concept and characteristics See text for a description Thelink will be able to perform frequency comparisons with up to ten ground stations simultaneously(only one active comparison is shown) pps pulse per second Tx transmitter SC spacecraft

Exp Astron

between on-board and on-ground clocks Two-way operation allows for com-pensation of the effects due to spacecraft location and its motion relative toground Its two bi-directional links operate in Ka-band (32345 GHz) andKu-band (147135 GHz) The high carrier frequency of the up- and down-links allows for a noticeable reduction of the ionospheric delay A third down-link channel is in S-band (22 GHz) and is used to determine the ionospheretotal electron content MWL uses a combination of carrier phase and codephase data to measure with high precision the absolute ionospheric delayas well as ionospheric delay variations during a ground contact The useof three frequency bands allows to resolve third order ionospheric termsKey modulation parameters are Ka band Tone 250 MHz Ku band PN100 MChips S-band 1 MChips

The EGE-MWL concept is a considerably improved version of the ACES-MWL Microwave carrier generation is directly from the 10 GHz microwaveoutput available from the MOLO and having exceptionally low close-in phasenoise without intermediate frequency dividers or other complex synthesiserstages commonly used in RF transponders The high frequency carriers arekey to achieve the desired frequency comparison accuracy

The Ka-band carrier is generated by tripling the FCDP-generated signalThe Ku-band is subsequently generated by division by two using a verystable regenerative divider Finally the assigned radio frequency is obtainedby adding a low frequency signal using an image reject mixer A mixedanalog-digital receiver design will enhance its performance by use of a flexibledigital signal processing and digital control loops The number of simultaneousground contacts is increased from four (as in the ACES mission) to ten or moreusing the digital concept while the number of physical receiver channels perband is reduced from four to two one for operation to ground and one for in-flight calibration The concept allows adaptation of receiver loop parametersto optimise for the different portions of a highly elliptic orbit while reducingsystematic uncertainties due to high Doppler and Doppler rate Real-timedata exchange to ground terminals is by additional data modulation on all RFsignals The spacecraft carries two different small antennae for perigee and forapogee with gains of 8 dBi 20 dBi beamwidths 100 (full earth coverage) 14respectively power is 1 W for each

The MWL end-to-end performance between the attached clocks includingspace- and ground terminals is predicted to be a relative Allan deviation of1 times 10minus16 at 1000 s (perigee) and 1 times 10minus17 at 10000 s (apogee) Thesevalues are based on conservative implementation losses and other instrumentallosses assumed as 6 dB at apogee and 10 dB at perigee including additionallosses due to atmosphere An attitude knowledge of 1 is required

The MWL link will also support precise orbitographyndashcomplementary tothe GNSS receiver Generation of ranging data is simultaneous to two-waytime transfer and is available under all weather conditions Orbit determina-tion accuracy is expected to be better than 1 cm well matching EGE clockrequirements The value is expected based on demonstrated results fromother missions (METEOR ERS-2 TOPEXPOSEIDON) Laser ranging to

Exp Astron

the CCR array is used a few times during the mission to calibrate the MWLranging

In case of non-availability of the ultrapure signal from MOLO an additional10 GHz output is available from the FCDP to support ground-to-ground clockcomparison in common-view mode even after the nominal mission periodwhen the neither the clocks nor MOLO may be available any longer Basedon the PRARE experimentrsquos 12 years of operation this can be assumed forthe FCDPMWL combination as well The potential extended operational pe-riod makes the proposed configuration particularly attractive to metrologicallaboratories that rely on availability of long-lasting data in support of thegeneration of UTC

The resources of MWL are as follows mass 18 kg volume 18 l power55 W TC 2 kBday TM 120 MBday

426 Auxiliary units

The star camera assembly on the EGE spacecraft will provide real-timeattitude data primarily for the orientation of the spacecraft (with accuracy of1 in real-time) and the proper reduction of the GNSSSLR data

A GNSS receiver is used for the precise orbit determination time andfrequency transfer coarse positioning (lt1 m) for real time use by the Attitudeand Orbit Control as well as the back-up for time tagging of all payloaddata For the time and frequency transfer the receiver is nominally using theonboard frequency reference The receiver assembly includes three specialantennas By the year 2015 more than 100 GNSS satellites are expected tobe available for precise orbit determination permitting orbit inaccuracy of 1ndash2 cm RMS for altitudes up to approx 10000 km At apogee a combinationof GNSS signals and MWL ranging will be used with expected inaccuracyof 10 cm

5 Basic spacecraft specifications

The mission foresees a single spacecraft with a mission duration limited toapprox 3 years for cost reasons Attitude control along three axes withless than 1 error on a timescale of 1000 s is required The control can beperformed with cold gas thrusters magnetotorquers or reaction wheels Theselection will be done on the basis of the vibrations induced by these systems(desired level arms lt 10minus6 g)

Cold gas thrusters or monopropellant thrusters will be required to maintainthe orbit parameters during the mission by small orbit correction manoeuvresand will also guarantee a proper tuning of the orbit parameters after the releasefrom the launcher The ground segment (telemetry control and data transfer)could be via non-ESA stations user-operated to save costs

Further resources are indicated in Table 2

Exp Astron

The satellite bus design drivers are nadir-pointing microwave antennasthermal stability for the payload instruments no strong magnetic sources lowmicrovibrations harsh radiation environment due to crossing of the radiationbelts once per orbit power demand by science payload (with the possibility toreduce power demand by stand-by of some components during specific missionphases eg eclipse) mass limit due to required high altitude with low costlaunch

The satellite bus design can be based on established platforms such asFLEXBUS or PRIMA used on several past missions Their design philosophydiffers one having body-mounted solar arrays the other wings The latter canprovide larger amounts of power The vibration level induced by the panelswhen they are fixed and the duration of solar array drive periods per orbitwill have to be considered in the selection The FLEXBUS concept was usedfor a mission already requiring a low level of residual accelerations and theplatform is optimised for this parameter

For most of the satellite components one can consider using compo-nents and electronic units from navigation or communication satellites whichtypically cruise at altitudes between 15000 and 25000 km altitude LISAPathfinder which will cross the radiation belts during spiralling to theLagrange point will face similar problems concerning radiation as EGE as wellas XMM which is in a Molniya-type orbit with similar radiation characteristics

The instruments will therefore be located in the satellite structure such asto realise ionizing radiation shielding by the spacecraft structure and non-sensitive equipment to the maximum extent possible the need for additionalradiation shielding (including trapped proton radiation) for especially sensitiveparts of the instruments must be determined by ground tests It should bementioned that an orbit with higher perigee and less radiation dose is also anoption as it leads to a tolerable reduction of the science goals

The required plusmn3C temperature stability at the interface to the payloadenclosure is feasible The thermal control will employ the usual means (anenclosure with high thermal inertia heaters and a suitable radiator) Thelocation of the radiator will need a special study given the non-standard orbit

Magnetic field present during the mission can affect the properties of theonboard clock Using SPENVIS the magnetic field has been calculated alongthe spacecraft orbit the field intensity presents maxima around 02 G anda minimum value of the order of 10minus3 G As discussed above this is not alimitation to the instrument performance

6 Discussion and conclusion

The accuracy of the measurements depends crucially on the stability andaccuracy of the clocks While the performance of ground clocks (even thoseeventually deployed in new ground stations dedicated to EGE) will likelyimprove significantly from the present level the space clock design will beldquofrozenrdquo approx 10 years before data acquisition Thus their performance

Exp Astron

level will be significantly below the future ground station clock performanceAs specification we have chosen a performance level that has been demon-strated with laboratory breadboards recently We see no fundamental reasonthat this laboratory performance cannot also be achieved in a space ion opticalclock The experience with the PHARAO cold atom clock has shown thatan engineering model of a laser-based cold atom clock has a performancesimilar to that of a laboratory prototype The perturbation of the ion clock fre-quency due to magnetic fields along the orbit can be modelled and suppressedwith shielding and the residual effects measured and compensated for Themoderate temperature stability (plusmn 3C) to be provided in the experimentalbay can be achieved with thermal shields Accurate temperature controlof particular units is done by active stabilization of the units themselves

One important aspect is radiation which may affect all optical and trapcomponents as well as the lasers and reference cavity The radiation envi-ronment on the EGE orbit is significantly harsher than the near-earth ISSorbit on which ACES will operate The radiation effects must therefore becharacterized comprehensively with appropriate tests and then the radiationshield for the EGE instruments designed

In summary the Einstein Gravity Explorer mission (EGE) is a satellitemission aimed at a precise measurement of the properties of space-time usingatomic clocks By testing one of the most fundamental predictions of EinsteinrsquosTheory of General Relativity the gravitational redshift and thereby searchingfor quantum effects in gravity it explores one of the most important frontiers infundamental physics The EGE mission reaches beyond fundamental physicsproviding an important spin-off in geophysics Furthermore the technologicalaspects of the mission are ground-breaking for future time-keeping and ref-erence frame systems in space and thus are also expected to be of wide useto science and technology in general The payload of the mission has beenconceived as a compromise between cost and performance needing to fit thecost envelope of an ESA medium-class mission1

Acknowledgement We thank S Cesare for helpful information

References

1 Will CM The confrontation between general relativity and experiment Living Rev Relativ9 3 (2006)

2 Ni W-T Empirical foundations of the relativistic gravity Int J Mod Phys D 14 901 (2005)doi101142S0218271805007139

1Finally we point out that a spacecraft with a significantly reduced and thus lower-cost payload (noclocks only MOLO FCDP MWL) on a circular orbit of appropriate height could be used for along-duration earth science mission dedicated specifically to the measurement of the geopotential

Exp Astron

For an optimized quantum-projection-noise-limited performance of thesingle-ion clock Ramsey interrogation with a cycle time equal to the lifetimeof the metastable D32 level of 50 ms will be performed leading to an Allandeviation of about 27 times 10minus15 (τ s)minus12 This is sim45 times 10minus17 at 1 h satisfyingrequired stabilities for the science objectives Current laboratory results are astability of 1 times 10minus15 at 30 s averaging down to 3 times 10minus16 at 1000 s (PTBGermany) Also comparisons between two independent traps showed agree-ment at the level of several parts in 10minus16 over a series of eight measurementruns

The 171Yb+ ion clock offers a second option namely the 2S12ndash2 F72 octu-pole transition at 476 nm (see Table 1 and Fig 6) [48] It has an extremelylong upper state lifetime low electric quadrupole and second order Zeemansystematic shifts and therefore is a candidate for an even higher performanceclock if a clock laser system with lower instability than the above is providedApart from the clock laser it shares the laser system of the quadrupoleclock Thus with moderate additional resources the Yb+ ion apparatus couldprovide two optical clocks in a single package and specifically perform directfrequency comparisons between the quadrupole and the octupole transitionwith a large sensitivity Aα

The ion clock will operate under microprocessor control providing thecontrol algorithms for the integrated pulse sequencing for cooling repumpingclock interrogation magnetic field switching cooling fluorescence monitoringThe controller will also provide error flags for critical processes and systemreset and correction where necessary This will include ion re-loading andmicromotion reduction algorithms Automatic monitoring of laser power andspectral quality with signal re-optimisation or laser failure determination isrequired Redundant laser units will be pre-aligned and multiplexed into thefibre delivery to the trap so that redundant units can be readily activated in thecase of diode laser failure

423 Microwave-optical local oscillator (MOLO)

This oscillator provides two ultrastable frequency outputs one in the opticalregion the other in the microwave region the two frequencies being coher-ently related The MOLO is composed of three subunits the clock laser thereference cavity and the frequency comb see Fig 8

The clock laser subunit delivers the light for the excitation of the opticalclock transition It is generated by an extended-cavity diode laser at a wave-length of 871 nm Most of the output wave of this laser is sent to the ionclock The laser subunit containing two lasers for redundancy will require3 kg 3 l 15 W The frequency instability of the clock laser must be superiorto the minimum clock instability 27 times 10minus15 (τ s)minus12 for τ up to the timeconstant (sim10 s) of the servo system that locks the laser to the ionrsquos resonancesignal In order to achieve this the laser frequency is stabilized to the reference

Exp Astron

871 nm clock laser

EO

Detector and servo

AOM Femtosecond

frequency comb

To optical clock 10 GHz To FCDP

Microwave-optical local oscillator

ULE cavity

Fig 8 Microwave-optical local oscillator (MOLO) consisting of clock laser reference ULE cavityand femtosecond frequency comb (OFC) Radiofrequency inputs and outputs are not shown EOelectro-optic modulator ULE ultra-low expansion AOM frequency shifter

cavity where the PoundndashDreverndashHall method is used About 10 μW of thelaser output are used for this

The system will exhibit sim04 Hz (sim1 times 10minus15 relative) linewidth and afrequency drift lt01 Hzs The cavity consists of two high-reflectivity mirrorsoptically contacted to a cylinder of diameter 70 mm length 150 mm made ofULE (a glass exhibiting ultra low thermal expansion coefficient) with a cavityfinesse of about 300000 Residual vibrations of the satellite (arms lt 1 times 10minus6 g)should not be a limiting factor to the desired linewidth for safety marginan optimized cavity support can reduce the sensitivity to accelerations Foracoustic and thermal isolation the cavity resides inside an aluminium vacuumchamber equipped with a small (3 ls) ion-getter pump Two stages of polishedaluminium shields are implemented around the cavity which are activelytemperature stabilized so that the cavity is precisely kept at the zero crossingtemperature of the ULE thermal expansion coefficient Heat application andremoval is by thermoelectric elements between the shield and the vacuumchamber Additionally the entire vacuum chamber is temperature-stabilized

During transportation and launch the cavity will be rigidly fixed insidethe vacuum chamber by pressing on it from the sides by piezo-mechanicalactuators allowing accelerations of several g After bringing the satellite onthe orbit the actuators will be released

The optical setup for frequency stabilization will be a classic Pound-Drever-Hall scheme in reflection using an electro-optic modulator (EO) Couplingof the clock laser radiation to the cavity is via single-mode polarization main-taining optical fiber The fiber in- and out-couplers are mounted on miniaturepiezo-motor driven multiaxis translation stages Microprocessor control ofthe translation stages performs a laser-to-fiber incoupling stabilization andlaser-to-cavity mode-matching optimization (positioning inaccuracy lt10 μmalignment inaccuracy lt10 μrad)

The frequency stability of the clock laser will be limited by thermal 1f noisein the mirrors with an fractional Allan deviation slightly below 1 times 10minus15 from1 s to a few 10 s Resources including control electronics are 15 l 12 kg 8 W

Exp Astron

A femtosecond fiber laser based optical frequency comb generator (OFCFig 8) will be used for the clock frequency comparisons and to obtain aspectrally ultrapure microwave local oscillator output for the operation ofPHARAO for the microwave link and for the GNSS receiver The fem-tosecond laser is based on erbium-doped fiber pumped by fiber-coupled laserdiodes It emits 100 fs pulses at a repetition rate of 250 MHz with an averagepower of 100 mW The emission spectrum is centered around 1560 nm Spec-tral broadening to more than one octave is obtained via nonlinear interactionin a microstructure fibre so that the range from about 1050 to 2100 nm iscovered

Second harmonic generation of a portion of the broadened spectrum allowsobtaining the required light for the measurement of the clock laser frequencyat 871 nm The OFC will be referenced to the clock laser frequency (afterintroduction a controllable shift via the frequency shifter AOM) by makingthe carrier envelope offset frequency vanish so that the repetition rate fr isa high-order subharmonic of the optical clock reference frequency f1 = m1 frThe link between the optical and the microwave domain is then simply thedistribution of the repetition rate frequency or more conveniently a high(order 40) harmonic of the comb repetition rate 10 GHz

It has been demonstrated by several groups that frequency comparison andfrequency division with an OFC can be performed to very high accuracy andstability beyond the limitations imposed by the optical clocks themselves Forexample a fiber laser OFC with an instability (Allan deviation) of 6 times 10minus17

in 1 s has been demonstrated [51] OFCs based on fiber technology possessgood long-term reliability and high power efficiency The projection for a spacequalified OFC based on current industrial terrestrial technology is 16 kg mass20 l volume and 38 W power requirement

424 Frequency comparison and distribution package (FCDP)

This package distributes the ultra-stable 10 GHz reference frequency outputof the MOLO to the MWL performs frequency comparisons between MOLOand PHARAO and generates a variety of different reference signals for theindividual instruments including the 250 MHz reference clock to the OFCAll signals are coherent to the MOLO In the FCDP a USO is locked to theMOLO signal by a digital PLL A divider chain generates timing pulses (1 pps1000 pps etc) and keeps a continuous local time scale From the digital phasecomparison data the phase noise density spectrum between the USO and theMOLO is available by telemetry (TM) to assess the overall frequency purityin-flight

As the FCDP distributes the full performance reference from the MOLOdirectly to the MWL there is minimum impact of the FCDP on overall clockcomparison performance The USO-phase lock loop uses a similar designas for ACES-FCDP The digital PLL with linear phase detector allows theUSO to be slightly offset from its reference which is selectable by groundcommand This results in an accurate on-board time scale despite any offset

Exp Astron

of the MOLO reference frequency which can be synchronised to UTC Thefrequency synthesis chain re-uses modules developed for ACES-FCDP andtheir frequency inaccuracy Expected relative uncertainty (Allan deviation)of the frequency comparison between two signals provided to the FCDP is47 times 10minus17 at 1000 s 95 times 10minus18 at 10000 s

During non-availability of the MOLO ie during system start-up andcontingency conditions the USO runs as flywheel (ie it is unlocked withinstability lt15 times 10minus13 from 1 to 100 s) It will still provide useable butless accurate references to the instruments including the reference to OFCCommon view ground-to-ground time- and frequency comparison and supplyof the remaining instruments remain possible at the USO provided accuracy

Mass volume and power consumption are 8 kg 7 l 12 W Controldata rateis 10 kBday 50 kBday (FCDP alone)

425 Microwave link (MWL)

The microwave link is a two-way triple-frequency microwave link (Fig 9) Itis designed to perform accurate frequency comparison as well as time-transfer

Ku-Band Up-linkPower Tx 5 WCarrier 13475 GHzPN-Code 100 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

S-Band Down-linkPower Tx 1 WCarrier 2248 MHzPN-Code 1 MChips1pps 1 time marker sData 5 kBits

Ku-Band Down-linkPower Tx 1 WCarrier 1470333 GHzPN-Code 100 MChips1pps 1 time marker sData 5 kBits

EGE SC

S

Ku

KaKa-Band Up-linkPower Tx 5 WCarrier 345 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

Ka-Band Down-LinkPower Tx 1 WCarrier 32 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 5 kBits

GS Antenna diameterEGE terminal 1 mACES terminal 06 m

Fig 9 EGE microwave two-way link concept and characteristics See text for a description Thelink will be able to perform frequency comparisons with up to ten ground stations simultaneously(only one active comparison is shown) pps pulse per second Tx transmitter SC spacecraft

Exp Astron

between on-board and on-ground clocks Two-way operation allows for com-pensation of the effects due to spacecraft location and its motion relative toground Its two bi-directional links operate in Ka-band (32345 GHz) andKu-band (147135 GHz) The high carrier frequency of the up- and down-links allows for a noticeable reduction of the ionospheric delay A third down-link channel is in S-band (22 GHz) and is used to determine the ionospheretotal electron content MWL uses a combination of carrier phase and codephase data to measure with high precision the absolute ionospheric delayas well as ionospheric delay variations during a ground contact The useof three frequency bands allows to resolve third order ionospheric termsKey modulation parameters are Ka band Tone 250 MHz Ku band PN100 MChips S-band 1 MChips

The EGE-MWL concept is a considerably improved version of the ACES-MWL Microwave carrier generation is directly from the 10 GHz microwaveoutput available from the MOLO and having exceptionally low close-in phasenoise without intermediate frequency dividers or other complex synthesiserstages commonly used in RF transponders The high frequency carriers arekey to achieve the desired frequency comparison accuracy

The Ka-band carrier is generated by tripling the FCDP-generated signalThe Ku-band is subsequently generated by division by two using a verystable regenerative divider Finally the assigned radio frequency is obtainedby adding a low frequency signal using an image reject mixer A mixedanalog-digital receiver design will enhance its performance by use of a flexibledigital signal processing and digital control loops The number of simultaneousground contacts is increased from four (as in the ACES mission) to ten or moreusing the digital concept while the number of physical receiver channels perband is reduced from four to two one for operation to ground and one for in-flight calibration The concept allows adaptation of receiver loop parametersto optimise for the different portions of a highly elliptic orbit while reducingsystematic uncertainties due to high Doppler and Doppler rate Real-timedata exchange to ground terminals is by additional data modulation on all RFsignals The spacecraft carries two different small antennae for perigee and forapogee with gains of 8 dBi 20 dBi beamwidths 100 (full earth coverage) 14respectively power is 1 W for each

The MWL end-to-end performance between the attached clocks includingspace- and ground terminals is predicted to be a relative Allan deviation of1 times 10minus16 at 1000 s (perigee) and 1 times 10minus17 at 10000 s (apogee) Thesevalues are based on conservative implementation losses and other instrumentallosses assumed as 6 dB at apogee and 10 dB at perigee including additionallosses due to atmosphere An attitude knowledge of 1 is required

The MWL link will also support precise orbitographyndashcomplementary tothe GNSS receiver Generation of ranging data is simultaneous to two-waytime transfer and is available under all weather conditions Orbit determina-tion accuracy is expected to be better than 1 cm well matching EGE clockrequirements The value is expected based on demonstrated results fromother missions (METEOR ERS-2 TOPEXPOSEIDON) Laser ranging to

Exp Astron

the CCR array is used a few times during the mission to calibrate the MWLranging

In case of non-availability of the ultrapure signal from MOLO an additional10 GHz output is available from the FCDP to support ground-to-ground clockcomparison in common-view mode even after the nominal mission periodwhen the neither the clocks nor MOLO may be available any longer Basedon the PRARE experimentrsquos 12 years of operation this can be assumed forthe FCDPMWL combination as well The potential extended operational pe-riod makes the proposed configuration particularly attractive to metrologicallaboratories that rely on availability of long-lasting data in support of thegeneration of UTC

The resources of MWL are as follows mass 18 kg volume 18 l power55 W TC 2 kBday TM 120 MBday

426 Auxiliary units

The star camera assembly on the EGE spacecraft will provide real-timeattitude data primarily for the orientation of the spacecraft (with accuracy of1 in real-time) and the proper reduction of the GNSSSLR data

A GNSS receiver is used for the precise orbit determination time andfrequency transfer coarse positioning (lt1 m) for real time use by the Attitudeand Orbit Control as well as the back-up for time tagging of all payloaddata For the time and frequency transfer the receiver is nominally using theonboard frequency reference The receiver assembly includes three specialantennas By the year 2015 more than 100 GNSS satellites are expected tobe available for precise orbit determination permitting orbit inaccuracy of 1ndash2 cm RMS for altitudes up to approx 10000 km At apogee a combinationof GNSS signals and MWL ranging will be used with expected inaccuracyof 10 cm

5 Basic spacecraft specifications

The mission foresees a single spacecraft with a mission duration limited toapprox 3 years for cost reasons Attitude control along three axes withless than 1 error on a timescale of 1000 s is required The control can beperformed with cold gas thrusters magnetotorquers or reaction wheels Theselection will be done on the basis of the vibrations induced by these systems(desired level arms lt 10minus6 g)

Cold gas thrusters or monopropellant thrusters will be required to maintainthe orbit parameters during the mission by small orbit correction manoeuvresand will also guarantee a proper tuning of the orbit parameters after the releasefrom the launcher The ground segment (telemetry control and data transfer)could be via non-ESA stations user-operated to save costs

Further resources are indicated in Table 2

Exp Astron

The satellite bus design drivers are nadir-pointing microwave antennasthermal stability for the payload instruments no strong magnetic sources lowmicrovibrations harsh radiation environment due to crossing of the radiationbelts once per orbit power demand by science payload (with the possibility toreduce power demand by stand-by of some components during specific missionphases eg eclipse) mass limit due to required high altitude with low costlaunch

The satellite bus design can be based on established platforms such asFLEXBUS or PRIMA used on several past missions Their design philosophydiffers one having body-mounted solar arrays the other wings The latter canprovide larger amounts of power The vibration level induced by the panelswhen they are fixed and the duration of solar array drive periods per orbitwill have to be considered in the selection The FLEXBUS concept was usedfor a mission already requiring a low level of residual accelerations and theplatform is optimised for this parameter

For most of the satellite components one can consider using compo-nents and electronic units from navigation or communication satellites whichtypically cruise at altitudes between 15000 and 25000 km altitude LISAPathfinder which will cross the radiation belts during spiralling to theLagrange point will face similar problems concerning radiation as EGE as wellas XMM which is in a Molniya-type orbit with similar radiation characteristics

The instruments will therefore be located in the satellite structure such asto realise ionizing radiation shielding by the spacecraft structure and non-sensitive equipment to the maximum extent possible the need for additionalradiation shielding (including trapped proton radiation) for especially sensitiveparts of the instruments must be determined by ground tests It should bementioned that an orbit with higher perigee and less radiation dose is also anoption as it leads to a tolerable reduction of the science goals

The required plusmn3C temperature stability at the interface to the payloadenclosure is feasible The thermal control will employ the usual means (anenclosure with high thermal inertia heaters and a suitable radiator) Thelocation of the radiator will need a special study given the non-standard orbit

Magnetic field present during the mission can affect the properties of theonboard clock Using SPENVIS the magnetic field has been calculated alongthe spacecraft orbit the field intensity presents maxima around 02 G anda minimum value of the order of 10minus3 G As discussed above this is not alimitation to the instrument performance

6 Discussion and conclusion

The accuracy of the measurements depends crucially on the stability andaccuracy of the clocks While the performance of ground clocks (even thoseeventually deployed in new ground stations dedicated to EGE) will likelyimprove significantly from the present level the space clock design will beldquofrozenrdquo approx 10 years before data acquisition Thus their performance

Exp Astron

level will be significantly below the future ground station clock performanceAs specification we have chosen a performance level that has been demon-strated with laboratory breadboards recently We see no fundamental reasonthat this laboratory performance cannot also be achieved in a space ion opticalclock The experience with the PHARAO cold atom clock has shown thatan engineering model of a laser-based cold atom clock has a performancesimilar to that of a laboratory prototype The perturbation of the ion clock fre-quency due to magnetic fields along the orbit can be modelled and suppressedwith shielding and the residual effects measured and compensated for Themoderate temperature stability (plusmn 3C) to be provided in the experimentalbay can be achieved with thermal shields Accurate temperature controlof particular units is done by active stabilization of the units themselves

One important aspect is radiation which may affect all optical and trapcomponents as well as the lasers and reference cavity The radiation envi-ronment on the EGE orbit is significantly harsher than the near-earth ISSorbit on which ACES will operate The radiation effects must therefore becharacterized comprehensively with appropriate tests and then the radiationshield for the EGE instruments designed

In summary the Einstein Gravity Explorer mission (EGE) is a satellitemission aimed at a precise measurement of the properties of space-time usingatomic clocks By testing one of the most fundamental predictions of EinsteinrsquosTheory of General Relativity the gravitational redshift and thereby searchingfor quantum effects in gravity it explores one of the most important frontiers infundamental physics The EGE mission reaches beyond fundamental physicsproviding an important spin-off in geophysics Furthermore the technologicalaspects of the mission are ground-breaking for future time-keeping and ref-erence frame systems in space and thus are also expected to be of wide useto science and technology in general The payload of the mission has beenconceived as a compromise between cost and performance needing to fit thecost envelope of an ESA medium-class mission1

Acknowledgement We thank S Cesare for helpful information

References

1 Will CM The confrontation between general relativity and experiment Living Rev Relativ9 3 (2006)

2 Ni W-T Empirical foundations of the relativistic gravity Int J Mod Phys D 14 901 (2005)doi101142S0218271805007139

1Finally we point out that a spacecraft with a significantly reduced and thus lower-cost payload (noclocks only MOLO FCDP MWL) on a circular orbit of appropriate height could be used for along-duration earth science mission dedicated specifically to the measurement of the geopotential

Exp Astron

871 nm clock laser

EO

Detector and servo

AOM Femtosecond

frequency comb

To optical clock 10 GHz To FCDP

Microwave-optical local oscillator

ULE cavity

Fig 8 Microwave-optical local oscillator (MOLO) consisting of clock laser reference ULE cavityand femtosecond frequency comb (OFC) Radiofrequency inputs and outputs are not shown EOelectro-optic modulator ULE ultra-low expansion AOM frequency shifter

cavity where the PoundndashDreverndashHall method is used About 10 μW of thelaser output are used for this

The system will exhibit sim04 Hz (sim1 times 10minus15 relative) linewidth and afrequency drift lt01 Hzs The cavity consists of two high-reflectivity mirrorsoptically contacted to a cylinder of diameter 70 mm length 150 mm made ofULE (a glass exhibiting ultra low thermal expansion coefficient) with a cavityfinesse of about 300000 Residual vibrations of the satellite (arms lt 1 times 10minus6 g)should not be a limiting factor to the desired linewidth for safety marginan optimized cavity support can reduce the sensitivity to accelerations Foracoustic and thermal isolation the cavity resides inside an aluminium vacuumchamber equipped with a small (3 ls) ion-getter pump Two stages of polishedaluminium shields are implemented around the cavity which are activelytemperature stabilized so that the cavity is precisely kept at the zero crossingtemperature of the ULE thermal expansion coefficient Heat application andremoval is by thermoelectric elements between the shield and the vacuumchamber Additionally the entire vacuum chamber is temperature-stabilized

During transportation and launch the cavity will be rigidly fixed insidethe vacuum chamber by pressing on it from the sides by piezo-mechanicalactuators allowing accelerations of several g After bringing the satellite onthe orbit the actuators will be released

The optical setup for frequency stabilization will be a classic Pound-Drever-Hall scheme in reflection using an electro-optic modulator (EO) Couplingof the clock laser radiation to the cavity is via single-mode polarization main-taining optical fiber The fiber in- and out-couplers are mounted on miniaturepiezo-motor driven multiaxis translation stages Microprocessor control ofthe translation stages performs a laser-to-fiber incoupling stabilization andlaser-to-cavity mode-matching optimization (positioning inaccuracy lt10 μmalignment inaccuracy lt10 μrad)

The frequency stability of the clock laser will be limited by thermal 1f noisein the mirrors with an fractional Allan deviation slightly below 1 times 10minus15 from1 s to a few 10 s Resources including control electronics are 15 l 12 kg 8 W

Exp Astron

A femtosecond fiber laser based optical frequency comb generator (OFCFig 8) will be used for the clock frequency comparisons and to obtain aspectrally ultrapure microwave local oscillator output for the operation ofPHARAO for the microwave link and for the GNSS receiver The fem-tosecond laser is based on erbium-doped fiber pumped by fiber-coupled laserdiodes It emits 100 fs pulses at a repetition rate of 250 MHz with an averagepower of 100 mW The emission spectrum is centered around 1560 nm Spec-tral broadening to more than one octave is obtained via nonlinear interactionin a microstructure fibre so that the range from about 1050 to 2100 nm iscovered

Second harmonic generation of a portion of the broadened spectrum allowsobtaining the required light for the measurement of the clock laser frequencyat 871 nm The OFC will be referenced to the clock laser frequency (afterintroduction a controllable shift via the frequency shifter AOM) by makingthe carrier envelope offset frequency vanish so that the repetition rate fr isa high-order subharmonic of the optical clock reference frequency f1 = m1 frThe link between the optical and the microwave domain is then simply thedistribution of the repetition rate frequency or more conveniently a high(order 40) harmonic of the comb repetition rate 10 GHz

It has been demonstrated by several groups that frequency comparison andfrequency division with an OFC can be performed to very high accuracy andstability beyond the limitations imposed by the optical clocks themselves Forexample a fiber laser OFC with an instability (Allan deviation) of 6 times 10minus17

in 1 s has been demonstrated [51] OFCs based on fiber technology possessgood long-term reliability and high power efficiency The projection for a spacequalified OFC based on current industrial terrestrial technology is 16 kg mass20 l volume and 38 W power requirement

424 Frequency comparison and distribution package (FCDP)

This package distributes the ultra-stable 10 GHz reference frequency outputof the MOLO to the MWL performs frequency comparisons between MOLOand PHARAO and generates a variety of different reference signals for theindividual instruments including the 250 MHz reference clock to the OFCAll signals are coherent to the MOLO In the FCDP a USO is locked to theMOLO signal by a digital PLL A divider chain generates timing pulses (1 pps1000 pps etc) and keeps a continuous local time scale From the digital phasecomparison data the phase noise density spectrum between the USO and theMOLO is available by telemetry (TM) to assess the overall frequency purityin-flight

As the FCDP distributes the full performance reference from the MOLOdirectly to the MWL there is minimum impact of the FCDP on overall clockcomparison performance The USO-phase lock loop uses a similar designas for ACES-FCDP The digital PLL with linear phase detector allows theUSO to be slightly offset from its reference which is selectable by groundcommand This results in an accurate on-board time scale despite any offset

Exp Astron

of the MOLO reference frequency which can be synchronised to UTC Thefrequency synthesis chain re-uses modules developed for ACES-FCDP andtheir frequency inaccuracy Expected relative uncertainty (Allan deviation)of the frequency comparison between two signals provided to the FCDP is47 times 10minus17 at 1000 s 95 times 10minus18 at 10000 s

During non-availability of the MOLO ie during system start-up andcontingency conditions the USO runs as flywheel (ie it is unlocked withinstability lt15 times 10minus13 from 1 to 100 s) It will still provide useable butless accurate references to the instruments including the reference to OFCCommon view ground-to-ground time- and frequency comparison and supplyof the remaining instruments remain possible at the USO provided accuracy

Mass volume and power consumption are 8 kg 7 l 12 W Controldata rateis 10 kBday 50 kBday (FCDP alone)

425 Microwave link (MWL)

The microwave link is a two-way triple-frequency microwave link (Fig 9) Itis designed to perform accurate frequency comparison as well as time-transfer

Ku-Band Up-linkPower Tx 5 WCarrier 13475 GHzPN-Code 100 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

S-Band Down-linkPower Tx 1 WCarrier 2248 MHzPN-Code 1 MChips1pps 1 time marker sData 5 kBits

Ku-Band Down-linkPower Tx 1 WCarrier 1470333 GHzPN-Code 100 MChips1pps 1 time marker sData 5 kBits

EGE SC

S

Ku

KaKa-Band Up-linkPower Tx 5 WCarrier 345 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

Ka-Band Down-LinkPower Tx 1 WCarrier 32 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 5 kBits

GS Antenna diameterEGE terminal 1 mACES terminal 06 m

Fig 9 EGE microwave two-way link concept and characteristics See text for a description Thelink will be able to perform frequency comparisons with up to ten ground stations simultaneously(only one active comparison is shown) pps pulse per second Tx transmitter SC spacecraft

Exp Astron

between on-board and on-ground clocks Two-way operation allows for com-pensation of the effects due to spacecraft location and its motion relative toground Its two bi-directional links operate in Ka-band (32345 GHz) andKu-band (147135 GHz) The high carrier frequency of the up- and down-links allows for a noticeable reduction of the ionospheric delay A third down-link channel is in S-band (22 GHz) and is used to determine the ionospheretotal electron content MWL uses a combination of carrier phase and codephase data to measure with high precision the absolute ionospheric delayas well as ionospheric delay variations during a ground contact The useof three frequency bands allows to resolve third order ionospheric termsKey modulation parameters are Ka band Tone 250 MHz Ku band PN100 MChips S-band 1 MChips

The EGE-MWL concept is a considerably improved version of the ACES-MWL Microwave carrier generation is directly from the 10 GHz microwaveoutput available from the MOLO and having exceptionally low close-in phasenoise without intermediate frequency dividers or other complex synthesiserstages commonly used in RF transponders The high frequency carriers arekey to achieve the desired frequency comparison accuracy

The Ka-band carrier is generated by tripling the FCDP-generated signalThe Ku-band is subsequently generated by division by two using a verystable regenerative divider Finally the assigned radio frequency is obtainedby adding a low frequency signal using an image reject mixer A mixedanalog-digital receiver design will enhance its performance by use of a flexibledigital signal processing and digital control loops The number of simultaneousground contacts is increased from four (as in the ACES mission) to ten or moreusing the digital concept while the number of physical receiver channels perband is reduced from four to two one for operation to ground and one for in-flight calibration The concept allows adaptation of receiver loop parametersto optimise for the different portions of a highly elliptic orbit while reducingsystematic uncertainties due to high Doppler and Doppler rate Real-timedata exchange to ground terminals is by additional data modulation on all RFsignals The spacecraft carries two different small antennae for perigee and forapogee with gains of 8 dBi 20 dBi beamwidths 100 (full earth coverage) 14respectively power is 1 W for each

The MWL end-to-end performance between the attached clocks includingspace- and ground terminals is predicted to be a relative Allan deviation of1 times 10minus16 at 1000 s (perigee) and 1 times 10minus17 at 10000 s (apogee) Thesevalues are based on conservative implementation losses and other instrumentallosses assumed as 6 dB at apogee and 10 dB at perigee including additionallosses due to atmosphere An attitude knowledge of 1 is required

The MWL link will also support precise orbitographyndashcomplementary tothe GNSS receiver Generation of ranging data is simultaneous to two-waytime transfer and is available under all weather conditions Orbit determina-tion accuracy is expected to be better than 1 cm well matching EGE clockrequirements The value is expected based on demonstrated results fromother missions (METEOR ERS-2 TOPEXPOSEIDON) Laser ranging to

Exp Astron

the CCR array is used a few times during the mission to calibrate the MWLranging

In case of non-availability of the ultrapure signal from MOLO an additional10 GHz output is available from the FCDP to support ground-to-ground clockcomparison in common-view mode even after the nominal mission periodwhen the neither the clocks nor MOLO may be available any longer Basedon the PRARE experimentrsquos 12 years of operation this can be assumed forthe FCDPMWL combination as well The potential extended operational pe-riod makes the proposed configuration particularly attractive to metrologicallaboratories that rely on availability of long-lasting data in support of thegeneration of UTC

The resources of MWL are as follows mass 18 kg volume 18 l power55 W TC 2 kBday TM 120 MBday

426 Auxiliary units

The star camera assembly on the EGE spacecraft will provide real-timeattitude data primarily for the orientation of the spacecraft (with accuracy of1 in real-time) and the proper reduction of the GNSSSLR data

A GNSS receiver is used for the precise orbit determination time andfrequency transfer coarse positioning (lt1 m) for real time use by the Attitudeand Orbit Control as well as the back-up for time tagging of all payloaddata For the time and frequency transfer the receiver is nominally using theonboard frequency reference The receiver assembly includes three specialantennas By the year 2015 more than 100 GNSS satellites are expected tobe available for precise orbit determination permitting orbit inaccuracy of 1ndash2 cm RMS for altitudes up to approx 10000 km At apogee a combinationof GNSS signals and MWL ranging will be used with expected inaccuracyof 10 cm

5 Basic spacecraft specifications

The mission foresees a single spacecraft with a mission duration limited toapprox 3 years for cost reasons Attitude control along three axes withless than 1 error on a timescale of 1000 s is required The control can beperformed with cold gas thrusters magnetotorquers or reaction wheels Theselection will be done on the basis of the vibrations induced by these systems(desired level arms lt 10minus6 g)

Cold gas thrusters or monopropellant thrusters will be required to maintainthe orbit parameters during the mission by small orbit correction manoeuvresand will also guarantee a proper tuning of the orbit parameters after the releasefrom the launcher The ground segment (telemetry control and data transfer)could be via non-ESA stations user-operated to save costs

Further resources are indicated in Table 2

Exp Astron

The satellite bus design drivers are nadir-pointing microwave antennasthermal stability for the payload instruments no strong magnetic sources lowmicrovibrations harsh radiation environment due to crossing of the radiationbelts once per orbit power demand by science payload (with the possibility toreduce power demand by stand-by of some components during specific missionphases eg eclipse) mass limit due to required high altitude with low costlaunch

The satellite bus design can be based on established platforms such asFLEXBUS or PRIMA used on several past missions Their design philosophydiffers one having body-mounted solar arrays the other wings The latter canprovide larger amounts of power The vibration level induced by the panelswhen they are fixed and the duration of solar array drive periods per orbitwill have to be considered in the selection The FLEXBUS concept was usedfor a mission already requiring a low level of residual accelerations and theplatform is optimised for this parameter

For most of the satellite components one can consider using compo-nents and electronic units from navigation or communication satellites whichtypically cruise at altitudes between 15000 and 25000 km altitude LISAPathfinder which will cross the radiation belts during spiralling to theLagrange point will face similar problems concerning radiation as EGE as wellas XMM which is in a Molniya-type orbit with similar radiation characteristics

The instruments will therefore be located in the satellite structure such asto realise ionizing radiation shielding by the spacecraft structure and non-sensitive equipment to the maximum extent possible the need for additionalradiation shielding (including trapped proton radiation) for especially sensitiveparts of the instruments must be determined by ground tests It should bementioned that an orbit with higher perigee and less radiation dose is also anoption as it leads to a tolerable reduction of the science goals

The required plusmn3C temperature stability at the interface to the payloadenclosure is feasible The thermal control will employ the usual means (anenclosure with high thermal inertia heaters and a suitable radiator) Thelocation of the radiator will need a special study given the non-standard orbit

Magnetic field present during the mission can affect the properties of theonboard clock Using SPENVIS the magnetic field has been calculated alongthe spacecraft orbit the field intensity presents maxima around 02 G anda minimum value of the order of 10minus3 G As discussed above this is not alimitation to the instrument performance

6 Discussion and conclusion

The accuracy of the measurements depends crucially on the stability andaccuracy of the clocks While the performance of ground clocks (even thoseeventually deployed in new ground stations dedicated to EGE) will likelyimprove significantly from the present level the space clock design will beldquofrozenrdquo approx 10 years before data acquisition Thus their performance

Exp Astron

level will be significantly below the future ground station clock performanceAs specification we have chosen a performance level that has been demon-strated with laboratory breadboards recently We see no fundamental reasonthat this laboratory performance cannot also be achieved in a space ion opticalclock The experience with the PHARAO cold atom clock has shown thatan engineering model of a laser-based cold atom clock has a performancesimilar to that of a laboratory prototype The perturbation of the ion clock fre-quency due to magnetic fields along the orbit can be modelled and suppressedwith shielding and the residual effects measured and compensated for Themoderate temperature stability (plusmn 3C) to be provided in the experimentalbay can be achieved with thermal shields Accurate temperature controlof particular units is done by active stabilization of the units themselves

One important aspect is radiation which may affect all optical and trapcomponents as well as the lasers and reference cavity The radiation envi-ronment on the EGE orbit is significantly harsher than the near-earth ISSorbit on which ACES will operate The radiation effects must therefore becharacterized comprehensively with appropriate tests and then the radiationshield for the EGE instruments designed

In summary the Einstein Gravity Explorer mission (EGE) is a satellitemission aimed at a precise measurement of the properties of space-time usingatomic clocks By testing one of the most fundamental predictions of EinsteinrsquosTheory of General Relativity the gravitational redshift and thereby searchingfor quantum effects in gravity it explores one of the most important frontiers infundamental physics The EGE mission reaches beyond fundamental physicsproviding an important spin-off in geophysics Furthermore the technologicalaspects of the mission are ground-breaking for future time-keeping and ref-erence frame systems in space and thus are also expected to be of wide useto science and technology in general The payload of the mission has beenconceived as a compromise between cost and performance needing to fit thecost envelope of an ESA medium-class mission1

Acknowledgement We thank S Cesare for helpful information

References

1 Will CM The confrontation between general relativity and experiment Living Rev Relativ9 3 (2006)

2 Ni W-T Empirical foundations of the relativistic gravity Int J Mod Phys D 14 901 (2005)doi101142S0218271805007139

1Finally we point out that a spacecraft with a significantly reduced and thus lower-cost payload (noclocks only MOLO FCDP MWL) on a circular orbit of appropriate height could be used for along-duration earth science mission dedicated specifically to the measurement of the geopotential

Exp Astron

A femtosecond fiber laser based optical frequency comb generator (OFCFig 8) will be used for the clock frequency comparisons and to obtain aspectrally ultrapure microwave local oscillator output for the operation ofPHARAO for the microwave link and for the GNSS receiver The fem-tosecond laser is based on erbium-doped fiber pumped by fiber-coupled laserdiodes It emits 100 fs pulses at a repetition rate of 250 MHz with an averagepower of 100 mW The emission spectrum is centered around 1560 nm Spec-tral broadening to more than one octave is obtained via nonlinear interactionin a microstructure fibre so that the range from about 1050 to 2100 nm iscovered

Second harmonic generation of a portion of the broadened spectrum allowsobtaining the required light for the measurement of the clock laser frequencyat 871 nm The OFC will be referenced to the clock laser frequency (afterintroduction a controllable shift via the frequency shifter AOM) by makingthe carrier envelope offset frequency vanish so that the repetition rate fr isa high-order subharmonic of the optical clock reference frequency f1 = m1 frThe link between the optical and the microwave domain is then simply thedistribution of the repetition rate frequency or more conveniently a high(order 40) harmonic of the comb repetition rate 10 GHz

It has been demonstrated by several groups that frequency comparison andfrequency division with an OFC can be performed to very high accuracy andstability beyond the limitations imposed by the optical clocks themselves Forexample a fiber laser OFC with an instability (Allan deviation) of 6 times 10minus17

in 1 s has been demonstrated [51] OFCs based on fiber technology possessgood long-term reliability and high power efficiency The projection for a spacequalified OFC based on current industrial terrestrial technology is 16 kg mass20 l volume and 38 W power requirement

424 Frequency comparison and distribution package (FCDP)

This package distributes the ultra-stable 10 GHz reference frequency outputof the MOLO to the MWL performs frequency comparisons between MOLOand PHARAO and generates a variety of different reference signals for theindividual instruments including the 250 MHz reference clock to the OFCAll signals are coherent to the MOLO In the FCDP a USO is locked to theMOLO signal by a digital PLL A divider chain generates timing pulses (1 pps1000 pps etc) and keeps a continuous local time scale From the digital phasecomparison data the phase noise density spectrum between the USO and theMOLO is available by telemetry (TM) to assess the overall frequency purityin-flight

As the FCDP distributes the full performance reference from the MOLOdirectly to the MWL there is minimum impact of the FCDP on overall clockcomparison performance The USO-phase lock loop uses a similar designas for ACES-FCDP The digital PLL with linear phase detector allows theUSO to be slightly offset from its reference which is selectable by groundcommand This results in an accurate on-board time scale despite any offset

Exp Astron

of the MOLO reference frequency which can be synchronised to UTC Thefrequency synthesis chain re-uses modules developed for ACES-FCDP andtheir frequency inaccuracy Expected relative uncertainty (Allan deviation)of the frequency comparison between two signals provided to the FCDP is47 times 10minus17 at 1000 s 95 times 10minus18 at 10000 s

During non-availability of the MOLO ie during system start-up andcontingency conditions the USO runs as flywheel (ie it is unlocked withinstability lt15 times 10minus13 from 1 to 100 s) It will still provide useable butless accurate references to the instruments including the reference to OFCCommon view ground-to-ground time- and frequency comparison and supplyof the remaining instruments remain possible at the USO provided accuracy

Mass volume and power consumption are 8 kg 7 l 12 W Controldata rateis 10 kBday 50 kBday (FCDP alone)

425 Microwave link (MWL)

The microwave link is a two-way triple-frequency microwave link (Fig 9) Itis designed to perform accurate frequency comparison as well as time-transfer

Ku-Band Up-linkPower Tx 5 WCarrier 13475 GHzPN-Code 100 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

S-Band Down-linkPower Tx 1 WCarrier 2248 MHzPN-Code 1 MChips1pps 1 time marker sData 5 kBits

Ku-Band Down-linkPower Tx 1 WCarrier 1470333 GHzPN-Code 100 MChips1pps 1 time marker sData 5 kBits

EGE SC

S

Ku

KaKa-Band Up-linkPower Tx 5 WCarrier 345 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

Ka-Band Down-LinkPower Tx 1 WCarrier 32 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 5 kBits

GS Antenna diameterEGE terminal 1 mACES terminal 06 m

Fig 9 EGE microwave two-way link concept and characteristics See text for a description Thelink will be able to perform frequency comparisons with up to ten ground stations simultaneously(only one active comparison is shown) pps pulse per second Tx transmitter SC spacecraft

Exp Astron

between on-board and on-ground clocks Two-way operation allows for com-pensation of the effects due to spacecraft location and its motion relative toground Its two bi-directional links operate in Ka-band (32345 GHz) andKu-band (147135 GHz) The high carrier frequency of the up- and down-links allows for a noticeable reduction of the ionospheric delay A third down-link channel is in S-band (22 GHz) and is used to determine the ionospheretotal electron content MWL uses a combination of carrier phase and codephase data to measure with high precision the absolute ionospheric delayas well as ionospheric delay variations during a ground contact The useof three frequency bands allows to resolve third order ionospheric termsKey modulation parameters are Ka band Tone 250 MHz Ku band PN100 MChips S-band 1 MChips

The EGE-MWL concept is a considerably improved version of the ACES-MWL Microwave carrier generation is directly from the 10 GHz microwaveoutput available from the MOLO and having exceptionally low close-in phasenoise without intermediate frequency dividers or other complex synthesiserstages commonly used in RF transponders The high frequency carriers arekey to achieve the desired frequency comparison accuracy

The Ka-band carrier is generated by tripling the FCDP-generated signalThe Ku-band is subsequently generated by division by two using a verystable regenerative divider Finally the assigned radio frequency is obtainedby adding a low frequency signal using an image reject mixer A mixedanalog-digital receiver design will enhance its performance by use of a flexibledigital signal processing and digital control loops The number of simultaneousground contacts is increased from four (as in the ACES mission) to ten or moreusing the digital concept while the number of physical receiver channels perband is reduced from four to two one for operation to ground and one for in-flight calibration The concept allows adaptation of receiver loop parametersto optimise for the different portions of a highly elliptic orbit while reducingsystematic uncertainties due to high Doppler and Doppler rate Real-timedata exchange to ground terminals is by additional data modulation on all RFsignals The spacecraft carries two different small antennae for perigee and forapogee with gains of 8 dBi 20 dBi beamwidths 100 (full earth coverage) 14respectively power is 1 W for each

The MWL end-to-end performance between the attached clocks includingspace- and ground terminals is predicted to be a relative Allan deviation of1 times 10minus16 at 1000 s (perigee) and 1 times 10minus17 at 10000 s (apogee) Thesevalues are based on conservative implementation losses and other instrumentallosses assumed as 6 dB at apogee and 10 dB at perigee including additionallosses due to atmosphere An attitude knowledge of 1 is required

The MWL link will also support precise orbitographyndashcomplementary tothe GNSS receiver Generation of ranging data is simultaneous to two-waytime transfer and is available under all weather conditions Orbit determina-tion accuracy is expected to be better than 1 cm well matching EGE clockrequirements The value is expected based on demonstrated results fromother missions (METEOR ERS-2 TOPEXPOSEIDON) Laser ranging to

Exp Astron

the CCR array is used a few times during the mission to calibrate the MWLranging

In case of non-availability of the ultrapure signal from MOLO an additional10 GHz output is available from the FCDP to support ground-to-ground clockcomparison in common-view mode even after the nominal mission periodwhen the neither the clocks nor MOLO may be available any longer Basedon the PRARE experimentrsquos 12 years of operation this can be assumed forthe FCDPMWL combination as well The potential extended operational pe-riod makes the proposed configuration particularly attractive to metrologicallaboratories that rely on availability of long-lasting data in support of thegeneration of UTC

The resources of MWL are as follows mass 18 kg volume 18 l power55 W TC 2 kBday TM 120 MBday

426 Auxiliary units

The star camera assembly on the EGE spacecraft will provide real-timeattitude data primarily for the orientation of the spacecraft (with accuracy of1 in real-time) and the proper reduction of the GNSSSLR data

A GNSS receiver is used for the precise orbit determination time andfrequency transfer coarse positioning (lt1 m) for real time use by the Attitudeand Orbit Control as well as the back-up for time tagging of all payloaddata For the time and frequency transfer the receiver is nominally using theonboard frequency reference The receiver assembly includes three specialantennas By the year 2015 more than 100 GNSS satellites are expected tobe available for precise orbit determination permitting orbit inaccuracy of 1ndash2 cm RMS for altitudes up to approx 10000 km At apogee a combinationof GNSS signals and MWL ranging will be used with expected inaccuracyof 10 cm

5 Basic spacecraft specifications

The mission foresees a single spacecraft with a mission duration limited toapprox 3 years for cost reasons Attitude control along three axes withless than 1 error on a timescale of 1000 s is required The control can beperformed with cold gas thrusters magnetotorquers or reaction wheels Theselection will be done on the basis of the vibrations induced by these systems(desired level arms lt 10minus6 g)

Cold gas thrusters or monopropellant thrusters will be required to maintainthe orbit parameters during the mission by small orbit correction manoeuvresand will also guarantee a proper tuning of the orbit parameters after the releasefrom the launcher The ground segment (telemetry control and data transfer)could be via non-ESA stations user-operated to save costs

Further resources are indicated in Table 2

Exp Astron

The satellite bus design drivers are nadir-pointing microwave antennasthermal stability for the payload instruments no strong magnetic sources lowmicrovibrations harsh radiation environment due to crossing of the radiationbelts once per orbit power demand by science payload (with the possibility toreduce power demand by stand-by of some components during specific missionphases eg eclipse) mass limit due to required high altitude with low costlaunch

The satellite bus design can be based on established platforms such asFLEXBUS or PRIMA used on several past missions Their design philosophydiffers one having body-mounted solar arrays the other wings The latter canprovide larger amounts of power The vibration level induced by the panelswhen they are fixed and the duration of solar array drive periods per orbitwill have to be considered in the selection The FLEXBUS concept was usedfor a mission already requiring a low level of residual accelerations and theplatform is optimised for this parameter

For most of the satellite components one can consider using compo-nents and electronic units from navigation or communication satellites whichtypically cruise at altitudes between 15000 and 25000 km altitude LISAPathfinder which will cross the radiation belts during spiralling to theLagrange point will face similar problems concerning radiation as EGE as wellas XMM which is in a Molniya-type orbit with similar radiation characteristics

The instruments will therefore be located in the satellite structure such asto realise ionizing radiation shielding by the spacecraft structure and non-sensitive equipment to the maximum extent possible the need for additionalradiation shielding (including trapped proton radiation) for especially sensitiveparts of the instruments must be determined by ground tests It should bementioned that an orbit with higher perigee and less radiation dose is also anoption as it leads to a tolerable reduction of the science goals

The required plusmn3C temperature stability at the interface to the payloadenclosure is feasible The thermal control will employ the usual means (anenclosure with high thermal inertia heaters and a suitable radiator) Thelocation of the radiator will need a special study given the non-standard orbit

Magnetic field present during the mission can affect the properties of theonboard clock Using SPENVIS the magnetic field has been calculated alongthe spacecraft orbit the field intensity presents maxima around 02 G anda minimum value of the order of 10minus3 G As discussed above this is not alimitation to the instrument performance

6 Discussion and conclusion

The accuracy of the measurements depends crucially on the stability andaccuracy of the clocks While the performance of ground clocks (even thoseeventually deployed in new ground stations dedicated to EGE) will likelyimprove significantly from the present level the space clock design will beldquofrozenrdquo approx 10 years before data acquisition Thus their performance

Exp Astron

level will be significantly below the future ground station clock performanceAs specification we have chosen a performance level that has been demon-strated with laboratory breadboards recently We see no fundamental reasonthat this laboratory performance cannot also be achieved in a space ion opticalclock The experience with the PHARAO cold atom clock has shown thatan engineering model of a laser-based cold atom clock has a performancesimilar to that of a laboratory prototype The perturbation of the ion clock fre-quency due to magnetic fields along the orbit can be modelled and suppressedwith shielding and the residual effects measured and compensated for Themoderate temperature stability (plusmn 3C) to be provided in the experimentalbay can be achieved with thermal shields Accurate temperature controlof particular units is done by active stabilization of the units themselves

One important aspect is radiation which may affect all optical and trapcomponents as well as the lasers and reference cavity The radiation envi-ronment on the EGE orbit is significantly harsher than the near-earth ISSorbit on which ACES will operate The radiation effects must therefore becharacterized comprehensively with appropriate tests and then the radiationshield for the EGE instruments designed

In summary the Einstein Gravity Explorer mission (EGE) is a satellitemission aimed at a precise measurement of the properties of space-time usingatomic clocks By testing one of the most fundamental predictions of EinsteinrsquosTheory of General Relativity the gravitational redshift and thereby searchingfor quantum effects in gravity it explores one of the most important frontiers infundamental physics The EGE mission reaches beyond fundamental physicsproviding an important spin-off in geophysics Furthermore the technologicalaspects of the mission are ground-breaking for future time-keeping and ref-erence frame systems in space and thus are also expected to be of wide useto science and technology in general The payload of the mission has beenconceived as a compromise between cost and performance needing to fit thecost envelope of an ESA medium-class mission1

Acknowledgement We thank S Cesare for helpful information

References

1 Will CM The confrontation between general relativity and experiment Living Rev Relativ9 3 (2006)

2 Ni W-T Empirical foundations of the relativistic gravity Int J Mod Phys D 14 901 (2005)doi101142S0218271805007139

1Finally we point out that a spacecraft with a significantly reduced and thus lower-cost payload (noclocks only MOLO FCDP MWL) on a circular orbit of appropriate height could be used for along-duration earth science mission dedicated specifically to the measurement of the geopotential

Exp Astron

of the MOLO reference frequency which can be synchronised to UTC Thefrequency synthesis chain re-uses modules developed for ACES-FCDP andtheir frequency inaccuracy Expected relative uncertainty (Allan deviation)of the frequency comparison between two signals provided to the FCDP is47 times 10minus17 at 1000 s 95 times 10minus18 at 10000 s

During non-availability of the MOLO ie during system start-up andcontingency conditions the USO runs as flywheel (ie it is unlocked withinstability lt15 times 10minus13 from 1 to 100 s) It will still provide useable butless accurate references to the instruments including the reference to OFCCommon view ground-to-ground time- and frequency comparison and supplyof the remaining instruments remain possible at the USO provided accuracy

Mass volume and power consumption are 8 kg 7 l 12 W Controldata rateis 10 kBday 50 kBday (FCDP alone)

425 Microwave link (MWL)

The microwave link is a two-way triple-frequency microwave link (Fig 9) Itis designed to perform accurate frequency comparison as well as time-transfer

Ku-Band Up-linkPower Tx 5 WCarrier 13475 GHzPN-Code 100 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

S-Band Down-linkPower Tx 1 WCarrier 2248 MHzPN-Code 1 MChips1pps 1 time marker sData 5 kBits

Ku-Band Down-linkPower Tx 1 WCarrier 1470333 GHzPN-Code 100 MChips1pps 1 time marker sData 5 kBits

EGE SC

S

Ku

KaKa-Band Up-linkPower Tx 5 WCarrier 345 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 1 kBitsChannelSC 10 Rx Channels

Ka-Band Down-LinkPower Tx 1 WCarrier 32 GHzTone 250 MHzPN-Code 20 MChips1pps 1 time marker sData 5 kBits

GS Antenna diameterEGE terminal 1 mACES terminal 06 m

Fig 9 EGE microwave two-way link concept and characteristics See text for a description Thelink will be able to perform frequency comparisons with up to ten ground stations simultaneously(only one active comparison is shown) pps pulse per second Tx transmitter SC spacecraft

Exp Astron

between on-board and on-ground clocks Two-way operation allows for com-pensation of the effects due to spacecraft location and its motion relative toground Its two bi-directional links operate in Ka-band (32345 GHz) andKu-band (147135 GHz) The high carrier frequency of the up- and down-links allows for a noticeable reduction of the ionospheric delay A third down-link channel is in S-band (22 GHz) and is used to determine the ionospheretotal electron content MWL uses a combination of carrier phase and codephase data to measure with high precision the absolute ionospheric delayas well as ionospheric delay variations during a ground contact The useof three frequency bands allows to resolve third order ionospheric termsKey modulation parameters are Ka band Tone 250 MHz Ku band PN100 MChips S-band 1 MChips

The EGE-MWL concept is a considerably improved version of the ACES-MWL Microwave carrier generation is directly from the 10 GHz microwaveoutput available from the MOLO and having exceptionally low close-in phasenoise without intermediate frequency dividers or other complex synthesiserstages commonly used in RF transponders The high frequency carriers arekey to achieve the desired frequency comparison accuracy

The Ka-band carrier is generated by tripling the FCDP-generated signalThe Ku-band is subsequently generated by division by two using a verystable regenerative divider Finally the assigned radio frequency is obtainedby adding a low frequency signal using an image reject mixer A mixedanalog-digital receiver design will enhance its performance by use of a flexibledigital signal processing and digital control loops The number of simultaneousground contacts is increased from four (as in the ACES mission) to ten or moreusing the digital concept while the number of physical receiver channels perband is reduced from four to two one for operation to ground and one for in-flight calibration The concept allows adaptation of receiver loop parametersto optimise for the different portions of a highly elliptic orbit while reducingsystematic uncertainties due to high Doppler and Doppler rate Real-timedata exchange to ground terminals is by additional data modulation on all RFsignals The spacecraft carries two different small antennae for perigee and forapogee with gains of 8 dBi 20 dBi beamwidths 100 (full earth coverage) 14respectively power is 1 W for each

The MWL end-to-end performance between the attached clocks includingspace- and ground terminals is predicted to be a relative Allan deviation of1 times 10minus16 at 1000 s (perigee) and 1 times 10minus17 at 10000 s (apogee) Thesevalues are based on conservative implementation losses and other instrumentallosses assumed as 6 dB at apogee and 10 dB at perigee including additionallosses due to atmosphere An attitude knowledge of 1 is required

The MWL link will also support precise orbitographyndashcomplementary tothe GNSS receiver Generation of ranging data is simultaneous to two-waytime transfer and is available under all weather conditions Orbit determina-tion accuracy is expected to be better than 1 cm well matching EGE clockrequirements The value is expected based on demonstrated results fromother missions (METEOR ERS-2 TOPEXPOSEIDON) Laser ranging to

Exp Astron

the CCR array is used a few times during the mission to calibrate the MWLranging

In case of non-availability of the ultrapure signal from MOLO an additional10 GHz output is available from the FCDP to support ground-to-ground clockcomparison in common-view mode even after the nominal mission periodwhen the neither the clocks nor MOLO may be available any longer Basedon the PRARE experimentrsquos 12 years of operation this can be assumed forthe FCDPMWL combination as well The potential extended operational pe-riod makes the proposed configuration particularly attractive to metrologicallaboratories that rely on availability of long-lasting data in support of thegeneration of UTC

The resources of MWL are as follows mass 18 kg volume 18 l power55 W TC 2 kBday TM 120 MBday

426 Auxiliary units

The star camera assembly on the EGE spacecraft will provide real-timeattitude data primarily for the orientation of the spacecraft (with accuracy of1 in real-time) and the proper reduction of the GNSSSLR data

A GNSS receiver is used for the precise orbit determination time andfrequency transfer coarse positioning (lt1 m) for real time use by the Attitudeand Orbit Control as well as the back-up for time tagging of all payloaddata For the time and frequency transfer the receiver is nominally using theonboard frequency reference The receiver assembly includes three specialantennas By the year 2015 more than 100 GNSS satellites are expected tobe available for precise orbit determination permitting orbit inaccuracy of 1ndash2 cm RMS for altitudes up to approx 10000 km At apogee a combinationof GNSS signals and MWL ranging will be used with expected inaccuracyof 10 cm

5 Basic spacecraft specifications

The mission foresees a single spacecraft with a mission duration limited toapprox 3 years for cost reasons Attitude control along three axes withless than 1 error on a timescale of 1000 s is required The control can beperformed with cold gas thrusters magnetotorquers or reaction wheels Theselection will be done on the basis of the vibrations induced by these systems(desired level arms lt 10minus6 g)

Cold gas thrusters or monopropellant thrusters will be required to maintainthe orbit parameters during the mission by small orbit correction manoeuvresand will also guarantee a proper tuning of the orbit parameters after the releasefrom the launcher The ground segment (telemetry control and data transfer)could be via non-ESA stations user-operated to save costs

Further resources are indicated in Table 2

Exp Astron

The satellite bus design drivers are nadir-pointing microwave antennasthermal stability for the payload instruments no strong magnetic sources lowmicrovibrations harsh radiation environment due to crossing of the radiationbelts once per orbit power demand by science payload (with the possibility toreduce power demand by stand-by of some components during specific missionphases eg eclipse) mass limit due to required high altitude with low costlaunch

The satellite bus design can be based on established platforms such asFLEXBUS or PRIMA used on several past missions Their design philosophydiffers one having body-mounted solar arrays the other wings The latter canprovide larger amounts of power The vibration level induced by the panelswhen they are fixed and the duration of solar array drive periods per orbitwill have to be considered in the selection The FLEXBUS concept was usedfor a mission already requiring a low level of residual accelerations and theplatform is optimised for this parameter

For most of the satellite components one can consider using compo-nents and electronic units from navigation or communication satellites whichtypically cruise at altitudes between 15000 and 25000 km altitude LISAPathfinder which will cross the radiation belts during spiralling to theLagrange point will face similar problems concerning radiation as EGE as wellas XMM which is in a Molniya-type orbit with similar radiation characteristics

The instruments will therefore be located in the satellite structure such asto realise ionizing radiation shielding by the spacecraft structure and non-sensitive equipment to the maximum extent possible the need for additionalradiation shielding (including trapped proton radiation) for especially sensitiveparts of the instruments must be determined by ground tests It should bementioned that an orbit with higher perigee and less radiation dose is also anoption as it leads to a tolerable reduction of the science goals

The required plusmn3C temperature stability at the interface to the payloadenclosure is feasible The thermal control will employ the usual means (anenclosure with high thermal inertia heaters and a suitable radiator) Thelocation of the radiator will need a special study given the non-standard orbit

Magnetic field present during the mission can affect the properties of theonboard clock Using SPENVIS the magnetic field has been calculated alongthe spacecraft orbit the field intensity presents maxima around 02 G anda minimum value of the order of 10minus3 G As discussed above this is not alimitation to the instrument performance

6 Discussion and conclusion

The accuracy of the measurements depends crucially on the stability andaccuracy of the clocks While the performance of ground clocks (even thoseeventually deployed in new ground stations dedicated to EGE) will likelyimprove significantly from the present level the space clock design will beldquofrozenrdquo approx 10 years before data acquisition Thus their performance

Exp Astron

level will be significantly below the future ground station clock performanceAs specification we have chosen a performance level that has been demon-strated with laboratory breadboards recently We see no fundamental reasonthat this laboratory performance cannot also be achieved in a space ion opticalclock The experience with the PHARAO cold atom clock has shown thatan engineering model of a laser-based cold atom clock has a performancesimilar to that of a laboratory prototype The perturbation of the ion clock fre-quency due to magnetic fields along the orbit can be modelled and suppressedwith shielding and the residual effects measured and compensated for Themoderate temperature stability (plusmn 3C) to be provided in the experimentalbay can be achieved with thermal shields Accurate temperature controlof particular units is done by active stabilization of the units themselves

One important aspect is radiation which may affect all optical and trapcomponents as well as the lasers and reference cavity The radiation envi-ronment on the EGE orbit is significantly harsher than the near-earth ISSorbit on which ACES will operate The radiation effects must therefore becharacterized comprehensively with appropriate tests and then the radiationshield for the EGE instruments designed

In summary the Einstein Gravity Explorer mission (EGE) is a satellitemission aimed at a precise measurement of the properties of space-time usingatomic clocks By testing one of the most fundamental predictions of EinsteinrsquosTheory of General Relativity the gravitational redshift and thereby searchingfor quantum effects in gravity it explores one of the most important frontiers infundamental physics The EGE mission reaches beyond fundamental physicsproviding an important spin-off in geophysics Furthermore the technologicalaspects of the mission are ground-breaking for future time-keeping and ref-erence frame systems in space and thus are also expected to be of wide useto science and technology in general The payload of the mission has beenconceived as a compromise between cost and performance needing to fit thecost envelope of an ESA medium-class mission1

Acknowledgement We thank S Cesare for helpful information

References

1 Will CM The confrontation between general relativity and experiment Living Rev Relativ9 3 (2006)

2 Ni W-T Empirical foundations of the relativistic gravity Int J Mod Phys D 14 901 (2005)doi101142S0218271805007139

1Finally we point out that a spacecraft with a significantly reduced and thus lower-cost payload (noclocks only MOLO FCDP MWL) on a circular orbit of appropriate height could be used for along-duration earth science mission dedicated specifically to the measurement of the geopotential

Exp Astron

between on-board and on-ground clocks Two-way operation allows for com-pensation of the effects due to spacecraft location and its motion relative toground Its two bi-directional links operate in Ka-band (32345 GHz) andKu-band (147135 GHz) The high carrier frequency of the up- and down-links allows for a noticeable reduction of the ionospheric delay A third down-link channel is in S-band (22 GHz) and is used to determine the ionospheretotal electron content MWL uses a combination of carrier phase and codephase data to measure with high precision the absolute ionospheric delayas well as ionospheric delay variations during a ground contact The useof three frequency bands allows to resolve third order ionospheric termsKey modulation parameters are Ka band Tone 250 MHz Ku band PN100 MChips S-band 1 MChips

The EGE-MWL concept is a considerably improved version of the ACES-MWL Microwave carrier generation is directly from the 10 GHz microwaveoutput available from the MOLO and having exceptionally low close-in phasenoise without intermediate frequency dividers or other complex synthesiserstages commonly used in RF transponders The high frequency carriers arekey to achieve the desired frequency comparison accuracy

The Ka-band carrier is generated by tripling the FCDP-generated signalThe Ku-band is subsequently generated by division by two using a verystable regenerative divider Finally the assigned radio frequency is obtainedby adding a low frequency signal using an image reject mixer A mixedanalog-digital receiver design will enhance its performance by use of a flexibledigital signal processing and digital control loops The number of simultaneousground contacts is increased from four (as in the ACES mission) to ten or moreusing the digital concept while the number of physical receiver channels perband is reduced from four to two one for operation to ground and one for in-flight calibration The concept allows adaptation of receiver loop parametersto optimise for the different portions of a highly elliptic orbit while reducingsystematic uncertainties due to high Doppler and Doppler rate Real-timedata exchange to ground terminals is by additional data modulation on all RFsignals The spacecraft carries two different small antennae for perigee and forapogee with gains of 8 dBi 20 dBi beamwidths 100 (full earth coverage) 14respectively power is 1 W for each

The MWL end-to-end performance between the attached clocks includingspace- and ground terminals is predicted to be a relative Allan deviation of1 times 10minus16 at 1000 s (perigee) and 1 times 10minus17 at 10000 s (apogee) Thesevalues are based on conservative implementation losses and other instrumentallosses assumed as 6 dB at apogee and 10 dB at perigee including additionallosses due to atmosphere An attitude knowledge of 1 is required

The MWL link will also support precise orbitographyndashcomplementary tothe GNSS receiver Generation of ranging data is simultaneous to two-waytime transfer and is available under all weather conditions Orbit determina-tion accuracy is expected to be better than 1 cm well matching EGE clockrequirements The value is expected based on demonstrated results fromother missions (METEOR ERS-2 TOPEXPOSEIDON) Laser ranging to

Exp Astron

the CCR array is used a few times during the mission to calibrate the MWLranging

In case of non-availability of the ultrapure signal from MOLO an additional10 GHz output is available from the FCDP to support ground-to-ground clockcomparison in common-view mode even after the nominal mission periodwhen the neither the clocks nor MOLO may be available any longer Basedon the PRARE experimentrsquos 12 years of operation this can be assumed forthe FCDPMWL combination as well The potential extended operational pe-riod makes the proposed configuration particularly attractive to metrologicallaboratories that rely on availability of long-lasting data in support of thegeneration of UTC

The resources of MWL are as follows mass 18 kg volume 18 l power55 W TC 2 kBday TM 120 MBday

426 Auxiliary units

The star camera assembly on the EGE spacecraft will provide real-timeattitude data primarily for the orientation of the spacecraft (with accuracy of1 in real-time) and the proper reduction of the GNSSSLR data

A GNSS receiver is used for the precise orbit determination time andfrequency transfer coarse positioning (lt1 m) for real time use by the Attitudeand Orbit Control as well as the back-up for time tagging of all payloaddata For the time and frequency transfer the receiver is nominally using theonboard frequency reference The receiver assembly includes three specialantennas By the year 2015 more than 100 GNSS satellites are expected tobe available for precise orbit determination permitting orbit inaccuracy of 1ndash2 cm RMS for altitudes up to approx 10000 km At apogee a combinationof GNSS signals and MWL ranging will be used with expected inaccuracyof 10 cm

5 Basic spacecraft specifications

The mission foresees a single spacecraft with a mission duration limited toapprox 3 years for cost reasons Attitude control along three axes withless than 1 error on a timescale of 1000 s is required The control can beperformed with cold gas thrusters magnetotorquers or reaction wheels Theselection will be done on the basis of the vibrations induced by these systems(desired level arms lt 10minus6 g)

Cold gas thrusters or monopropellant thrusters will be required to maintainthe orbit parameters during the mission by small orbit correction manoeuvresand will also guarantee a proper tuning of the orbit parameters after the releasefrom the launcher The ground segment (telemetry control and data transfer)could be via non-ESA stations user-operated to save costs

Further resources are indicated in Table 2

Exp Astron

The satellite bus design drivers are nadir-pointing microwave antennasthermal stability for the payload instruments no strong magnetic sources lowmicrovibrations harsh radiation environment due to crossing of the radiationbelts once per orbit power demand by science payload (with the possibility toreduce power demand by stand-by of some components during specific missionphases eg eclipse) mass limit due to required high altitude with low costlaunch

The satellite bus design can be based on established platforms such asFLEXBUS or PRIMA used on several past missions Their design philosophydiffers one having body-mounted solar arrays the other wings The latter canprovide larger amounts of power The vibration level induced by the panelswhen they are fixed and the duration of solar array drive periods per orbitwill have to be considered in the selection The FLEXBUS concept was usedfor a mission already requiring a low level of residual accelerations and theplatform is optimised for this parameter

For most of the satellite components one can consider using compo-nents and electronic units from navigation or communication satellites whichtypically cruise at altitudes between 15000 and 25000 km altitude LISAPathfinder which will cross the radiation belts during spiralling to theLagrange point will face similar problems concerning radiation as EGE as wellas XMM which is in a Molniya-type orbit with similar radiation characteristics

The instruments will therefore be located in the satellite structure such asto realise ionizing radiation shielding by the spacecraft structure and non-sensitive equipment to the maximum extent possible the need for additionalradiation shielding (including trapped proton radiation) for especially sensitiveparts of the instruments must be determined by ground tests It should bementioned that an orbit with higher perigee and less radiation dose is also anoption as it leads to a tolerable reduction of the science goals

The required plusmn3C temperature stability at the interface to the payloadenclosure is feasible The thermal control will employ the usual means (anenclosure with high thermal inertia heaters and a suitable radiator) Thelocation of the radiator will need a special study given the non-standard orbit

Magnetic field present during the mission can affect the properties of theonboard clock Using SPENVIS the magnetic field has been calculated alongthe spacecraft orbit the field intensity presents maxima around 02 G anda minimum value of the order of 10minus3 G As discussed above this is not alimitation to the instrument performance

6 Discussion and conclusion

The accuracy of the measurements depends crucially on the stability andaccuracy of the clocks While the performance of ground clocks (even thoseeventually deployed in new ground stations dedicated to EGE) will likelyimprove significantly from the present level the space clock design will beldquofrozenrdquo approx 10 years before data acquisition Thus their performance

Exp Astron

level will be significantly below the future ground station clock performanceAs specification we have chosen a performance level that has been demon-strated with laboratory breadboards recently We see no fundamental reasonthat this laboratory performance cannot also be achieved in a space ion opticalclock The experience with the PHARAO cold atom clock has shown thatan engineering model of a laser-based cold atom clock has a performancesimilar to that of a laboratory prototype The perturbation of the ion clock fre-quency due to magnetic fields along the orbit can be modelled and suppressedwith shielding and the residual effects measured and compensated for Themoderate temperature stability (plusmn 3C) to be provided in the experimentalbay can be achieved with thermal shields Accurate temperature controlof particular units is done by active stabilization of the units themselves

One important aspect is radiation which may affect all optical and trapcomponents as well as the lasers and reference cavity The radiation envi-ronment on the EGE orbit is significantly harsher than the near-earth ISSorbit on which ACES will operate The radiation effects must therefore becharacterized comprehensively with appropriate tests and then the radiationshield for the EGE instruments designed

In summary the Einstein Gravity Explorer mission (EGE) is a satellitemission aimed at a precise measurement of the properties of space-time usingatomic clocks By testing one of the most fundamental predictions of EinsteinrsquosTheory of General Relativity the gravitational redshift and thereby searchingfor quantum effects in gravity it explores one of the most important frontiers infundamental physics The EGE mission reaches beyond fundamental physicsproviding an important spin-off in geophysics Furthermore the technologicalaspects of the mission are ground-breaking for future time-keeping and ref-erence frame systems in space and thus are also expected to be of wide useto science and technology in general The payload of the mission has beenconceived as a compromise between cost and performance needing to fit thecost envelope of an ESA medium-class mission1

Acknowledgement We thank S Cesare for helpful information

References

1 Will CM The confrontation between general relativity and experiment Living Rev Relativ9 3 (2006)

2 Ni W-T Empirical foundations of the relativistic gravity Int J Mod Phys D 14 901 (2005)doi101142S0218271805007139

1Finally we point out that a spacecraft with a significantly reduced and thus lower-cost payload (noclocks only MOLO FCDP MWL) on a circular orbit of appropriate height could be used for along-duration earth science mission dedicated specifically to the measurement of the geopotential

Exp Astron

the CCR array is used a few times during the mission to calibrate the MWLranging

In case of non-availability of the ultrapure signal from MOLO an additional10 GHz output is available from the FCDP to support ground-to-ground clockcomparison in common-view mode even after the nominal mission periodwhen the neither the clocks nor MOLO may be available any longer Basedon the PRARE experimentrsquos 12 years of operation this can be assumed forthe FCDPMWL combination as well The potential extended operational pe-riod makes the proposed configuration particularly attractive to metrologicallaboratories that rely on availability of long-lasting data in support of thegeneration of UTC

The resources of MWL are as follows mass 18 kg volume 18 l power55 W TC 2 kBday TM 120 MBday

426 Auxiliary units

The star camera assembly on the EGE spacecraft will provide real-timeattitude data primarily for the orientation of the spacecraft (with accuracy of1 in real-time) and the proper reduction of the GNSSSLR data

A GNSS receiver is used for the precise orbit determination time andfrequency transfer coarse positioning (lt1 m) for real time use by the Attitudeand Orbit Control as well as the back-up for time tagging of all payloaddata For the time and frequency transfer the receiver is nominally using theonboard frequency reference The receiver assembly includes three specialantennas By the year 2015 more than 100 GNSS satellites are expected tobe available for precise orbit determination permitting orbit inaccuracy of 1ndash2 cm RMS for altitudes up to approx 10000 km At apogee a combinationof GNSS signals and MWL ranging will be used with expected inaccuracyof 10 cm

5 Basic spacecraft specifications

The mission foresees a single spacecraft with a mission duration limited toapprox 3 years for cost reasons Attitude control along three axes withless than 1 error on a timescale of 1000 s is required The control can beperformed with cold gas thrusters magnetotorquers or reaction wheels Theselection will be done on the basis of the vibrations induced by these systems(desired level arms lt 10minus6 g)

Cold gas thrusters or monopropellant thrusters will be required to maintainthe orbit parameters during the mission by small orbit correction manoeuvresand will also guarantee a proper tuning of the orbit parameters after the releasefrom the launcher The ground segment (telemetry control and data transfer)could be via non-ESA stations user-operated to save costs

Further resources are indicated in Table 2

Exp Astron

The satellite bus design drivers are nadir-pointing microwave antennasthermal stability for the payload instruments no strong magnetic sources lowmicrovibrations harsh radiation environment due to crossing of the radiationbelts once per orbit power demand by science payload (with the possibility toreduce power demand by stand-by of some components during specific missionphases eg eclipse) mass limit due to required high altitude with low costlaunch

The satellite bus design can be based on established platforms such asFLEXBUS or PRIMA used on several past missions Their design philosophydiffers one having body-mounted solar arrays the other wings The latter canprovide larger amounts of power The vibration level induced by the panelswhen they are fixed and the duration of solar array drive periods per orbitwill have to be considered in the selection The FLEXBUS concept was usedfor a mission already requiring a low level of residual accelerations and theplatform is optimised for this parameter

For most of the satellite components one can consider using compo-nents and electronic units from navigation or communication satellites whichtypically cruise at altitudes between 15000 and 25000 km altitude LISAPathfinder which will cross the radiation belts during spiralling to theLagrange point will face similar problems concerning radiation as EGE as wellas XMM which is in a Molniya-type orbit with similar radiation characteristics

The instruments will therefore be located in the satellite structure such asto realise ionizing radiation shielding by the spacecraft structure and non-sensitive equipment to the maximum extent possible the need for additionalradiation shielding (including trapped proton radiation) for especially sensitiveparts of the instruments must be determined by ground tests It should bementioned that an orbit with higher perigee and less radiation dose is also anoption as it leads to a tolerable reduction of the science goals

The required plusmn3C temperature stability at the interface to the payloadenclosure is feasible The thermal control will employ the usual means (anenclosure with high thermal inertia heaters and a suitable radiator) Thelocation of the radiator will need a special study given the non-standard orbit

Magnetic field present during the mission can affect the properties of theonboard clock Using SPENVIS the magnetic field has been calculated alongthe spacecraft orbit the field intensity presents maxima around 02 G anda minimum value of the order of 10minus3 G As discussed above this is not alimitation to the instrument performance

6 Discussion and conclusion

The accuracy of the measurements depends crucially on the stability andaccuracy of the clocks While the performance of ground clocks (even thoseeventually deployed in new ground stations dedicated to EGE) will likelyimprove significantly from the present level the space clock design will beldquofrozenrdquo approx 10 years before data acquisition Thus their performance

Exp Astron

level will be significantly below the future ground station clock performanceAs specification we have chosen a performance level that has been demon-strated with laboratory breadboards recently We see no fundamental reasonthat this laboratory performance cannot also be achieved in a space ion opticalclock The experience with the PHARAO cold atom clock has shown thatan engineering model of a laser-based cold atom clock has a performancesimilar to that of a laboratory prototype The perturbation of the ion clock fre-quency due to magnetic fields along the orbit can be modelled and suppressedwith shielding and the residual effects measured and compensated for Themoderate temperature stability (plusmn 3C) to be provided in the experimentalbay can be achieved with thermal shields Accurate temperature controlof particular units is done by active stabilization of the units themselves

One important aspect is radiation which may affect all optical and trapcomponents as well as the lasers and reference cavity The radiation envi-ronment on the EGE orbit is significantly harsher than the near-earth ISSorbit on which ACES will operate The radiation effects must therefore becharacterized comprehensively with appropriate tests and then the radiationshield for the EGE instruments designed

In summary the Einstein Gravity Explorer mission (EGE) is a satellitemission aimed at a precise measurement of the properties of space-time usingatomic clocks By testing one of the most fundamental predictions of EinsteinrsquosTheory of General Relativity the gravitational redshift and thereby searchingfor quantum effects in gravity it explores one of the most important frontiers infundamental physics The EGE mission reaches beyond fundamental physicsproviding an important spin-off in geophysics Furthermore the technologicalaspects of the mission are ground-breaking for future time-keeping and ref-erence frame systems in space and thus are also expected to be of wide useto science and technology in general The payload of the mission has beenconceived as a compromise between cost and performance needing to fit thecost envelope of an ESA medium-class mission1

Acknowledgement We thank S Cesare for helpful information

References

1 Will CM The confrontation between general relativity and experiment Living Rev Relativ9 3 (2006)

2 Ni W-T Empirical foundations of the relativistic gravity Int J Mod Phys D 14 901 (2005)doi101142S0218271805007139

1Finally we point out that a spacecraft with a significantly reduced and thus lower-cost payload (noclocks only MOLO FCDP MWL) on a circular orbit of appropriate height could be used for along-duration earth science mission dedicated specifically to the measurement of the geopotential

Exp Astron

The satellite bus design drivers are nadir-pointing microwave antennasthermal stability for the payload instruments no strong magnetic sources lowmicrovibrations harsh radiation environment due to crossing of the radiationbelts once per orbit power demand by science payload (with the possibility toreduce power demand by stand-by of some components during specific missionphases eg eclipse) mass limit due to required high altitude with low costlaunch

The satellite bus design can be based on established platforms such asFLEXBUS or PRIMA used on several past missions Their design philosophydiffers one having body-mounted solar arrays the other wings The latter canprovide larger amounts of power The vibration level induced by the panelswhen they are fixed and the duration of solar array drive periods per orbitwill have to be considered in the selection The FLEXBUS concept was usedfor a mission already requiring a low level of residual accelerations and theplatform is optimised for this parameter

For most of the satellite components one can consider using compo-nents and electronic units from navigation or communication satellites whichtypically cruise at altitudes between 15000 and 25000 km altitude LISAPathfinder which will cross the radiation belts during spiralling to theLagrange point will face similar problems concerning radiation as EGE as wellas XMM which is in a Molniya-type orbit with similar radiation characteristics

The instruments will therefore be located in the satellite structure such asto realise ionizing radiation shielding by the spacecraft structure and non-sensitive equipment to the maximum extent possible the need for additionalradiation shielding (including trapped proton radiation) for especially sensitiveparts of the instruments must be determined by ground tests It should bementioned that an orbit with higher perigee and less radiation dose is also anoption as it leads to a tolerable reduction of the science goals

The required plusmn3C temperature stability at the interface to the payloadenclosure is feasible The thermal control will employ the usual means (anenclosure with high thermal inertia heaters and a suitable radiator) Thelocation of the radiator will need a special study given the non-standard orbit

Magnetic field present during the mission can affect the properties of theonboard clock Using SPENVIS the magnetic field has been calculated alongthe spacecraft orbit the field intensity presents maxima around 02 G anda minimum value of the order of 10minus3 G As discussed above this is not alimitation to the instrument performance

6 Discussion and conclusion

The accuracy of the measurements depends crucially on the stability andaccuracy of the clocks While the performance of ground clocks (even thoseeventually deployed in new ground stations dedicated to EGE) will likelyimprove significantly from the present level the space clock design will beldquofrozenrdquo approx 10 years before data acquisition Thus their performance

Exp Astron

level will be significantly below the future ground station clock performanceAs specification we have chosen a performance level that has been demon-strated with laboratory breadboards recently We see no fundamental reasonthat this laboratory performance cannot also be achieved in a space ion opticalclock The experience with the PHARAO cold atom clock has shown thatan engineering model of a laser-based cold atom clock has a performancesimilar to that of a laboratory prototype The perturbation of the ion clock fre-quency due to magnetic fields along the orbit can be modelled and suppressedwith shielding and the residual effects measured and compensated for Themoderate temperature stability (plusmn 3C) to be provided in the experimentalbay can be achieved with thermal shields Accurate temperature controlof particular units is done by active stabilization of the units themselves

One important aspect is radiation which may affect all optical and trapcomponents as well as the lasers and reference cavity The radiation envi-ronment on the EGE orbit is significantly harsher than the near-earth ISSorbit on which ACES will operate The radiation effects must therefore becharacterized comprehensively with appropriate tests and then the radiationshield for the EGE instruments designed

In summary the Einstein Gravity Explorer mission (EGE) is a satellitemission aimed at a precise measurement of the properties of space-time usingatomic clocks By testing one of the most fundamental predictions of EinsteinrsquosTheory of General Relativity the gravitational redshift and thereby searchingfor quantum effects in gravity it explores one of the most important frontiers infundamental physics The EGE mission reaches beyond fundamental physicsproviding an important spin-off in geophysics Furthermore the technologicalaspects of the mission are ground-breaking for future time-keeping and ref-erence frame systems in space and thus are also expected to be of wide useto science and technology in general The payload of the mission has beenconceived as a compromise between cost and performance needing to fit thecost envelope of an ESA medium-class mission1

Acknowledgement We thank S Cesare for helpful information

References

1 Will CM The confrontation between general relativity and experiment Living Rev Relativ9 3 (2006)

2 Ni W-T Empirical foundations of the relativistic gravity Int J Mod Phys D 14 901 (2005)doi101142S0218271805007139

1Finally we point out that a spacecraft with a significantly reduced and thus lower-cost payload (noclocks only MOLO FCDP MWL) on a circular orbit of appropriate height could be used for along-duration earth science mission dedicated specifically to the measurement of the geopotential

Exp Astron

level will be significantly below the future ground station clock performanceAs specification we have chosen a performance level that has been demon-strated with laboratory breadboards recently We see no fundamental reasonthat this laboratory performance cannot also be achieved in a space ion opticalclock The experience with the PHARAO cold atom clock has shown thatan engineering model of a laser-based cold atom clock has a performancesimilar to that of a laboratory prototype The perturbation of the ion clock fre-quency due to magnetic fields along the orbit can be modelled and suppressedwith shielding and the residual effects measured and compensated for Themoderate temperature stability (plusmn 3C) to be provided in the experimentalbay can be achieved with thermal shields Accurate temperature controlof particular units is done by active stabilization of the units themselves

One important aspect is radiation which may affect all optical and trapcomponents as well as the lasers and reference cavity The radiation envi-ronment on the EGE orbit is significantly harsher than the near-earth ISSorbit on which ACES will operate The radiation effects must therefore becharacterized comprehensively with appropriate tests and then the radiationshield for the EGE instruments designed

In summary the Einstein Gravity Explorer mission (EGE) is a satellitemission aimed at a precise measurement of the properties of space-time usingatomic clocks By testing one of the most fundamental predictions of EinsteinrsquosTheory of General Relativity the gravitational redshift and thereby searchingfor quantum effects in gravity it explores one of the most important frontiers infundamental physics The EGE mission reaches beyond fundamental physicsproviding an important spin-off in geophysics Furthermore the technologicalaspects of the mission are ground-breaking for future time-keeping and ref-erence frame systems in space and thus are also expected to be of wide useto science and technology in general The payload of the mission has beenconceived as a compromise between cost and performance needing to fit thecost envelope of an ESA medium-class mission1

Acknowledgement We thank S Cesare for helpful information

References

1 Will CM The confrontation between general relativity and experiment Living Rev Relativ9 3 (2006)

2 Ni W-T Empirical foundations of the relativistic gravity Int J Mod Phys D 14 901 (2005)doi101142S0218271805007139

1Finally we point out that a spacecraft with a significantly reduced and thus lower-cost payload (noclocks only MOLO FCDP MWL) on a circular orbit of appropriate height could be used for along-duration earth science mission dedicated specifically to the measurement of the geopotential

Exp Astron

3 Bertotti B et al A test of general relativity using radio links with the Cassini spacecraftNature 425 374 (2003) doi101038nature01997

4 Vessot RCF et al Test of relativistic gravitation with a space-borne hydrogen maser PhysRev Lett 45 2801 (1980) doi101103PhysRevLett452081

5 Laurent P et al Design of the cold atom PHARAO space clock and initial test results ApplPhys B 84 683 (2006) doi101007s00340-006-2396-6

6 Cacciapuoti L et al Atomic clock ensemble in space scientific objectives and mission statusNucl Phys B Proc Suppl 166 303 (2007) doi101016jnuclphysbps200612033

7 Flambaum VV Kozlov MG Limit on the cosmological variation of mpmefrom the Inversion Spectrum of Ammonia Phys Rev Lett 98 240801 (2007) doi101103PhysRevLett98240801

8 Reynold E et al Indication of a cosmological variation of the proton-electron mass ratiobased on laboratory measurement and reanalysis of H2 spectra Phys Rev Lett 96 151101(2006) doi101103PhysRevLett96151101

9 Tzanavaris T et al Limits on variations in fundamental constants from 21-cm andultraviolet quasar absorption lines Phys Rev Lett 95 041301 (2005) doi 101103PhysRevLett95041301

10 Kanekar N et al Constraints on changes in fundamental constants from a cosmo-logically distant OH absorber or emitter Phys Rev Lett 95 261301 (2005) doi101103PhysRevLett95261301

11 Flambaum VV Variation of the fundamental constants theory and observations Int JMod Phys A 22 4937 (2007) doi101142S0217751X07038293

12 Dent T et al Primordial nucleosynthesis as a probe of fundamental physics parametersPhys Rev D Part Fields Gravit Cosmol 76 063513 (2007) doi101103PhysRevD76063513

13 Dmitriev VF et al Cosmological variation of the deuteron binding energy strong interac-tion and quark masses from big bang uncleosynthesis Phys Rev D 69 063506 (2004) andrefs therein

14 Wetterich C Crossover quintessence and cosmological history of fundamental ldquoconstantsrdquoPhys Lett B 561 10 (2003) doi101016S0370-2693(03)00383-6

15 Wetterich C Probing quintessence with time variation of couplings Astropart Phys 10 2(2003)

16 Sandvik HB et al A simple cosmology with a varying fine structure constant Phys RevLett 88 031302 (2002) doi101103PhysRevLett88031302

17 Fortier TM et al Precision atomic spectroscopy for improved limits on variationof the fine structure constant and local position invariance PRL 98 070801 (2007)doi101103PhysRevLett98070801

18 Ashby N et al Testing local position invariance with four cesium-fountain primaryfrequency standards and four NIST hydrogen masers PRL 98 070802 (2007) doi101103PhysRevLett98070802

19 Hoffmann B Noon-midnight red shift Phys Rev 121 337 (1961) doi101103PhysRev121337

20 LoPresto JC et al Solar gravitational redshift from the infrared oxygen triplet AstrophysJ 376 757 (1991)

21 Krisher TP et al The Galileo solar redshift experiment Phys Rev Lett 70 2213 (1993)doi101103PhysRevLett702213

22 Soffel M et al The IAU 2000 resolutions for astrometry celestial mechanics and metrol-ogy in the relativistic framework explanatory supplement Astron J 126 2687 (2003)doi101086378162

23 Moyer TD Formulation for Observed and Computed Values of Deep Space Network DataTypes for Navigation Wiley New York (2003)

24 IERS Conventions In McCarthy DD Petit G (eds) IERS Technical Note No 32Frankfurt am Main Verlag des Bundesamts fuumlr Kartographie und Geodaumlsie 2004 (2003)

25 Kaplan GH The IAU Resolutions on Astronomical Reference Systems Time Scales andEarth Rotation Models US Naval Circular No 179 (Washington USNO) 2005 astro-ph0602086 (2004)

26 Blanchet L et al Relativistic theory for time and frequency transfer to order cminus3 AstronAstrophys 370 320 (2000) doi1010510004-636120010233

Exp Astron

27 Linet B Teyssandier P Time transfer and frequency shift to the order 1c4 in the fieldof an axisymmetric rotating body Phys Rev D Part Fields 66 024045 (2002) doi101103PhysRevD66024045

28 Le Poncin-Lafitte C Lambert SB Testing general relativity with the ACES missionarxivastro-ph0610463v1

29 Le Poncin-Lafitte C Lambert SB Numerical study of relativistic frequency shift for thecold-atom clock experiment in space Class Quantum Gravity 24 801 (2007)

30 Reinhardt S et al Test of relativistic time dilation with fast optical atomic clocks at differentvelocities Nat Phys 3 861 (2007) doi101038nphys778

31 Ehlers G Laumlmmerzahl C (eds) Special Relativity Will It Survive the Next 101 YearsSpringer (2006)

32 Kostelecky VA (ed) Proc First Second and Third Meetings on CPT and Lorentz Symme-try (World Scientific 1999 2002 2005)

33 Kostelecky A Lane CD Constraints on Lorentz violation from clock com-parison experiments Phys Rev D Part Fields 60 116010 (1999) doi101103PhysRevD60116010

34 Mansouri R Sexl RU A test theory of special relativity III Second-order tests GenRelativ Gravit 8 809 (1977) doi101007BF00759585

35 Braxmaier C et al Tests of relativity using a cryogenic optical resonator Phys Rev Lett88 010401 (2002) doi101103PhysRevLett88010401

36 Laumlmmerzahl C et al OPTIS a satellite-based test of special and general relativity ClassQuantum Gravity 18 2499 (2001) doi1010880264-93811813312

37 Scheithauer S Laumlmmerzahl C Analytical solution for the deformation of a cylinderunder tidal gravitational forces Class Quantum Gravity 23 7273 (2006) doi1010880264-93812324006

38 Wolf P et al Cold atom clock test of Lorentz invariance in the matter sector Phys RevLett 96 060801 (2006) doi101103PhysRevLett96060801

39 Fu LL Cazenave A Satellite Altimetry and Earth Sciences Academic New York (2001)40 Reigber C et al An Earth gravity field model complete to degree and order 150 from Grace

Eigen-Grace02S J Geodynamics 39 1 (2005)41 Tapley BD et al GRACE measurements of mass variability in the earth system Science

305 503 (2004) doi101126science109919242 Rummel R et al IAG-Symposia vol 120 p 261 Springer Heidelberg (2000)43 Bjerhammar A On a relativistic geodesy Bull Geod 59 207 (1985) doi101007BF0252032744 Francis CR et al The ERS-2 Spacecraft and its payload ESA Bulletin-European Space

Agency 83 13 (1995) httpearthesaintobjectindexcfmfobjectid=400745 Oskay WH et al Single-atom optical clock with high accuracy Phys Rev Lett 97 020801

(2006) doi101103PhysRevLett9702080146 Schneider T et al Sub-Hertz optical frequency comparisons between two trapped 171Yb+

ions Phys Rev Lett 94 230801 (2005) doi101103PhysRevLett9423080147 Margolis HS et al Hertz-level measurement of the optical clock frequency in a single 88Sr+

ion Science 306 1355 (2004) doi101126science110549748 Tamm C et al 171Yb+ single-ion optical frequency standard at 688 THz IEEE Trans

Instrum Meas 56 601 (2007) doi101109TIM200789114049 Hosaka K et al An optical frequency standard based on the electric octupole transition in

171Yb+ IEEE Trans Instrum Meas 54 759 (2005) doi101109TIM200484331950 Ludlow AD et al Sr Lattice clock at 1 times 10minus16 Fractional uncertainty by remote optical

evaluation with a Ca clock Science 319 1805 (2008) doi101126science115334151 Hollberg L From Quantum to Cosmos Workshop Airlie (Va USA) 6ndash10 July 200852 Newbury NR et al Proc 2007 Joint Mtg IEEE Intl Freq Cont Symp and EFTF Conf

980 (2007) httptfnistgovtimefreqgeneralpdf2207pdf