arX

iv1

212

1840

v1 [

phys

ics

ins-

det]

8 D

ec 2

012

The Polarized H and D Atomic Beam Source for ANKE at COSY-Julicha)

MMikirtychyants1 2 b) R Engels1 KGrigoryev1 2 HKleines3 PKravtsov2 S Lorenz4 c) MNekipelov1 2 d)

VNelyubin2 e) F Rathmann1 J Sarkadi1 HPaetz gen Schieck5 H Seyfarth1 E Steffens4 H Stroher1 and

A Vasilyev21)Institut fur Kernphysik Forschungszentrum Julich 52425 Julich Germany2)High Energy Physics Department StPetersburg Nuclear Physics Institute 188300 Gatchina

Russia3)Zentrallabor fur Elektronik Forschungszentrum Julich 52425 Julich Germany4)Physikalisches Institut Friedrich-Alexander-Universitat Erlangen-Nurnberg 91058 Erlangen

Germany5)Institut fur Kernphysik Universitat zu Koln 50937 Koln Germany

(Dated 11 December 2012)

A polarized atomic beam source was developed for the polarized internal storage-cell gas target at the magnetspectrometer ANKE of COSY-Julich The intensities of the beams injected into the storage cell measuredwith a compression tube are 75 middot 1016 hydrogen atomss (two hyperfine states) and 39 middot 1016 deuteriumatomss (three hyperfine states) For the hydrogen beam the achieved vector polarizations are pz asymp plusmn092For the deuterium beam the obtained combinations of vector and tensor (pzz) polarizations are pz asymp plusmn090(with a constant pzz asymp +086) and pzz = +090 or pzz = minus171 (both with vanishing pz) The paper includesa detailed technical description of the apparatus and of the investigations performed during the development

PACS numbers 2925Pj 2470+s

I INTRODUCTION

Single-polarized experiments making use of the po-larized proton and deuteron beams of the cooler storagering COSY-Julich and unpolarized targets are extendedto double-polarized studies1 by the installation of an in-ternal polarized hydrogen or deuterium storage-cell gastarget Utilizing pure gas of polarized hydrogen or deu-terium these targets circumvent the problem of dilutionby unpolarized nucleons and they permit fast change ofthe polarization direction In order to compensate for therelatively low areal density these targets are placed insidestorage rings where they are traversed by the orbitingbeam typically a million times per second As conceivedalready some forty years ago2 a substantial enhancementof the areal target-gas density (or luminosity) comparedto gas-jet targets by about two orders of magnitude isachieved when the polarized atoms are injected into anopen-ended T-shaped storage cell A review describingthe capabilities of polarized gas-jet and storage-cell gastargets is found in Ref3

a)Work financially supported by German Ministry for Educationand Research (BMBF) under contract Nos RUS-649-96 and 06ER 831 by Forschungszentrum Julich (FFampE) under contract No41149451 by Deutsche Forschungsgemeinschaft under contract No436 RUS 113430 and by the Russian Ministry of Sciencesb)Electronic mail mmikirtychyantsfz-juelichde Now at In-stitut fur Experimentalphysik Ruhr-Universitat Bochum 44801Bochum Germanyc)Now at Osram GmbH 93049 Regensburg Germanyd)Now at Wissenschaftlich-Technische Ingenieurberatung GmbH52428 Julich Germanye)Now at Department of Physics University of Virginia Char-lottesville VA 22904 USA

The polarized internal target (PIT) developed for themagnet spectrometer ANKE4 in COSY Julich5 consistsof the polarized atomic beam source (ABS) the stor-age cell6 and the Lamb-shift polarimeter (LSP)78 Inthe development of the PIT for ANKE the experienceof other groups in the operation of polarized gas targetscould be used Essentially these are the Madison source9used by the PINTEX collaboration at IUCF1011 and theFILTEX source12 used by the HERMES collaboration atDESYHamburg13Section II presents the general layout of the ABS and

it describes the major elements the pumping systemthe layout of the baffles the dissociator the nozzle-skimmer setup the system of sextupole magnets thehigh-frequency transition units and the slow control sys-tem Section III contains studies with the free hydro-gen jet from the nozzle In Secs IV to VII measure-ments and results are presented concerning the propertiesof the beam behind the last magnet the hydrogen anddeuterium beam intensities (IV) hydrogen beam profiles(V) the degree of dissociation of the hydrogen beam(VI) and the polarization values for the hydrogen anddeuterium beam (VII) Finally in Sec VIII the conclu-sions and an outlook to the future efforts are given Theprocedure of discharge-tube and nozzle conditioning isdescribed in Appendix A

II DESCRIPTION OF THE SETUP

A General layout of the ABS

The limited space at the target position between thefirst beam-bending dipole magnet D1 and the centralspectrometer dipole magnet D2 of the ANKE facility4 on

2

the one hand enforces the ABS to be mounted in a verti-cal position contrary to the layout of the earlier sourcesOn the other hand this position allows less complicatedsupports of the components and easier alignment Fur-thermore the height of the COSY tunnel restricts thevertical dimension of the ABS which requires a verycompact layout of the source Initially a slight inclina-tion from the vertical ABS orientation had been foreseento avoid drizzling of powder created during operation ofthe dissociator down into the storage cell Productionof such a powder had been observed earlier (SiO2

1415rdquogreen powderrdquo16) No such powder however was foundon a glass window at the exit of the ABS in verticalposition after weeks of dissociator operation Thus theinclination could be regarded as unnecessary

The layout of the ABS is presented in Fig 1 Thetwo main cylindrical vacuum vessels are fixed above andbelow a central support plate The inner diameter of theupper vessel17 is about 390mm It houses the chambersI II and III which are separated by two conical bafflesThe dissociator tube and the coldhead18 with the heatbridge for nozzle cooling are mounted on a three-leggedplate Three screws allow one to adjust the radial nozzleposition and the distance to the skimmer

The upper baffle between chamber I and II with thebeam skimmer is rigidly connected to the upper flange ofthe vessel The lower baffle with the diaphragm in frontof the first sextupole magnet separating the chambersII and III can be moved axially from outside to opti-mize the pumping conditions on the one hand betweenthe skimmer and the diaphragm and on the other in thenarrow region between the diaphragm and the front faceof the first magnet The first set of three sextupole mag-nets and the medium field rf transition (MFT) unit inchamber III are carried by two rods attached to the cen-tral support plate The lower vacuum vessel chamberIV has a smaller inner diameter of 200mm It housesthe second set of three sextupole magnets and a cylin-drical beam chopper rotating around a horizontal axisThe separate chamber V provides space for the weak andstrong field rf transition units (WFT and SFT units re-spectively) The ABS can be separated from the ANKEtarget chamber and the COSY vacuum system by a gatevalve19 embedded by a dedicated construction into theend flange of chamber V

Two strong turbomolecular pumps are installed at op-posite flanges of chamber I perpendicular to the beamdirection one on the beam-up side of chamber II Thechambers III IV and V are evacuated by cryopumpsDue to space limitation around the ABS shutters on thecryopumps as used in other sources were omitted Thegas originating from regeneration of the cryopumps ispumped via bypass tubes by turbomolecular pumps onthe upper chambers During regeneration as in othercases of pressure increase the gate valve on chamber V isclosed to avoid gas flow into the ANKE target chamber

Details of the pumping system the baffles the disso-ciator the area around the nozzle the magnet system

FIG 1 Cut along the ABS axis (1 dissociator 2 one ofthree adjustment screws for nozzle positioning 3 Cu heat-bridge for nozzle cooling with a flexible Cu strands connectionto the coldhead 4 first set of sextupole magnets 5 mediumfield rf transition unit 6 central support plate 7 one of tworotational feed-throughs enabling shift of the lower baffle 8second set of sextupole magnets 9 rotating beam chopper10 weak and strong field rf transition units 11 vacuum gatevalve between ABS and ANKE target chamber 12 schemat-ical indication of the storage cell) The labels I to V denotethe chambers of the differential pumping system

3

the rf transition units and the slow control system aredescribed in the subsequent subsections

B Pumping System

The system of pumps on the chambers I to V of theABS (Fig 1) is shown in Fig 2 the types of the pumpstheir pumping speeds and the achieved pressures arelisted in Table I Chamber I with the highest gas loaddue to the effect of the skimmer is pumped by two strongturbomolecular pumps Each of them is backed by asmaller turbomolecular pump Their exhausts are con-

FIG 2 The system of pumps on the chambers I to V of theABS (Fig 1) The specifications of the pumps are listed inTab I The figure also contains the bypass system for the gasload from regeneration of the cryopumps

nected to a common pump of the same type The totalcompression ratios of the serially connected turbomolec-ular pumps yields sufficient pumping speed for a primarymolecular gas flow up to 3mbar ls into the dissociatorThe line of pumps is backed by two oil free membranepumps According to the lower gas load chamber II isevacuated by a simpler line consisting of two turbomolec-ular pumps and a membrane pump All turbomolecularpumps are operated with use of synthetic oil20 Com-pared to mineral oil synthetic oil allows longer pump-

ing of hydrogen before oil exchange becomes mandatoryStrong cryopumps are utilized on chambers III and IVwhile the lowest chamber with the WFT and SFT unitsis evacuated by a smaller cryopump All cryopumps areequipped with temperature-controlled heating units forregeneration on both cooling stages21 Heating up toroom temperature while pumping the resulting gas loadby the bypass system and cooling down again takes about25 to 3 hours

C Baffles

The layout of the baffles had been defined by the nec-essary narrow vertical extension of the ABS and the re-quirement to provide sufficient pumping speed in viewof the heavy gas load to the vacuum chambers I and IIFurthermore the construction aimed at the possibilityof axial movements from outside to optimize the beamparameters by variation of the distances between nozzleskimmer and diaphragm The resulting shape for theupper baffle is shown in Fig 3 Except for details inthe openings the lower baffle carrying the diaphragmis identical The layout of the upper vessel and the baf-fles was done under the boundary conditions that on theone hand the baffles have to be movable within the ves-sel and on the other hand the slits between cylindricalsurfaces of the baffles and the inner surface of the vesselhas to be narrow to reach a small gas conductance Thediameter of the inner vessel surface is 3892mm with alongitudinal and non-circular tolerance of +02mm theouter diameters of both baffels are (3887minus02)mm The

TABLE I List of the devices employed in the ABS pumpingsystem composed of turbomolecular pumps (TP) membranepumps (MP) and cryopumps (CP) with nominal the individ-ual capacities CH2

the pumping speeds SH2 and the achieved

pressures at a primary gas flow of 10 mbar ls

Cham- Device Type CH2SH2

Pressure

ber [bar l] [ls] [mbar]

I TP1-2 TPH 2200a 2800 10minus4

TP4-57 TMH 260a 180

MP1-2 MVP 100-3a 1812c

II TP3 TPH 2200a 2800 10minus6

TP6 TMH 260a 180

MP3 MVP 100-3a 1812c

III CP1 COOLVAC 3000b 28d 5000 10minus7

IV CP2 COOLVAC 1500b 28d 5000 5 middot 10minus8

V CP3 COOLVAC 800b 43d 1000 5 middot 10minus8

a Pfeiffer Vacuum GmbH 35614 Asslar Germanyb Leybold Vakuum GmbH 50968 Koln Germanyc Pumping speed at 1000 mbar10 mbard At 10minus6 mbar

4

conductances of the slits of le5 ls are small comparedwith the applied pumping speed Because of the compli-cated shape identical raw pieces of cast Al22 were ma-chined to the final dimensions Contrary to the lower baf-

FIG 3 3D drawing of the upper baffle separating the vac-uum chambers I and II with the two wide cuts in front ofthe turbopumps (1) and the openings for the viewport (2)the skimmer (3) four of the 16 ball bearings (4) and the foursupporting rods (5)

fle the upper baffle until now has to be installed togetherwith the flange of the upper vessel at a fixed axial posi-tion (cf Fig 1) In order to reach full flexibility in vary-ing the nozzle skimmer and collimator relative positionsfrom outside the installation of rotational feedthroughsin the flange of the upper vacuum vessel is necessary aforeseen but not yet implemented feature

D Dissociator

To dissociate molecular hydrogen or deuterium to neu-tral atoms an rf discharge is employed which is fed bya 13560 MHz generator23 delivering up to 600 W into a50Ω load The layout of the dissociator shown in Fig 4is similar to that of the FILTEX design2425 The dis-charge tube (empty11times15mm)26 is surrounded by two coax-ial tubes (empty204times18mm and empty28times2mm) all three aremade from borosilicate glass27 The coolant streams fromthe inlet connection down between the discharge tubeand the middle tube and after flow reversal at the lowerend (Fig 5 label 2) it streams up in the outer slit to theoutlet connection In a closed loop the coolant inlet tem-perature (typically 15 C for a 50 water ndash 50 ethanolmixture) is stabilized by a cooling thermostate28 whichwould allow coolant temperatures down to minus80 C Therf coil and the capacitor at fixed relative positions canbe positioned from outside by means of a sliding rf con-nection29 and the feed-through ground connection This

FIG 4 3D drawing of the dissociator (1 gas inlet 2 slidingground connection 3 coolant inlet 4 coolant outlet 5 rfinput 6 sliding rf connection 7 grounded capacitor plate8 rf coil 9 rf-fed capacitor plate 10 isolating plastic sup-port rings 11 grounded limiter plate 12 lower end of thecoolant-guiding tubes 13 tube support and connection tothe coldhead (details are given in Fig 5) 14 lower end of thedischarge tube)

enables variation of the plasma-nozzle distance to opti-mize the atomic beam intensity while the plasma is burn-ing The treatment of the discharge tube and the nozzleprior to installation is described in Appendix A

E Nozzle

The nozzle cooled via the heat bridge and the sur-rounding components are shown in Fig 5 The nozzlemade from 995 Al has a simple conical shape with thetip cut Comparative measurements show that nozzleswith sharp edges as used eg in the Madison source9 donot yield higher atomic beam intensities First a sharpedge is more difficult to produce due to the softness ofpure Al Second the low heat conductance of a sharpedge leads to appreciable temperatures of the nozzle tipcaused by recombination of atoms on the nozzle surface

5

The temperature at the bottom of the nozzle is measuredwith a Pt-100 sensor and it is stabilized with an accuracyof plusmn05K utilizing a heater Measurements with temper-ature sensors placed along the outer nozzle surface haveshown a temperature increase from 60K at the nozzlebottom to sim200K at the sharp nozzle tip In the follow-ing the nozzle temperature is defined as that measuredwith this Pt-100 sensorWith the present system of sextupole magnets the

maximum atomic beam intensity feeding the storagecell is obtained with a nozzle-orifice diameter of 23mmand a nozzle-tip to skimmer-tip distance of 15mm at askimmer-tip diameter of 44mm and a skimmer-tip to di-aphragm distance of 17mm The 2mm thick diaphragmwith a conical bore opening from 95mm to 99mm to-wards the first permanent sextupole magnet shields themagnet from heating by atoms recombining on its sur-face The slit between the diaphragm and the front faceof the magnet enables pumping of gas from the entranceto the magnetThe Teflon washer and the stainless steel support sep-

arate the cold lower end of the heat bridge from thewarm lower end of the dissociator The dimensions of

FIG 5 Technical drawing including the lower end of theheat bridge and the dissociator the nozzle surroundings andthe first sextupole magnet (in scale 1 discharge and coolant-guiding tubes 2 coolant-reversal piece 3 heat flow reducingTeflon washer 4 sliding heat connection 5 stainless steelconnector 6 groove for nozzle-heating element 7 lower endof the Cu heat bridge 8 nozzle fixture 9 nozzle 10 baf-fle separating the chambers I and II with a viewport 11stainless steel beam skimmer 12 Cu diaphragm 13 firstsextupole magnet and 14 baffle separating the chambers IIand III

these two components and the sliding heat connector aworked-over sliding high current connector similar to therf connector in the dissociator define the temperatureof the lower end of the discharge tube relative to thatof the nozzle The discharge tube adapted at its lowerend to the nozzle by a chamfered edge is pressed to thenozzle by a viton O-ring at its upper end The two O-rings around the discharge tube in the lower part of thedissociator seal against the atmosphere By this designonly minor forces are exerted to the discharge tubeThe removable viewport in the baffle and the window

flange in the upper vacuum vessel (on the right-hand sideof chamber II in Fig 1) allows one to observe the nozzlestatus from the outside and to exchange nozzles withoutremoval of the dissociator from the setupThe heat bridge from the coldhead to the nozzle is

made from electrolytic Cu The flexible link between thecoldhead and the heat bridge consisting of about 200high-purity Cu strands of 1 mm diameter allows for thethermal expansions of the cold and the warm compo-nents The total cross section of the strands and theirheat conductance is smaller than that of a massive Cubody This deficiency however is reduced by clampingthe flexible link directly to the coldhead At its operat-ing temperature of about 30K the thermal conductivityof Cu is about 11 9 and 5 times higher than that at300 100 and 60K respectively30 Thus the reductionof the conductance of the entire heat bridge by the flexi-ble link is minimized by placing it at the coldhead Withthe present system cooling the nozzle down from roomtemperature to 60K needs about 15 hours The heat-ing element facilitates warming up to room temperaturewithin about one hourFurthermore avoiding the maze of cold Cu strands

around the nozzle ie a labyrinthic cold surface com-pared to an earlier solution31 leads to improved pumpingconditions in the nozzle-skimmer area where the highestgas load has to be pumped offIn an earlier phase of the ABS development attempts

have been made to use a cryogenic Ne heat-pipe of 20Wcooling power instead of the usual solid Cu bridge toachieve faster cooling and warming of the nozzle becauseof the lower heat capacity32 An observed instability inthe necessary operation mode however lead to difficul-ties in nozzle-temperature stabilization In view of thefact that the cooling and warming-up times reached withthe Cu bridge were satisfying and that its use avoids theadditional precautions imposed by the heat-pipe opera-tion it has been replaced by the Cu bridge

F Magnet System

The design of the magnet system was made for a setof sextupole magnets consisting of permanently magne-tized segments made from NdFeB compounds deliveringpole-tip fields around 15 T Tracking calculations fromthe nozzle to the feeding tube of the storage cell were

6

performed with the use of a computer code originallydeveloped for the FILTEX ABS24 The boundary condi-tions by the layout of the target setup were the availabledistance of about 1250mm from the nozzle to the feeding-tube entrance of 10mm diameter and the distance fromthe exit of the last magnet to the feeding-tube entrance of300mm necessary to install the SFT and WFT units andthe gate valve between the ABS and the target chamberThe laboratory velocity distribution of the atoms in

the supersonic beam from the nozzle is described by amodified Maxwellian distribution

F(~vd Tb) =( m

2 k Tb

)32exp

[ minusm

2 k Tb(~v minus ~vd)

2

]

(1)

where m is the mass of the atoms and k is the Boltzmannconstant According to time-of-flight studies33 the driftvelocity along the beam axis vd and the beam temper-ature Tb for a primary molecular gas flow of 1 mbar lsand a nozzle-orifice diameter of 2mm follow a linear de-pendence on the nozzle temperature Tn For hydrogenvd[ms] = 1351 + 61 middot Tn[K] and Tb = 029 middot Tn and fordeuterium vd[ms] = 1070+345middotTn[K] and Tb = 025middotTnAs starting conditions of a track a random generator

selects a point in the nozzle orifice one within the di-aphragm in front of the first magnet and an atom ve-locity |v| In linear molecular flow approximation (cfthe discussion in Ref34) this defines ~v for the track be-tween the nozzle and the first magnet According to thegeometrical boundary conditions and the velocity distri-bution of Eq (1) the event is either rejected or used inthe further track calculation Within the magnet theevolution of the track is calculated stepwise by numeri-cal integration of the equation of motion over integrationtimes of 2micros corresponding to track lengths of 36mmfor a typical particle velocity of 1800ms The pureradial force acting on an atom within the field of the

sextupole magnet is ~Fr = minusmicroeff middot δBδr middot ~rr The ef-fective magnetic moment resulting from the Breit-Rabidiagram (eg Ref35) as microeff = δWδB is positive (neg-ative) for atoms in the hyperfine states with the electron

spin parallel (antiparallel) to ~B in the magnet aperturewhich therefore are deflected towards (away from) thebeam axis In the drift sections between the two magnetgroups and between the last magnet and the feeding tubethe trajectories are assumed as straight linesA variety of systems were studied all under the as-

sumption of Tn = 60K and pole-tip fields of 15T Asystem utilizing 6 magnets was found to yield satisfyingboth separation of the atoms in the microeff lt 0 and microeff gt 0states and focusing of the microeff gt 0 states into the feedingtube Optimization of the parameters led to the systemlisted in Table II (The tracking calculations yielding themagnet dimensions for the order to the manufacturer hadbeen performed for a slightly different geometry) Thetable gives the two distances at which intensity mea-surements with the compression tube were performedThe Fig 6 shows the projection of the trajectories of Hatoms in the microeff gt 0 states calculated for this system

TABLE II Final dimensions and axial positions of the sourcecomponents (pole-tip field strenghts Blowast

0 as measured afterdelivery36 inner diameters (empty0) outer diameters (empty1) axialdimensions (ℓ) and distances (∆) between the componentsThe lower three lines give the two distances and the dimen-sions of the compression tube used in the intensity measure-ments

component Blowast

0 [T] empty0 [mm] empty1 [mm] ℓ [mm] ∆ [mm]

Nozzle orifice 23 33150

Skimmer 44304a 130169

Diaphragm 9599a 2036

Magnet 1 1630 10401412a 3998 400194

Magnet 2 1689 15982212a 6404 650194

Magnet 3 1628 2804 9400 70014297

Magnet 4 1583 3004 9402 38011010

Magnet 5 1607 3006 9400 5501150

Magnet 6 1611 3002 9404 550030003370

Compr tube 100 110 1000

a Conical openingthe first number denotes the measureddiameter of the entrance the second that of the exit aperture

One recognizes two groups of trajectories one with anintermediate focus and another one with focusing intothe feeding tube The present result like those of othergroups (see eg Ref33) confirms the expectation37 thatthe transmission as function of the atom velocity shouldshow two maxima one below and one above the mostprobable velocityThe transmission Tr of the system is defined as the

fraction of tracks ending within the entrance of the feed-ing tube to those passing the diaphragm in front of thefirst sextupole magnet For the four hyperfine states ofhydrogen38 the calculations yield Tr(|1〉) sim Tr(|2〉) =042 (for both microeff gt 0) and Tr(|3〉) = 0001 andTr(|4〉) = 00004 (for both microeff lt 0)The performed tracking calculations do not account for

intra-beam and residual-gas scattering The calculatedtransmissions thus only allowed one to estimate upperlimits of the expected atomic beam intensity Iin into thefeeding tube For a primary molecular flow q(H2) theintensity Iin(H) with atoms mainly in the states |1〉 and|2〉 (microeff gt 0) was expected as

Iin(H) = q(H2) middot 2α middot Ω

2πmiddot 14

i=4sum

i=1

Tr(|i〉) (2)

For the degree of dissociation α a routine value of 08(see eg Ref9) was assumed Ω = 0022π is the solid

7

angle covered by the collimator aperture The factor 14reflects the assumption that the four substates in theatomic beam from the nozzle are equally populated Forq(H2) = 1mbar ls or 27middot1019H2 moleculess one expectsIin(H) sim 1 middot 1017H atomssAs described in the subsequent section the rf tran-

sition units are used to change the relative occupationnumbers of the states The trajectory code allows oneto simulate this change by assigning a microeff of one of thestates to the atoms before they pass a magnet As an ex-ample the medium-field transition unit (MFT) behindmagnet No 3 (see Fig 1) brings H atoms from state |2〉into state |3〉 This is simulated by assigning microeff(|2〉) gt 0to the atoms in the magnets 1minus3 and microeff(|3〉) lt 0 in themagnets 4 minus 6 where they get deflected from the beamaxis This results in a small value Tr(|2〉) = 0017 Fromthis value and the above value Tr(|1〉) = 042 the vectorpolarization is expected as

pz =Tr(|1〉)minus Tr(|2〉)Tr(|1〉) + Tr(|2〉) = 091 (3)

under the assumption of 100 efficiency of the transitionunitThe design and the properties of the permanent sex-

tupole magnets39 were discussed in an earlier paper36To achieve the pole-tip field values of sim15T each mag-net was produced from 24 segments employing three dif-

FIG 6 Projection of the 3-dimensional trajectories of hydro-gen atoms in hyperfine states |1〉 and |2〉 (effective magneticmoment microeff gt 0)) from the nozzle (empty = 2mm Tn = 60K)to the storage cell calculated for the magnet arrangement ofTable II and pole-tip fields of 15 T The positions and lat-eral dimensions of the six magnets and the feeding tube areindicated (in red)

ferent types of NdFeB compounds The expected pole-tip values (Table II) and the precise radial dependenceB(r) sim r2 within the magnet apertures were confirmedFor the first time the predicted high multipole compo-nents40 up to a 102-pole structure very near to the aper-ture surface could be measured36After the field measurements the magnets were encap-

sulated to prevent diffusion of hydrogen into the magnetmaterial which might deteriorate the magnetic proper-ties and to avoid the pumping of gas from the sinteredmagnet bodies The housings were made from thin stain-less steel cans of 02mm thickness for the conical andcylindrical walls within the magnet apertures and 03mmfor the front and end covers During the final welding toclose the housings with magnets installed the local tem-perature of the magnet material had to be kept belowthe Curie temperature of 60 C This was achieved bywelding with the use of a pulsed 15Hz NdYAG laserdelivering 11 J in a pulse of 2ms41 Overlapping weldspots of sim03mm diameter set around the adjacent cir-cular 02mm thick weld lips allowed one to finish thehousings with leak rates sim 10minus10mbar ls Inside thehousings the magnets were fixed to suppress axial androtational movements caused by the force of the adja-cent magnets Finally the free slits within the housingswere filled by sim20mbar krypton to enable leak tests bymass spectroscopy

G Radio Frequency Transition Units

The ABS is equipped with three types of transitionunits a weak field a medium field and a strong field rftransition unit (WFT MFT and SFT units) Togetherwith the selecting properties of the sextupole magnetsthey enable one to achieve all vector and tensor polar-izations of the atomic hydrogen and deuterium gas inthe storage cell In all three units transitions betweenthe hyperfine states split according to the Breit-Rabi di-agram by a static magnetic field (see eg Ref35) areinduced by the magnetic component (Brf) of an rf fieldleading to changes in the population of the states Thestatic field Bstat consists of two parallel components ahomogeneous field Bhom and a superimposed weaker gra-dient field Bgrad both orthogonal to the beam directionThe field gradient along the beam direction is requiredto satisfy the condition of adiabatic passage3542The assemblies of the WFT and the MFT units are

similar43 The layouts follow those of the units devel-oped for the HERMES experiment44 In both units therf field is produced by a coil with the axis along the beamdirection and consequently Brf orthogonal to Bstat TheMFT unit is shown in Fig 7 Figure 8 schematicallyshows one of the grooved aluminum frames with thewindings producing the gradient field A WFT unitis operated in a weak magnetic field Bstat le10G for hy-drogen and le5G for deuterium where the total atomicspin F is a good quantum number In hydrogen the

8

F = 1 levels |1〉 |2〉 and |3〉 with magnetic quantumnumbers mF = +1 0 and minus1 respectively can be re-garded as equally spaced In deuterium the same holdsfor the four F = 32 levels |1〉 |2〉 |3〉 and |4〉) withmF = +32 +12 minus12 and minus32 respectively andfor the two F = 12 levels |5〉 and |6〉 with mF = minus12and +12 respectively The magnetic component of therf dipole field induces transitions between each pair ofneighboring mF states with ∆mF = plusmn1 |∆mF| = 2transitions are forbidden The interchange of the popu-lation between the states |1〉 and |3〉 in hydrogen eg iscaused by a two-quantum transition via the intermedi-ate state |2〉 In the classical description of the adiabaticpassage method42 the population change should not de-pend on the sign of the magnetic field gradient relativeto the beam direction An exact quantum-mechanicaltreatment4546 however indicates that cleaner popula-tion changes from state |1〉 to |3〉 in hydrogen and fromstate |1〉 to |4〉 in deuterium are obtained with a nega-tive field gradient ie a Brf field decreasing in the beamdirection Deviations from adiabaticity are discussed inRef4547

The MFT unit is operated at higher values of Bstatwhere the differences in the energy spacings of pairs of hy-

FIG 7 Three-quarter-section view of the MFT unit with thesupport structure (1 self-supporting rf coil with spacers 2pick-up loop 3 Al tubes defining the length of the transition-inducing rf field 4 Cu cavity 5 coil around the pole shoe(6) providing the static field Bstat 7 slit between pole shoeand cavity wall housing the gradient-field coil 8 componentsof the static magnet yoke also serving as shielding againstexternal fields 9 cavity-positioning element 10 Cu padscooled by means of water-carrying tubes The cavity withthe rf coil and the pick-up loop can be taken out from thesurrounding components

z

Bstatic

transition

region

FIG 8 Arrangement of the windings producing the staticgradient field Bgrad shown in the left-hand side of the figureIn all transition units the field lies in the direction of the statichomogeneous field the field gradient dBdz lies in the beamdirection which defines the z axis In z direction the tran-sition reagion (indicated by the dashed lines) is confined tothe range of constant gradient by the Al tubes in orthogonaldirection by the beam diameter

perfine states with ∆mF = plusmn1 allow one to select singletransitions Originally developed for an polarized alkaliion source48 the MFT unit now is a standard compo-nent in polarized hydrogen and deuterium sources as dis-cussed eg in Ref49 Appropriate choice ofBhom Bgradand the rf frequency allows one to induce selected tran-sitions |1〉 harr |2〉 and |2〉 harr |3〉 in hydrogen or |1〉 harr |2〉|2〉 harr |3〉 and |3〉 harr |4〉 in deuterium Furthermore thechoice of the field gradient allows one to achieve consecu-tive transitions As an example a negative field gradientin the MFT unit behind the first set of magnets ie aB field decreasing in beam direction at a fixed rf fre-quency leads to the sequence of the transitions |3〉 rarr |4〉|2〉 rarr |3〉 and finally |1〉 rarr |2〉 in deuterium leaving thestate |1〉 empty

The SFT unit is used to induce transitions betweenstates in the upper and lower hyperfine multiplet in hy-drogen and deuterium Contrary to the historical nameindicating a strong magnetic field the SFT unit is op-erated with magnetic fields comparable to those used inthe MFT unit The transition frequencies are comparablewith those of the hyperfine splitting (1420MHz for hy-drogen and 327MHz for deuterium) and thus are muchhigher than those in the WFT and MFT units The rffield in a SFT unit is produced by a twin-line resonatorinside a Cu box tuned to the λ4 resonance50 The SFTunit51 is shown in Fig 9 Again the layout follows that ofthe unit used in the HERMES experiment44 Two vari-able capacitors at the free ends of the conducting rodsfed by the rf power with a relative phase shift of 180 allow one to tune the device

9

FIG 9 Three-quarter-section view of the rf cavity of theSFT unit for deuterium (1 the two resonant-field creatingconductors 2 the adjustible capacitor plates 3 Cu cavity)The inner dimensions of the cavity are 56mm along Bstat36mm orthogonal to it and 36mm height The cross sectionof the conductors is 14times 4mm2

H Slow Control System

Industrial components providing reliable and long-term support were selected for the control system of thewhole setup consisting of the ABS and the diagnosticstools the storage cell positioning system the Lamb-shiftpolarimeter and the supply system of a calibrated flowof unpolarized molecular gas The interlock system hasbeen implemented on the basis of SIEMENS SIMATICS7-300 family of programmable logic controllers In orderto unify the interfacing to the control computer all front-end equipment is connected via the PROFIBUS DP field-bus The process control software was implemented us-ing the Windows-based WinCC toolkit from SIEMENSThe system controls the operation of the pumps and thevalves It reads the pressure gauges and controls theregeneration cycles of the cryopumps Via a control net-work the temperature of the nozzle is stabilized withinplusmn05K Furthermore all power-supply units rf genera-tors and amplifiers are set and controlled The wholevariety of components to be controlled the logical struc-ture of the control and interlock system and a separatedevice for parameter studies are described in Ref52

III STUDIES OF THE FREE HYDROGEN JET

A Atomic beam profile near the nozzle

A novel device has been used to measure the profile ofan atomic beam via the deposition of recombination heaton thin wires in a two-dimensional grid5354 Atoms stuckon the surface of gold-plated tungsten wires of 5microm diam-eter recombine and are reemitted as molecules The re-combination heat (4476 eV per hydrogen molecule) leadsto a change of temperature and thus resistance alongeach wire The measurement of the resistance changes of

FIG 10 Two-dimensional profile of the atomic hydrogenbeam 10mm from the nozzle deduced from recombinationheating of gold-plated tungsten wires of 5microm in a 8times 8 wiregrid

all the wires in the grid allows one to deduce the centerand the profile of the beam Figure 10 shows the beamprofile resulting with a 8times8 wire grid positioned betweenskimmer and collimator performed as a first proof of themethod Later such a device has been used to comparemeasured and calculated beam profiles along the beamaxis between nozzle and skimmer34

B Degree of dissociation of the free atomic jet

The dissociation of the primary molecules is achievedby the interaction of the electrons and the hydrogen ordeuterium molecules in the plasma of the dissociatorThe degree of dissociation of the beam from the nozzledepends on the rf power applied to maintain the plasmathe primary molecular gas flow into the dissociator andthe temperature of the nozzle and the lower end of thedischarge tube These dependencies have been studiedbefore installation of the sextupole magnets with a setupcontaining a crossed-beam quadrupole mass spectrome-ter5556

α =ρa

ρa + 2 middot ρm (4)

The admixture of molecules in an atomic beam is de-scribed by the degree of dissociation where ρa and ρmare the densities of atomic and molecular hydrogen ordeuterium in the beam Other authors (eg Ref13) usethe atomic and molecular intensities Ia and Im in the def-inition of the degree of dissociation (αI) in Eq (4) Thetwo definitions of are related by

ImIa

=vmva

middot 1minus α

2α=

1minus αI

2αI (5)

10

This quantity was determined with the quadrupolemass spectrometer (QMS) in a conventional way as

α =Slowasta

Slowasta + 2 kv kion kdetSm

(6)

Here Slowasta = SaminusδSm denotes the atomic signal corrected

for dissociative ionization The parameter δ = 00141was obtained following the method described in Ref31The coefficient kv = vmva accounting for the difference

in atom and molecule velocity was chosen as 1radic2 under

the assumption of thermalization of the beam emergingfrom the nozzle Furthermore kion = 064 57 accounts forthe differences in ionization cross section for atomic andmolecular hydrogen and kdet = 084 for the detectionprobability55 As an example of the parameter stud-ies Fig 11 shows the deduced dependencies on the rfpower for a set of primary molecular hydrogen gas flowsFor typical flow values q(H2) le 10mbar ls a saturationvalue around 08 was obtained

0 50 100 150 200 250 300 350 400

00

02

04

06

08

10

Deg

ree

of d

isso

ciat

ion

()

Dissociator rf power [W]

q=07 mbar ls q=08 mbar ls q=10 mbar ls q=15 mbar ls

FIG 11 Degree of dissociation α of the free hydrogen jet asfunction of the applied rf power for different primary molec-ular hydrogen flows and a nozzle temperature of 70K

IV BEAM INTENSITY

The intensity of the polarized beam from the ABS to-gether with the layout of the storage cell determines theareal density of the target gas The intensity of the beamhas been measured with the use of a compression-tubesetup5859 shown in Fig 12 to optimize the ABS opera-tion parameters The measurements were performed at a300mm distance from the compression-tube entrance tothe last magnet and an inner tube diameter of 100mmas set in the tracking calculations The length of thecompression tube of 100mm was made equal to that ofthe foreseen feeding tube of the storage cell The narrowtube around the compression tube on a support based onthe lower flange separates the volume around the tube

FIG 12 Side view of the compression-tube setup made fromstandard ultra-high-vacuum components with a partial cutalong the axis (1 compression tube 2 support of thecompression tube based on the lower flange 3 narrow tubearound 1 closing the upper volume and allowing axial shiftsof the tube by the support 4 compression volume 5 hot-cathode pressure gauge 6 xy manipulator 7 z manipulator8 glass viewport 9 electromagnetic valve for gas inlet

from the compression volume The xy manipulator servesfor centering the tubes and for intensity-profile measure-ments The construction allows axial shifts of the com-pression tube by the z manipulator and the use of tubesof different diametersThe intensity of the beam entering the compression

volume through the compression tube is measured viathe pressure in the compression volume It is determinedby the equilibrium between the incoming beam intensityIin and the outgoing intensity Iout Under the assump-tion of a pure atomic beam and complete recombinationin the compression volume

Iin(atomss) = 2 middot Iout(moleculess)

= 2 middot∆P middot Ctube

= 2 middot∆P middot 103 middot 1020 middot d3

l

radic

T

M (7)

Here ∆P is the difference between the pressure measuredin the compreesion volume and that in the ABS cham-ber V The conductance of the compression tube Ctubeis determined by the inner diameter d of the tube itslength l the gas temperature T and the molar massM of the gas (given in cm and K respectively)60 The

11

CG

RVC200

H2

F

EV

EV

EV

EV

EV

MP

V1V2

EVR116

HC

V0

EV EVA016UDV136

HCCPLC

PS

220V AC

220V AC

Interface

24V DC

FIG 13 Source of calibrated molecular gas flow (V0 com-pression chamber V1 gas-storage chamber feeding V0 viathe needle valve UDV136a V2 chamber of calibrated volumeused to determine that of V1) The pressure in V1 is mea-sured by the capacitance gauge CG and is kept constant bythe dosing valve EVR116 with the gauge controller RVC200The whole setup including the evacuation elements can beoperated manually or by the programmable logic controllerPLC either within the ABS control system52 or as a separatesystem

a All the valves and the gauge controller are supplied by PfeifferVacuum GmbH Dndash35614 Asslar Germany (manufacturerBalzers AG Liechtenstein)

factor 2 takes into account that the same pressure is mea-sured in the hot-cathode gauge for 2 middotIin (H atomss) and1 middot Iin (H2 moleculess) For d = 10mm l = 100mmT = 290K and M = 2 for hydrogen pressure differences∆P on the order of 10minus4mbar are expected for atomichydrogen beam intensities in the order of 1017 atomssThe relation between Iin and ∆P for hydrogen has beendetermined experimentally with the use of a source ofcalibrated molecular hydrogen gas flow5859 depicted inFig 13 The measured dependence with a linear fit isshown in Fig 14 The calibration curve allows one to de-termine absolute values of Iin of hydrogen and deuteriumbeams The calibration for deuterium was deduced fromthe one for hydrogen by scaling with a factor 1

radic2 ac-

cording to Eq (7)

The dependences of Iin on the dissociator-operationparameters primary molecular hydrogen flow q(H2) noz-zle temperature Tn and dissociator power Pdiss have beenstudied to find the optimum values They are shown inthe Figs 15 16 and 17 respectively for different nozzle-orifice diameters The figures show that for the hydrogenbeam (states |1〉 and |2〉) with the standard operation pa-rameters qH2

= 11 mbar ls Tn = 70K Pdiss = 350Wand with a nozzle-orifice diameter of 23mm an inten-sity of Iin(H) = (75 plusmn 02) middot 1016 particless is achievedquite close to the earlier estimate from Eq (2) Besidesthe dominant atomic component of H atoms this value

04 06 08 10 12 14 16 18 20 2201

02

03

04

05

06

07

08

09

10

11

12

Gas

flow

from

the

stor

age

volu

me

[1017

ats

]

Pressure in compression volume [10-4 mbar]

FIG 14 Calibration curve for hydrogen used to deduce fromthe measured pressures the intensities of the hydrogen anddeuterium beam injected into the compression tube

00 05 10 15 20 2500

10

20

30

40

50

60

70

80 D = 20 mm D = 23 mm D = 25 mm

Bea

m in

tens

ity [1

016 a

ts]

Primary gas flow [mbar ls]

FIG 15 Intensity of the hydrogen beam (states |1〉 and |2〉)injected into the compression tube as function of the primarymolecular gas flow for different nozzle diameters D (nozzletemperature 60K dissociator power 300W)

includes small admixtures of H atoms in state |3〉 andmolecular hydrogen The first kind can be estimatedwith the use of the calculated transmissions (Sec II F)as 0017084 asymp 2 The amount of the second admix-ture has been measured as described below

For the deuterium beam (states |1〉 |2〉 and |3〉) theoptimization procedure gave an intensity of Iin(D) =(39 plusmn 02) middot 1016 particless achieved with q(D2) = 09mbar ls Tn = 65K and Pdiss = 300W slightly lowerthan those for hydrogen

12

50 60 70 80 90 100 110

40

50

60

70

80 D = 20 mm D = 23 mm D = 25 mm

Bea

m in

tens

ity [1

016 a

ts]

Nozzle temperature [K]

FIG 16 Intensity of the hydrogen beam (states |1〉 and |2〉)injected into the compression tube as function of the nozzletemperature for different nozzle diameters D (primary molec-ular gas flow 1mbar ls dissociator power 300W)

100 200 300 400 50010

20

30

40

50

60

70

80 D = 20 mm

Bea

m in

tens

ity [1

016 a

ts]

Dissociator power [W]

FIG 17 Intensity of the hydrogen beam (states |1〉 and |2〉)into the compression tube as function of the dissociator powerfor a nozzle diameter of 2 mm (nozzle temperature 60K pri-mary molecular gas flow 1mbar ls)

V HYDROGEN BEAM PROFILES

Beam profiles were measured at various positions atvarious positions behind the last sextupole magnet withthe use of

bull a compression tube of reduced dimensions (5mmdiameter)

bull a crossed-beam quadrupole mass spectrometer andbull a supplementary method of reduction of MoO3 by

hydrogen

A Measurements with the compression tube

For the determination of the beam dimensions at twopositions 300mm and 337mm behind the last magnetthe compression tube setup (Fig 12) was used makinguse of the possibility of axial movement by the z ma-nipulator and of that to install a narrower and shortercompression tube of 5mm diameter and 50mm lengthto enhance the spatial resolution The xy manipula-tor provided a lateral displacement of the compressiontube by plusmn10 mm in x and y direction The center co-ordinates of the geometrical axis of the source had beendetermined with the use of a bi-directional laser cen-tered inside the bore of the central support plate (seeFig 1) The relative intensity distributions in the xzand yz planes given by the measured pressure in thecompression volume are shown in Fig 18 Fits by Gaus-sian distributions to the data yield full widths at halfmaximum Γx = (642plusmn 009)mm Γy = (699plusmn 006)mmfor the distributions measured at z = 300mm and Γx =(627plusmn 008)mm Γy = (658plusmn 008)mm at 337mm

00

02

04

06

08

10

12

00

02

04

06

08

10

12

0 5 10 15 20 25

00

02

04

06

08

10

12

14

0 5 10 15 20 25

00

02

04

06

08

10

12

14

Pre

ssur

e in

the

com

pres

sion

tube

[au

]

Z = 300 mm

(a) (b)

(c) (d)

Z = 337 mm

X [mm]

Y [mm]

FIG 18 Cross sections of the beam profile in the mid-planemeasured with compression tube of 5mm diameter and 50mmlength Measurements in the xz-plane (a c) and yz-plane (bd) performed at two different positions z = 300mm (a b)and z = 337mm (c d) behind the last sextupole magnet ofthe ABS The shaded area represents position and dimensionsof the compression tube used in intensity measurements

The center of gravity of the measured profile definedas

rc =

sum

ij

radic

x2i + y2j middot P (xi yj)

sum

ij

P (xi yj) (8)

where xi and yj give the position of the compression-tubeaxis and P (xi yj) is the pressure measured in the com-

13

pression volume The resulting rc shows a deviation of012 mm from the geometrical axis of the source Further-more the data measured with the narrow compressiontube of 25mm radius can be used to derive the fractionof the beam entering the compression tube of 5mm ra-dius used in the intensity measurement of Sec IV Theratio

η =

rile25 mmsum

0P (xi yj)

rile10 mmsum

0P (xi yj)

(9)

where ri is the distance of the compression-tube axis tothe beam axis yields η asymp 07

B Measurements with the QMS

The beam-profile studies of Sec VA were extendedwith a setup utilizing a crossed-beam quadrupole massspectrometer (QMS) in the setup of Fig 19 Contraryto the measurements with the compression tube thosewith the QMS allow to separate the atomic and molec-ular fractions in the beam A 2mm diameter aperturewas installed at the entrance of the sensitive volume of

FIG 19 Setup for the measurements of the beam profilewith the QMS (1) xy-table enabling two-dimensional dis-placement of the entrance window of the QMS against thegeometrical axis of the ABS (2) the QMS (3) manually op-erated beam shutter The beam dump is an axially mountedcryo pump

the QMS to improve the resolution compared with thatachieved by the compression tube of 5mm diameter usedin measurements of the preceding section The layout ofthe setup presented in Fig 19 shows that in the presentcase the profile could not be measured at a distance ofz = 300mm to the last magnet Instead measurementswere performed at z = 567mm and with installation ofan extension tube at z = 697mm The xy manipulatorenabled displacements of the aperture axis from the geo-metrical axis of the source in any direction within limitsset by the bore diameter of the xy manipulatorThe first measured distribution of the atomic hydro-

gen (Fig 20) showed a distinct deviation from azimuthalsymmetry indicating an insufficient relative alignment ofnozzle and skimmer The three threaded rods support-ing the dissociator with the nozzle via the three-legged

FIG 20 Two-dimensional distribution of the atomic hydro-gen component of the beam at z = 567mm before the nozzle-to-skimmer adjustment showing a disinct deviation from az-imuthal symmetry

plate (label 2 in Fig 1) allow one to vary the position ofthe nozzle relative to that of the skimmer while the sourceis running This possibility has been used to find a nozzleposition which results in an azimuthally symmetric distri-bution The achieved symmetric distribution is shown inFig 21 and profiles of the atomic hadrogen component inthe beam measured in x and y direction at z = 567mmand z = 697mm are presented in Fig 22 Fits by Gaus-sian distributions to the data yield full widths at halfmaximum Γx = (736plusmn 043)mm Γy = (668plusmn 080)mmfor the distributions measured at z = 567mm and Γx =(669plusmn 022)mm Γy = (638plusmn 027)mm at 697mm

C Reduction of MoO3 by hydrogen

In addition to the compression tube and the QMS tech-nique a supplementary attempt was made to determinethe beam profile by exposing molybdenium trioxide (a

14

FIG 21 The distribution corresponding to that of Fig 20 af-ter nozzle-to-skimmer adjustment resulting in azimuthal sym-metry

0

20

40

60

80

0

20

40

60

80

2 4 6 8 10 12 14 16 18 20 22

0

20

40

60

80

6 8 10 12 14 16 18 20 22 24

0

20

40

60

80

QM

S si

gnal

s [a

u]

(a)

Z = 567 mm

Z = 697 mm

(c)

(b)

X [mm]

(d)

Y [mm]

FIG 22 Profiles of the atomic hydrogen component inthe beam measured with the QMS 567mm and 697mmbehind the last magnet

yellowish powder) on a glass plate to the beam The prin-ciple of this method is based on the reduction of MoO3 toa lower oxide of blue colour It first was used in the ex-periment to measure the magnetic moment of the hydro-gen atom by splitting of the beam in an inhomogeneousmagnetic field 61

This method is much simpler than the time-consumingmeasurements described in Secs VA and VB It givesqualitative results as presented in Fig 23 A quantitativeanalysis however requires development of the measuringtechnique (eg preparation of glass plates study of theoptimum exposure time digital image processing)

FIG 23 Photo of the glass plate covered with molybdenumtrioxide MoO3 exposed to the atomic hydrogen beam

D Summary of the profile measurements

Table III summarizes results of the measurements ofthe ABS beam profile with the compression-tube and theQMS setup The larger errors of the widths measuredwith the QMS are due to the lack of measurements withthe dissociator switched off and the necessity to estimatethe background signal from the existing data Within theerrors the measured widths do not show a dependenceon the distance from the last magnet This facilitatesto position the feeding tube of the storage cell in a widerange of a distances to the last magnet The average val-ues Γx = (638 plusmn 060)mm and Γy = (684 plusmn 033)mmagree within the errors and yield a common width ofΓxy = (673 plusmn 029)mm The two-dimensional Gaus-sian distribution of this width allows one to estimate thefraction η of the beam intensity injected into the com-pression tube or a feeding tube For a tube of 10mmdiameter η = 078plusmn 003 comparable with η asymp 07 givenin Sec VA

TABLE III Dimensions (FWHM) of the atomic hydrogenbeam measured with the compression tube (CT) and thecrossed-beam quadrupole mass spectrometer (QMS) at dis-tances z to the last magnet along perpendicular directions xand y

z[mm] Γx[mm] Γy[mm]

CT 300 642 plusmn 009 699plusmn 006

CT 337 627 plusmn 008 658plusmn 008

QMS 567 736 plusmn 043 668plusmn 080

QMS 697 669 plusmn 022 638plusmn 027

15

VI DEGREE OF DISSOCIATION

Besides the intensity of the atomic beam it is impor-tant to determine the molecular fraction in the beamMolecules injected into the feeding tube reduce the po-larization of the target gas

A Measurements with crossed-beam QMS

In addition to the data on the profile of the atomichydrogen beam (Sec VB) data on the distributions ofmolecular hydrogen in the beam were taken too at thepositions z = 567mm and 697mm behind the last mag-net The relation between the degree of dissociation andthe QMS signals by the atomic and molecular beam com-ponent was given above by Eq (6) The coefficient kv =vmva however is chosen here under the assumptionthat the average velocity of the atoms is determined bythe nozzle temperature of 65K and that of the moleculesby scattering and recombination on the ABS chamberwalls at 290K This yields kv =

radic

2 middot 65290 = 067 ingood agreement with Ref13 where this coefficient wasdetermined by the measured velocity distributions undersimilar conditionsThe measured profiles of the atomic fraction (identical

to those of Fig 21) those of the molecular fraction andthose of the degree of dissociation deduced from Eq (6)are collected in Fig 24As it is seen from the figure the distribution of the

degree of dissociation shows a dip around the central linedue to the higher density of molecular hydrogen originat-ing from the nozzle The mean value in an aperture of

0

20

40

60

80

100

120

2 4 6 8 10 12 14 16 18 20 22

0

20

40

60

80

100

120

6 8 10 12 14 16 18 20 22 24

QM

S si

gnal

s [a

u]

(a)

(b)

(c)

00

02

04

06

08

10

12

Deg

ree

of d

isso

ciat

ion

()

X [mm]

Y [mm]

(d)

00

02

04

06

08

10

12

Z = 567 mm

Z = 697 mm

FIG 24 Spatial distributions of H1 (bull)H2 () and degree ofdissociation () averaged over 3mm wide bands in the xz andyz planes respectively (here the z-axis is the geometrical axisof the ABS)

10mm diameter results as α = 095 plusmn 004

B Measurements with the Lamb-shift polarimeter

A cup in the quench chamber of the Lamb-shift po-larimeter (LSP) described in Ref8 allows one to mea-sure the currents Icup(H1) and Icup(H2) of the H+

1 andH+

2 ions extracted from the ionizer and separated bythe Wien filter with the cesium evaporation and the spinfilter switched off The relation between the degree ofdissociation α and the measured currents is

α =Icup(H1)minus r1

r2Icup(H2)

Icup(H1)minus r1r2Icup(H2) + 2kv

r2Icup(H2)

(10)

Among the three coefficients kv = 067 as for the mea-surement with the QMS For the electron energy of about100keV the ratio r1 of dissociative to non-dissociativeionization of H2 is8

r1 =σ(H2 rarr 2H+

1 )

σ(H2 rarr H+2 )

= 0095plusmn 0008 (11)

and the ratio between the ionization cross sections is857

r2 =σion(H2)

σion(H1)= 17plusmn 01 (12)

At the standard operation parameters of the source(Sec IV) the measured currents are Icup(H1) = (125 plusmn05) nA and Icup(H2) = (64 plusmn 01) nA yielding α =(096plusmn004) in excellent agreement with the value result-ing from the measurements with the QMS (Sec VIA)

VII BEAM POLARIZATION

The Lamb-shift polarimeter was designed built andtested at Universitat zu Koln7 It was used to measureand to optimize the polarization of the atomic hydrogenand deuterium beams delivered by the ABS Details arefound in Ref7The vector polarization pz for hydrogen is defined by

the relative hyperfine-state occupation numbers N(mI)

pz =N(+ 1

2 )minusN(minus 12 )

N(+ 12 ) +N(minus 1

2 ) (13)

for deuterium

pz =N(+1)minusN(minus1)

N(+1) +N(0) +N(minus1) (14)

Deuterium tensor polarization pzz is given by

pzz =N(+1) +N(minus1)minus 2 middotN(0)

N(+1) +N(0) +N(minus1) (15)

16

These polarizations can be derived from the measuredLyman-α peak strengths S by application of a number ofcorrection factors78

Typical Lyman α spectra measured with the polarizedhydrogen and deuterium beam from the ABS are shownin the Figs 25 and 26

MFT 2-3

200 300 400 5000

40

80

120

160

Magnetic field in the spinfilter [au]

PM

T si

gnal

[au

]

mI = -12

(a)

mI = +12

100 200 300 400 5000

40

80

120

160(b)

MFT 2-3WFT 1-3

mI = +12

mI = -12

FIG 25 Lyman-α spectra measured with the polarized hy-drogen beam (a) population change from state |2〉 to state|3〉 induced by the MFT unit (b) same as (a) with subsequentpopulation change from state |1〉 to state |3〉 induced by theWFT unit

MFT 3-4SFT 2-6

0

2

4

6

8

mI = -1m

I = 0

(a)

mI = +1

0

2

4

6

8(b)

MFT 3-4WFT 1-4 2-3

mI = +1 m

I = 0

mI = -1

mI = +1

mI = 0

mI = -1

mI = +1

mI = 0

mI = -1

0 1000 2000 3000

0

1

2

3

4 (c)

PMT

sign

al [a

u]

Magnetic field in the spinfilter [au]

WFT 1-4SFT 2-6

0 1000 2000 30000

2

4

6

8WFT 1-4SFT 3-5

(d)

FIG 26 Lyman-α spectra measured with the polarized deu-terium beam (a) and (b) vector polarization resulting fromsubsequent transitions MFT (3 rarr 4) and SFT (2 rarr 6) andWFT (1 rarr 4 2 rarr 3) respectively (c) and (d) tensor polar-ization resulting from subsequent transitions WFT (1 rarr 4)and SFT (2 rarr 6) and SFT (3 rarr 5) respectively

The polarization values for the hydrogen and the deu-terium beam derived from the Lyman-α peak-strengthratios with application of the necessary corrections aresummarized in Table IV

The vector polarization for hydrogen of the first linereflects the population of state |1〉 and state |2〉 accordingto the Eqs (3) and (13) The value of 091 deduced fromthe calculated transmission values is confirmed by themeasured one

TABLE IV The vector polarization pz of the hydrogen beamand the vector and the tensor polarization pzz of the deu-terium beam from the ABS measured with the Lamb-shiftpolarimeter

populated pz pzz

state(s)

Hydrogen |1〉 +089plusmn 001 -

|3〉 minus096plusmn 001 -

Deuterium |1〉+ |6〉 +088plusmn 001 +088plusmn 003

|3〉+ |4〉 minus091plusmn 001 +085plusmn 002

|3〉+ |6〉 +0005plusmn 0003 +090plusmn 001

|2〉+ |5〉 +0005plusmn 0003 minus171plusmn 003

VIII CONCLUSIONS AND OUTLOOK

In this paper we present the detailed description ofthe major components of the atomic beam source (ABS)for the polarized internal gas target of the magnet spec-trometer ANKE in COSY-Julich The ABS was builtfor the purpose of extending the physics program ofANKE from unpolarized and single-polarized investiga-tions with stored beams towards double-polarized exper-iments1 thus facilitating nuclear reaction studies involv-

ing ~p~p ~p~d ~d~p and ~d~d initial statesThe mechanical design took into account that at

ANKE the source has to be mounted vertically and trans-versely movable together with the transverse motion ofthe spectrometer magnet D2 The design of the system ofsextupole magnets took advantage of the developments inthe field of rare-earth permanent magnets (NdFeB) Ded-icated tools and methods were developed to determineand to optimize the source parameters ie intensity de-gree of dissociation and polarization Special emphasiswas put on the measurements of the spatial distributionsof the atomic and molecular beam near the focus wherethe feeding tube of the storage cell is located The ABShas been used in a number of investigations at ANKEthe commissioning effort to prepare the target for the usewith polarized H is described in Ref62 Performed stud-ies of the deuteron-charge exchange reaction are summa-rized in Ref6364 studies in near-threshold pion produc-tion are reported about in Ref65The ABS resides at the ANKE target position for a few

months per year only thus during the remaining timeit is used for other studies It had been observed thatthe nuclear polarization in recombined hydrogen is par-tially retained after recombination66 as well as evidencefor nuclear tensor polarization in recombined deuteriummolecules67 In order to investigate this recombinationprocess in more detail a special setup has been developedin the framework of an ISTC project68 and the recombi-nation process for different cell-wall coatings and differ-ent polarizations of the injected hydrogen or deuterium

17

atoms as function of cell-wall temperature strength ofthe magnetic holding field and gas pressure in the cell ispresently investigated69ndash71

Appendix A Preparation of Discharge Tubes and Nozzles

1 Tube Treatment

One end of the discharge tube is machined at a 45

angle while the other is kept flat Both ends are thenremelted and the tubes are tempered at 150 C Thetubes are further treated according to the procedure de-scribed in Ref72 which includes successive cleaning withacetone methanol distilled water and subsequent rins-ing by a 21 acid mixture of concentrated HF (40) andHCl (32) for 5 min The tubes are then flushed bydistilled water and dried

2 Nozzle Treatment

The nozzles are cleaned in an ultrasonic bath oftrichlorethylene acetone methanol and finally distilledwater all at 50 C Anodizing takes place in sulfuric acidto form a thin layer of Al2O3 as described in Ref72 Af-terwards they are immersed in distilled water for 30minat 95 C

ACKNOWLEDGMENTS

The authors want to thank OWB Schult Institut furKernphysik (IKP) Julich who initiated the polarizationprogram of ANKE Thanks go to the design office themechanical workshop and especially to WR Ermer allIKP Valuable advice was received from the PINTEX col-laboration at IUCF from the target group at HERMESespecially NKoch and from DToporkov BINP Novosi-birsk The support by VKoptev PNPI Gatchina whoregrettably passed away in January 2012 is gratefullyacknowledged Thanks go also to R Poprawe and col-leagues Fraunhofer-Institut fur Lasertechnik Aachenwhere the encapsulations of the magnets were laser-welded

1AKacharava F Rathmann and CWilkin Spin Physics fromCOSY to FAIR COSY Experiment Proposal No 152 (2005)Available under httparXivnucl-ex0511028

2WHaeberli in Proc 2nd Int Symp on Polarization Phenomenaof Nucleons Karlsruhe 1965 Eds P Huber and H SchopperExperientia Supplementum 12 64 (Birkhauser Verlag 1966)

3E Steffens and WHaeberli Rep Progr Phys 66 1887 (2003)4SBarsov et al Nucl Instr and Meth A 462 364 (2001)5RMaier Nucl Instr and Meth A 390 1 (1997)6KGrigoryev et al Proc 14th International Workshop on Po-larized Sources Targets and Polarimetry (PSTP 2011) 12-16 September 2011 StPetersburg Russia eds KGrigoryevPKravtsov and AVasilyev ISBN 978-5-86763-282-3 61 (2011)

7REngels et al Rev Sci Instrum 74 4607 (2003)8REngels et al Rev Sci Instrum 76 053305 (2005)9TWise et al Nucl Instr and Meth A 336 410 (1993)

10WADezarn et al Nucl Instr and Meth A 362 36 (1995)11T Rinckel et al Nucl Instr and Meth A 439 117 (2000)12F Stock et al Nucl Instr and Meth A 343 334 (1994)13ANass et al Nucl Instr and Meth A 505 633 (2003)14VDerenchuk et al Proc Conf Polarized Ion Sources and Polar-ized Gas Targets Madison WI 1993 Eds LWAnderson andWHaeberli AIP Conf Proc 293 72 (American Institute ofPhysics 1994)

15HOkamura et al see Ref14 p 8416KHatanaka et al Nucl Instr and Meth A 384 575 (1997)17Manufacturer Schiffer Metall- amp Vakuumtechnik 52428 JulichGermany

18Single-stage type RGS120 refrigerating capacity 120W at 80Kand 20W at 30K Leybold Vacuum GmbH 50968 Koln Ger-many

19Mini UHV gate valve series 010 VAT Germany GmbH 85630Grasbrunn Germany

20Type F3 fomblin oil Pfeiffer Vacuum GmbH 35614 Asslar Ger-many

21Model HU 1 Leybold Vacuum GmbH 50968 Koln Germany22Manufacturer SK Industriemodell GmbH 52531 Ubach-Palenberg Germany

23Type PFG 600 RF with automatic matchbox PFM 1500 A-INDHuttinger Elektronik GmbH 79110 Freiburg Germany

24WKorsch PhD Thesis Philipps Universitat Marburg (1990)25F Stock et al Int Workshop on Polarized Beams and PolarizedGas Targets Koeln Germany 1995 Eds HPaetz gen Schieckand L Sydow (World Scientific Publ Co 1996) p 260

26The first number denotes the outer diameter and the second onethe wall thickness

27Type Duran 8330 equivalent to Corning 7740 (Pyrex) SchottAG 55122 Mainz Germany

28Ultra-Kryomat RUL 80-D Lauda DrRWobser GmbH 97912Lauda-Konigshofen Germany

29ODU-Kontakt GmbH 84444 Muhldorf Germany30Handbook of Chemistry and Physics Ed RCEast (The Chem-ical Rubber Co 1973) p E-10

31NKoch and E Steffens Rev Sci Instrum 70 1631 (1999)32AVassiliev et al Petersburg Nuclear Physics Institute ReportNP-32-1997 No 2175 (1997)

33BLorentz Diploma Thesis Ruprecht-Karls-Universitat Heidel-berg (1993)

34ANass and E Steffens Nucl Instr and Meth A 598 653(2009)

35WHaeberli Ann Rev Nucl Sci 17 373 (1967)36AVassiliev et al Rev Sci Instr 71 3331 (2000)37WKubischta Proc Workshop on Polarized Gas Targetsfor Storage Rings Heidelberg 23-26 September 1991 EdsHGGaul E Steffens and K Zapfe (Max-Planck-Institut furKernphysik Heidelberg)

38The labeling of the hyperfine states as |1〉 = |mj = +12 mI =+12〉 |2〉 = | + 12minus12〉 |3〉 = | minus 12+12〉 and |4〉 =|minus12minus12〉 for hydrogen and |1〉 = |+12+1〉 |2〉 = |+12 0〉|3〉 = | + 12minus1〉 |4〉 = | minus 12minus1〉 |5〉 = | minus 12 0〉 and

3

the rf transition units and the slow control system aredescribed in the subsequent subsections

B Pumping System

The system of pumps on the chambers I to V of theABS (Fig 1) is shown in Fig 2 the types of the pumpstheir pumping speeds and the achieved pressures arelisted in Table I Chamber I with the highest gas loaddue to the effect of the skimmer is pumped by two strongturbomolecular pumps Each of them is backed by asmaller turbomolecular pump Their exhausts are con-

FIG 2 The system of pumps on the chambers I to V of theABS (Fig 1) The specifications of the pumps are listed inTab I The figure also contains the bypass system for the gasload from regeneration of the cryopumps

nected to a common pump of the same type The totalcompression ratios of the serially connected turbomolec-ular pumps yields sufficient pumping speed for a primarymolecular gas flow up to 3mbar ls into the dissociatorThe line of pumps is backed by two oil free membranepumps According to the lower gas load chamber II isevacuated by a simpler line consisting of two turbomolec-ular pumps and a membrane pump All turbomolecularpumps are operated with use of synthetic oil20 Com-pared to mineral oil synthetic oil allows longer pump-

ing of hydrogen before oil exchange becomes mandatoryStrong cryopumps are utilized on chambers III and IVwhile the lowest chamber with the WFT and SFT unitsis evacuated by a smaller cryopump All cryopumps areequipped with temperature-controlled heating units forregeneration on both cooling stages21 Heating up toroom temperature while pumping the resulting gas loadby the bypass system and cooling down again takes about25 to 3 hours

C Baffles

The layout of the baffles had been defined by the nec-essary narrow vertical extension of the ABS and the re-quirement to provide sufficient pumping speed in viewof the heavy gas load to the vacuum chambers I and IIFurthermore the construction aimed at the possibilityof axial movements from outside to optimize the beamparameters by variation of the distances between nozzleskimmer and diaphragm The resulting shape for theupper baffle is shown in Fig 3 Except for details inthe openings the lower baffle carrying the diaphragmis identical The layout of the upper vessel and the baf-fles was done under the boundary conditions that on theone hand the baffles have to be movable within the ves-sel and on the other hand the slits between cylindricalsurfaces of the baffles and the inner surface of the vesselhas to be narrow to reach a small gas conductance Thediameter of the inner vessel surface is 3892mm with alongitudinal and non-circular tolerance of +02mm theouter diameters of both baffels are (3887minus02)mm The

TABLE I List of the devices employed in the ABS pumpingsystem composed of turbomolecular pumps (TP) membranepumps (MP) and cryopumps (CP) with nominal the individ-ual capacities CH2

the pumping speeds SH2 and the achieved

pressures at a primary gas flow of 10 mbar ls

Cham- Device Type CH2SH2

Pressure

ber [bar l] [ls] [mbar]

I TP1-2 TPH 2200a 2800 10minus4

TP4-57 TMH 260a 180

MP1-2 MVP 100-3a 1812c

II TP3 TPH 2200a 2800 10minus6

TP6 TMH 260a 180

MP3 MVP 100-3a 1812c

III CP1 COOLVAC 3000b 28d 5000 10minus7

IV CP2 COOLVAC 1500b 28d 5000 5 middot 10minus8

V CP3 COOLVAC 800b 43d 1000 5 middot 10minus8

a Pfeiffer Vacuum GmbH 35614 Asslar Germanyb Leybold Vakuum GmbH 50968 Koln Germanyc Pumping speed at 1000 mbar10 mbard At 10minus6 mbar

4

conductances of the slits of le5 ls are small comparedwith the applied pumping speed Because of the compli-cated shape identical raw pieces of cast Al22 were ma-chined to the final dimensions Contrary to the lower baf-

FIG 3 3D drawing of the upper baffle separating the vac-uum chambers I and II with the two wide cuts in front ofthe turbopumps (1) and the openings for the viewport (2)the skimmer (3) four of the 16 ball bearings (4) and the foursupporting rods (5)

fle the upper baffle until now has to be installed togetherwith the flange of the upper vessel at a fixed axial posi-tion (cf Fig 1) In order to reach full flexibility in vary-ing the nozzle skimmer and collimator relative positionsfrom outside the installation of rotational feedthroughsin the flange of the upper vacuum vessel is necessary aforeseen but not yet implemented feature

D Dissociator

To dissociate molecular hydrogen or deuterium to neu-tral atoms an rf discharge is employed which is fed bya 13560 MHz generator23 delivering up to 600 W into a50Ω load The layout of the dissociator shown in Fig 4is similar to that of the FILTEX design2425 The dis-charge tube (empty11times15mm)26 is surrounded by two coax-ial tubes (empty204times18mm and empty28times2mm) all three aremade from borosilicate glass27 The coolant streams fromthe inlet connection down between the discharge tubeand the middle tube and after flow reversal at the lowerend (Fig 5 label 2) it streams up in the outer slit to theoutlet connection In a closed loop the coolant inlet tem-perature (typically 15 C for a 50 water ndash 50 ethanolmixture) is stabilized by a cooling thermostate28 whichwould allow coolant temperatures down to minus80 C Therf coil and the capacitor at fixed relative positions canbe positioned from outside by means of a sliding rf con-nection29 and the feed-through ground connection This

FIG 4 3D drawing of the dissociator (1 gas inlet 2 slidingground connection 3 coolant inlet 4 coolant outlet 5 rfinput 6 sliding rf connection 7 grounded capacitor plate8 rf coil 9 rf-fed capacitor plate 10 isolating plastic sup-port rings 11 grounded limiter plate 12 lower end of thecoolant-guiding tubes 13 tube support and connection tothe coldhead (details are given in Fig 5) 14 lower end of thedischarge tube)

enables variation of the plasma-nozzle distance to opti-mize the atomic beam intensity while the plasma is burn-ing The treatment of the discharge tube and the nozzleprior to installation is described in Appendix A

E Nozzle

The nozzle cooled via the heat bridge and the sur-rounding components are shown in Fig 5 The nozzlemade from 995 Al has a simple conical shape with thetip cut Comparative measurements show that nozzleswith sharp edges as used eg in the Madison source9 donot yield higher atomic beam intensities First a sharpedge is more difficult to produce due to the softness ofpure Al Second the low heat conductance of a sharpedge leads to appreciable temperatures of the nozzle tipcaused by recombination of atoms on the nozzle surface

5

The temperature at the bottom of the nozzle is measuredwith a Pt-100 sensor and it is stabilized with an accuracyof plusmn05K utilizing a heater Measurements with temper-ature sensors placed along the outer nozzle surface haveshown a temperature increase from 60K at the nozzlebottom to sim200K at the sharp nozzle tip In the follow-ing the nozzle temperature is defined as that measuredwith this Pt-100 sensorWith the present system of sextupole magnets the

maximum atomic beam intensity feeding the storagecell is obtained with a nozzle-orifice diameter of 23mmand a nozzle-tip to skimmer-tip distance of 15mm at askimmer-tip diameter of 44mm and a skimmer-tip to di-aphragm distance of 17mm The 2mm thick diaphragmwith a conical bore opening from 95mm to 99mm to-wards the first permanent sextupole magnet shields themagnet from heating by atoms recombining on its sur-face The slit between the diaphragm and the front faceof the magnet enables pumping of gas from the entranceto the magnetThe Teflon washer and the stainless steel support sep-

arate the cold lower end of the heat bridge from thewarm lower end of the dissociator The dimensions of

FIG 5 Technical drawing including the lower end of theheat bridge and the dissociator the nozzle surroundings andthe first sextupole magnet (in scale 1 discharge and coolant-guiding tubes 2 coolant-reversal piece 3 heat flow reducingTeflon washer 4 sliding heat connection 5 stainless steelconnector 6 groove for nozzle-heating element 7 lower endof the Cu heat bridge 8 nozzle fixture 9 nozzle 10 baf-fle separating the chambers I and II with a viewport 11stainless steel beam skimmer 12 Cu diaphragm 13 firstsextupole magnet and 14 baffle separating the chambers IIand III

these two components and the sliding heat connector aworked-over sliding high current connector similar to therf connector in the dissociator define the temperatureof the lower end of the discharge tube relative to thatof the nozzle The discharge tube adapted at its lowerend to the nozzle by a chamfered edge is pressed to thenozzle by a viton O-ring at its upper end The two O-rings around the discharge tube in the lower part of thedissociator seal against the atmosphere By this designonly minor forces are exerted to the discharge tubeThe removable viewport in the baffle and the window

flange in the upper vacuum vessel (on the right-hand sideof chamber II in Fig 1) allows one to observe the nozzlestatus from the outside and to exchange nozzles withoutremoval of the dissociator from the setupThe heat bridge from the coldhead to the nozzle is

made from electrolytic Cu The flexible link between thecoldhead and the heat bridge consisting of about 200high-purity Cu strands of 1 mm diameter allows for thethermal expansions of the cold and the warm compo-nents The total cross section of the strands and theirheat conductance is smaller than that of a massive Cubody This deficiency however is reduced by clampingthe flexible link directly to the coldhead At its operat-ing temperature of about 30K the thermal conductivityof Cu is about 11 9 and 5 times higher than that at300 100 and 60K respectively30 Thus the reductionof the conductance of the entire heat bridge by the flexi-ble link is minimized by placing it at the coldhead Withthe present system cooling the nozzle down from roomtemperature to 60K needs about 15 hours The heat-ing element facilitates warming up to room temperaturewithin about one hourFurthermore avoiding the maze of cold Cu strands

around the nozzle ie a labyrinthic cold surface com-pared to an earlier solution31 leads to improved pumpingconditions in the nozzle-skimmer area where the highestgas load has to be pumped offIn an earlier phase of the ABS development attempts

have been made to use a cryogenic Ne heat-pipe of 20Wcooling power instead of the usual solid Cu bridge toachieve faster cooling and warming of the nozzle becauseof the lower heat capacity32 An observed instability inthe necessary operation mode however lead to difficul-ties in nozzle-temperature stabilization In view of thefact that the cooling and warming-up times reached withthe Cu bridge were satisfying and that its use avoids theadditional precautions imposed by the heat-pipe opera-tion it has been replaced by the Cu bridge

F Magnet System

The design of the magnet system was made for a setof sextupole magnets consisting of permanently magne-tized segments made from NdFeB compounds deliveringpole-tip fields around 15 T Tracking calculations fromthe nozzle to the feeding tube of the storage cell were

6

performed with the use of a computer code originallydeveloped for the FILTEX ABS24 The boundary condi-tions by the layout of the target setup were the availabledistance of about 1250mm from the nozzle to the feeding-tube entrance of 10mm diameter and the distance fromthe exit of the last magnet to the feeding-tube entrance of300mm necessary to install the SFT and WFT units andthe gate valve between the ABS and the target chamberThe laboratory velocity distribution of the atoms in

the supersonic beam from the nozzle is described by amodified Maxwellian distribution

F(~vd Tb) =( m

2 k Tb

)32exp

[ minusm

2 k Tb(~v minus ~vd)

2

]

(1)

where m is the mass of the atoms and k is the Boltzmannconstant According to time-of-flight studies33 the driftvelocity along the beam axis vd and the beam temper-ature Tb for a primary molecular gas flow of 1 mbar lsand a nozzle-orifice diameter of 2mm follow a linear de-pendence on the nozzle temperature Tn For hydrogenvd[ms] = 1351 + 61 middot Tn[K] and Tb = 029 middot Tn and fordeuterium vd[ms] = 1070+345middotTn[K] and Tb = 025middotTnAs starting conditions of a track a random generator

selects a point in the nozzle orifice one within the di-aphragm in front of the first magnet and an atom ve-locity |v| In linear molecular flow approximation (cfthe discussion in Ref34) this defines ~v for the track be-tween the nozzle and the first magnet According to thegeometrical boundary conditions and the velocity distri-bution of Eq (1) the event is either rejected or used inthe further track calculation Within the magnet theevolution of the track is calculated stepwise by numeri-cal integration of the equation of motion over integrationtimes of 2micros corresponding to track lengths of 36mmfor a typical particle velocity of 1800ms The pureradial force acting on an atom within the field of the

sextupole magnet is ~Fr = minusmicroeff middot δBδr middot ~rr The ef-fective magnetic moment resulting from the Breit-Rabidiagram (eg Ref35) as microeff = δWδB is positive (neg-ative) for atoms in the hyperfine states with the electron

spin parallel (antiparallel) to ~B in the magnet aperturewhich therefore are deflected towards (away from) thebeam axis In the drift sections between the two magnetgroups and between the last magnet and the feeding tubethe trajectories are assumed as straight linesA variety of systems were studied all under the as-

sumption of Tn = 60K and pole-tip fields of 15T Asystem utilizing 6 magnets was found to yield satisfyingboth separation of the atoms in the microeff lt 0 and microeff gt 0states and focusing of the microeff gt 0 states into the feedingtube Optimization of the parameters led to the systemlisted in Table II (The tracking calculations yielding themagnet dimensions for the order to the manufacturer hadbeen performed for a slightly different geometry) Thetable gives the two distances at which intensity mea-surements with the compression tube were performedThe Fig 6 shows the projection of the trajectories of Hatoms in the microeff gt 0 states calculated for this system

TABLE II Final dimensions and axial positions of the sourcecomponents (pole-tip field strenghts Blowast

0 as measured afterdelivery36 inner diameters (empty0) outer diameters (empty1) axialdimensions (ℓ) and distances (∆) between the componentsThe lower three lines give the two distances and the dimen-sions of the compression tube used in the intensity measure-ments

component Blowast

0 [T] empty0 [mm] empty1 [mm] ℓ [mm] ∆ [mm]

Nozzle orifice 23 33150

Skimmer 44304a 130169

Diaphragm 9599a 2036

Magnet 1 1630 10401412a 3998 400194

Magnet 2 1689 15982212a 6404 650194

Magnet 3 1628 2804 9400 70014297

Magnet 4 1583 3004 9402 38011010

Magnet 5 1607 3006 9400 5501150

Magnet 6 1611 3002 9404 550030003370

Compr tube 100 110 1000

a Conical openingthe first number denotes the measureddiameter of the entrance the second that of the exit aperture

One recognizes two groups of trajectories one with anintermediate focus and another one with focusing intothe feeding tube The present result like those of othergroups (see eg Ref33) confirms the expectation37 thatthe transmission as function of the atom velocity shouldshow two maxima one below and one above the mostprobable velocityThe transmission Tr of the system is defined as the

fraction of tracks ending within the entrance of the feed-ing tube to those passing the diaphragm in front of thefirst sextupole magnet For the four hyperfine states ofhydrogen38 the calculations yield Tr(|1〉) sim Tr(|2〉) =042 (for both microeff gt 0) and Tr(|3〉) = 0001 andTr(|4〉) = 00004 (for both microeff lt 0)The performed tracking calculations do not account for

intra-beam and residual-gas scattering The calculatedtransmissions thus only allowed one to estimate upperlimits of the expected atomic beam intensity Iin into thefeeding tube For a primary molecular flow q(H2) theintensity Iin(H) with atoms mainly in the states |1〉 and|2〉 (microeff gt 0) was expected as

Iin(H) = q(H2) middot 2α middot Ω

2πmiddot 14

i=4sum

i=1

Tr(|i〉) (2)

For the degree of dissociation α a routine value of 08(see eg Ref9) was assumed Ω = 0022π is the solid

7

angle covered by the collimator aperture The factor 14reflects the assumption that the four substates in theatomic beam from the nozzle are equally populated Forq(H2) = 1mbar ls or 27middot1019H2 moleculess one expectsIin(H) sim 1 middot 1017H atomssAs described in the subsequent section the rf tran-

sition units are used to change the relative occupationnumbers of the states The trajectory code allows oneto simulate this change by assigning a microeff of one of thestates to the atoms before they pass a magnet As an ex-ample the medium-field transition unit (MFT) behindmagnet No 3 (see Fig 1) brings H atoms from state |2〉into state |3〉 This is simulated by assigning microeff(|2〉) gt 0to the atoms in the magnets 1minus3 and microeff(|3〉) lt 0 in themagnets 4 minus 6 where they get deflected from the beamaxis This results in a small value Tr(|2〉) = 0017 Fromthis value and the above value Tr(|1〉) = 042 the vectorpolarization is expected as

pz =Tr(|1〉)minus Tr(|2〉)Tr(|1〉) + Tr(|2〉) = 091 (3)

under the assumption of 100 efficiency of the transitionunitThe design and the properties of the permanent sex-

tupole magnets39 were discussed in an earlier paper36To achieve the pole-tip field values of sim15T each mag-net was produced from 24 segments employing three dif-

FIG 6 Projection of the 3-dimensional trajectories of hydro-gen atoms in hyperfine states |1〉 and |2〉 (effective magneticmoment microeff gt 0)) from the nozzle (empty = 2mm Tn = 60K)to the storage cell calculated for the magnet arrangement ofTable II and pole-tip fields of 15 T The positions and lat-eral dimensions of the six magnets and the feeding tube areindicated (in red)

ferent types of NdFeB compounds The expected pole-tip values (Table II) and the precise radial dependenceB(r) sim r2 within the magnet apertures were confirmedFor the first time the predicted high multipole compo-nents40 up to a 102-pole structure very near to the aper-ture surface could be measured36After the field measurements the magnets were encap-

sulated to prevent diffusion of hydrogen into the magnetmaterial which might deteriorate the magnetic proper-ties and to avoid the pumping of gas from the sinteredmagnet bodies The housings were made from thin stain-less steel cans of 02mm thickness for the conical andcylindrical walls within the magnet apertures and 03mmfor the front and end covers During the final welding toclose the housings with magnets installed the local tem-perature of the magnet material had to be kept belowthe Curie temperature of 60 C This was achieved bywelding with the use of a pulsed 15Hz NdYAG laserdelivering 11 J in a pulse of 2ms41 Overlapping weldspots of sim03mm diameter set around the adjacent cir-cular 02mm thick weld lips allowed one to finish thehousings with leak rates sim 10minus10mbar ls Inside thehousings the magnets were fixed to suppress axial androtational movements caused by the force of the adja-cent magnets Finally the free slits within the housingswere filled by sim20mbar krypton to enable leak tests bymass spectroscopy

G Radio Frequency Transition Units

The ABS is equipped with three types of transitionunits a weak field a medium field and a strong field rftransition unit (WFT MFT and SFT units) Togetherwith the selecting properties of the sextupole magnetsthey enable one to achieve all vector and tensor polar-izations of the atomic hydrogen and deuterium gas inthe storage cell In all three units transitions betweenthe hyperfine states split according to the Breit-Rabi di-agram by a static magnetic field (see eg Ref35) areinduced by the magnetic component (Brf) of an rf fieldleading to changes in the population of the states Thestatic field Bstat consists of two parallel components ahomogeneous field Bhom and a superimposed weaker gra-dient field Bgrad both orthogonal to the beam directionThe field gradient along the beam direction is requiredto satisfy the condition of adiabatic passage3542The assemblies of the WFT and the MFT units are

similar43 The layouts follow those of the units devel-oped for the HERMES experiment44 In both units therf field is produced by a coil with the axis along the beamdirection and consequently Brf orthogonal to Bstat TheMFT unit is shown in Fig 7 Figure 8 schematicallyshows one of the grooved aluminum frames with thewindings producing the gradient field A WFT unitis operated in a weak magnetic field Bstat le10G for hy-drogen and le5G for deuterium where the total atomicspin F is a good quantum number In hydrogen the

8

F = 1 levels |1〉 |2〉 and |3〉 with magnetic quantumnumbers mF = +1 0 and minus1 respectively can be re-garded as equally spaced In deuterium the same holdsfor the four F = 32 levels |1〉 |2〉 |3〉 and |4〉) withmF = +32 +12 minus12 and minus32 respectively andfor the two F = 12 levels |5〉 and |6〉 with mF = minus12and +12 respectively The magnetic component of therf dipole field induces transitions between each pair ofneighboring mF states with ∆mF = plusmn1 |∆mF| = 2transitions are forbidden The interchange of the popu-lation between the states |1〉 and |3〉 in hydrogen eg iscaused by a two-quantum transition via the intermedi-ate state |2〉 In the classical description of the adiabaticpassage method42 the population change should not de-pend on the sign of the magnetic field gradient relativeto the beam direction An exact quantum-mechanicaltreatment4546 however indicates that cleaner popula-tion changes from state |1〉 to |3〉 in hydrogen and fromstate |1〉 to |4〉 in deuterium are obtained with a nega-tive field gradient ie a Brf field decreasing in the beamdirection Deviations from adiabaticity are discussed inRef4547

The MFT unit is operated at higher values of Bstatwhere the differences in the energy spacings of pairs of hy-

FIG 7 Three-quarter-section view of the MFT unit with thesupport structure (1 self-supporting rf coil with spacers 2pick-up loop 3 Al tubes defining the length of the transition-inducing rf field 4 Cu cavity 5 coil around the pole shoe(6) providing the static field Bstat 7 slit between pole shoeand cavity wall housing the gradient-field coil 8 componentsof the static magnet yoke also serving as shielding againstexternal fields 9 cavity-positioning element 10 Cu padscooled by means of water-carrying tubes The cavity withthe rf coil and the pick-up loop can be taken out from thesurrounding components

z

Bstatic

transition

region

FIG 8 Arrangement of the windings producing the staticgradient field Bgrad shown in the left-hand side of the figureIn all transition units the field lies in the direction of the statichomogeneous field the field gradient dBdz lies in the beamdirection which defines the z axis In z direction the tran-sition reagion (indicated by the dashed lines) is confined tothe range of constant gradient by the Al tubes in orthogonaldirection by the beam diameter

perfine states with ∆mF = plusmn1 allow one to select singletransitions Originally developed for an polarized alkaliion source48 the MFT unit now is a standard compo-nent in polarized hydrogen and deuterium sources as dis-cussed eg in Ref49 Appropriate choice ofBhom Bgradand the rf frequency allows one to induce selected tran-sitions |1〉 harr |2〉 and |2〉 harr |3〉 in hydrogen or |1〉 harr |2〉|2〉 harr |3〉 and |3〉 harr |4〉 in deuterium Furthermore thechoice of the field gradient allows one to achieve consecu-tive transitions As an example a negative field gradientin the MFT unit behind the first set of magnets ie aB field decreasing in beam direction at a fixed rf fre-quency leads to the sequence of the transitions |3〉 rarr |4〉|2〉 rarr |3〉 and finally |1〉 rarr |2〉 in deuterium leaving thestate |1〉 empty

The SFT unit is used to induce transitions betweenstates in the upper and lower hyperfine multiplet in hy-drogen and deuterium Contrary to the historical nameindicating a strong magnetic field the SFT unit is op-erated with magnetic fields comparable to those used inthe MFT unit The transition frequencies are comparablewith those of the hyperfine splitting (1420MHz for hy-drogen and 327MHz for deuterium) and thus are muchhigher than those in the WFT and MFT units The rffield in a SFT unit is produced by a twin-line resonatorinside a Cu box tuned to the λ4 resonance50 The SFTunit51 is shown in Fig 9 Again the layout follows that ofthe unit used in the HERMES experiment44 Two vari-able capacitors at the free ends of the conducting rodsfed by the rf power with a relative phase shift of 180 allow one to tune the device

9

FIG 9 Three-quarter-section view of the rf cavity of theSFT unit for deuterium (1 the two resonant-field creatingconductors 2 the adjustible capacitor plates 3 Cu cavity)The inner dimensions of the cavity are 56mm along Bstat36mm orthogonal to it and 36mm height The cross sectionof the conductors is 14times 4mm2

H Slow Control System

Industrial components providing reliable and long-term support were selected for the control system of thewhole setup consisting of the ABS and the diagnosticstools the storage cell positioning system the Lamb-shiftpolarimeter and the supply system of a calibrated flowof unpolarized molecular gas The interlock system hasbeen implemented on the basis of SIEMENS SIMATICS7-300 family of programmable logic controllers In orderto unify the interfacing to the control computer all front-end equipment is connected via the PROFIBUS DP field-bus The process control software was implemented us-ing the Windows-based WinCC toolkit from SIEMENSThe system controls the operation of the pumps and thevalves It reads the pressure gauges and controls theregeneration cycles of the cryopumps Via a control net-work the temperature of the nozzle is stabilized withinplusmn05K Furthermore all power-supply units rf genera-tors and amplifiers are set and controlled The wholevariety of components to be controlled the logical struc-ture of the control and interlock system and a separatedevice for parameter studies are described in Ref52

III STUDIES OF THE FREE HYDROGEN JET

A Atomic beam profile near the nozzle

A novel device has been used to measure the profile ofan atomic beam via the deposition of recombination heaton thin wires in a two-dimensional grid5354 Atoms stuckon the surface of gold-plated tungsten wires of 5microm diam-eter recombine and are reemitted as molecules The re-combination heat (4476 eV per hydrogen molecule) leadsto a change of temperature and thus resistance alongeach wire The measurement of the resistance changes of

FIG 10 Two-dimensional profile of the atomic hydrogenbeam 10mm from the nozzle deduced from recombinationheating of gold-plated tungsten wires of 5microm in a 8times 8 wiregrid

all the wires in the grid allows one to deduce the centerand the profile of the beam Figure 10 shows the beamprofile resulting with a 8times8 wire grid positioned betweenskimmer and collimator performed as a first proof of themethod Later such a device has been used to comparemeasured and calculated beam profiles along the beamaxis between nozzle and skimmer34

B Degree of dissociation of the free atomic jet

The dissociation of the primary molecules is achievedby the interaction of the electrons and the hydrogen ordeuterium molecules in the plasma of the dissociatorThe degree of dissociation of the beam from the nozzledepends on the rf power applied to maintain the plasmathe primary molecular gas flow into the dissociator andthe temperature of the nozzle and the lower end of thedischarge tube These dependencies have been studiedbefore installation of the sextupole magnets with a setupcontaining a crossed-beam quadrupole mass spectrome-ter5556

α =ρa

ρa + 2 middot ρm (4)

The admixture of molecules in an atomic beam is de-scribed by the degree of dissociation where ρa and ρmare the densities of atomic and molecular hydrogen ordeuterium in the beam Other authors (eg Ref13) usethe atomic and molecular intensities Ia and Im in the def-inition of the degree of dissociation (αI) in Eq (4) Thetwo definitions of are related by

ImIa

=vmva

middot 1minus α

2α=

1minus αI

2αI (5)

10

This quantity was determined with the quadrupolemass spectrometer (QMS) in a conventional way as

α =Slowasta

Slowasta + 2 kv kion kdetSm

(6)

Here Slowasta = SaminusδSm denotes the atomic signal corrected

for dissociative ionization The parameter δ = 00141was obtained following the method described in Ref31The coefficient kv = vmva accounting for the difference

in atom and molecule velocity was chosen as 1radic2 under

the assumption of thermalization of the beam emergingfrom the nozzle Furthermore kion = 064 57 accounts forthe differences in ionization cross section for atomic andmolecular hydrogen and kdet = 084 for the detectionprobability55 As an example of the parameter stud-ies Fig 11 shows the deduced dependencies on the rfpower for a set of primary molecular hydrogen gas flowsFor typical flow values q(H2) le 10mbar ls a saturationvalue around 08 was obtained

0 50 100 150 200 250 300 350 400

00

02

04

06

08

10

Deg

ree

of d

isso

ciat

ion

()

Dissociator rf power [W]

q=07 mbar ls q=08 mbar ls q=10 mbar ls q=15 mbar ls

FIG 11 Degree of dissociation α of the free hydrogen jet asfunction of the applied rf power for different primary molec-ular hydrogen flows and a nozzle temperature of 70K

IV BEAM INTENSITY

The intensity of the polarized beam from the ABS to-gether with the layout of the storage cell determines theareal density of the target gas The intensity of the beamhas been measured with the use of a compression-tubesetup5859 shown in Fig 12 to optimize the ABS opera-tion parameters The measurements were performed at a300mm distance from the compression-tube entrance tothe last magnet and an inner tube diameter of 100mmas set in the tracking calculations The length of thecompression tube of 100mm was made equal to that ofthe foreseen feeding tube of the storage cell The narrowtube around the compression tube on a support based onthe lower flange separates the volume around the tube

FIG 12 Side view of the compression-tube setup made fromstandard ultra-high-vacuum components with a partial cutalong the axis (1 compression tube 2 support of thecompression tube based on the lower flange 3 narrow tubearound 1 closing the upper volume and allowing axial shiftsof the tube by the support 4 compression volume 5 hot-cathode pressure gauge 6 xy manipulator 7 z manipulator8 glass viewport 9 electromagnetic valve for gas inlet

from the compression volume The xy manipulator servesfor centering the tubes and for intensity-profile measure-ments The construction allows axial shifts of the com-pression tube by the z manipulator and the use of tubesof different diametersThe intensity of the beam entering the compression

volume through the compression tube is measured viathe pressure in the compression volume It is determinedby the equilibrium between the incoming beam intensityIin and the outgoing intensity Iout Under the assump-tion of a pure atomic beam and complete recombinationin the compression volume

Iin(atomss) = 2 middot Iout(moleculess)

= 2 middot∆P middot Ctube

= 2 middot∆P middot 103 middot 1020 middot d3

l

radic

T

M (7)

Here ∆P is the difference between the pressure measuredin the compreesion volume and that in the ABS cham-ber V The conductance of the compression tube Ctubeis determined by the inner diameter d of the tube itslength l the gas temperature T and the molar massM of the gas (given in cm and K respectively)60 The

11

CG

RVC200

H2

F

EV

EV

EV

EV

EV

MP

V1V2

EVR116

HC

V0

EV EVA016UDV136

HCCPLC

PS

220V AC

220V AC

Interface

24V DC

FIG 13 Source of calibrated molecular gas flow (V0 com-pression chamber V1 gas-storage chamber feeding V0 viathe needle valve UDV136a V2 chamber of calibrated volumeused to determine that of V1) The pressure in V1 is mea-sured by the capacitance gauge CG and is kept constant bythe dosing valve EVR116 with the gauge controller RVC200The whole setup including the evacuation elements can beoperated manually or by the programmable logic controllerPLC either within the ABS control system52 or as a separatesystem

a All the valves and the gauge controller are supplied by PfeifferVacuum GmbH Dndash35614 Asslar Germany (manufacturerBalzers AG Liechtenstein)

factor 2 takes into account that the same pressure is mea-sured in the hot-cathode gauge for 2 middotIin (H atomss) and1 middot Iin (H2 moleculess) For d = 10mm l = 100mmT = 290K and M = 2 for hydrogen pressure differences∆P on the order of 10minus4mbar are expected for atomichydrogen beam intensities in the order of 1017 atomssThe relation between Iin and ∆P for hydrogen has beendetermined experimentally with the use of a source ofcalibrated molecular hydrogen gas flow5859 depicted inFig 13 The measured dependence with a linear fit isshown in Fig 14 The calibration curve allows one to de-termine absolute values of Iin of hydrogen and deuteriumbeams The calibration for deuterium was deduced fromthe one for hydrogen by scaling with a factor 1

radic2 ac-

cording to Eq (7)

The dependences of Iin on the dissociator-operationparameters primary molecular hydrogen flow q(H2) noz-zle temperature Tn and dissociator power Pdiss have beenstudied to find the optimum values They are shown inthe Figs 15 16 and 17 respectively for different nozzle-orifice diameters The figures show that for the hydrogenbeam (states |1〉 and |2〉) with the standard operation pa-rameters qH2

= 11 mbar ls Tn = 70K Pdiss = 350Wand with a nozzle-orifice diameter of 23mm an inten-sity of Iin(H) = (75 plusmn 02) middot 1016 particless is achievedquite close to the earlier estimate from Eq (2) Besidesthe dominant atomic component of H atoms this value

04 06 08 10 12 14 16 18 20 2201

02

03

04

05

06

07

08

09

10

11

12

Gas

flow

from

the

stor

age

volu

me

[1017

ats

]

Pressure in compression volume [10-4 mbar]

FIG 14 Calibration curve for hydrogen used to deduce fromthe measured pressures the intensities of the hydrogen anddeuterium beam injected into the compression tube

00 05 10 15 20 2500

10

20

30

40

50

60

70

80 D = 20 mm D = 23 mm D = 25 mm

Bea

m in

tens

ity [1

016 a

ts]

Primary gas flow [mbar ls]

FIG 15 Intensity of the hydrogen beam (states |1〉 and |2〉)injected into the compression tube as function of the primarymolecular gas flow for different nozzle diameters D (nozzletemperature 60K dissociator power 300W)

includes small admixtures of H atoms in state |3〉 andmolecular hydrogen The first kind can be estimatedwith the use of the calculated transmissions (Sec II F)as 0017084 asymp 2 The amount of the second admix-ture has been measured as described below

For the deuterium beam (states |1〉 |2〉 and |3〉) theoptimization procedure gave an intensity of Iin(D) =(39 plusmn 02) middot 1016 particless achieved with q(D2) = 09mbar ls Tn = 65K and Pdiss = 300W slightly lowerthan those for hydrogen

12

50 60 70 80 90 100 110

40

50

60

70

80 D = 20 mm D = 23 mm D = 25 mm

Bea

m in

tens

ity [1

016 a

ts]

Nozzle temperature [K]

FIG 16 Intensity of the hydrogen beam (states |1〉 and |2〉)injected into the compression tube as function of the nozzletemperature for different nozzle diameters D (primary molec-ular gas flow 1mbar ls dissociator power 300W)

100 200 300 400 50010

20

30

40

50

60

70

80 D = 20 mm

Bea

m in

tens

ity [1

016 a

ts]

Dissociator power [W]

FIG 17 Intensity of the hydrogen beam (states |1〉 and |2〉)into the compression tube as function of the dissociator powerfor a nozzle diameter of 2 mm (nozzle temperature 60K pri-mary molecular gas flow 1mbar ls)

V HYDROGEN BEAM PROFILES

Beam profiles were measured at various positions atvarious positions behind the last sextupole magnet withthe use of

bull a compression tube of reduced dimensions (5mmdiameter)

bull a crossed-beam quadrupole mass spectrometer andbull a supplementary method of reduction of MoO3 by

hydrogen

A Measurements with the compression tube

For the determination of the beam dimensions at twopositions 300mm and 337mm behind the last magnetthe compression tube setup (Fig 12) was used makinguse of the possibility of axial movement by the z ma-nipulator and of that to install a narrower and shortercompression tube of 5mm diameter and 50mm lengthto enhance the spatial resolution The xy manipula-tor provided a lateral displacement of the compressiontube by plusmn10 mm in x and y direction The center co-ordinates of the geometrical axis of the source had beendetermined with the use of a bi-directional laser cen-tered inside the bore of the central support plate (seeFig 1) The relative intensity distributions in the xzand yz planes given by the measured pressure in thecompression volume are shown in Fig 18 Fits by Gaus-sian distributions to the data yield full widths at halfmaximum Γx = (642plusmn 009)mm Γy = (699plusmn 006)mmfor the distributions measured at z = 300mm and Γx =(627plusmn 008)mm Γy = (658plusmn 008)mm at 337mm

00

02

04

06

08

10

12

00

02

04

06

08

10

12

0 5 10 15 20 25

00

02

04

06

08

10

12

14

0 5 10 15 20 25

00

02

04

06

08

10

12

14

Pre

ssur

e in

the

com

pres

sion

tube

[au

]

Z = 300 mm

(a) (b)

(c) (d)

Z = 337 mm

X [mm]

Y [mm]

FIG 18 Cross sections of the beam profile in the mid-planemeasured with compression tube of 5mm diameter and 50mmlength Measurements in the xz-plane (a c) and yz-plane (bd) performed at two different positions z = 300mm (a b)and z = 337mm (c d) behind the last sextupole magnet ofthe ABS The shaded area represents position and dimensionsof the compression tube used in intensity measurements

The center of gravity of the measured profile definedas

rc =

sum

ij

radic

x2i + y2j middot P (xi yj)

sum

ij

P (xi yj) (8)

where xi and yj give the position of the compression-tubeaxis and P (xi yj) is the pressure measured in the com-

13

pression volume The resulting rc shows a deviation of012 mm from the geometrical axis of the source Further-more the data measured with the narrow compressiontube of 25mm radius can be used to derive the fractionof the beam entering the compression tube of 5mm ra-dius used in the intensity measurement of Sec IV Theratio

η =

rile25 mmsum

0P (xi yj)

rile10 mmsum

0P (xi yj)

(9)

where ri is the distance of the compression-tube axis tothe beam axis yields η asymp 07

B Measurements with the QMS

The beam-profile studies of Sec VA were extendedwith a setup utilizing a crossed-beam quadrupole massspectrometer (QMS) in the setup of Fig 19 Contraryto the measurements with the compression tube thosewith the QMS allow to separate the atomic and molec-ular fractions in the beam A 2mm diameter aperturewas installed at the entrance of the sensitive volume of

FIG 19 Setup for the measurements of the beam profilewith the QMS (1) xy-table enabling two-dimensional dis-placement of the entrance window of the QMS against thegeometrical axis of the ABS (2) the QMS (3) manually op-erated beam shutter The beam dump is an axially mountedcryo pump

the QMS to improve the resolution compared with thatachieved by the compression tube of 5mm diameter usedin measurements of the preceding section The layout ofthe setup presented in Fig 19 shows that in the presentcase the profile could not be measured at a distance ofz = 300mm to the last magnet Instead measurementswere performed at z = 567mm and with installation ofan extension tube at z = 697mm The xy manipulatorenabled displacements of the aperture axis from the geo-metrical axis of the source in any direction within limitsset by the bore diameter of the xy manipulatorThe first measured distribution of the atomic hydro-

gen (Fig 20) showed a distinct deviation from azimuthalsymmetry indicating an insufficient relative alignment ofnozzle and skimmer The three threaded rods support-ing the dissociator with the nozzle via the three-legged

FIG 20 Two-dimensional distribution of the atomic hydro-gen component of the beam at z = 567mm before the nozzle-to-skimmer adjustment showing a disinct deviation from az-imuthal symmetry

plate (label 2 in Fig 1) allow one to vary the position ofthe nozzle relative to that of the skimmer while the sourceis running This possibility has been used to find a nozzleposition which results in an azimuthally symmetric distri-bution The achieved symmetric distribution is shown inFig 21 and profiles of the atomic hadrogen component inthe beam measured in x and y direction at z = 567mmand z = 697mm are presented in Fig 22 Fits by Gaus-sian distributions to the data yield full widths at halfmaximum Γx = (736plusmn 043)mm Γy = (668plusmn 080)mmfor the distributions measured at z = 567mm and Γx =(669plusmn 022)mm Γy = (638plusmn 027)mm at 697mm

C Reduction of MoO3 by hydrogen

In addition to the compression tube and the QMS tech-nique a supplementary attempt was made to determinethe beam profile by exposing molybdenium trioxide (a

14

FIG 21 The distribution corresponding to that of Fig 20 af-ter nozzle-to-skimmer adjustment resulting in azimuthal sym-metry

0

20

40

60

80

0

20

40

60

80

2 4 6 8 10 12 14 16 18 20 22

0

20

40

60

80

6 8 10 12 14 16 18 20 22 24

0

20

40

60

80

QM

S si

gnal

s [a

u]

(a)

Z = 567 mm

Z = 697 mm

(c)

(b)

X [mm]

(d)

Y [mm]

FIG 22 Profiles of the atomic hydrogen component inthe beam measured with the QMS 567mm and 697mmbehind the last magnet

yellowish powder) on a glass plate to the beam The prin-ciple of this method is based on the reduction of MoO3 toa lower oxide of blue colour It first was used in the ex-periment to measure the magnetic moment of the hydro-gen atom by splitting of the beam in an inhomogeneousmagnetic field 61

This method is much simpler than the time-consumingmeasurements described in Secs VA and VB It givesqualitative results as presented in Fig 23 A quantitativeanalysis however requires development of the measuringtechnique (eg preparation of glass plates study of theoptimum exposure time digital image processing)

FIG 23 Photo of the glass plate covered with molybdenumtrioxide MoO3 exposed to the atomic hydrogen beam

D Summary of the profile measurements

Table III summarizes results of the measurements ofthe ABS beam profile with the compression-tube and theQMS setup The larger errors of the widths measuredwith the QMS are due to the lack of measurements withthe dissociator switched off and the necessity to estimatethe background signal from the existing data Within theerrors the measured widths do not show a dependenceon the distance from the last magnet This facilitatesto position the feeding tube of the storage cell in a widerange of a distances to the last magnet The average val-ues Γx = (638 plusmn 060)mm and Γy = (684 plusmn 033)mmagree within the errors and yield a common width ofΓxy = (673 plusmn 029)mm The two-dimensional Gaus-sian distribution of this width allows one to estimate thefraction η of the beam intensity injected into the com-pression tube or a feeding tube For a tube of 10mmdiameter η = 078plusmn 003 comparable with η asymp 07 givenin Sec VA

TABLE III Dimensions (FWHM) of the atomic hydrogenbeam measured with the compression tube (CT) and thecrossed-beam quadrupole mass spectrometer (QMS) at dis-tances z to the last magnet along perpendicular directions xand y

z[mm] Γx[mm] Γy[mm]

CT 300 642 plusmn 009 699plusmn 006

CT 337 627 plusmn 008 658plusmn 008

QMS 567 736 plusmn 043 668plusmn 080

QMS 697 669 plusmn 022 638plusmn 027

15

VI DEGREE OF DISSOCIATION

Besides the intensity of the atomic beam it is impor-tant to determine the molecular fraction in the beamMolecules injected into the feeding tube reduce the po-larization of the target gas

A Measurements with crossed-beam QMS

In addition to the data on the profile of the atomichydrogen beam (Sec VB) data on the distributions ofmolecular hydrogen in the beam were taken too at thepositions z = 567mm and 697mm behind the last mag-net The relation between the degree of dissociation andthe QMS signals by the atomic and molecular beam com-ponent was given above by Eq (6) The coefficient kv =vmva however is chosen here under the assumptionthat the average velocity of the atoms is determined bythe nozzle temperature of 65K and that of the moleculesby scattering and recombination on the ABS chamberwalls at 290K This yields kv =

radic

2 middot 65290 = 067 ingood agreement with Ref13 where this coefficient wasdetermined by the measured velocity distributions undersimilar conditionsThe measured profiles of the atomic fraction (identical

to those of Fig 21) those of the molecular fraction andthose of the degree of dissociation deduced from Eq (6)are collected in Fig 24As it is seen from the figure the distribution of the

degree of dissociation shows a dip around the central linedue to the higher density of molecular hydrogen originat-ing from the nozzle The mean value in an aperture of

0

20

40

60

80

100

120

2 4 6 8 10 12 14 16 18 20 22

0

20

40

60

80

100

120

6 8 10 12 14 16 18 20 22 24

QM

S si

gnal

s [a

u]

(a)

(b)

(c)

00

02

04

06

08

10

12

Deg

ree

of d

isso

ciat

ion

()

X [mm]

Y [mm]

(d)

00

02

04

06

08

10

12

Z = 567 mm

Z = 697 mm

FIG 24 Spatial distributions of H1 (bull)H2 () and degree ofdissociation () averaged over 3mm wide bands in the xz andyz planes respectively (here the z-axis is the geometrical axisof the ABS)

10mm diameter results as α = 095 plusmn 004

B Measurements with the Lamb-shift polarimeter

A cup in the quench chamber of the Lamb-shift po-larimeter (LSP) described in Ref8 allows one to mea-sure the currents Icup(H1) and Icup(H2) of the H+

1 andH+

2 ions extracted from the ionizer and separated bythe Wien filter with the cesium evaporation and the spinfilter switched off The relation between the degree ofdissociation α and the measured currents is

α =Icup(H1)minus r1

r2Icup(H2)

Icup(H1)minus r1r2Icup(H2) + 2kv

r2Icup(H2)

(10)

Among the three coefficients kv = 067 as for the mea-surement with the QMS For the electron energy of about100keV the ratio r1 of dissociative to non-dissociativeionization of H2 is8

r1 =σ(H2 rarr 2H+

1 )

σ(H2 rarr H+2 )

= 0095plusmn 0008 (11)

and the ratio between the ionization cross sections is857

r2 =σion(H2)

σion(H1)= 17plusmn 01 (12)

At the standard operation parameters of the source(Sec IV) the measured currents are Icup(H1) = (125 plusmn05) nA and Icup(H2) = (64 plusmn 01) nA yielding α =(096plusmn004) in excellent agreement with the value result-ing from the measurements with the QMS (Sec VIA)

VII BEAM POLARIZATION

The Lamb-shift polarimeter was designed built andtested at Universitat zu Koln7 It was used to measureand to optimize the polarization of the atomic hydrogenand deuterium beams delivered by the ABS Details arefound in Ref7The vector polarization pz for hydrogen is defined by

the relative hyperfine-state occupation numbers N(mI)

pz =N(+ 1

2 )minusN(minus 12 )

N(+ 12 ) +N(minus 1

2 ) (13)

for deuterium

pz =N(+1)minusN(minus1)

N(+1) +N(0) +N(minus1) (14)

Deuterium tensor polarization pzz is given by

pzz =N(+1) +N(minus1)minus 2 middotN(0)

N(+1) +N(0) +N(minus1) (15)

16

These polarizations can be derived from the measuredLyman-α peak strengths S by application of a number ofcorrection factors78

Typical Lyman α spectra measured with the polarizedhydrogen and deuterium beam from the ABS are shownin the Figs 25 and 26

MFT 2-3

200 300 400 5000

40

80

120

160

Magnetic field in the spinfilter [au]

PM

T si

gnal

[au

]

mI = -12

(a)

mI = +12

100 200 300 400 5000

40

80

120

160(b)

MFT 2-3WFT 1-3

mI = +12

mI = -12

FIG 25 Lyman-α spectra measured with the polarized hy-drogen beam (a) population change from state |2〉 to state|3〉 induced by the MFT unit (b) same as (a) with subsequentpopulation change from state |1〉 to state |3〉 induced by theWFT unit

MFT 3-4SFT 2-6

0

2

4

6

8

mI = -1m

I = 0

(a)

mI = +1

0

2

4

6

8(b)

MFT 3-4WFT 1-4 2-3

mI = +1 m

I = 0

mI = -1

mI = +1

mI = 0

mI = -1

mI = +1

mI = 0

mI = -1

0 1000 2000 3000

0

1

2

3

4 (c)

PMT

sign

al [a

u]

Magnetic field in the spinfilter [au]

WFT 1-4SFT 2-6

0 1000 2000 30000

2

4

6

8WFT 1-4SFT 3-5

(d)

FIG 26 Lyman-α spectra measured with the polarized deu-terium beam (a) and (b) vector polarization resulting fromsubsequent transitions MFT (3 rarr 4) and SFT (2 rarr 6) andWFT (1 rarr 4 2 rarr 3) respectively (c) and (d) tensor polar-ization resulting from subsequent transitions WFT (1 rarr 4)and SFT (2 rarr 6) and SFT (3 rarr 5) respectively

The polarization values for the hydrogen and the deu-terium beam derived from the Lyman-α peak-strengthratios with application of the necessary corrections aresummarized in Table IV

The vector polarization for hydrogen of the first linereflects the population of state |1〉 and state |2〉 accordingto the Eqs (3) and (13) The value of 091 deduced fromthe calculated transmission values is confirmed by themeasured one

TABLE IV The vector polarization pz of the hydrogen beamand the vector and the tensor polarization pzz of the deu-terium beam from the ABS measured with the Lamb-shiftpolarimeter

populated pz pzz

state(s)

Hydrogen |1〉 +089plusmn 001 -

|3〉 minus096plusmn 001 -

Deuterium |1〉+ |6〉 +088plusmn 001 +088plusmn 003

|3〉+ |4〉 minus091plusmn 001 +085plusmn 002

|3〉+ |6〉 +0005plusmn 0003 +090plusmn 001

|2〉+ |5〉 +0005plusmn 0003 minus171plusmn 003

VIII CONCLUSIONS AND OUTLOOK

In this paper we present the detailed description ofthe major components of the atomic beam source (ABS)for the polarized internal gas target of the magnet spec-trometer ANKE in COSY-Julich The ABS was builtfor the purpose of extending the physics program ofANKE from unpolarized and single-polarized investiga-tions with stored beams towards double-polarized exper-iments1 thus facilitating nuclear reaction studies involv-

ing ~p~p ~p~d ~d~p and ~d~d initial statesThe mechanical design took into account that at

ANKE the source has to be mounted vertically and trans-versely movable together with the transverse motion ofthe spectrometer magnet D2 The design of the system ofsextupole magnets took advantage of the developments inthe field of rare-earth permanent magnets (NdFeB) Ded-icated tools and methods were developed to determineand to optimize the source parameters ie intensity de-gree of dissociation and polarization Special emphasiswas put on the measurements of the spatial distributionsof the atomic and molecular beam near the focus wherethe feeding tube of the storage cell is located The ABShas been used in a number of investigations at ANKEthe commissioning effort to prepare the target for the usewith polarized H is described in Ref62 Performed stud-ies of the deuteron-charge exchange reaction are summa-rized in Ref6364 studies in near-threshold pion produc-tion are reported about in Ref65The ABS resides at the ANKE target position for a few

months per year only thus during the remaining timeit is used for other studies It had been observed thatthe nuclear polarization in recombined hydrogen is par-tially retained after recombination66 as well as evidencefor nuclear tensor polarization in recombined deuteriummolecules67 In order to investigate this recombinationprocess in more detail a special setup has been developedin the framework of an ISTC project68 and the recombi-nation process for different cell-wall coatings and differ-ent polarizations of the injected hydrogen or deuterium

17

atoms as function of cell-wall temperature strength ofthe magnetic holding field and gas pressure in the cell ispresently investigated69ndash71

Appendix A Preparation of Discharge Tubes and Nozzles

1 Tube Treatment

One end of the discharge tube is machined at a 45

angle while the other is kept flat Both ends are thenremelted and the tubes are tempered at 150 C Thetubes are further treated according to the procedure de-scribed in Ref72 which includes successive cleaning withacetone methanol distilled water and subsequent rins-ing by a 21 acid mixture of concentrated HF (40) andHCl (32) for 5 min The tubes are then flushed bydistilled water and dried

2 Nozzle Treatment

The nozzles are cleaned in an ultrasonic bath oftrichlorethylene acetone methanol and finally distilledwater all at 50 C Anodizing takes place in sulfuric acidto form a thin layer of Al2O3 as described in Ref72 Af-terwards they are immersed in distilled water for 30minat 95 C

ACKNOWLEDGMENTS

The authors want to thank OWB Schult Institut furKernphysik (IKP) Julich who initiated the polarizationprogram of ANKE Thanks go to the design office themechanical workshop and especially to WR Ermer allIKP Valuable advice was received from the PINTEX col-laboration at IUCF from the target group at HERMESespecially NKoch and from DToporkov BINP Novosi-birsk The support by VKoptev PNPI Gatchina whoregrettably passed away in January 2012 is gratefullyacknowledged Thanks go also to R Poprawe and col-leagues Fraunhofer-Institut fur Lasertechnik Aachenwhere the encapsulations of the magnets were laser-welded

1AKacharava F Rathmann and CWilkin Spin Physics fromCOSY to FAIR COSY Experiment Proposal No 152 (2005)Available under httparXivnucl-ex0511028

2WHaeberli in Proc 2nd Int Symp on Polarization Phenomenaof Nucleons Karlsruhe 1965 Eds P Huber and H SchopperExperientia Supplementum 12 64 (Birkhauser Verlag 1966)

3E Steffens and WHaeberli Rep Progr Phys 66 1887 (2003)4SBarsov et al Nucl Instr and Meth A 462 364 (2001)5RMaier Nucl Instr and Meth A 390 1 (1997)6KGrigoryev et al Proc 14th International Workshop on Po-larized Sources Targets and Polarimetry (PSTP 2011) 12-16 September 2011 StPetersburg Russia eds KGrigoryevPKravtsov and AVasilyev ISBN 978-5-86763-282-3 61 (2011)

7REngels et al Rev Sci Instrum 74 4607 (2003)8REngels et al Rev Sci Instrum 76 053305 (2005)9TWise et al Nucl Instr and Meth A 336 410 (1993)

10WADezarn et al Nucl Instr and Meth A 362 36 (1995)11T Rinckel et al Nucl Instr and Meth A 439 117 (2000)12F Stock et al Nucl Instr and Meth A 343 334 (1994)13ANass et al Nucl Instr and Meth A 505 633 (2003)14VDerenchuk et al Proc Conf Polarized Ion Sources and Polar-ized Gas Targets Madison WI 1993 Eds LWAnderson andWHaeberli AIP Conf Proc 293 72 (American Institute ofPhysics 1994)

15HOkamura et al see Ref14 p 8416KHatanaka et al Nucl Instr and Meth A 384 575 (1997)17Manufacturer Schiffer Metall- amp Vakuumtechnik 52428 JulichGermany

18Single-stage type RGS120 refrigerating capacity 120W at 80Kand 20W at 30K Leybold Vacuum GmbH 50968 Koln Ger-many

19Mini UHV gate valve series 010 VAT Germany GmbH 85630Grasbrunn Germany

20Type F3 fomblin oil Pfeiffer Vacuum GmbH 35614 Asslar Ger-many

21Model HU 1 Leybold Vacuum GmbH 50968 Koln Germany22Manufacturer SK Industriemodell GmbH 52531 Ubach-Palenberg Germany

23Type PFG 600 RF with automatic matchbox PFM 1500 A-INDHuttinger Elektronik GmbH 79110 Freiburg Germany

24WKorsch PhD Thesis Philipps Universitat Marburg (1990)25F Stock et al Int Workshop on Polarized Beams and PolarizedGas Targets Koeln Germany 1995 Eds HPaetz gen Schieckand L Sydow (World Scientific Publ Co 1996) p 260

26The first number denotes the outer diameter and the second onethe wall thickness

27Type Duran 8330 equivalent to Corning 7740 (Pyrex) SchottAG 55122 Mainz Germany

28Ultra-Kryomat RUL 80-D Lauda DrRWobser GmbH 97912Lauda-Konigshofen Germany

29ODU-Kontakt GmbH 84444 Muhldorf Germany30Handbook of Chemistry and Physics Ed RCEast (The Chem-ical Rubber Co 1973) p E-10

31NKoch and E Steffens Rev Sci Instrum 70 1631 (1999)32AVassiliev et al Petersburg Nuclear Physics Institute ReportNP-32-1997 No 2175 (1997)

33BLorentz Diploma Thesis Ruprecht-Karls-Universitat Heidel-berg (1993)

34ANass and E Steffens Nucl Instr and Meth A 598 653(2009)

35WHaeberli Ann Rev Nucl Sci 17 373 (1967)36AVassiliev et al Rev Sci Instr 71 3331 (2000)37WKubischta Proc Workshop on Polarized Gas Targetsfor Storage Rings Heidelberg 23-26 September 1991 EdsHGGaul E Steffens and K Zapfe (Max-Planck-Institut furKernphysik Heidelberg)

38The labeling of the hyperfine states as |1〉 = |mj = +12 mI =+12〉 |2〉 = | + 12minus12〉 |3〉 = | minus 12+12〉 and |4〉 =|minus12minus12〉 for hydrogen and |1〉 = |+12+1〉 |2〉 = |+12 0〉|3〉 = | + 12minus1〉 |4〉 = | minus 12minus1〉 |5〉 = | minus 12 0〉 and

4

conductances of the slits of le5 ls are small comparedwith the applied pumping speed Because of the compli-cated shape identical raw pieces of cast Al22 were ma-chined to the final dimensions Contrary to the lower baf-

FIG 3 3D drawing of the upper baffle separating the vac-uum chambers I and II with the two wide cuts in front ofthe turbopumps (1) and the openings for the viewport (2)the skimmer (3) four of the 16 ball bearings (4) and the foursupporting rods (5)

fle the upper baffle until now has to be installed togetherwith the flange of the upper vessel at a fixed axial posi-tion (cf Fig 1) In order to reach full flexibility in vary-ing the nozzle skimmer and collimator relative positionsfrom outside the installation of rotational feedthroughsin the flange of the upper vacuum vessel is necessary aforeseen but not yet implemented feature

D Dissociator

To dissociate molecular hydrogen or deuterium to neu-tral atoms an rf discharge is employed which is fed bya 13560 MHz generator23 delivering up to 600 W into a50Ω load The layout of the dissociator shown in Fig 4is similar to that of the FILTEX design2425 The dis-charge tube (empty11times15mm)26 is surrounded by two coax-ial tubes (empty204times18mm and empty28times2mm) all three aremade from borosilicate glass27 The coolant streams fromthe inlet connection down between the discharge tubeand the middle tube and after flow reversal at the lowerend (Fig 5 label 2) it streams up in the outer slit to theoutlet connection In a closed loop the coolant inlet tem-perature (typically 15 C for a 50 water ndash 50 ethanolmixture) is stabilized by a cooling thermostate28 whichwould allow coolant temperatures down to minus80 C Therf coil and the capacitor at fixed relative positions canbe positioned from outside by means of a sliding rf con-nection29 and the feed-through ground connection This

FIG 4 3D drawing of the dissociator (1 gas inlet 2 slidingground connection 3 coolant inlet 4 coolant outlet 5 rfinput 6 sliding rf connection 7 grounded capacitor plate8 rf coil 9 rf-fed capacitor plate 10 isolating plastic sup-port rings 11 grounded limiter plate 12 lower end of thecoolant-guiding tubes 13 tube support and connection tothe coldhead (details are given in Fig 5) 14 lower end of thedischarge tube)

enables variation of the plasma-nozzle distance to opti-mize the atomic beam intensity while the plasma is burn-ing The treatment of the discharge tube and the nozzleprior to installation is described in Appendix A

E Nozzle

The nozzle cooled via the heat bridge and the sur-rounding components are shown in Fig 5 The nozzlemade from 995 Al has a simple conical shape with thetip cut Comparative measurements show that nozzleswith sharp edges as used eg in the Madison source9 donot yield higher atomic beam intensities First a sharpedge is more difficult to produce due to the softness ofpure Al Second the low heat conductance of a sharpedge leads to appreciable temperatures of the nozzle tipcaused by recombination of atoms on the nozzle surface

5

The temperature at the bottom of the nozzle is measuredwith a Pt-100 sensor and it is stabilized with an accuracyof plusmn05K utilizing a heater Measurements with temper-ature sensors placed along the outer nozzle surface haveshown a temperature increase from 60K at the nozzlebottom to sim200K at the sharp nozzle tip In the follow-ing the nozzle temperature is defined as that measuredwith this Pt-100 sensorWith the present system of sextupole magnets the

maximum atomic beam intensity feeding the storagecell is obtained with a nozzle-orifice diameter of 23mmand a nozzle-tip to skimmer-tip distance of 15mm at askimmer-tip diameter of 44mm and a skimmer-tip to di-aphragm distance of 17mm The 2mm thick diaphragmwith a conical bore opening from 95mm to 99mm to-wards the first permanent sextupole magnet shields themagnet from heating by atoms recombining on its sur-face The slit between the diaphragm and the front faceof the magnet enables pumping of gas from the entranceto the magnetThe Teflon washer and the stainless steel support sep-

arate the cold lower end of the heat bridge from thewarm lower end of the dissociator The dimensions of

FIG 5 Technical drawing including the lower end of theheat bridge and the dissociator the nozzle surroundings andthe first sextupole magnet (in scale 1 discharge and coolant-guiding tubes 2 coolant-reversal piece 3 heat flow reducingTeflon washer 4 sliding heat connection 5 stainless steelconnector 6 groove for nozzle-heating element 7 lower endof the Cu heat bridge 8 nozzle fixture 9 nozzle 10 baf-fle separating the chambers I and II with a viewport 11stainless steel beam skimmer 12 Cu diaphragm 13 firstsextupole magnet and 14 baffle separating the chambers IIand III

these two components and the sliding heat connector aworked-over sliding high current connector similar to therf connector in the dissociator define the temperatureof the lower end of the discharge tube relative to thatof the nozzle The discharge tube adapted at its lowerend to the nozzle by a chamfered edge is pressed to thenozzle by a viton O-ring at its upper end The two O-rings around the discharge tube in the lower part of thedissociator seal against the atmosphere By this designonly minor forces are exerted to the discharge tubeThe removable viewport in the baffle and the window

flange in the upper vacuum vessel (on the right-hand sideof chamber II in Fig 1) allows one to observe the nozzlestatus from the outside and to exchange nozzles withoutremoval of the dissociator from the setupThe heat bridge from the coldhead to the nozzle is

made from electrolytic Cu The flexible link between thecoldhead and the heat bridge consisting of about 200high-purity Cu strands of 1 mm diameter allows for thethermal expansions of the cold and the warm compo-nents The total cross section of the strands and theirheat conductance is smaller than that of a massive Cubody This deficiency however is reduced by clampingthe flexible link directly to the coldhead At its operat-ing temperature of about 30K the thermal conductivityof Cu is about 11 9 and 5 times higher than that at300 100 and 60K respectively30 Thus the reductionof the conductance of the entire heat bridge by the flexi-ble link is minimized by placing it at the coldhead Withthe present system cooling the nozzle down from roomtemperature to 60K needs about 15 hours The heat-ing element facilitates warming up to room temperaturewithin about one hourFurthermore avoiding the maze of cold Cu strands

around the nozzle ie a labyrinthic cold surface com-pared to an earlier solution31 leads to improved pumpingconditions in the nozzle-skimmer area where the highestgas load has to be pumped offIn an earlier phase of the ABS development attempts

have been made to use a cryogenic Ne heat-pipe of 20Wcooling power instead of the usual solid Cu bridge toachieve faster cooling and warming of the nozzle becauseof the lower heat capacity32 An observed instability inthe necessary operation mode however lead to difficul-ties in nozzle-temperature stabilization In view of thefact that the cooling and warming-up times reached withthe Cu bridge were satisfying and that its use avoids theadditional precautions imposed by the heat-pipe opera-tion it has been replaced by the Cu bridge

F Magnet System

The design of the magnet system was made for a setof sextupole magnets consisting of permanently magne-tized segments made from NdFeB compounds deliveringpole-tip fields around 15 T Tracking calculations fromthe nozzle to the feeding tube of the storage cell were

6

performed with the use of a computer code originallydeveloped for the FILTEX ABS24 The boundary condi-tions by the layout of the target setup were the availabledistance of about 1250mm from the nozzle to the feeding-tube entrance of 10mm diameter and the distance fromthe exit of the last magnet to the feeding-tube entrance of300mm necessary to install the SFT and WFT units andthe gate valve between the ABS and the target chamberThe laboratory velocity distribution of the atoms in

the supersonic beam from the nozzle is described by amodified Maxwellian distribution

F(~vd Tb) =( m

2 k Tb

)32exp

[ minusm

2 k Tb(~v minus ~vd)

2

]

(1)

where m is the mass of the atoms and k is the Boltzmannconstant According to time-of-flight studies33 the driftvelocity along the beam axis vd and the beam temper-ature Tb for a primary molecular gas flow of 1 mbar lsand a nozzle-orifice diameter of 2mm follow a linear de-pendence on the nozzle temperature Tn For hydrogenvd[ms] = 1351 + 61 middot Tn[K] and Tb = 029 middot Tn and fordeuterium vd[ms] = 1070+345middotTn[K] and Tb = 025middotTnAs starting conditions of a track a random generator

selects a point in the nozzle orifice one within the di-aphragm in front of the first magnet and an atom ve-locity |v| In linear molecular flow approximation (cfthe discussion in Ref34) this defines ~v for the track be-tween the nozzle and the first magnet According to thegeometrical boundary conditions and the velocity distri-bution of Eq (1) the event is either rejected or used inthe further track calculation Within the magnet theevolution of the track is calculated stepwise by numeri-cal integration of the equation of motion over integrationtimes of 2micros corresponding to track lengths of 36mmfor a typical particle velocity of 1800ms The pureradial force acting on an atom within the field of the

sextupole magnet is ~Fr = minusmicroeff middot δBδr middot ~rr The ef-fective magnetic moment resulting from the Breit-Rabidiagram (eg Ref35) as microeff = δWδB is positive (neg-ative) for atoms in the hyperfine states with the electron

spin parallel (antiparallel) to ~B in the magnet aperturewhich therefore are deflected towards (away from) thebeam axis In the drift sections between the two magnetgroups and between the last magnet and the feeding tubethe trajectories are assumed as straight linesA variety of systems were studied all under the as-

sumption of Tn = 60K and pole-tip fields of 15T Asystem utilizing 6 magnets was found to yield satisfyingboth separation of the atoms in the microeff lt 0 and microeff gt 0states and focusing of the microeff gt 0 states into the feedingtube Optimization of the parameters led to the systemlisted in Table II (The tracking calculations yielding themagnet dimensions for the order to the manufacturer hadbeen performed for a slightly different geometry) Thetable gives the two distances at which intensity mea-surements with the compression tube were performedThe Fig 6 shows the projection of the trajectories of Hatoms in the microeff gt 0 states calculated for this system

TABLE II Final dimensions and axial positions of the sourcecomponents (pole-tip field strenghts Blowast

0 as measured afterdelivery36 inner diameters (empty0) outer diameters (empty1) axialdimensions (ℓ) and distances (∆) between the componentsThe lower three lines give the two distances and the dimen-sions of the compression tube used in the intensity measure-ments

component Blowast

0 [T] empty0 [mm] empty1 [mm] ℓ [mm] ∆ [mm]

Nozzle orifice 23 33150

Skimmer 44304a 130169

Diaphragm 9599a 2036

Magnet 1 1630 10401412a 3998 400194

Magnet 2 1689 15982212a 6404 650194

Magnet 3 1628 2804 9400 70014297

Magnet 4 1583 3004 9402 38011010

Magnet 5 1607 3006 9400 5501150

Magnet 6 1611 3002 9404 550030003370

Compr tube 100 110 1000

a Conical openingthe first number denotes the measureddiameter of the entrance the second that of the exit aperture

One recognizes two groups of trajectories one with anintermediate focus and another one with focusing intothe feeding tube The present result like those of othergroups (see eg Ref33) confirms the expectation37 thatthe transmission as function of the atom velocity shouldshow two maxima one below and one above the mostprobable velocityThe transmission Tr of the system is defined as the

fraction of tracks ending within the entrance of the feed-ing tube to those passing the diaphragm in front of thefirst sextupole magnet For the four hyperfine states ofhydrogen38 the calculations yield Tr(|1〉) sim Tr(|2〉) =042 (for both microeff gt 0) and Tr(|3〉) = 0001 andTr(|4〉) = 00004 (for both microeff lt 0)The performed tracking calculations do not account for

intra-beam and residual-gas scattering The calculatedtransmissions thus only allowed one to estimate upperlimits of the expected atomic beam intensity Iin into thefeeding tube For a primary molecular flow q(H2) theintensity Iin(H) with atoms mainly in the states |1〉 and|2〉 (microeff gt 0) was expected as

Iin(H) = q(H2) middot 2α middot Ω

2πmiddot 14

i=4sum

i=1

Tr(|i〉) (2)

For the degree of dissociation α a routine value of 08(see eg Ref9) was assumed Ω = 0022π is the solid

7

angle covered by the collimator aperture The factor 14reflects the assumption that the four substates in theatomic beam from the nozzle are equally populated Forq(H2) = 1mbar ls or 27middot1019H2 moleculess one expectsIin(H) sim 1 middot 1017H atomssAs described in the subsequent section the rf tran-

sition units are used to change the relative occupationnumbers of the states The trajectory code allows oneto simulate this change by assigning a microeff of one of thestates to the atoms before they pass a magnet As an ex-ample the medium-field transition unit (MFT) behindmagnet No 3 (see Fig 1) brings H atoms from state |2〉into state |3〉 This is simulated by assigning microeff(|2〉) gt 0to the atoms in the magnets 1minus3 and microeff(|3〉) lt 0 in themagnets 4 minus 6 where they get deflected from the beamaxis This results in a small value Tr(|2〉) = 0017 Fromthis value and the above value Tr(|1〉) = 042 the vectorpolarization is expected as

pz =Tr(|1〉)minus Tr(|2〉)Tr(|1〉) + Tr(|2〉) = 091 (3)

under the assumption of 100 efficiency of the transitionunitThe design and the properties of the permanent sex-

tupole magnets39 were discussed in an earlier paper36To achieve the pole-tip field values of sim15T each mag-net was produced from 24 segments employing three dif-

FIG 6 Projection of the 3-dimensional trajectories of hydro-gen atoms in hyperfine states |1〉 and |2〉 (effective magneticmoment microeff gt 0)) from the nozzle (empty = 2mm Tn = 60K)to the storage cell calculated for the magnet arrangement ofTable II and pole-tip fields of 15 T The positions and lat-eral dimensions of the six magnets and the feeding tube areindicated (in red)

ferent types of NdFeB compounds The expected pole-tip values (Table II) and the precise radial dependenceB(r) sim r2 within the magnet apertures were confirmedFor the first time the predicted high multipole compo-nents40 up to a 102-pole structure very near to the aper-ture surface could be measured36After the field measurements the magnets were encap-

sulated to prevent diffusion of hydrogen into the magnetmaterial which might deteriorate the magnetic proper-ties and to avoid the pumping of gas from the sinteredmagnet bodies The housings were made from thin stain-less steel cans of 02mm thickness for the conical andcylindrical walls within the magnet apertures and 03mmfor the front and end covers During the final welding toclose the housings with magnets installed the local tem-perature of the magnet material had to be kept belowthe Curie temperature of 60 C This was achieved bywelding with the use of a pulsed 15Hz NdYAG laserdelivering 11 J in a pulse of 2ms41 Overlapping weldspots of sim03mm diameter set around the adjacent cir-cular 02mm thick weld lips allowed one to finish thehousings with leak rates sim 10minus10mbar ls Inside thehousings the magnets were fixed to suppress axial androtational movements caused by the force of the adja-cent magnets Finally the free slits within the housingswere filled by sim20mbar krypton to enable leak tests bymass spectroscopy

G Radio Frequency Transition Units

The ABS is equipped with three types of transitionunits a weak field a medium field and a strong field rftransition unit (WFT MFT and SFT units) Togetherwith the selecting properties of the sextupole magnetsthey enable one to achieve all vector and tensor polar-izations of the atomic hydrogen and deuterium gas inthe storage cell In all three units transitions betweenthe hyperfine states split according to the Breit-Rabi di-agram by a static magnetic field (see eg Ref35) areinduced by the magnetic component (Brf) of an rf fieldleading to changes in the population of the states Thestatic field Bstat consists of two parallel components ahomogeneous field Bhom and a superimposed weaker gra-dient field Bgrad both orthogonal to the beam directionThe field gradient along the beam direction is requiredto satisfy the condition of adiabatic passage3542The assemblies of the WFT and the MFT units are

similar43 The layouts follow those of the units devel-oped for the HERMES experiment44 In both units therf field is produced by a coil with the axis along the beamdirection and consequently Brf orthogonal to Bstat TheMFT unit is shown in Fig 7 Figure 8 schematicallyshows one of the grooved aluminum frames with thewindings producing the gradient field A WFT unitis operated in a weak magnetic field Bstat le10G for hy-drogen and le5G for deuterium where the total atomicspin F is a good quantum number In hydrogen the

8

F = 1 levels |1〉 |2〉 and |3〉 with magnetic quantumnumbers mF = +1 0 and minus1 respectively can be re-garded as equally spaced In deuterium the same holdsfor the four F = 32 levels |1〉 |2〉 |3〉 and |4〉) withmF = +32 +12 minus12 and minus32 respectively andfor the two F = 12 levels |5〉 and |6〉 with mF = minus12and +12 respectively The magnetic component of therf dipole field induces transitions between each pair ofneighboring mF states with ∆mF = plusmn1 |∆mF| = 2transitions are forbidden The interchange of the popu-lation between the states |1〉 and |3〉 in hydrogen eg iscaused by a two-quantum transition via the intermedi-ate state |2〉 In the classical description of the adiabaticpassage method42 the population change should not de-pend on the sign of the magnetic field gradient relativeto the beam direction An exact quantum-mechanicaltreatment4546 however indicates that cleaner popula-tion changes from state |1〉 to |3〉 in hydrogen and fromstate |1〉 to |4〉 in deuterium are obtained with a nega-tive field gradient ie a Brf field decreasing in the beamdirection Deviations from adiabaticity are discussed inRef4547

The MFT unit is operated at higher values of Bstatwhere the differences in the energy spacings of pairs of hy-

FIG 7 Three-quarter-section view of the MFT unit with thesupport structure (1 self-supporting rf coil with spacers 2pick-up loop 3 Al tubes defining the length of the transition-inducing rf field 4 Cu cavity 5 coil around the pole shoe(6) providing the static field Bstat 7 slit between pole shoeand cavity wall housing the gradient-field coil 8 componentsof the static magnet yoke also serving as shielding againstexternal fields 9 cavity-positioning element 10 Cu padscooled by means of water-carrying tubes The cavity withthe rf coil and the pick-up loop can be taken out from thesurrounding components

z

Bstatic

transition

region

FIG 8 Arrangement of the windings producing the staticgradient field Bgrad shown in the left-hand side of the figureIn all transition units the field lies in the direction of the statichomogeneous field the field gradient dBdz lies in the beamdirection which defines the z axis In z direction the tran-sition reagion (indicated by the dashed lines) is confined tothe range of constant gradient by the Al tubes in orthogonaldirection by the beam diameter

perfine states with ∆mF = plusmn1 allow one to select singletransitions Originally developed for an polarized alkaliion source48 the MFT unit now is a standard compo-nent in polarized hydrogen and deuterium sources as dis-cussed eg in Ref49 Appropriate choice ofBhom Bgradand the rf frequency allows one to induce selected tran-sitions |1〉 harr |2〉 and |2〉 harr |3〉 in hydrogen or |1〉 harr |2〉|2〉 harr |3〉 and |3〉 harr |4〉 in deuterium Furthermore thechoice of the field gradient allows one to achieve consecu-tive transitions As an example a negative field gradientin the MFT unit behind the first set of magnets ie aB field decreasing in beam direction at a fixed rf fre-quency leads to the sequence of the transitions |3〉 rarr |4〉|2〉 rarr |3〉 and finally |1〉 rarr |2〉 in deuterium leaving thestate |1〉 empty

The SFT unit is used to induce transitions betweenstates in the upper and lower hyperfine multiplet in hy-drogen and deuterium Contrary to the historical nameindicating a strong magnetic field the SFT unit is op-erated with magnetic fields comparable to those used inthe MFT unit The transition frequencies are comparablewith those of the hyperfine splitting (1420MHz for hy-drogen and 327MHz for deuterium) and thus are muchhigher than those in the WFT and MFT units The rffield in a SFT unit is produced by a twin-line resonatorinside a Cu box tuned to the λ4 resonance50 The SFTunit51 is shown in Fig 9 Again the layout follows that ofthe unit used in the HERMES experiment44 Two vari-able capacitors at the free ends of the conducting rodsfed by the rf power with a relative phase shift of 180 allow one to tune the device

9

FIG 9 Three-quarter-section view of the rf cavity of theSFT unit for deuterium (1 the two resonant-field creatingconductors 2 the adjustible capacitor plates 3 Cu cavity)The inner dimensions of the cavity are 56mm along Bstat36mm orthogonal to it and 36mm height The cross sectionof the conductors is 14times 4mm2

H Slow Control System

Industrial components providing reliable and long-term support were selected for the control system of thewhole setup consisting of the ABS and the diagnosticstools the storage cell positioning system the Lamb-shiftpolarimeter and the supply system of a calibrated flowof unpolarized molecular gas The interlock system hasbeen implemented on the basis of SIEMENS SIMATICS7-300 family of programmable logic controllers In orderto unify the interfacing to the control computer all front-end equipment is connected via the PROFIBUS DP field-bus The process control software was implemented us-ing the Windows-based WinCC toolkit from SIEMENSThe system controls the operation of the pumps and thevalves It reads the pressure gauges and controls theregeneration cycles of the cryopumps Via a control net-work the temperature of the nozzle is stabilized withinplusmn05K Furthermore all power-supply units rf genera-tors and amplifiers are set and controlled The wholevariety of components to be controlled the logical struc-ture of the control and interlock system and a separatedevice for parameter studies are described in Ref52

III STUDIES OF THE FREE HYDROGEN JET

A Atomic beam profile near the nozzle

A novel device has been used to measure the profile ofan atomic beam via the deposition of recombination heaton thin wires in a two-dimensional grid5354 Atoms stuckon the surface of gold-plated tungsten wires of 5microm diam-eter recombine and are reemitted as molecules The re-combination heat (4476 eV per hydrogen molecule) leadsto a change of temperature and thus resistance alongeach wire The measurement of the resistance changes of

FIG 10 Two-dimensional profile of the atomic hydrogenbeam 10mm from the nozzle deduced from recombinationheating of gold-plated tungsten wires of 5microm in a 8times 8 wiregrid

all the wires in the grid allows one to deduce the centerand the profile of the beam Figure 10 shows the beamprofile resulting with a 8times8 wire grid positioned betweenskimmer and collimator performed as a first proof of themethod Later such a device has been used to comparemeasured and calculated beam profiles along the beamaxis between nozzle and skimmer34

B Degree of dissociation of the free atomic jet

The dissociation of the primary molecules is achievedby the interaction of the electrons and the hydrogen ordeuterium molecules in the plasma of the dissociatorThe degree of dissociation of the beam from the nozzledepends on the rf power applied to maintain the plasmathe primary molecular gas flow into the dissociator andthe temperature of the nozzle and the lower end of thedischarge tube These dependencies have been studiedbefore installation of the sextupole magnets with a setupcontaining a crossed-beam quadrupole mass spectrome-ter5556

α =ρa

ρa + 2 middot ρm (4)

The admixture of molecules in an atomic beam is de-scribed by the degree of dissociation where ρa and ρmare the densities of atomic and molecular hydrogen ordeuterium in the beam Other authors (eg Ref13) usethe atomic and molecular intensities Ia and Im in the def-inition of the degree of dissociation (αI) in Eq (4) Thetwo definitions of are related by

ImIa

=vmva

middot 1minus α

2α=

1minus αI

2αI (5)

10

This quantity was determined with the quadrupolemass spectrometer (QMS) in a conventional way as

α =Slowasta

Slowasta + 2 kv kion kdetSm

(6)

Here Slowasta = SaminusδSm denotes the atomic signal corrected

for dissociative ionization The parameter δ = 00141was obtained following the method described in Ref31The coefficient kv = vmva accounting for the difference

in atom and molecule velocity was chosen as 1radic2 under

the assumption of thermalization of the beam emergingfrom the nozzle Furthermore kion = 064 57 accounts forthe differences in ionization cross section for atomic andmolecular hydrogen and kdet = 084 for the detectionprobability55 As an example of the parameter stud-ies Fig 11 shows the deduced dependencies on the rfpower for a set of primary molecular hydrogen gas flowsFor typical flow values q(H2) le 10mbar ls a saturationvalue around 08 was obtained

0 50 100 150 200 250 300 350 400

00

02

04

06

08

10

Deg

ree

of d

isso

ciat

ion

()

Dissociator rf power [W]

q=07 mbar ls q=08 mbar ls q=10 mbar ls q=15 mbar ls

FIG 11 Degree of dissociation α of the free hydrogen jet asfunction of the applied rf power for different primary molec-ular hydrogen flows and a nozzle temperature of 70K

IV BEAM INTENSITY

The intensity of the polarized beam from the ABS to-gether with the layout of the storage cell determines theareal density of the target gas The intensity of the beamhas been measured with the use of a compression-tubesetup5859 shown in Fig 12 to optimize the ABS opera-tion parameters The measurements were performed at a300mm distance from the compression-tube entrance tothe last magnet and an inner tube diameter of 100mmas set in the tracking calculations The length of thecompression tube of 100mm was made equal to that ofthe foreseen feeding tube of the storage cell The narrowtube around the compression tube on a support based onthe lower flange separates the volume around the tube

FIG 12 Side view of the compression-tube setup made fromstandard ultra-high-vacuum components with a partial cutalong the axis (1 compression tube 2 support of thecompression tube based on the lower flange 3 narrow tubearound 1 closing the upper volume and allowing axial shiftsof the tube by the support 4 compression volume 5 hot-cathode pressure gauge 6 xy manipulator 7 z manipulator8 glass viewport 9 electromagnetic valve for gas inlet

from the compression volume The xy manipulator servesfor centering the tubes and for intensity-profile measure-ments The construction allows axial shifts of the com-pression tube by the z manipulator and the use of tubesof different diametersThe intensity of the beam entering the compression

volume through the compression tube is measured viathe pressure in the compression volume It is determinedby the equilibrium between the incoming beam intensityIin and the outgoing intensity Iout Under the assump-tion of a pure atomic beam and complete recombinationin the compression volume

Iin(atomss) = 2 middot Iout(moleculess)

= 2 middot∆P middot Ctube

= 2 middot∆P middot 103 middot 1020 middot d3

l

radic

T

M (7)

Here ∆P is the difference between the pressure measuredin the compreesion volume and that in the ABS cham-ber V The conductance of the compression tube Ctubeis determined by the inner diameter d of the tube itslength l the gas temperature T and the molar massM of the gas (given in cm and K respectively)60 The

11

CG

RVC200

H2

F

EV

EV

EV

EV

EV

MP

V1V2

EVR116

HC

V0

EV EVA016UDV136

HCCPLC

PS

220V AC

220V AC

Interface

24V DC

FIG 13 Source of calibrated molecular gas flow (V0 com-pression chamber V1 gas-storage chamber feeding V0 viathe needle valve UDV136a V2 chamber of calibrated volumeused to determine that of V1) The pressure in V1 is mea-sured by the capacitance gauge CG and is kept constant bythe dosing valve EVR116 with the gauge controller RVC200The whole setup including the evacuation elements can beoperated manually or by the programmable logic controllerPLC either within the ABS control system52 or as a separatesystem

a All the valves and the gauge controller are supplied by PfeifferVacuum GmbH Dndash35614 Asslar Germany (manufacturerBalzers AG Liechtenstein)

factor 2 takes into account that the same pressure is mea-sured in the hot-cathode gauge for 2 middotIin (H atomss) and1 middot Iin (H2 moleculess) For d = 10mm l = 100mmT = 290K and M = 2 for hydrogen pressure differences∆P on the order of 10minus4mbar are expected for atomichydrogen beam intensities in the order of 1017 atomssThe relation between Iin and ∆P for hydrogen has beendetermined experimentally with the use of a source ofcalibrated molecular hydrogen gas flow5859 depicted inFig 13 The measured dependence with a linear fit isshown in Fig 14 The calibration curve allows one to de-termine absolute values of Iin of hydrogen and deuteriumbeams The calibration for deuterium was deduced fromthe one for hydrogen by scaling with a factor 1

radic2 ac-

cording to Eq (7)

The dependences of Iin on the dissociator-operationparameters primary molecular hydrogen flow q(H2) noz-zle temperature Tn and dissociator power Pdiss have beenstudied to find the optimum values They are shown inthe Figs 15 16 and 17 respectively for different nozzle-orifice diameters The figures show that for the hydrogenbeam (states |1〉 and |2〉) with the standard operation pa-rameters qH2

= 11 mbar ls Tn = 70K Pdiss = 350Wand with a nozzle-orifice diameter of 23mm an inten-sity of Iin(H) = (75 plusmn 02) middot 1016 particless is achievedquite close to the earlier estimate from Eq (2) Besidesthe dominant atomic component of H atoms this value

04 06 08 10 12 14 16 18 20 2201

02

03

04

05

06

07

08

09

10

11

12

Gas

flow

from

the

stor

age

volu

me

[1017

ats

]

Pressure in compression volume [10-4 mbar]

FIG 14 Calibration curve for hydrogen used to deduce fromthe measured pressures the intensities of the hydrogen anddeuterium beam injected into the compression tube

00 05 10 15 20 2500

10

20

30

40

50

60

70

80 D = 20 mm D = 23 mm D = 25 mm

Bea

m in

tens

ity [1

016 a

ts]

Primary gas flow [mbar ls]

FIG 15 Intensity of the hydrogen beam (states |1〉 and |2〉)injected into the compression tube as function of the primarymolecular gas flow for different nozzle diameters D (nozzletemperature 60K dissociator power 300W)

includes small admixtures of H atoms in state |3〉 andmolecular hydrogen The first kind can be estimatedwith the use of the calculated transmissions (Sec II F)as 0017084 asymp 2 The amount of the second admix-ture has been measured as described below

For the deuterium beam (states |1〉 |2〉 and |3〉) theoptimization procedure gave an intensity of Iin(D) =(39 plusmn 02) middot 1016 particless achieved with q(D2) = 09mbar ls Tn = 65K and Pdiss = 300W slightly lowerthan those for hydrogen

12

50 60 70 80 90 100 110

40

50

60

70

80 D = 20 mm D = 23 mm D = 25 mm

Bea

m in

tens

ity [1

016 a

ts]

Nozzle temperature [K]

FIG 16 Intensity of the hydrogen beam (states |1〉 and |2〉)injected into the compression tube as function of the nozzletemperature for different nozzle diameters D (primary molec-ular gas flow 1mbar ls dissociator power 300W)

100 200 300 400 50010

20

30

40

50

60

70

80 D = 20 mm

Bea

m in

tens

ity [1

016 a

ts]

Dissociator power [W]

FIG 17 Intensity of the hydrogen beam (states |1〉 and |2〉)into the compression tube as function of the dissociator powerfor a nozzle diameter of 2 mm (nozzle temperature 60K pri-mary molecular gas flow 1mbar ls)

V HYDROGEN BEAM PROFILES

Beam profiles were measured at various positions atvarious positions behind the last sextupole magnet withthe use of

bull a compression tube of reduced dimensions (5mmdiameter)

bull a crossed-beam quadrupole mass spectrometer andbull a supplementary method of reduction of MoO3 by

hydrogen

A Measurements with the compression tube

For the determination of the beam dimensions at twopositions 300mm and 337mm behind the last magnetthe compression tube setup (Fig 12) was used makinguse of the possibility of axial movement by the z ma-nipulator and of that to install a narrower and shortercompression tube of 5mm diameter and 50mm lengthto enhance the spatial resolution The xy manipula-tor provided a lateral displacement of the compressiontube by plusmn10 mm in x and y direction The center co-ordinates of the geometrical axis of the source had beendetermined with the use of a bi-directional laser cen-tered inside the bore of the central support plate (seeFig 1) The relative intensity distributions in the xzand yz planes given by the measured pressure in thecompression volume are shown in Fig 18 Fits by Gaus-sian distributions to the data yield full widths at halfmaximum Γx = (642plusmn 009)mm Γy = (699plusmn 006)mmfor the distributions measured at z = 300mm and Γx =(627plusmn 008)mm Γy = (658plusmn 008)mm at 337mm

00

02

04

06

08

10

12

00

02

04

06

08

10

12

0 5 10 15 20 25

00

02

04

06

08

10

12

14

0 5 10 15 20 25

00

02

04

06

08

10

12

14

Pre

ssur

e in

the

com

pres

sion

tube

[au

]

Z = 300 mm

(a) (b)

(c) (d)

Z = 337 mm

X [mm]

Y [mm]

FIG 18 Cross sections of the beam profile in the mid-planemeasured with compression tube of 5mm diameter and 50mmlength Measurements in the xz-plane (a c) and yz-plane (bd) performed at two different positions z = 300mm (a b)and z = 337mm (c d) behind the last sextupole magnet ofthe ABS The shaded area represents position and dimensionsof the compression tube used in intensity measurements

The center of gravity of the measured profile definedas

rc =

sum

ij

radic

x2i + y2j middot P (xi yj)

sum

ij

P (xi yj) (8)

where xi and yj give the position of the compression-tubeaxis and P (xi yj) is the pressure measured in the com-

13

pression volume The resulting rc shows a deviation of012 mm from the geometrical axis of the source Further-more the data measured with the narrow compressiontube of 25mm radius can be used to derive the fractionof the beam entering the compression tube of 5mm ra-dius used in the intensity measurement of Sec IV Theratio

η =

rile25 mmsum

0P (xi yj)

rile10 mmsum

0P (xi yj)

(9)

where ri is the distance of the compression-tube axis tothe beam axis yields η asymp 07

B Measurements with the QMS

The beam-profile studies of Sec VA were extendedwith a setup utilizing a crossed-beam quadrupole massspectrometer (QMS) in the setup of Fig 19 Contraryto the measurements with the compression tube thosewith the QMS allow to separate the atomic and molec-ular fractions in the beam A 2mm diameter aperturewas installed at the entrance of the sensitive volume of

FIG 19 Setup for the measurements of the beam profilewith the QMS (1) xy-table enabling two-dimensional dis-placement of the entrance window of the QMS against thegeometrical axis of the ABS (2) the QMS (3) manually op-erated beam shutter The beam dump is an axially mountedcryo pump

the QMS to improve the resolution compared with thatachieved by the compression tube of 5mm diameter usedin measurements of the preceding section The layout ofthe setup presented in Fig 19 shows that in the presentcase the profile could not be measured at a distance ofz = 300mm to the last magnet Instead measurementswere performed at z = 567mm and with installation ofan extension tube at z = 697mm The xy manipulatorenabled displacements of the aperture axis from the geo-metrical axis of the source in any direction within limitsset by the bore diameter of the xy manipulatorThe first measured distribution of the atomic hydro-

gen (Fig 20) showed a distinct deviation from azimuthalsymmetry indicating an insufficient relative alignment ofnozzle and skimmer The three threaded rods support-ing the dissociator with the nozzle via the three-legged

FIG 20 Two-dimensional distribution of the atomic hydro-gen component of the beam at z = 567mm before the nozzle-to-skimmer adjustment showing a disinct deviation from az-imuthal symmetry

plate (label 2 in Fig 1) allow one to vary the position ofthe nozzle relative to that of the skimmer while the sourceis running This possibility has been used to find a nozzleposition which results in an azimuthally symmetric distri-bution The achieved symmetric distribution is shown inFig 21 and profiles of the atomic hadrogen component inthe beam measured in x and y direction at z = 567mmand z = 697mm are presented in Fig 22 Fits by Gaus-sian distributions to the data yield full widths at halfmaximum Γx = (736plusmn 043)mm Γy = (668plusmn 080)mmfor the distributions measured at z = 567mm and Γx =(669plusmn 022)mm Γy = (638plusmn 027)mm at 697mm

C Reduction of MoO3 by hydrogen

In addition to the compression tube and the QMS tech-nique a supplementary attempt was made to determinethe beam profile by exposing molybdenium trioxide (a

14

FIG 21 The distribution corresponding to that of Fig 20 af-ter nozzle-to-skimmer adjustment resulting in azimuthal sym-metry

0

20

40

60

80

0

20

40

60

80

2 4 6 8 10 12 14 16 18 20 22

0

20

40

60

80

6 8 10 12 14 16 18 20 22 24

0

20

40

60

80

QM

S si

gnal

s [a

u]

(a)

Z = 567 mm

Z = 697 mm

(c)

(b)

X [mm]

(d)

Y [mm]

FIG 22 Profiles of the atomic hydrogen component inthe beam measured with the QMS 567mm and 697mmbehind the last magnet

yellowish powder) on a glass plate to the beam The prin-ciple of this method is based on the reduction of MoO3 toa lower oxide of blue colour It first was used in the ex-periment to measure the magnetic moment of the hydro-gen atom by splitting of the beam in an inhomogeneousmagnetic field 61

This method is much simpler than the time-consumingmeasurements described in Secs VA and VB It givesqualitative results as presented in Fig 23 A quantitativeanalysis however requires development of the measuringtechnique (eg preparation of glass plates study of theoptimum exposure time digital image processing)

FIG 23 Photo of the glass plate covered with molybdenumtrioxide MoO3 exposed to the atomic hydrogen beam

D Summary of the profile measurements

Table III summarizes results of the measurements ofthe ABS beam profile with the compression-tube and theQMS setup The larger errors of the widths measuredwith the QMS are due to the lack of measurements withthe dissociator switched off and the necessity to estimatethe background signal from the existing data Within theerrors the measured widths do not show a dependenceon the distance from the last magnet This facilitatesto position the feeding tube of the storage cell in a widerange of a distances to the last magnet The average val-ues Γx = (638 plusmn 060)mm and Γy = (684 plusmn 033)mmagree within the errors and yield a common width ofΓxy = (673 plusmn 029)mm The two-dimensional Gaus-sian distribution of this width allows one to estimate thefraction η of the beam intensity injected into the com-pression tube or a feeding tube For a tube of 10mmdiameter η = 078plusmn 003 comparable with η asymp 07 givenin Sec VA

TABLE III Dimensions (FWHM) of the atomic hydrogenbeam measured with the compression tube (CT) and thecrossed-beam quadrupole mass spectrometer (QMS) at dis-tances z to the last magnet along perpendicular directions xand y

z[mm] Γx[mm] Γy[mm]

CT 300 642 plusmn 009 699plusmn 006

CT 337 627 plusmn 008 658plusmn 008

QMS 567 736 plusmn 043 668plusmn 080

QMS 697 669 plusmn 022 638plusmn 027

15

VI DEGREE OF DISSOCIATION

Besides the intensity of the atomic beam it is impor-tant to determine the molecular fraction in the beamMolecules injected into the feeding tube reduce the po-larization of the target gas

A Measurements with crossed-beam QMS

In addition to the data on the profile of the atomichydrogen beam (Sec VB) data on the distributions ofmolecular hydrogen in the beam were taken too at thepositions z = 567mm and 697mm behind the last mag-net The relation between the degree of dissociation andthe QMS signals by the atomic and molecular beam com-ponent was given above by Eq (6) The coefficient kv =vmva however is chosen here under the assumptionthat the average velocity of the atoms is determined bythe nozzle temperature of 65K and that of the moleculesby scattering and recombination on the ABS chamberwalls at 290K This yields kv =

radic

2 middot 65290 = 067 ingood agreement with Ref13 where this coefficient wasdetermined by the measured velocity distributions undersimilar conditionsThe measured profiles of the atomic fraction (identical

to those of Fig 21) those of the molecular fraction andthose of the degree of dissociation deduced from Eq (6)are collected in Fig 24As it is seen from the figure the distribution of the

degree of dissociation shows a dip around the central linedue to the higher density of molecular hydrogen originat-ing from the nozzle The mean value in an aperture of

0

20

40

60

80

100

120

2 4 6 8 10 12 14 16 18 20 22

0

20

40

60

80

100

120

6 8 10 12 14 16 18 20 22 24

QM

S si

gnal

s [a

u]

(a)

(b)

(c)

00

02

04

06

08

10

12

Deg

ree

of d

isso

ciat

ion

()

X [mm]

Y [mm]

(d)

00

02

04

06

08

10

12

Z = 567 mm

Z = 697 mm

FIG 24 Spatial distributions of H1 (bull)H2 () and degree ofdissociation () averaged over 3mm wide bands in the xz andyz planes respectively (here the z-axis is the geometrical axisof the ABS)

10mm diameter results as α = 095 plusmn 004

B Measurements with the Lamb-shift polarimeter

A cup in the quench chamber of the Lamb-shift po-larimeter (LSP) described in Ref8 allows one to mea-sure the currents Icup(H1) and Icup(H2) of the H+

1 andH+

2 ions extracted from the ionizer and separated bythe Wien filter with the cesium evaporation and the spinfilter switched off The relation between the degree ofdissociation α and the measured currents is

α =Icup(H1)minus r1

r2Icup(H2)

Icup(H1)minus r1r2Icup(H2) + 2kv

r2Icup(H2)

(10)

Among the three coefficients kv = 067 as for the mea-surement with the QMS For the electron energy of about100keV the ratio r1 of dissociative to non-dissociativeionization of H2 is8

r1 =σ(H2 rarr 2H+

1 )

σ(H2 rarr H+2 )

= 0095plusmn 0008 (11)

and the ratio between the ionization cross sections is857

r2 =σion(H2)

σion(H1)= 17plusmn 01 (12)

At the standard operation parameters of the source(Sec IV) the measured currents are Icup(H1) = (125 plusmn05) nA and Icup(H2) = (64 plusmn 01) nA yielding α =(096plusmn004) in excellent agreement with the value result-ing from the measurements with the QMS (Sec VIA)

VII BEAM POLARIZATION

The Lamb-shift polarimeter was designed built andtested at Universitat zu Koln7 It was used to measureand to optimize the polarization of the atomic hydrogenand deuterium beams delivered by the ABS Details arefound in Ref7The vector polarization pz for hydrogen is defined by

the relative hyperfine-state occupation numbers N(mI)

pz =N(+ 1

2 )minusN(minus 12 )

N(+ 12 ) +N(minus 1

2 ) (13)

for deuterium

pz =N(+1)minusN(minus1)

N(+1) +N(0) +N(minus1) (14)

Deuterium tensor polarization pzz is given by

pzz =N(+1) +N(minus1)minus 2 middotN(0)

N(+1) +N(0) +N(minus1) (15)

16

These polarizations can be derived from the measuredLyman-α peak strengths S by application of a number ofcorrection factors78

Typical Lyman α spectra measured with the polarizedhydrogen and deuterium beam from the ABS are shownin the Figs 25 and 26

MFT 2-3

200 300 400 5000

40

80

120

160

Magnetic field in the spinfilter [au]

PM

T si

gnal

[au

]

mI = -12

(a)

mI = +12

100 200 300 400 5000

40

80

120

160(b)

MFT 2-3WFT 1-3

mI = +12

mI = -12

FIG 25 Lyman-α spectra measured with the polarized hy-drogen beam (a) population change from state |2〉 to state|3〉 induced by the MFT unit (b) same as (a) with subsequentpopulation change from state |1〉 to state |3〉 induced by theWFT unit

MFT 3-4SFT 2-6

0

2

4

6

8

mI = -1m

I = 0

(a)

mI = +1

0

2

4

6

8(b)

MFT 3-4WFT 1-4 2-3

mI = +1 m

I = 0

mI = -1

mI = +1

mI = 0

mI = -1

mI = +1

mI = 0

mI = -1

0 1000 2000 3000

0

1

2

3

4 (c)

PMT

sign

al [a

u]

Magnetic field in the spinfilter [au]

WFT 1-4SFT 2-6

0 1000 2000 30000

2

4

6

8WFT 1-4SFT 3-5

(d)

FIG 26 Lyman-α spectra measured with the polarized deu-terium beam (a) and (b) vector polarization resulting fromsubsequent transitions MFT (3 rarr 4) and SFT (2 rarr 6) andWFT (1 rarr 4 2 rarr 3) respectively (c) and (d) tensor polar-ization resulting from subsequent transitions WFT (1 rarr 4)and SFT (2 rarr 6) and SFT (3 rarr 5) respectively

The polarization values for the hydrogen and the deu-terium beam derived from the Lyman-α peak-strengthratios with application of the necessary corrections aresummarized in Table IV

The vector polarization for hydrogen of the first linereflects the population of state |1〉 and state |2〉 accordingto the Eqs (3) and (13) The value of 091 deduced fromthe calculated transmission values is confirmed by themeasured one

TABLE IV The vector polarization pz of the hydrogen beamand the vector and the tensor polarization pzz of the deu-terium beam from the ABS measured with the Lamb-shiftpolarimeter

populated pz pzz

state(s)

Hydrogen |1〉 +089plusmn 001 -

|3〉 minus096plusmn 001 -

Deuterium |1〉+ |6〉 +088plusmn 001 +088plusmn 003

|3〉+ |4〉 minus091plusmn 001 +085plusmn 002

|3〉+ |6〉 +0005plusmn 0003 +090plusmn 001

|2〉+ |5〉 +0005plusmn 0003 minus171plusmn 003

VIII CONCLUSIONS AND OUTLOOK

In this paper we present the detailed description ofthe major components of the atomic beam source (ABS)for the polarized internal gas target of the magnet spec-trometer ANKE in COSY-Julich The ABS was builtfor the purpose of extending the physics program ofANKE from unpolarized and single-polarized investiga-tions with stored beams towards double-polarized exper-iments1 thus facilitating nuclear reaction studies involv-

ing ~p~p ~p~d ~d~p and ~d~d initial statesThe mechanical design took into account that at

ANKE the source has to be mounted vertically and trans-versely movable together with the transverse motion ofthe spectrometer magnet D2 The design of the system ofsextupole magnets took advantage of the developments inthe field of rare-earth permanent magnets (NdFeB) Ded-icated tools and methods were developed to determineand to optimize the source parameters ie intensity de-gree of dissociation and polarization Special emphasiswas put on the measurements of the spatial distributionsof the atomic and molecular beam near the focus wherethe feeding tube of the storage cell is located The ABShas been used in a number of investigations at ANKEthe commissioning effort to prepare the target for the usewith polarized H is described in Ref62 Performed stud-ies of the deuteron-charge exchange reaction are summa-rized in Ref6364 studies in near-threshold pion produc-tion are reported about in Ref65The ABS resides at the ANKE target position for a few

months per year only thus during the remaining timeit is used for other studies It had been observed thatthe nuclear polarization in recombined hydrogen is par-tially retained after recombination66 as well as evidencefor nuclear tensor polarization in recombined deuteriummolecules67 In order to investigate this recombinationprocess in more detail a special setup has been developedin the framework of an ISTC project68 and the recombi-nation process for different cell-wall coatings and differ-ent polarizations of the injected hydrogen or deuterium

17

atoms as function of cell-wall temperature strength ofthe magnetic holding field and gas pressure in the cell ispresently investigated69ndash71

Appendix A Preparation of Discharge Tubes and Nozzles

1 Tube Treatment

One end of the discharge tube is machined at a 45

angle while the other is kept flat Both ends are thenremelted and the tubes are tempered at 150 C Thetubes are further treated according to the procedure de-scribed in Ref72 which includes successive cleaning withacetone methanol distilled water and subsequent rins-ing by a 21 acid mixture of concentrated HF (40) andHCl (32) for 5 min The tubes are then flushed bydistilled water and dried

2 Nozzle Treatment

The nozzles are cleaned in an ultrasonic bath oftrichlorethylene acetone methanol and finally distilledwater all at 50 C Anodizing takes place in sulfuric acidto form a thin layer of Al2O3 as described in Ref72 Af-terwards they are immersed in distilled water for 30minat 95 C

ACKNOWLEDGMENTS

The authors want to thank OWB Schult Institut furKernphysik (IKP) Julich who initiated the polarizationprogram of ANKE Thanks go to the design office themechanical workshop and especially to WR Ermer allIKP Valuable advice was received from the PINTEX col-laboration at IUCF from the target group at HERMESespecially NKoch and from DToporkov BINP Novosi-birsk The support by VKoptev PNPI Gatchina whoregrettably passed away in January 2012 is gratefullyacknowledged Thanks go also to R Poprawe and col-leagues Fraunhofer-Institut fur Lasertechnik Aachenwhere the encapsulations of the magnets were laser-welded

1AKacharava F Rathmann and CWilkin Spin Physics fromCOSY to FAIR COSY Experiment Proposal No 152 (2005)Available under httparXivnucl-ex0511028

2WHaeberli in Proc 2nd Int Symp on Polarization Phenomenaof Nucleons Karlsruhe 1965 Eds P Huber and H SchopperExperientia Supplementum 12 64 (Birkhauser Verlag 1966)

3E Steffens and WHaeberli Rep Progr Phys 66 1887 (2003)4SBarsov et al Nucl Instr and Meth A 462 364 (2001)5RMaier Nucl Instr and Meth A 390 1 (1997)6KGrigoryev et al Proc 14th International Workshop on Po-larized Sources Targets and Polarimetry (PSTP 2011) 12-16 September 2011 StPetersburg Russia eds KGrigoryevPKravtsov and AVasilyev ISBN 978-5-86763-282-3 61 (2011)

7REngels et al Rev Sci Instrum 74 4607 (2003)8REngels et al Rev Sci Instrum 76 053305 (2005)9TWise et al Nucl Instr and Meth A 336 410 (1993)

10WADezarn et al Nucl Instr and Meth A 362 36 (1995)11T Rinckel et al Nucl Instr and Meth A 439 117 (2000)12F Stock et al Nucl Instr and Meth A 343 334 (1994)13ANass et al Nucl Instr and Meth A 505 633 (2003)14VDerenchuk et al Proc Conf Polarized Ion Sources and Polar-ized Gas Targets Madison WI 1993 Eds LWAnderson andWHaeberli AIP Conf Proc 293 72 (American Institute ofPhysics 1994)

15HOkamura et al see Ref14 p 8416KHatanaka et al Nucl Instr and Meth A 384 575 (1997)17Manufacturer Schiffer Metall- amp Vakuumtechnik 52428 JulichGermany

18Single-stage type RGS120 refrigerating capacity 120W at 80Kand 20W at 30K Leybold Vacuum GmbH 50968 Koln Ger-many

19Mini UHV gate valve series 010 VAT Germany GmbH 85630Grasbrunn Germany

20Type F3 fomblin oil Pfeiffer Vacuum GmbH 35614 Asslar Ger-many

21Model HU 1 Leybold Vacuum GmbH 50968 Koln Germany22Manufacturer SK Industriemodell GmbH 52531 Ubach-Palenberg Germany

23Type PFG 600 RF with automatic matchbox PFM 1500 A-INDHuttinger Elektronik GmbH 79110 Freiburg Germany

24WKorsch PhD Thesis Philipps Universitat Marburg (1990)25F Stock et al Int Workshop on Polarized Beams and PolarizedGas Targets Koeln Germany 1995 Eds HPaetz gen Schieckand L Sydow (World Scientific Publ Co 1996) p 260

26The first number denotes the outer diameter and the second onethe wall thickness

27Type Duran 8330 equivalent to Corning 7740 (Pyrex) SchottAG 55122 Mainz Germany

28Ultra-Kryomat RUL 80-D Lauda DrRWobser GmbH 97912Lauda-Konigshofen Germany

29ODU-Kontakt GmbH 84444 Muhldorf Germany30Handbook of Chemistry and Physics Ed RCEast (The Chem-ical Rubber Co 1973) p E-10

31NKoch and E Steffens Rev Sci Instrum 70 1631 (1999)32AVassiliev et al Petersburg Nuclear Physics Institute ReportNP-32-1997 No 2175 (1997)

33BLorentz Diploma Thesis Ruprecht-Karls-Universitat Heidel-berg (1993)

34ANass and E Steffens Nucl Instr and Meth A 598 653(2009)

35WHaeberli Ann Rev Nucl Sci 17 373 (1967)36AVassiliev et al Rev Sci Instr 71 3331 (2000)37WKubischta Proc Workshop on Polarized Gas Targetsfor Storage Rings Heidelberg 23-26 September 1991 EdsHGGaul E Steffens and K Zapfe (Max-Planck-Institut furKernphysik Heidelberg)

38The labeling of the hyperfine states as |1〉 = |mj = +12 mI =+12〉 |2〉 = | + 12minus12〉 |3〉 = | minus 12+12〉 and |4〉 =|minus12minus12〉 for hydrogen and |1〉 = |+12+1〉 |2〉 = |+12 0〉|3〉 = | + 12minus1〉 |4〉 = | minus 12minus1〉 |5〉 = | minus 12 0〉 and

5

The temperature at the bottom of the nozzle is measuredwith a Pt-100 sensor and it is stabilized with an accuracyof plusmn05K utilizing a heater Measurements with temper-ature sensors placed along the outer nozzle surface haveshown a temperature increase from 60K at the nozzlebottom to sim200K at the sharp nozzle tip In the follow-ing the nozzle temperature is defined as that measuredwith this Pt-100 sensorWith the present system of sextupole magnets the

maximum atomic beam intensity feeding the storagecell is obtained with a nozzle-orifice diameter of 23mmand a nozzle-tip to skimmer-tip distance of 15mm at askimmer-tip diameter of 44mm and a skimmer-tip to di-aphragm distance of 17mm The 2mm thick diaphragmwith a conical bore opening from 95mm to 99mm to-wards the first permanent sextupole magnet shields themagnet from heating by atoms recombining on its sur-face The slit between the diaphragm and the front faceof the magnet enables pumping of gas from the entranceto the magnetThe Teflon washer and the stainless steel support sep-

arate the cold lower end of the heat bridge from thewarm lower end of the dissociator The dimensions of

FIG 5 Technical drawing including the lower end of theheat bridge and the dissociator the nozzle surroundings andthe first sextupole magnet (in scale 1 discharge and coolant-guiding tubes 2 coolant-reversal piece 3 heat flow reducingTeflon washer 4 sliding heat connection 5 stainless steelconnector 6 groove for nozzle-heating element 7 lower endof the Cu heat bridge 8 nozzle fixture 9 nozzle 10 baf-fle separating the chambers I and II with a viewport 11stainless steel beam skimmer 12 Cu diaphragm 13 firstsextupole magnet and 14 baffle separating the chambers IIand III

these two components and the sliding heat connector aworked-over sliding high current connector similar to therf connector in the dissociator define the temperatureof the lower end of the discharge tube relative to thatof the nozzle The discharge tube adapted at its lowerend to the nozzle by a chamfered edge is pressed to thenozzle by a viton O-ring at its upper end The two O-rings around the discharge tube in the lower part of thedissociator seal against the atmosphere By this designonly minor forces are exerted to the discharge tubeThe removable viewport in the baffle and the window

flange in the upper vacuum vessel (on the right-hand sideof chamber II in Fig 1) allows one to observe the nozzlestatus from the outside and to exchange nozzles withoutremoval of the dissociator from the setupThe heat bridge from the coldhead to the nozzle is

made from electrolytic Cu The flexible link between thecoldhead and the heat bridge consisting of about 200high-purity Cu strands of 1 mm diameter allows for thethermal expansions of the cold and the warm compo-nents The total cross section of the strands and theirheat conductance is smaller than that of a massive Cubody This deficiency however is reduced by clampingthe flexible link directly to the coldhead At its operat-ing temperature of about 30K the thermal conductivityof Cu is about 11 9 and 5 times higher than that at300 100 and 60K respectively30 Thus the reductionof the conductance of the entire heat bridge by the flexi-ble link is minimized by placing it at the coldhead Withthe present system cooling the nozzle down from roomtemperature to 60K needs about 15 hours The heat-ing element facilitates warming up to room temperaturewithin about one hourFurthermore avoiding the maze of cold Cu strands

around the nozzle ie a labyrinthic cold surface com-pared to an earlier solution31 leads to improved pumpingconditions in the nozzle-skimmer area where the highestgas load has to be pumped offIn an earlier phase of the ABS development attempts

have been made to use a cryogenic Ne heat-pipe of 20Wcooling power instead of the usual solid Cu bridge toachieve faster cooling and warming of the nozzle becauseof the lower heat capacity32 An observed instability inthe necessary operation mode however lead to difficul-ties in nozzle-temperature stabilization In view of thefact that the cooling and warming-up times reached withthe Cu bridge were satisfying and that its use avoids theadditional precautions imposed by the heat-pipe opera-tion it has been replaced by the Cu bridge

F Magnet System

The design of the magnet system was made for a setof sextupole magnets consisting of permanently magne-tized segments made from NdFeB compounds deliveringpole-tip fields around 15 T Tracking calculations fromthe nozzle to the feeding tube of the storage cell were

6

performed with the use of a computer code originallydeveloped for the FILTEX ABS24 The boundary condi-tions by the layout of the target setup were the availabledistance of about 1250mm from the nozzle to the feeding-tube entrance of 10mm diameter and the distance fromthe exit of the last magnet to the feeding-tube entrance of300mm necessary to install the SFT and WFT units andthe gate valve between the ABS and the target chamberThe laboratory velocity distribution of the atoms in

the supersonic beam from the nozzle is described by amodified Maxwellian distribution

F(~vd Tb) =( m

2 k Tb

)32exp

[ minusm

2 k Tb(~v minus ~vd)

2

]

(1)

where m is the mass of the atoms and k is the Boltzmannconstant According to time-of-flight studies33 the driftvelocity along the beam axis vd and the beam temper-ature Tb for a primary molecular gas flow of 1 mbar lsand a nozzle-orifice diameter of 2mm follow a linear de-pendence on the nozzle temperature Tn For hydrogenvd[ms] = 1351 + 61 middot Tn[K] and Tb = 029 middot Tn and fordeuterium vd[ms] = 1070+345middotTn[K] and Tb = 025middotTnAs starting conditions of a track a random generator

selects a point in the nozzle orifice one within the di-aphragm in front of the first magnet and an atom ve-locity |v| In linear molecular flow approximation (cfthe discussion in Ref34) this defines ~v for the track be-tween the nozzle and the first magnet According to thegeometrical boundary conditions and the velocity distri-bution of Eq (1) the event is either rejected or used inthe further track calculation Within the magnet theevolution of the track is calculated stepwise by numeri-cal integration of the equation of motion over integrationtimes of 2micros corresponding to track lengths of 36mmfor a typical particle velocity of 1800ms The pureradial force acting on an atom within the field of the

sextupole magnet is ~Fr = minusmicroeff middot δBδr middot ~rr The ef-fective magnetic moment resulting from the Breit-Rabidiagram (eg Ref35) as microeff = δWδB is positive (neg-ative) for atoms in the hyperfine states with the electron

spin parallel (antiparallel) to ~B in the magnet aperturewhich therefore are deflected towards (away from) thebeam axis In the drift sections between the two magnetgroups and between the last magnet and the feeding tubethe trajectories are assumed as straight linesA variety of systems were studied all under the as-

sumption of Tn = 60K and pole-tip fields of 15T Asystem utilizing 6 magnets was found to yield satisfyingboth separation of the atoms in the microeff lt 0 and microeff gt 0states and focusing of the microeff gt 0 states into the feedingtube Optimization of the parameters led to the systemlisted in Table II (The tracking calculations yielding themagnet dimensions for the order to the manufacturer hadbeen performed for a slightly different geometry) Thetable gives the two distances at which intensity mea-surements with the compression tube were performedThe Fig 6 shows the projection of the trajectories of Hatoms in the microeff gt 0 states calculated for this system

TABLE II Final dimensions and axial positions of the sourcecomponents (pole-tip field strenghts Blowast

0 as measured afterdelivery36 inner diameters (empty0) outer diameters (empty1) axialdimensions (ℓ) and distances (∆) between the componentsThe lower three lines give the two distances and the dimen-sions of the compression tube used in the intensity measure-ments

component Blowast

0 [T] empty0 [mm] empty1 [mm] ℓ [mm] ∆ [mm]

Nozzle orifice 23 33150

Skimmer 44304a 130169

Diaphragm 9599a 2036

Magnet 1 1630 10401412a 3998 400194

Magnet 2 1689 15982212a 6404 650194

Magnet 3 1628 2804 9400 70014297

Magnet 4 1583 3004 9402 38011010

Magnet 5 1607 3006 9400 5501150

Magnet 6 1611 3002 9404 550030003370

Compr tube 100 110 1000

a Conical openingthe first number denotes the measureddiameter of the entrance the second that of the exit aperture

One recognizes two groups of trajectories one with anintermediate focus and another one with focusing intothe feeding tube The present result like those of othergroups (see eg Ref33) confirms the expectation37 thatthe transmission as function of the atom velocity shouldshow two maxima one below and one above the mostprobable velocityThe transmission Tr of the system is defined as the

fraction of tracks ending within the entrance of the feed-ing tube to those passing the diaphragm in front of thefirst sextupole magnet For the four hyperfine states ofhydrogen38 the calculations yield Tr(|1〉) sim Tr(|2〉) =042 (for both microeff gt 0) and Tr(|3〉) = 0001 andTr(|4〉) = 00004 (for both microeff lt 0)The performed tracking calculations do not account for

intra-beam and residual-gas scattering The calculatedtransmissions thus only allowed one to estimate upperlimits of the expected atomic beam intensity Iin into thefeeding tube For a primary molecular flow q(H2) theintensity Iin(H) with atoms mainly in the states |1〉 and|2〉 (microeff gt 0) was expected as

Iin(H) = q(H2) middot 2α middot Ω

2πmiddot 14

i=4sum

i=1

Tr(|i〉) (2)

For the degree of dissociation α a routine value of 08(see eg Ref9) was assumed Ω = 0022π is the solid

7

angle covered by the collimator aperture The factor 14reflects the assumption that the four substates in theatomic beam from the nozzle are equally populated Forq(H2) = 1mbar ls or 27middot1019H2 moleculess one expectsIin(H) sim 1 middot 1017H atomssAs described in the subsequent section the rf tran-

sition units are used to change the relative occupationnumbers of the states The trajectory code allows oneto simulate this change by assigning a microeff of one of thestates to the atoms before they pass a magnet As an ex-ample the medium-field transition unit (MFT) behindmagnet No 3 (see Fig 1) brings H atoms from state |2〉into state |3〉 This is simulated by assigning microeff(|2〉) gt 0to the atoms in the magnets 1minus3 and microeff(|3〉) lt 0 in themagnets 4 minus 6 where they get deflected from the beamaxis This results in a small value Tr(|2〉) = 0017 Fromthis value and the above value Tr(|1〉) = 042 the vectorpolarization is expected as

pz =Tr(|1〉)minus Tr(|2〉)Tr(|1〉) + Tr(|2〉) = 091 (3)

under the assumption of 100 efficiency of the transitionunitThe design and the properties of the permanent sex-

tupole magnets39 were discussed in an earlier paper36To achieve the pole-tip field values of sim15T each mag-net was produced from 24 segments employing three dif-

FIG 6 Projection of the 3-dimensional trajectories of hydro-gen atoms in hyperfine states |1〉 and |2〉 (effective magneticmoment microeff gt 0)) from the nozzle (empty = 2mm Tn = 60K)to the storage cell calculated for the magnet arrangement ofTable II and pole-tip fields of 15 T The positions and lat-eral dimensions of the six magnets and the feeding tube areindicated (in red)

ferent types of NdFeB compounds The expected pole-tip values (Table II) and the precise radial dependenceB(r) sim r2 within the magnet apertures were confirmedFor the first time the predicted high multipole compo-nents40 up to a 102-pole structure very near to the aper-ture surface could be measured36After the field measurements the magnets were encap-

sulated to prevent diffusion of hydrogen into the magnetmaterial which might deteriorate the magnetic proper-ties and to avoid the pumping of gas from the sinteredmagnet bodies The housings were made from thin stain-less steel cans of 02mm thickness for the conical andcylindrical walls within the magnet apertures and 03mmfor the front and end covers During the final welding toclose the housings with magnets installed the local tem-perature of the magnet material had to be kept belowthe Curie temperature of 60 C This was achieved bywelding with the use of a pulsed 15Hz NdYAG laserdelivering 11 J in a pulse of 2ms41 Overlapping weldspots of sim03mm diameter set around the adjacent cir-cular 02mm thick weld lips allowed one to finish thehousings with leak rates sim 10minus10mbar ls Inside thehousings the magnets were fixed to suppress axial androtational movements caused by the force of the adja-cent magnets Finally the free slits within the housingswere filled by sim20mbar krypton to enable leak tests bymass spectroscopy

G Radio Frequency Transition Units

The ABS is equipped with three types of transitionunits a weak field a medium field and a strong field rftransition unit (WFT MFT and SFT units) Togetherwith the selecting properties of the sextupole magnetsthey enable one to achieve all vector and tensor polar-izations of the atomic hydrogen and deuterium gas inthe storage cell In all three units transitions betweenthe hyperfine states split according to the Breit-Rabi di-agram by a static magnetic field (see eg Ref35) areinduced by the magnetic component (Brf) of an rf fieldleading to changes in the population of the states Thestatic field Bstat consists of two parallel components ahomogeneous field Bhom and a superimposed weaker gra-dient field Bgrad both orthogonal to the beam directionThe field gradient along the beam direction is requiredto satisfy the condition of adiabatic passage3542The assemblies of the WFT and the MFT units are

similar43 The layouts follow those of the units devel-oped for the HERMES experiment44 In both units therf field is produced by a coil with the axis along the beamdirection and consequently Brf orthogonal to Bstat TheMFT unit is shown in Fig 7 Figure 8 schematicallyshows one of the grooved aluminum frames with thewindings producing the gradient field A WFT unitis operated in a weak magnetic field Bstat le10G for hy-drogen and le5G for deuterium where the total atomicspin F is a good quantum number In hydrogen the

8

F = 1 levels |1〉 |2〉 and |3〉 with magnetic quantumnumbers mF = +1 0 and minus1 respectively can be re-garded as equally spaced In deuterium the same holdsfor the four F = 32 levels |1〉 |2〉 |3〉 and |4〉) withmF = +32 +12 minus12 and minus32 respectively andfor the two F = 12 levels |5〉 and |6〉 with mF = minus12and +12 respectively The magnetic component of therf dipole field induces transitions between each pair ofneighboring mF states with ∆mF = plusmn1 |∆mF| = 2transitions are forbidden The interchange of the popu-lation between the states |1〉 and |3〉 in hydrogen eg iscaused by a two-quantum transition via the intermedi-ate state |2〉 In the classical description of the adiabaticpassage method42 the population change should not de-pend on the sign of the magnetic field gradient relativeto the beam direction An exact quantum-mechanicaltreatment4546 however indicates that cleaner popula-tion changes from state |1〉 to |3〉 in hydrogen and fromstate |1〉 to |4〉 in deuterium are obtained with a nega-tive field gradient ie a Brf field decreasing in the beamdirection Deviations from adiabaticity are discussed inRef4547

The MFT unit is operated at higher values of Bstatwhere the differences in the energy spacings of pairs of hy-

FIG 7 Three-quarter-section view of the MFT unit with thesupport structure (1 self-supporting rf coil with spacers 2pick-up loop 3 Al tubes defining the length of the transition-inducing rf field 4 Cu cavity 5 coil around the pole shoe(6) providing the static field Bstat 7 slit between pole shoeand cavity wall housing the gradient-field coil 8 componentsof the static magnet yoke also serving as shielding againstexternal fields 9 cavity-positioning element 10 Cu padscooled by means of water-carrying tubes The cavity withthe rf coil and the pick-up loop can be taken out from thesurrounding components

z

Bstatic

transition

region

FIG 8 Arrangement of the windings producing the staticgradient field Bgrad shown in the left-hand side of the figureIn all transition units the field lies in the direction of the statichomogeneous field the field gradient dBdz lies in the beamdirection which defines the z axis In z direction the tran-sition reagion (indicated by the dashed lines) is confined tothe range of constant gradient by the Al tubes in orthogonaldirection by the beam diameter

perfine states with ∆mF = plusmn1 allow one to select singletransitions Originally developed for an polarized alkaliion source48 the MFT unit now is a standard compo-nent in polarized hydrogen and deuterium sources as dis-cussed eg in Ref49 Appropriate choice ofBhom Bgradand the rf frequency allows one to induce selected tran-sitions |1〉 harr |2〉 and |2〉 harr |3〉 in hydrogen or |1〉 harr |2〉|2〉 harr |3〉 and |3〉 harr |4〉 in deuterium Furthermore thechoice of the field gradient allows one to achieve consecu-tive transitions As an example a negative field gradientin the MFT unit behind the first set of magnets ie aB field decreasing in beam direction at a fixed rf fre-quency leads to the sequence of the transitions |3〉 rarr |4〉|2〉 rarr |3〉 and finally |1〉 rarr |2〉 in deuterium leaving thestate |1〉 empty

The SFT unit is used to induce transitions betweenstates in the upper and lower hyperfine multiplet in hy-drogen and deuterium Contrary to the historical nameindicating a strong magnetic field the SFT unit is op-erated with magnetic fields comparable to those used inthe MFT unit The transition frequencies are comparablewith those of the hyperfine splitting (1420MHz for hy-drogen and 327MHz for deuterium) and thus are muchhigher than those in the WFT and MFT units The rffield in a SFT unit is produced by a twin-line resonatorinside a Cu box tuned to the λ4 resonance50 The SFTunit51 is shown in Fig 9 Again the layout follows that ofthe unit used in the HERMES experiment44 Two vari-able capacitors at the free ends of the conducting rodsfed by the rf power with a relative phase shift of 180 allow one to tune the device

9

FIG 9 Three-quarter-section view of the rf cavity of theSFT unit for deuterium (1 the two resonant-field creatingconductors 2 the adjustible capacitor plates 3 Cu cavity)The inner dimensions of the cavity are 56mm along Bstat36mm orthogonal to it and 36mm height The cross sectionof the conductors is 14times 4mm2

H Slow Control System

Industrial components providing reliable and long-term support were selected for the control system of thewhole setup consisting of the ABS and the diagnosticstools the storage cell positioning system the Lamb-shiftpolarimeter and the supply system of a calibrated flowof unpolarized molecular gas The interlock system hasbeen implemented on the basis of SIEMENS SIMATICS7-300 family of programmable logic controllers In orderto unify the interfacing to the control computer all front-end equipment is connected via the PROFIBUS DP field-bus The process control software was implemented us-ing the Windows-based WinCC toolkit from SIEMENSThe system controls the operation of the pumps and thevalves It reads the pressure gauges and controls theregeneration cycles of the cryopumps Via a control net-work the temperature of the nozzle is stabilized withinplusmn05K Furthermore all power-supply units rf genera-tors and amplifiers are set and controlled The wholevariety of components to be controlled the logical struc-ture of the control and interlock system and a separatedevice for parameter studies are described in Ref52

III STUDIES OF THE FREE HYDROGEN JET

A Atomic beam profile near the nozzle

A novel device has been used to measure the profile ofan atomic beam via the deposition of recombination heaton thin wires in a two-dimensional grid5354 Atoms stuckon the surface of gold-plated tungsten wires of 5microm diam-eter recombine and are reemitted as molecules The re-combination heat (4476 eV per hydrogen molecule) leadsto a change of temperature and thus resistance alongeach wire The measurement of the resistance changes of

FIG 10 Two-dimensional profile of the atomic hydrogenbeam 10mm from the nozzle deduced from recombinationheating of gold-plated tungsten wires of 5microm in a 8times 8 wiregrid

all the wires in the grid allows one to deduce the centerand the profile of the beam Figure 10 shows the beamprofile resulting with a 8times8 wire grid positioned betweenskimmer and collimator performed as a first proof of themethod Later such a device has been used to comparemeasured and calculated beam profiles along the beamaxis between nozzle and skimmer34

B Degree of dissociation of the free atomic jet

The dissociation of the primary molecules is achievedby the interaction of the electrons and the hydrogen ordeuterium molecules in the plasma of the dissociatorThe degree of dissociation of the beam from the nozzledepends on the rf power applied to maintain the plasmathe primary molecular gas flow into the dissociator andthe temperature of the nozzle and the lower end of thedischarge tube These dependencies have been studiedbefore installation of the sextupole magnets with a setupcontaining a crossed-beam quadrupole mass spectrome-ter5556

α =ρa

ρa + 2 middot ρm (4)

The admixture of molecules in an atomic beam is de-scribed by the degree of dissociation where ρa and ρmare the densities of atomic and molecular hydrogen ordeuterium in the beam Other authors (eg Ref13) usethe atomic and molecular intensities Ia and Im in the def-inition of the degree of dissociation (αI) in Eq (4) Thetwo definitions of are related by

ImIa

=vmva

middot 1minus α

2α=

1minus αI

2αI (5)

10

This quantity was determined with the quadrupolemass spectrometer (QMS) in a conventional way as

α =Slowasta

Slowasta + 2 kv kion kdetSm

(6)

Here Slowasta = SaminusδSm denotes the atomic signal corrected

for dissociative ionization The parameter δ = 00141was obtained following the method described in Ref31The coefficient kv = vmva accounting for the difference

in atom and molecule velocity was chosen as 1radic2 under

the assumption of thermalization of the beam emergingfrom the nozzle Furthermore kion = 064 57 accounts forthe differences in ionization cross section for atomic andmolecular hydrogen and kdet = 084 for the detectionprobability55 As an example of the parameter stud-ies Fig 11 shows the deduced dependencies on the rfpower for a set of primary molecular hydrogen gas flowsFor typical flow values q(H2) le 10mbar ls a saturationvalue around 08 was obtained

0 50 100 150 200 250 300 350 400

00

02

04

06

08

10

Deg

ree

of d

isso

ciat

ion

()

Dissociator rf power [W]

q=07 mbar ls q=08 mbar ls q=10 mbar ls q=15 mbar ls

FIG 11 Degree of dissociation α of the free hydrogen jet asfunction of the applied rf power for different primary molec-ular hydrogen flows and a nozzle temperature of 70K

IV BEAM INTENSITY

The intensity of the polarized beam from the ABS to-gether with the layout of the storage cell determines theareal density of the target gas The intensity of the beamhas been measured with the use of a compression-tubesetup5859 shown in Fig 12 to optimize the ABS opera-tion parameters The measurements were performed at a300mm distance from the compression-tube entrance tothe last magnet and an inner tube diameter of 100mmas set in the tracking calculations The length of thecompression tube of 100mm was made equal to that ofthe foreseen feeding tube of the storage cell The narrowtube around the compression tube on a support based onthe lower flange separates the volume around the tube

FIG 12 Side view of the compression-tube setup made fromstandard ultra-high-vacuum components with a partial cutalong the axis (1 compression tube 2 support of thecompression tube based on the lower flange 3 narrow tubearound 1 closing the upper volume and allowing axial shiftsof the tube by the support 4 compression volume 5 hot-cathode pressure gauge 6 xy manipulator 7 z manipulator8 glass viewport 9 electromagnetic valve for gas inlet

from the compression volume The xy manipulator servesfor centering the tubes and for intensity-profile measure-ments The construction allows axial shifts of the com-pression tube by the z manipulator and the use of tubesof different diametersThe intensity of the beam entering the compression

volume through the compression tube is measured viathe pressure in the compression volume It is determinedby the equilibrium between the incoming beam intensityIin and the outgoing intensity Iout Under the assump-tion of a pure atomic beam and complete recombinationin the compression volume

Iin(atomss) = 2 middot Iout(moleculess)

= 2 middot∆P middot Ctube

= 2 middot∆P middot 103 middot 1020 middot d3

l

radic

T

M (7)

Here ∆P is the difference between the pressure measuredin the compreesion volume and that in the ABS cham-ber V The conductance of the compression tube Ctubeis determined by the inner diameter d of the tube itslength l the gas temperature T and the molar massM of the gas (given in cm and K respectively)60 The

11

CG

RVC200

H2

F

EV

EV

EV

EV

EV

MP

V1V2

EVR116

HC

V0

EV EVA016UDV136

HCCPLC

PS

220V AC

220V AC

Interface

24V DC

FIG 13 Source of calibrated molecular gas flow (V0 com-pression chamber V1 gas-storage chamber feeding V0 viathe needle valve UDV136a V2 chamber of calibrated volumeused to determine that of V1) The pressure in V1 is mea-sured by the capacitance gauge CG and is kept constant bythe dosing valve EVR116 with the gauge controller RVC200The whole setup including the evacuation elements can beoperated manually or by the programmable logic controllerPLC either within the ABS control system52 or as a separatesystem

a All the valves and the gauge controller are supplied by PfeifferVacuum GmbH Dndash35614 Asslar Germany (manufacturerBalzers AG Liechtenstein)

factor 2 takes into account that the same pressure is mea-sured in the hot-cathode gauge for 2 middotIin (H atomss) and1 middot Iin (H2 moleculess) For d = 10mm l = 100mmT = 290K and M = 2 for hydrogen pressure differences∆P on the order of 10minus4mbar are expected for atomichydrogen beam intensities in the order of 1017 atomssThe relation between Iin and ∆P for hydrogen has beendetermined experimentally with the use of a source ofcalibrated molecular hydrogen gas flow5859 depicted inFig 13 The measured dependence with a linear fit isshown in Fig 14 The calibration curve allows one to de-termine absolute values of Iin of hydrogen and deuteriumbeams The calibration for deuterium was deduced fromthe one for hydrogen by scaling with a factor 1

radic2 ac-

cording to Eq (7)

The dependences of Iin on the dissociator-operationparameters primary molecular hydrogen flow q(H2) noz-zle temperature Tn and dissociator power Pdiss have beenstudied to find the optimum values They are shown inthe Figs 15 16 and 17 respectively for different nozzle-orifice diameters The figures show that for the hydrogenbeam (states |1〉 and |2〉) with the standard operation pa-rameters qH2

= 11 mbar ls Tn = 70K Pdiss = 350Wand with a nozzle-orifice diameter of 23mm an inten-sity of Iin(H) = (75 plusmn 02) middot 1016 particless is achievedquite close to the earlier estimate from Eq (2) Besidesthe dominant atomic component of H atoms this value

04 06 08 10 12 14 16 18 20 2201

02

03

04

05

06

07

08

09

10

11

12

Gas

flow

from

the

stor

age

volu

me

[1017

ats

]

Pressure in compression volume [10-4 mbar]

FIG 14 Calibration curve for hydrogen used to deduce fromthe measured pressures the intensities of the hydrogen anddeuterium beam injected into the compression tube

00 05 10 15 20 2500

10

20

30

40

50

60

70

80 D = 20 mm D = 23 mm D = 25 mm

Bea

m in

tens

ity [1

016 a

ts]

Primary gas flow [mbar ls]

FIG 15 Intensity of the hydrogen beam (states |1〉 and |2〉)injected into the compression tube as function of the primarymolecular gas flow for different nozzle diameters D (nozzletemperature 60K dissociator power 300W)

includes small admixtures of H atoms in state |3〉 andmolecular hydrogen The first kind can be estimatedwith the use of the calculated transmissions (Sec II F)as 0017084 asymp 2 The amount of the second admix-ture has been measured as described below

For the deuterium beam (states |1〉 |2〉 and |3〉) theoptimization procedure gave an intensity of Iin(D) =(39 plusmn 02) middot 1016 particless achieved with q(D2) = 09mbar ls Tn = 65K and Pdiss = 300W slightly lowerthan those for hydrogen

12

50 60 70 80 90 100 110

40

50

60

70

80 D = 20 mm D = 23 mm D = 25 mm

Bea

m in

tens

ity [1

016 a

ts]

Nozzle temperature [K]

FIG 16 Intensity of the hydrogen beam (states |1〉 and |2〉)injected into the compression tube as function of the nozzletemperature for different nozzle diameters D (primary molec-ular gas flow 1mbar ls dissociator power 300W)

100 200 300 400 50010

20

30

40

50

60

70

80 D = 20 mm

Bea

m in

tens

ity [1

016 a

ts]

Dissociator power [W]

FIG 17 Intensity of the hydrogen beam (states |1〉 and |2〉)into the compression tube as function of the dissociator powerfor a nozzle diameter of 2 mm (nozzle temperature 60K pri-mary molecular gas flow 1mbar ls)

V HYDROGEN BEAM PROFILES

Beam profiles were measured at various positions atvarious positions behind the last sextupole magnet withthe use of

bull a compression tube of reduced dimensions (5mmdiameter)

bull a crossed-beam quadrupole mass spectrometer andbull a supplementary method of reduction of MoO3 by

hydrogen

A Measurements with the compression tube

For the determination of the beam dimensions at twopositions 300mm and 337mm behind the last magnetthe compression tube setup (Fig 12) was used makinguse of the possibility of axial movement by the z ma-nipulator and of that to install a narrower and shortercompression tube of 5mm diameter and 50mm lengthto enhance the spatial resolution The xy manipula-tor provided a lateral displacement of the compressiontube by plusmn10 mm in x and y direction The center co-ordinates of the geometrical axis of the source had beendetermined with the use of a bi-directional laser cen-tered inside the bore of the central support plate (seeFig 1) The relative intensity distributions in the xzand yz planes given by the measured pressure in thecompression volume are shown in Fig 18 Fits by Gaus-sian distributions to the data yield full widths at halfmaximum Γx = (642plusmn 009)mm Γy = (699plusmn 006)mmfor the distributions measured at z = 300mm and Γx =(627plusmn 008)mm Γy = (658plusmn 008)mm at 337mm

00

02

04

06

08

10

12

00

02

04

06

08

10

12

0 5 10 15 20 25

00

02

04

06

08

10

12

14

0 5 10 15 20 25

00

02

04

06

08

10

12

14

Pre

ssur

e in

the

com

pres

sion

tube

[au

]

Z = 300 mm

(a) (b)

(c) (d)

Z = 337 mm

X [mm]

Y [mm]

FIG 18 Cross sections of the beam profile in the mid-planemeasured with compression tube of 5mm diameter and 50mmlength Measurements in the xz-plane (a c) and yz-plane (bd) performed at two different positions z = 300mm (a b)and z = 337mm (c d) behind the last sextupole magnet ofthe ABS The shaded area represents position and dimensionsof the compression tube used in intensity measurements

The center of gravity of the measured profile definedas

rc =

sum

ij

radic

x2i + y2j middot P (xi yj)

sum

ij

P (xi yj) (8)

where xi and yj give the position of the compression-tubeaxis and P (xi yj) is the pressure measured in the com-

13

pression volume The resulting rc shows a deviation of012 mm from the geometrical axis of the source Further-more the data measured with the narrow compressiontube of 25mm radius can be used to derive the fractionof the beam entering the compression tube of 5mm ra-dius used in the intensity measurement of Sec IV Theratio

η =

rile25 mmsum

0P (xi yj)

rile10 mmsum

0P (xi yj)

(9)

where ri is the distance of the compression-tube axis tothe beam axis yields η asymp 07

B Measurements with the QMS

The beam-profile studies of Sec VA were extendedwith a setup utilizing a crossed-beam quadrupole massspectrometer (QMS) in the setup of Fig 19 Contraryto the measurements with the compression tube thosewith the QMS allow to separate the atomic and molec-ular fractions in the beam A 2mm diameter aperturewas installed at the entrance of the sensitive volume of

FIG 19 Setup for the measurements of the beam profilewith the QMS (1) xy-table enabling two-dimensional dis-placement of the entrance window of the QMS against thegeometrical axis of the ABS (2) the QMS (3) manually op-erated beam shutter The beam dump is an axially mountedcryo pump

the QMS to improve the resolution compared with thatachieved by the compression tube of 5mm diameter usedin measurements of the preceding section The layout ofthe setup presented in Fig 19 shows that in the presentcase the profile could not be measured at a distance ofz = 300mm to the last magnet Instead measurementswere performed at z = 567mm and with installation ofan extension tube at z = 697mm The xy manipulatorenabled displacements of the aperture axis from the geo-metrical axis of the source in any direction within limitsset by the bore diameter of the xy manipulatorThe first measured distribution of the atomic hydro-

gen (Fig 20) showed a distinct deviation from azimuthalsymmetry indicating an insufficient relative alignment ofnozzle and skimmer The three threaded rods support-ing the dissociator with the nozzle via the three-legged

FIG 20 Two-dimensional distribution of the atomic hydro-gen component of the beam at z = 567mm before the nozzle-to-skimmer adjustment showing a disinct deviation from az-imuthal symmetry

plate (label 2 in Fig 1) allow one to vary the position ofthe nozzle relative to that of the skimmer while the sourceis running This possibility has been used to find a nozzleposition which results in an azimuthally symmetric distri-bution The achieved symmetric distribution is shown inFig 21 and profiles of the atomic hadrogen component inthe beam measured in x and y direction at z = 567mmand z = 697mm are presented in Fig 22 Fits by Gaus-sian distributions to the data yield full widths at halfmaximum Γx = (736plusmn 043)mm Γy = (668plusmn 080)mmfor the distributions measured at z = 567mm and Γx =(669plusmn 022)mm Γy = (638plusmn 027)mm at 697mm

C Reduction of MoO3 by hydrogen

In addition to the compression tube and the QMS tech-nique a supplementary attempt was made to determinethe beam profile by exposing molybdenium trioxide (a

14

FIG 21 The distribution corresponding to that of Fig 20 af-ter nozzle-to-skimmer adjustment resulting in azimuthal sym-metry

0

20

40

60

80

0

20

40

60

80

2 4 6 8 10 12 14 16 18 20 22

0

20

40

60

80

6 8 10 12 14 16 18 20 22 24

0

20

40

60

80

QM

S si

gnal

s [a

u]

(a)

Z = 567 mm

Z = 697 mm

(c)

(b)

X [mm]

(d)

Y [mm]

FIG 22 Profiles of the atomic hydrogen component inthe beam measured with the QMS 567mm and 697mmbehind the last magnet

yellowish powder) on a glass plate to the beam The prin-ciple of this method is based on the reduction of MoO3 toa lower oxide of blue colour It first was used in the ex-periment to measure the magnetic moment of the hydro-gen atom by splitting of the beam in an inhomogeneousmagnetic field 61

This method is much simpler than the time-consumingmeasurements described in Secs VA and VB It givesqualitative results as presented in Fig 23 A quantitativeanalysis however requires development of the measuringtechnique (eg preparation of glass plates study of theoptimum exposure time digital image processing)

FIG 23 Photo of the glass plate covered with molybdenumtrioxide MoO3 exposed to the atomic hydrogen beam

D Summary of the profile measurements

Table III summarizes results of the measurements ofthe ABS beam profile with the compression-tube and theQMS setup The larger errors of the widths measuredwith the QMS are due to the lack of measurements withthe dissociator switched off and the necessity to estimatethe background signal from the existing data Within theerrors the measured widths do not show a dependenceon the distance from the last magnet This facilitatesto position the feeding tube of the storage cell in a widerange of a distances to the last magnet The average val-ues Γx = (638 plusmn 060)mm and Γy = (684 plusmn 033)mmagree within the errors and yield a common width ofΓxy = (673 plusmn 029)mm The two-dimensional Gaus-sian distribution of this width allows one to estimate thefraction η of the beam intensity injected into the com-pression tube or a feeding tube For a tube of 10mmdiameter η = 078plusmn 003 comparable with η asymp 07 givenin Sec VA

TABLE III Dimensions (FWHM) of the atomic hydrogenbeam measured with the compression tube (CT) and thecrossed-beam quadrupole mass spectrometer (QMS) at dis-tances z to the last magnet along perpendicular directions xand y

z[mm] Γx[mm] Γy[mm]

CT 300 642 plusmn 009 699plusmn 006

CT 337 627 plusmn 008 658plusmn 008

QMS 567 736 plusmn 043 668plusmn 080

QMS 697 669 plusmn 022 638plusmn 027

15

VI DEGREE OF DISSOCIATION

Besides the intensity of the atomic beam it is impor-tant to determine the molecular fraction in the beamMolecules injected into the feeding tube reduce the po-larization of the target gas

A Measurements with crossed-beam QMS

In addition to the data on the profile of the atomichydrogen beam (Sec VB) data on the distributions ofmolecular hydrogen in the beam were taken too at thepositions z = 567mm and 697mm behind the last mag-net The relation between the degree of dissociation andthe QMS signals by the atomic and molecular beam com-ponent was given above by Eq (6) The coefficient kv =vmva however is chosen here under the assumptionthat the average velocity of the atoms is determined bythe nozzle temperature of 65K and that of the moleculesby scattering and recombination on the ABS chamberwalls at 290K This yields kv =

radic

2 middot 65290 = 067 ingood agreement with Ref13 where this coefficient wasdetermined by the measured velocity distributions undersimilar conditionsThe measured profiles of the atomic fraction (identical

to those of Fig 21) those of the molecular fraction andthose of the degree of dissociation deduced from Eq (6)are collected in Fig 24As it is seen from the figure the distribution of the

degree of dissociation shows a dip around the central linedue to the higher density of molecular hydrogen originat-ing from the nozzle The mean value in an aperture of

0

20

40

60

80

100

120

2 4 6 8 10 12 14 16 18 20 22

0

20

40

60

80

100

120

6 8 10 12 14 16 18 20 22 24

QM

S si

gnal

s [a

u]

(a)

(b)

(c)

00

02

04

06

08

10

12

Deg

ree

of d

isso

ciat

ion

()

X [mm]

Y [mm]

(d)

00

02

04

06

08

10

12

Z = 567 mm

Z = 697 mm

FIG 24 Spatial distributions of H1 (bull)H2 () and degree ofdissociation () averaged over 3mm wide bands in the xz andyz planes respectively (here the z-axis is the geometrical axisof the ABS)

10mm diameter results as α = 095 plusmn 004

B Measurements with the Lamb-shift polarimeter

A cup in the quench chamber of the Lamb-shift po-larimeter (LSP) described in Ref8 allows one to mea-sure the currents Icup(H1) and Icup(H2) of the H+

1 andH+

2 ions extracted from the ionizer and separated bythe Wien filter with the cesium evaporation and the spinfilter switched off The relation between the degree ofdissociation α and the measured currents is

α =Icup(H1)minus r1

r2Icup(H2)

Icup(H1)minus r1r2Icup(H2) + 2kv

r2Icup(H2)

(10)

Among the three coefficients kv = 067 as for the mea-surement with the QMS For the electron energy of about100keV the ratio r1 of dissociative to non-dissociativeionization of H2 is8

r1 =σ(H2 rarr 2H+

1 )

σ(H2 rarr H+2 )

= 0095plusmn 0008 (11)

and the ratio between the ionization cross sections is857

r2 =σion(H2)

σion(H1)= 17plusmn 01 (12)

At the standard operation parameters of the source(Sec IV) the measured currents are Icup(H1) = (125 plusmn05) nA and Icup(H2) = (64 plusmn 01) nA yielding α =(096plusmn004) in excellent agreement with the value result-ing from the measurements with the QMS (Sec VIA)

VII BEAM POLARIZATION

The Lamb-shift polarimeter was designed built andtested at Universitat zu Koln7 It was used to measureand to optimize the polarization of the atomic hydrogenand deuterium beams delivered by the ABS Details arefound in Ref7The vector polarization pz for hydrogen is defined by

the relative hyperfine-state occupation numbers N(mI)

pz =N(+ 1

2 )minusN(minus 12 )

N(+ 12 ) +N(minus 1

2 ) (13)

for deuterium

pz =N(+1)minusN(minus1)

N(+1) +N(0) +N(minus1) (14)

Deuterium tensor polarization pzz is given by

pzz =N(+1) +N(minus1)minus 2 middotN(0)

N(+1) +N(0) +N(minus1) (15)

16

These polarizations can be derived from the measuredLyman-α peak strengths S by application of a number ofcorrection factors78

Typical Lyman α spectra measured with the polarizedhydrogen and deuterium beam from the ABS are shownin the Figs 25 and 26

MFT 2-3

200 300 400 5000

40

80

120

160

Magnetic field in the spinfilter [au]

PM

T si

gnal

[au

]

mI = -12

(a)

mI = +12

100 200 300 400 5000

40

80

120

160(b)

MFT 2-3WFT 1-3

mI = +12

mI = -12

FIG 25 Lyman-α spectra measured with the polarized hy-drogen beam (a) population change from state |2〉 to state|3〉 induced by the MFT unit (b) same as (a) with subsequentpopulation change from state |1〉 to state |3〉 induced by theWFT unit

MFT 3-4SFT 2-6

0

2

4

6

8

mI = -1m

I = 0

(a)

mI = +1

0

2

4

6

8(b)

MFT 3-4WFT 1-4 2-3

mI = +1 m

I = 0

mI = -1

mI = +1

mI = 0

mI = -1

mI = +1

mI = 0

mI = -1

0 1000 2000 3000

0

1

2

3

4 (c)

PMT

sign

al [a

u]

Magnetic field in the spinfilter [au]

WFT 1-4SFT 2-6

0 1000 2000 30000

2

4

6

8WFT 1-4SFT 3-5

(d)

FIG 26 Lyman-α spectra measured with the polarized deu-terium beam (a) and (b) vector polarization resulting fromsubsequent transitions MFT (3 rarr 4) and SFT (2 rarr 6) andWFT (1 rarr 4 2 rarr 3) respectively (c) and (d) tensor polar-ization resulting from subsequent transitions WFT (1 rarr 4)and SFT (2 rarr 6) and SFT (3 rarr 5) respectively

The polarization values for the hydrogen and the deu-terium beam derived from the Lyman-α peak-strengthratios with application of the necessary corrections aresummarized in Table IV

The vector polarization for hydrogen of the first linereflects the population of state |1〉 and state |2〉 accordingto the Eqs (3) and (13) The value of 091 deduced fromthe calculated transmission values is confirmed by themeasured one

TABLE IV The vector polarization pz of the hydrogen beamand the vector and the tensor polarization pzz of the deu-terium beam from the ABS measured with the Lamb-shiftpolarimeter

populated pz pzz

state(s)

Hydrogen |1〉 +089plusmn 001 -

|3〉 minus096plusmn 001 -

Deuterium |1〉+ |6〉 +088plusmn 001 +088plusmn 003

|3〉+ |4〉 minus091plusmn 001 +085plusmn 002

|3〉+ |6〉 +0005plusmn 0003 +090plusmn 001

|2〉+ |5〉 +0005plusmn 0003 minus171plusmn 003

VIII CONCLUSIONS AND OUTLOOK

In this paper we present the detailed description ofthe major components of the atomic beam source (ABS)for the polarized internal gas target of the magnet spec-trometer ANKE in COSY-Julich The ABS was builtfor the purpose of extending the physics program ofANKE from unpolarized and single-polarized investiga-tions with stored beams towards double-polarized exper-iments1 thus facilitating nuclear reaction studies involv-

ing ~p~p ~p~d ~d~p and ~d~d initial statesThe mechanical design took into account that at

ANKE the source has to be mounted vertically and trans-versely movable together with the transverse motion ofthe spectrometer magnet D2 The design of the system ofsextupole magnets took advantage of the developments inthe field of rare-earth permanent magnets (NdFeB) Ded-icated tools and methods were developed to determineand to optimize the source parameters ie intensity de-gree of dissociation and polarization Special emphasiswas put on the measurements of the spatial distributionsof the atomic and molecular beam near the focus wherethe feeding tube of the storage cell is located The ABShas been used in a number of investigations at ANKEthe commissioning effort to prepare the target for the usewith polarized H is described in Ref62 Performed stud-ies of the deuteron-charge exchange reaction are summa-rized in Ref6364 studies in near-threshold pion produc-tion are reported about in Ref65The ABS resides at the ANKE target position for a few

months per year only thus during the remaining timeit is used for other studies It had been observed thatthe nuclear polarization in recombined hydrogen is par-tially retained after recombination66 as well as evidencefor nuclear tensor polarization in recombined deuteriummolecules67 In order to investigate this recombinationprocess in more detail a special setup has been developedin the framework of an ISTC project68 and the recombi-nation process for different cell-wall coatings and differ-ent polarizations of the injected hydrogen or deuterium

17

atoms as function of cell-wall temperature strength ofthe magnetic holding field and gas pressure in the cell ispresently investigated69ndash71

Appendix A Preparation of Discharge Tubes and Nozzles

1 Tube Treatment

One end of the discharge tube is machined at a 45

angle while the other is kept flat Both ends are thenremelted and the tubes are tempered at 150 C Thetubes are further treated according to the procedure de-scribed in Ref72 which includes successive cleaning withacetone methanol distilled water and subsequent rins-ing by a 21 acid mixture of concentrated HF (40) andHCl (32) for 5 min The tubes are then flushed bydistilled water and dried

2 Nozzle Treatment

The nozzles are cleaned in an ultrasonic bath oftrichlorethylene acetone methanol and finally distilledwater all at 50 C Anodizing takes place in sulfuric acidto form a thin layer of Al2O3 as described in Ref72 Af-terwards they are immersed in distilled water for 30minat 95 C

ACKNOWLEDGMENTS

The authors want to thank OWB Schult Institut furKernphysik (IKP) Julich who initiated the polarizationprogram of ANKE Thanks go to the design office themechanical workshop and especially to WR Ermer allIKP Valuable advice was received from the PINTEX col-laboration at IUCF from the target group at HERMESespecially NKoch and from DToporkov BINP Novosi-birsk The support by VKoptev PNPI Gatchina whoregrettably passed away in January 2012 is gratefullyacknowledged Thanks go also to R Poprawe and col-leagues Fraunhofer-Institut fur Lasertechnik Aachenwhere the encapsulations of the magnets were laser-welded

1AKacharava F Rathmann and CWilkin Spin Physics fromCOSY to FAIR COSY Experiment Proposal No 152 (2005)Available under httparXivnucl-ex0511028

2WHaeberli in Proc 2nd Int Symp on Polarization Phenomenaof Nucleons Karlsruhe 1965 Eds P Huber and H SchopperExperientia Supplementum 12 64 (Birkhauser Verlag 1966)

3E Steffens and WHaeberli Rep Progr Phys 66 1887 (2003)4SBarsov et al Nucl Instr and Meth A 462 364 (2001)5RMaier Nucl Instr and Meth A 390 1 (1997)6KGrigoryev et al Proc 14th International Workshop on Po-larized Sources Targets and Polarimetry (PSTP 2011) 12-16 September 2011 StPetersburg Russia eds KGrigoryevPKravtsov and AVasilyev ISBN 978-5-86763-282-3 61 (2011)

7REngels et al Rev Sci Instrum 74 4607 (2003)8REngels et al Rev Sci Instrum 76 053305 (2005)9TWise et al Nucl Instr and Meth A 336 410 (1993)

10WADezarn et al Nucl Instr and Meth A 362 36 (1995)11T Rinckel et al Nucl Instr and Meth A 439 117 (2000)12F Stock et al Nucl Instr and Meth A 343 334 (1994)13ANass et al Nucl Instr and Meth A 505 633 (2003)14VDerenchuk et al Proc Conf Polarized Ion Sources and Polar-ized Gas Targets Madison WI 1993 Eds LWAnderson andWHaeberli AIP Conf Proc 293 72 (American Institute ofPhysics 1994)

15HOkamura et al see Ref14 p 8416KHatanaka et al Nucl Instr and Meth A 384 575 (1997)17Manufacturer Schiffer Metall- amp Vakuumtechnik 52428 JulichGermany

18Single-stage type RGS120 refrigerating capacity 120W at 80Kand 20W at 30K Leybold Vacuum GmbH 50968 Koln Ger-many

19Mini UHV gate valve series 010 VAT Germany GmbH 85630Grasbrunn Germany

20Type F3 fomblin oil Pfeiffer Vacuum GmbH 35614 Asslar Ger-many

21Model HU 1 Leybold Vacuum GmbH 50968 Koln Germany22Manufacturer SK Industriemodell GmbH 52531 Ubach-Palenberg Germany

23Type PFG 600 RF with automatic matchbox PFM 1500 A-INDHuttinger Elektronik GmbH 79110 Freiburg Germany

24WKorsch PhD Thesis Philipps Universitat Marburg (1990)25F Stock et al Int Workshop on Polarized Beams and PolarizedGas Targets Koeln Germany 1995 Eds HPaetz gen Schieckand L Sydow (World Scientific Publ Co 1996) p 260

26The first number denotes the outer diameter and the second onethe wall thickness

27Type Duran 8330 equivalent to Corning 7740 (Pyrex) SchottAG 55122 Mainz Germany

28Ultra-Kryomat RUL 80-D Lauda DrRWobser GmbH 97912Lauda-Konigshofen Germany

29ODU-Kontakt GmbH 84444 Muhldorf Germany30Handbook of Chemistry and Physics Ed RCEast (The Chem-ical Rubber Co 1973) p E-10

31NKoch and E Steffens Rev Sci Instrum 70 1631 (1999)32AVassiliev et al Petersburg Nuclear Physics Institute ReportNP-32-1997 No 2175 (1997)

33BLorentz Diploma Thesis Ruprecht-Karls-Universitat Heidel-berg (1993)

34ANass and E Steffens Nucl Instr and Meth A 598 653(2009)

35WHaeberli Ann Rev Nucl Sci 17 373 (1967)36AVassiliev et al Rev Sci Instr 71 3331 (2000)37WKubischta Proc Workshop on Polarized Gas Targetsfor Storage Rings Heidelberg 23-26 September 1991 EdsHGGaul E Steffens and K Zapfe (Max-Planck-Institut furKernphysik Heidelberg)

38The labeling of the hyperfine states as |1〉 = |mj = +12 mI =+12〉 |2〉 = | + 12minus12〉 |3〉 = | minus 12+12〉 and |4〉 =|minus12minus12〉 for hydrogen and |1〉 = |+12+1〉 |2〉 = |+12 0〉|3〉 = | + 12minus1〉 |4〉 = | minus 12minus1〉 |5〉 = | minus 12 0〉 and

6

performed with the use of a computer code originallydeveloped for the FILTEX ABS24 The boundary condi-tions by the layout of the target setup were the availabledistance of about 1250mm from the nozzle to the feeding-tube entrance of 10mm diameter and the distance fromthe exit of the last magnet to the feeding-tube entrance of300mm necessary to install the SFT and WFT units andthe gate valve between the ABS and the target chamberThe laboratory velocity distribution of the atoms in

the supersonic beam from the nozzle is described by amodified Maxwellian distribution

F(~vd Tb) =( m

2 k Tb

)32exp

[ minusm

2 k Tb(~v minus ~vd)

2

]

(1)

where m is the mass of the atoms and k is the Boltzmannconstant According to time-of-flight studies33 the driftvelocity along the beam axis vd and the beam temper-ature Tb for a primary molecular gas flow of 1 mbar lsand a nozzle-orifice diameter of 2mm follow a linear de-pendence on the nozzle temperature Tn For hydrogenvd[ms] = 1351 + 61 middot Tn[K] and Tb = 029 middot Tn and fordeuterium vd[ms] = 1070+345middotTn[K] and Tb = 025middotTnAs starting conditions of a track a random generator

selects a point in the nozzle orifice one within the di-aphragm in front of the first magnet and an atom ve-locity |v| In linear molecular flow approximation (cfthe discussion in Ref34) this defines ~v for the track be-tween the nozzle and the first magnet According to thegeometrical boundary conditions and the velocity distri-bution of Eq (1) the event is either rejected or used inthe further track calculation Within the magnet theevolution of the track is calculated stepwise by numeri-cal integration of the equation of motion over integrationtimes of 2micros corresponding to track lengths of 36mmfor a typical particle velocity of 1800ms The pureradial force acting on an atom within the field of the

sextupole magnet is ~Fr = minusmicroeff middot δBδr middot ~rr The ef-fective magnetic moment resulting from the Breit-Rabidiagram (eg Ref35) as microeff = δWδB is positive (neg-ative) for atoms in the hyperfine states with the electron

spin parallel (antiparallel) to ~B in the magnet aperturewhich therefore are deflected towards (away from) thebeam axis In the drift sections between the two magnetgroups and between the last magnet and the feeding tubethe trajectories are assumed as straight linesA variety of systems were studied all under the as-

sumption of Tn = 60K and pole-tip fields of 15T Asystem utilizing 6 magnets was found to yield satisfyingboth separation of the atoms in the microeff lt 0 and microeff gt 0states and focusing of the microeff gt 0 states into the feedingtube Optimization of the parameters led to the systemlisted in Table II (The tracking calculations yielding themagnet dimensions for the order to the manufacturer hadbeen performed for a slightly different geometry) Thetable gives the two distances at which intensity mea-surements with the compression tube were performedThe Fig 6 shows the projection of the trajectories of Hatoms in the microeff gt 0 states calculated for this system

TABLE II Final dimensions and axial positions of the sourcecomponents (pole-tip field strenghts Blowast

0 as measured afterdelivery36 inner diameters (empty0) outer diameters (empty1) axialdimensions (ℓ) and distances (∆) between the componentsThe lower three lines give the two distances and the dimen-sions of the compression tube used in the intensity measure-ments

component Blowast

0 [T] empty0 [mm] empty1 [mm] ℓ [mm] ∆ [mm]

Nozzle orifice 23 33150

Skimmer 44304a 130169

Diaphragm 9599a 2036

Magnet 1 1630 10401412a 3998 400194

Magnet 2 1689 15982212a 6404 650194

Magnet 3 1628 2804 9400 70014297

Magnet 4 1583 3004 9402 38011010

Magnet 5 1607 3006 9400 5501150

Magnet 6 1611 3002 9404 550030003370

Compr tube 100 110 1000

a Conical openingthe first number denotes the measureddiameter of the entrance the second that of the exit aperture

One recognizes two groups of trajectories one with anintermediate focus and another one with focusing intothe feeding tube The present result like those of othergroups (see eg Ref33) confirms the expectation37 thatthe transmission as function of the atom velocity shouldshow two maxima one below and one above the mostprobable velocityThe transmission Tr of the system is defined as the

fraction of tracks ending within the entrance of the feed-ing tube to those passing the diaphragm in front of thefirst sextupole magnet For the four hyperfine states ofhydrogen38 the calculations yield Tr(|1〉) sim Tr(|2〉) =042 (for both microeff gt 0) and Tr(|3〉) = 0001 andTr(|4〉) = 00004 (for both microeff lt 0)The performed tracking calculations do not account for

intra-beam and residual-gas scattering The calculatedtransmissions thus only allowed one to estimate upperlimits of the expected atomic beam intensity Iin into thefeeding tube For a primary molecular flow q(H2) theintensity Iin(H) with atoms mainly in the states |1〉 and|2〉 (microeff gt 0) was expected as

Iin(H) = q(H2) middot 2α middot Ω

2πmiddot 14

i=4sum

i=1

Tr(|i〉) (2)

For the degree of dissociation α a routine value of 08(see eg Ref9) was assumed Ω = 0022π is the solid

7

angle covered by the collimator aperture The factor 14reflects the assumption that the four substates in theatomic beam from the nozzle are equally populated Forq(H2) = 1mbar ls or 27middot1019H2 moleculess one expectsIin(H) sim 1 middot 1017H atomssAs described in the subsequent section the rf tran-

sition units are used to change the relative occupationnumbers of the states The trajectory code allows oneto simulate this change by assigning a microeff of one of thestates to the atoms before they pass a magnet As an ex-ample the medium-field transition unit (MFT) behindmagnet No 3 (see Fig 1) brings H atoms from state |2〉into state |3〉 This is simulated by assigning microeff(|2〉) gt 0to the atoms in the magnets 1minus3 and microeff(|3〉) lt 0 in themagnets 4 minus 6 where they get deflected from the beamaxis This results in a small value Tr(|2〉) = 0017 Fromthis value and the above value Tr(|1〉) = 042 the vectorpolarization is expected as

pz =Tr(|1〉)minus Tr(|2〉)Tr(|1〉) + Tr(|2〉) = 091 (3)

under the assumption of 100 efficiency of the transitionunitThe design and the properties of the permanent sex-

tupole magnets39 were discussed in an earlier paper36To achieve the pole-tip field values of sim15T each mag-net was produced from 24 segments employing three dif-

FIG 6 Projection of the 3-dimensional trajectories of hydro-gen atoms in hyperfine states |1〉 and |2〉 (effective magneticmoment microeff gt 0)) from the nozzle (empty = 2mm Tn = 60K)to the storage cell calculated for the magnet arrangement ofTable II and pole-tip fields of 15 T The positions and lat-eral dimensions of the six magnets and the feeding tube areindicated (in red)

ferent types of NdFeB compounds The expected pole-tip values (Table II) and the precise radial dependenceB(r) sim r2 within the magnet apertures were confirmedFor the first time the predicted high multipole compo-nents40 up to a 102-pole structure very near to the aper-ture surface could be measured36After the field measurements the magnets were encap-

sulated to prevent diffusion of hydrogen into the magnetmaterial which might deteriorate the magnetic proper-ties and to avoid the pumping of gas from the sinteredmagnet bodies The housings were made from thin stain-less steel cans of 02mm thickness for the conical andcylindrical walls within the magnet apertures and 03mmfor the front and end covers During the final welding toclose the housings with magnets installed the local tem-perature of the magnet material had to be kept belowthe Curie temperature of 60 C This was achieved bywelding with the use of a pulsed 15Hz NdYAG laserdelivering 11 J in a pulse of 2ms41 Overlapping weldspots of sim03mm diameter set around the adjacent cir-cular 02mm thick weld lips allowed one to finish thehousings with leak rates sim 10minus10mbar ls Inside thehousings the magnets were fixed to suppress axial androtational movements caused by the force of the adja-cent magnets Finally the free slits within the housingswere filled by sim20mbar krypton to enable leak tests bymass spectroscopy

G Radio Frequency Transition Units

The ABS is equipped with three types of transitionunits a weak field a medium field and a strong field rftransition unit (WFT MFT and SFT units) Togetherwith the selecting properties of the sextupole magnetsthey enable one to achieve all vector and tensor polar-izations of the atomic hydrogen and deuterium gas inthe storage cell In all three units transitions betweenthe hyperfine states split according to the Breit-Rabi di-agram by a static magnetic field (see eg Ref35) areinduced by the magnetic component (Brf) of an rf fieldleading to changes in the population of the states Thestatic field Bstat consists of two parallel components ahomogeneous field Bhom and a superimposed weaker gra-dient field Bgrad both orthogonal to the beam directionThe field gradient along the beam direction is requiredto satisfy the condition of adiabatic passage3542The assemblies of the WFT and the MFT units are

similar43 The layouts follow those of the units devel-oped for the HERMES experiment44 In both units therf field is produced by a coil with the axis along the beamdirection and consequently Brf orthogonal to Bstat TheMFT unit is shown in Fig 7 Figure 8 schematicallyshows one of the grooved aluminum frames with thewindings producing the gradient field A WFT unitis operated in a weak magnetic field Bstat le10G for hy-drogen and le5G for deuterium where the total atomicspin F is a good quantum number In hydrogen the

8

F = 1 levels |1〉 |2〉 and |3〉 with magnetic quantumnumbers mF = +1 0 and minus1 respectively can be re-garded as equally spaced In deuterium the same holdsfor the four F = 32 levels |1〉 |2〉 |3〉 and |4〉) withmF = +32 +12 minus12 and minus32 respectively andfor the two F = 12 levels |5〉 and |6〉 with mF = minus12and +12 respectively The magnetic component of therf dipole field induces transitions between each pair ofneighboring mF states with ∆mF = plusmn1 |∆mF| = 2transitions are forbidden The interchange of the popu-lation between the states |1〉 and |3〉 in hydrogen eg iscaused by a two-quantum transition via the intermedi-ate state |2〉 In the classical description of the adiabaticpassage method42 the population change should not de-pend on the sign of the magnetic field gradient relativeto the beam direction An exact quantum-mechanicaltreatment4546 however indicates that cleaner popula-tion changes from state |1〉 to |3〉 in hydrogen and fromstate |1〉 to |4〉 in deuterium are obtained with a nega-tive field gradient ie a Brf field decreasing in the beamdirection Deviations from adiabaticity are discussed inRef4547

The MFT unit is operated at higher values of Bstatwhere the differences in the energy spacings of pairs of hy-

FIG 7 Three-quarter-section view of the MFT unit with thesupport structure (1 self-supporting rf coil with spacers 2pick-up loop 3 Al tubes defining the length of the transition-inducing rf field 4 Cu cavity 5 coil around the pole shoe(6) providing the static field Bstat 7 slit between pole shoeand cavity wall housing the gradient-field coil 8 componentsof the static magnet yoke also serving as shielding againstexternal fields 9 cavity-positioning element 10 Cu padscooled by means of water-carrying tubes The cavity withthe rf coil and the pick-up loop can be taken out from thesurrounding components

z

Bstatic

transition

region

FIG 8 Arrangement of the windings producing the staticgradient field Bgrad shown in the left-hand side of the figureIn all transition units the field lies in the direction of the statichomogeneous field the field gradient dBdz lies in the beamdirection which defines the z axis In z direction the tran-sition reagion (indicated by the dashed lines) is confined tothe range of constant gradient by the Al tubes in orthogonaldirection by the beam diameter

perfine states with ∆mF = plusmn1 allow one to select singletransitions Originally developed for an polarized alkaliion source48 the MFT unit now is a standard compo-nent in polarized hydrogen and deuterium sources as dis-cussed eg in Ref49 Appropriate choice ofBhom Bgradand the rf frequency allows one to induce selected tran-sitions |1〉 harr |2〉 and |2〉 harr |3〉 in hydrogen or |1〉 harr |2〉|2〉 harr |3〉 and |3〉 harr |4〉 in deuterium Furthermore thechoice of the field gradient allows one to achieve consecu-tive transitions As an example a negative field gradientin the MFT unit behind the first set of magnets ie aB field decreasing in beam direction at a fixed rf fre-quency leads to the sequence of the transitions |3〉 rarr |4〉|2〉 rarr |3〉 and finally |1〉 rarr |2〉 in deuterium leaving thestate |1〉 empty

The SFT unit is used to induce transitions betweenstates in the upper and lower hyperfine multiplet in hy-drogen and deuterium Contrary to the historical nameindicating a strong magnetic field the SFT unit is op-erated with magnetic fields comparable to those used inthe MFT unit The transition frequencies are comparablewith those of the hyperfine splitting (1420MHz for hy-drogen and 327MHz for deuterium) and thus are muchhigher than those in the WFT and MFT units The rffield in a SFT unit is produced by a twin-line resonatorinside a Cu box tuned to the λ4 resonance50 The SFTunit51 is shown in Fig 9 Again the layout follows that ofthe unit used in the HERMES experiment44 Two vari-able capacitors at the free ends of the conducting rodsfed by the rf power with a relative phase shift of 180 allow one to tune the device

9

FIG 9 Three-quarter-section view of the rf cavity of theSFT unit for deuterium (1 the two resonant-field creatingconductors 2 the adjustible capacitor plates 3 Cu cavity)The inner dimensions of the cavity are 56mm along Bstat36mm orthogonal to it and 36mm height The cross sectionof the conductors is 14times 4mm2

H Slow Control System

Industrial components providing reliable and long-term support were selected for the control system of thewhole setup consisting of the ABS and the diagnosticstools the storage cell positioning system the Lamb-shiftpolarimeter and the supply system of a calibrated flowof unpolarized molecular gas The interlock system hasbeen implemented on the basis of SIEMENS SIMATICS7-300 family of programmable logic controllers In orderto unify the interfacing to the control computer all front-end equipment is connected via the PROFIBUS DP field-bus The process control software was implemented us-ing the Windows-based WinCC toolkit from SIEMENSThe system controls the operation of the pumps and thevalves It reads the pressure gauges and controls theregeneration cycles of the cryopumps Via a control net-work the temperature of the nozzle is stabilized withinplusmn05K Furthermore all power-supply units rf genera-tors and amplifiers are set and controlled The wholevariety of components to be controlled the logical struc-ture of the control and interlock system and a separatedevice for parameter studies are described in Ref52

III STUDIES OF THE FREE HYDROGEN JET

A Atomic beam profile near the nozzle

A novel device has been used to measure the profile ofan atomic beam via the deposition of recombination heaton thin wires in a two-dimensional grid5354 Atoms stuckon the surface of gold-plated tungsten wires of 5microm diam-eter recombine and are reemitted as molecules The re-combination heat (4476 eV per hydrogen molecule) leadsto a change of temperature and thus resistance alongeach wire The measurement of the resistance changes of

FIG 10 Two-dimensional profile of the atomic hydrogenbeam 10mm from the nozzle deduced from recombinationheating of gold-plated tungsten wires of 5microm in a 8times 8 wiregrid

all the wires in the grid allows one to deduce the centerand the profile of the beam Figure 10 shows the beamprofile resulting with a 8times8 wire grid positioned betweenskimmer and collimator performed as a first proof of themethod Later such a device has been used to comparemeasured and calculated beam profiles along the beamaxis between nozzle and skimmer34

B Degree of dissociation of the free atomic jet

The dissociation of the primary molecules is achievedby the interaction of the electrons and the hydrogen ordeuterium molecules in the plasma of the dissociatorThe degree of dissociation of the beam from the nozzledepends on the rf power applied to maintain the plasmathe primary molecular gas flow into the dissociator andthe temperature of the nozzle and the lower end of thedischarge tube These dependencies have been studiedbefore installation of the sextupole magnets with a setupcontaining a crossed-beam quadrupole mass spectrome-ter5556

α =ρa

ρa + 2 middot ρm (4)

The admixture of molecules in an atomic beam is de-scribed by the degree of dissociation where ρa and ρmare the densities of atomic and molecular hydrogen ordeuterium in the beam Other authors (eg Ref13) usethe atomic and molecular intensities Ia and Im in the def-inition of the degree of dissociation (αI) in Eq (4) Thetwo definitions of are related by

ImIa

=vmva

middot 1minus α

2α=

1minus αI

2αI (5)

10

This quantity was determined with the quadrupolemass spectrometer (QMS) in a conventional way as

α =Slowasta

Slowasta + 2 kv kion kdetSm

(6)

Here Slowasta = SaminusδSm denotes the atomic signal corrected

for dissociative ionization The parameter δ = 00141was obtained following the method described in Ref31The coefficient kv = vmva accounting for the difference

in atom and molecule velocity was chosen as 1radic2 under

the assumption of thermalization of the beam emergingfrom the nozzle Furthermore kion = 064 57 accounts forthe differences in ionization cross section for atomic andmolecular hydrogen and kdet = 084 for the detectionprobability55 As an example of the parameter stud-ies Fig 11 shows the deduced dependencies on the rfpower for a set of primary molecular hydrogen gas flowsFor typical flow values q(H2) le 10mbar ls a saturationvalue around 08 was obtained

0 50 100 150 200 250 300 350 400

00

02

04

06

08

10

Deg

ree

of d

isso

ciat

ion

()

Dissociator rf power [W]

q=07 mbar ls q=08 mbar ls q=10 mbar ls q=15 mbar ls

FIG 11 Degree of dissociation α of the free hydrogen jet asfunction of the applied rf power for different primary molec-ular hydrogen flows and a nozzle temperature of 70K

IV BEAM INTENSITY

The intensity of the polarized beam from the ABS to-gether with the layout of the storage cell determines theareal density of the target gas The intensity of the beamhas been measured with the use of a compression-tubesetup5859 shown in Fig 12 to optimize the ABS opera-tion parameters The measurements were performed at a300mm distance from the compression-tube entrance tothe last magnet and an inner tube diameter of 100mmas set in the tracking calculations The length of thecompression tube of 100mm was made equal to that ofthe foreseen feeding tube of the storage cell The narrowtube around the compression tube on a support based onthe lower flange separates the volume around the tube

FIG 12 Side view of the compression-tube setup made fromstandard ultra-high-vacuum components with a partial cutalong the axis (1 compression tube 2 support of thecompression tube based on the lower flange 3 narrow tubearound 1 closing the upper volume and allowing axial shiftsof the tube by the support 4 compression volume 5 hot-cathode pressure gauge 6 xy manipulator 7 z manipulator8 glass viewport 9 electromagnetic valve for gas inlet

from the compression volume The xy manipulator servesfor centering the tubes and for intensity-profile measure-ments The construction allows axial shifts of the com-pression tube by the z manipulator and the use of tubesof different diametersThe intensity of the beam entering the compression

volume through the compression tube is measured viathe pressure in the compression volume It is determinedby the equilibrium between the incoming beam intensityIin and the outgoing intensity Iout Under the assump-tion of a pure atomic beam and complete recombinationin the compression volume

Iin(atomss) = 2 middot Iout(moleculess)

= 2 middot∆P middot Ctube

= 2 middot∆P middot 103 middot 1020 middot d3

l

radic

T

M (7)

Here ∆P is the difference between the pressure measuredin the compreesion volume and that in the ABS cham-ber V The conductance of the compression tube Ctubeis determined by the inner diameter d of the tube itslength l the gas temperature T and the molar massM of the gas (given in cm and K respectively)60 The

11

CG

RVC200

H2

F

EV

EV

EV

EV

EV

MP

V1V2

EVR116

HC

V0

EV EVA016UDV136

HCCPLC

PS

220V AC

220V AC

Interface

24V DC

FIG 13 Source of calibrated molecular gas flow (V0 com-pression chamber V1 gas-storage chamber feeding V0 viathe needle valve UDV136a V2 chamber of calibrated volumeused to determine that of V1) The pressure in V1 is mea-sured by the capacitance gauge CG and is kept constant bythe dosing valve EVR116 with the gauge controller RVC200The whole setup including the evacuation elements can beoperated manually or by the programmable logic controllerPLC either within the ABS control system52 or as a separatesystem

a All the valves and the gauge controller are supplied by PfeifferVacuum GmbH Dndash35614 Asslar Germany (manufacturerBalzers AG Liechtenstein)

factor 2 takes into account that the same pressure is mea-sured in the hot-cathode gauge for 2 middotIin (H atomss) and1 middot Iin (H2 moleculess) For d = 10mm l = 100mmT = 290K and M = 2 for hydrogen pressure differences∆P on the order of 10minus4mbar are expected for atomichydrogen beam intensities in the order of 1017 atomssThe relation between Iin and ∆P for hydrogen has beendetermined experimentally with the use of a source ofcalibrated molecular hydrogen gas flow5859 depicted inFig 13 The measured dependence with a linear fit isshown in Fig 14 The calibration curve allows one to de-termine absolute values of Iin of hydrogen and deuteriumbeams The calibration for deuterium was deduced fromthe one for hydrogen by scaling with a factor 1

radic2 ac-

cording to Eq (7)

The dependences of Iin on the dissociator-operationparameters primary molecular hydrogen flow q(H2) noz-zle temperature Tn and dissociator power Pdiss have beenstudied to find the optimum values They are shown inthe Figs 15 16 and 17 respectively for different nozzle-orifice diameters The figures show that for the hydrogenbeam (states |1〉 and |2〉) with the standard operation pa-rameters qH2

= 11 mbar ls Tn = 70K Pdiss = 350Wand with a nozzle-orifice diameter of 23mm an inten-sity of Iin(H) = (75 plusmn 02) middot 1016 particless is achievedquite close to the earlier estimate from Eq (2) Besidesthe dominant atomic component of H atoms this value

04 06 08 10 12 14 16 18 20 2201

02

03

04

05

06

07

08

09

10

11

12

Gas

flow

from

the

stor

age

volu

me

[1017

ats

]

Pressure in compression volume [10-4 mbar]

FIG 14 Calibration curve for hydrogen used to deduce fromthe measured pressures the intensities of the hydrogen anddeuterium beam injected into the compression tube

00 05 10 15 20 2500

10

20

30

40

50

60

70

80 D = 20 mm D = 23 mm D = 25 mm

Bea

m in

tens

ity [1

016 a

ts]

Primary gas flow [mbar ls]

FIG 15 Intensity of the hydrogen beam (states |1〉 and |2〉)injected into the compression tube as function of the primarymolecular gas flow for different nozzle diameters D (nozzletemperature 60K dissociator power 300W)

includes small admixtures of H atoms in state |3〉 andmolecular hydrogen The first kind can be estimatedwith the use of the calculated transmissions (Sec II F)as 0017084 asymp 2 The amount of the second admix-ture has been measured as described below

For the deuterium beam (states |1〉 |2〉 and |3〉) theoptimization procedure gave an intensity of Iin(D) =(39 plusmn 02) middot 1016 particless achieved with q(D2) = 09mbar ls Tn = 65K and Pdiss = 300W slightly lowerthan those for hydrogen

12

50 60 70 80 90 100 110

40

50

60

70

80 D = 20 mm D = 23 mm D = 25 mm

Bea

m in

tens

ity [1

016 a

ts]

Nozzle temperature [K]

FIG 16 Intensity of the hydrogen beam (states |1〉 and |2〉)injected into the compression tube as function of the nozzletemperature for different nozzle diameters D (primary molec-ular gas flow 1mbar ls dissociator power 300W)

100 200 300 400 50010

20

30

40

50

60

70

80 D = 20 mm

Bea

m in

tens

ity [1

016 a

ts]

Dissociator power [W]

FIG 17 Intensity of the hydrogen beam (states |1〉 and |2〉)into the compression tube as function of the dissociator powerfor a nozzle diameter of 2 mm (nozzle temperature 60K pri-mary molecular gas flow 1mbar ls)

V HYDROGEN BEAM PROFILES

Beam profiles were measured at various positions atvarious positions behind the last sextupole magnet withthe use of

bull a compression tube of reduced dimensions (5mmdiameter)

bull a crossed-beam quadrupole mass spectrometer andbull a supplementary method of reduction of MoO3 by

hydrogen

A Measurements with the compression tube

For the determination of the beam dimensions at twopositions 300mm and 337mm behind the last magnetthe compression tube setup (Fig 12) was used makinguse of the possibility of axial movement by the z ma-nipulator and of that to install a narrower and shortercompression tube of 5mm diameter and 50mm lengthto enhance the spatial resolution The xy manipula-tor provided a lateral displacement of the compressiontube by plusmn10 mm in x and y direction The center co-ordinates of the geometrical axis of the source had beendetermined with the use of a bi-directional laser cen-tered inside the bore of the central support plate (seeFig 1) The relative intensity distributions in the xzand yz planes given by the measured pressure in thecompression volume are shown in Fig 18 Fits by Gaus-sian distributions to the data yield full widths at halfmaximum Γx = (642plusmn 009)mm Γy = (699plusmn 006)mmfor the distributions measured at z = 300mm and Γx =(627plusmn 008)mm Γy = (658plusmn 008)mm at 337mm

00

02

04

06

08

10

12

00

02

04

06

08

10

12

0 5 10 15 20 25

00

02

04

06

08

10

12

14

0 5 10 15 20 25

00

02

04

06

08

10

12

14

Pre

ssur

e in

the

com

pres

sion

tube

[au

]

Z = 300 mm

(a) (b)

(c) (d)

Z = 337 mm

X [mm]

Y [mm]

FIG 18 Cross sections of the beam profile in the mid-planemeasured with compression tube of 5mm diameter and 50mmlength Measurements in the xz-plane (a c) and yz-plane (bd) performed at two different positions z = 300mm (a b)and z = 337mm (c d) behind the last sextupole magnet ofthe ABS The shaded area represents position and dimensionsof the compression tube used in intensity measurements

The center of gravity of the measured profile definedas

rc =

sum

ij

radic

x2i + y2j middot P (xi yj)

sum

ij

P (xi yj) (8)

where xi and yj give the position of the compression-tubeaxis and P (xi yj) is the pressure measured in the com-

13

pression volume The resulting rc shows a deviation of012 mm from the geometrical axis of the source Further-more the data measured with the narrow compressiontube of 25mm radius can be used to derive the fractionof the beam entering the compression tube of 5mm ra-dius used in the intensity measurement of Sec IV Theratio

η =

rile25 mmsum

0P (xi yj)

rile10 mmsum

0P (xi yj)

(9)

where ri is the distance of the compression-tube axis tothe beam axis yields η asymp 07

B Measurements with the QMS

The beam-profile studies of Sec VA were extendedwith a setup utilizing a crossed-beam quadrupole massspectrometer (QMS) in the setup of Fig 19 Contraryto the measurements with the compression tube thosewith the QMS allow to separate the atomic and molec-ular fractions in the beam A 2mm diameter aperturewas installed at the entrance of the sensitive volume of

FIG 19 Setup for the measurements of the beam profilewith the QMS (1) xy-table enabling two-dimensional dis-placement of the entrance window of the QMS against thegeometrical axis of the ABS (2) the QMS (3) manually op-erated beam shutter The beam dump is an axially mountedcryo pump

the QMS to improve the resolution compared with thatachieved by the compression tube of 5mm diameter usedin measurements of the preceding section The layout ofthe setup presented in Fig 19 shows that in the presentcase the profile could not be measured at a distance ofz = 300mm to the last magnet Instead measurementswere performed at z = 567mm and with installation ofan extension tube at z = 697mm The xy manipulatorenabled displacements of the aperture axis from the geo-metrical axis of the source in any direction within limitsset by the bore diameter of the xy manipulatorThe first measured distribution of the atomic hydro-

gen (Fig 20) showed a distinct deviation from azimuthalsymmetry indicating an insufficient relative alignment ofnozzle and skimmer The three threaded rods support-ing the dissociator with the nozzle via the three-legged

FIG 20 Two-dimensional distribution of the atomic hydro-gen component of the beam at z = 567mm before the nozzle-to-skimmer adjustment showing a disinct deviation from az-imuthal symmetry

plate (label 2 in Fig 1) allow one to vary the position ofthe nozzle relative to that of the skimmer while the sourceis running This possibility has been used to find a nozzleposition which results in an azimuthally symmetric distri-bution The achieved symmetric distribution is shown inFig 21 and profiles of the atomic hadrogen component inthe beam measured in x and y direction at z = 567mmand z = 697mm are presented in Fig 22 Fits by Gaus-sian distributions to the data yield full widths at halfmaximum Γx = (736plusmn 043)mm Γy = (668plusmn 080)mmfor the distributions measured at z = 567mm and Γx =(669plusmn 022)mm Γy = (638plusmn 027)mm at 697mm

C Reduction of MoO3 by hydrogen

In addition to the compression tube and the QMS tech-nique a supplementary attempt was made to determinethe beam profile by exposing molybdenium trioxide (a

14

FIG 21 The distribution corresponding to that of Fig 20 af-ter nozzle-to-skimmer adjustment resulting in azimuthal sym-metry

0

20

40

60

80

0

20

40

60

80

2 4 6 8 10 12 14 16 18 20 22

0

20

40

60

80

6 8 10 12 14 16 18 20 22 24

0

20

40

60

80

QM

S si

gnal

s [a

u]

(a)

Z = 567 mm

Z = 697 mm

(c)

(b)

X [mm]

(d)

Y [mm]

FIG 22 Profiles of the atomic hydrogen component inthe beam measured with the QMS 567mm and 697mmbehind the last magnet

yellowish powder) on a glass plate to the beam The prin-ciple of this method is based on the reduction of MoO3 toa lower oxide of blue colour It first was used in the ex-periment to measure the magnetic moment of the hydro-gen atom by splitting of the beam in an inhomogeneousmagnetic field 61

This method is much simpler than the time-consumingmeasurements described in Secs VA and VB It givesqualitative results as presented in Fig 23 A quantitativeanalysis however requires development of the measuringtechnique (eg preparation of glass plates study of theoptimum exposure time digital image processing)

FIG 23 Photo of the glass plate covered with molybdenumtrioxide MoO3 exposed to the atomic hydrogen beam

D Summary of the profile measurements

Table III summarizes results of the measurements ofthe ABS beam profile with the compression-tube and theQMS setup The larger errors of the widths measuredwith the QMS are due to the lack of measurements withthe dissociator switched off and the necessity to estimatethe background signal from the existing data Within theerrors the measured widths do not show a dependenceon the distance from the last magnet This facilitatesto position the feeding tube of the storage cell in a widerange of a distances to the last magnet The average val-ues Γx = (638 plusmn 060)mm and Γy = (684 plusmn 033)mmagree within the errors and yield a common width ofΓxy = (673 plusmn 029)mm The two-dimensional Gaus-sian distribution of this width allows one to estimate thefraction η of the beam intensity injected into the com-pression tube or a feeding tube For a tube of 10mmdiameter η = 078plusmn 003 comparable with η asymp 07 givenin Sec VA

TABLE III Dimensions (FWHM) of the atomic hydrogenbeam measured with the compression tube (CT) and thecrossed-beam quadrupole mass spectrometer (QMS) at dis-tances z to the last magnet along perpendicular directions xand y

z[mm] Γx[mm] Γy[mm]

CT 300 642 plusmn 009 699plusmn 006

CT 337 627 plusmn 008 658plusmn 008

QMS 567 736 plusmn 043 668plusmn 080

QMS 697 669 plusmn 022 638plusmn 027

15

VI DEGREE OF DISSOCIATION

Besides the intensity of the atomic beam it is impor-tant to determine the molecular fraction in the beamMolecules injected into the feeding tube reduce the po-larization of the target gas

A Measurements with crossed-beam QMS

In addition to the data on the profile of the atomichydrogen beam (Sec VB) data on the distributions ofmolecular hydrogen in the beam were taken too at thepositions z = 567mm and 697mm behind the last mag-net The relation between the degree of dissociation andthe QMS signals by the atomic and molecular beam com-ponent was given above by Eq (6) The coefficient kv =vmva however is chosen here under the assumptionthat the average velocity of the atoms is determined bythe nozzle temperature of 65K and that of the moleculesby scattering and recombination on the ABS chamberwalls at 290K This yields kv =

radic

2 middot 65290 = 067 ingood agreement with Ref13 where this coefficient wasdetermined by the measured velocity distributions undersimilar conditionsThe measured profiles of the atomic fraction (identical

to those of Fig 21) those of the molecular fraction andthose of the degree of dissociation deduced from Eq (6)are collected in Fig 24As it is seen from the figure the distribution of the

degree of dissociation shows a dip around the central linedue to the higher density of molecular hydrogen originat-ing from the nozzle The mean value in an aperture of

0

20

40

60

80

100

120

2 4 6 8 10 12 14 16 18 20 22

0

20

40

60

80

100

120

6 8 10 12 14 16 18 20 22 24

QM

S si

gnal

s [a

u]

(a)

(b)

(c)

00

02

04

06

08

10

12

Deg

ree

of d

isso

ciat

ion

()

X [mm]

Y [mm]

(d)

00

02

04

06

08

10

12

Z = 567 mm

Z = 697 mm

FIG 24 Spatial distributions of H1 (bull)H2 () and degree ofdissociation () averaged over 3mm wide bands in the xz andyz planes respectively (here the z-axis is the geometrical axisof the ABS)

10mm diameter results as α = 095 plusmn 004

B Measurements with the Lamb-shift polarimeter

A cup in the quench chamber of the Lamb-shift po-larimeter (LSP) described in Ref8 allows one to mea-sure the currents Icup(H1) and Icup(H2) of the H+

1 andH+

2 ions extracted from the ionizer and separated bythe Wien filter with the cesium evaporation and the spinfilter switched off The relation between the degree ofdissociation α and the measured currents is

α =Icup(H1)minus r1

r2Icup(H2)

Icup(H1)minus r1r2Icup(H2) + 2kv

r2Icup(H2)

(10)

Among the three coefficients kv = 067 as for the mea-surement with the QMS For the electron energy of about100keV the ratio r1 of dissociative to non-dissociativeionization of H2 is8

r1 =σ(H2 rarr 2H+

1 )

σ(H2 rarr H+2 )

= 0095plusmn 0008 (11)

and the ratio between the ionization cross sections is857

r2 =σion(H2)

σion(H1)= 17plusmn 01 (12)

At the standard operation parameters of the source(Sec IV) the measured currents are Icup(H1) = (125 plusmn05) nA and Icup(H2) = (64 plusmn 01) nA yielding α =(096plusmn004) in excellent agreement with the value result-ing from the measurements with the QMS (Sec VIA)

VII BEAM POLARIZATION

The Lamb-shift polarimeter was designed built andtested at Universitat zu Koln7 It was used to measureand to optimize the polarization of the atomic hydrogenand deuterium beams delivered by the ABS Details arefound in Ref7The vector polarization pz for hydrogen is defined by

the relative hyperfine-state occupation numbers N(mI)

pz =N(+ 1

2 )minusN(minus 12 )

N(+ 12 ) +N(minus 1

2 ) (13)

for deuterium

pz =N(+1)minusN(minus1)

N(+1) +N(0) +N(minus1) (14)

Deuterium tensor polarization pzz is given by

pzz =N(+1) +N(minus1)minus 2 middotN(0)

N(+1) +N(0) +N(minus1) (15)

16

These polarizations can be derived from the measuredLyman-α peak strengths S by application of a number ofcorrection factors78

Typical Lyman α spectra measured with the polarizedhydrogen and deuterium beam from the ABS are shownin the Figs 25 and 26

MFT 2-3

200 300 400 5000

40

80

120

160

Magnetic field in the spinfilter [au]

PM

T si

gnal

[au

]

mI = -12

(a)

mI = +12

100 200 300 400 5000

40

80

120

160(b)

MFT 2-3WFT 1-3

mI = +12

mI = -12

FIG 25 Lyman-α spectra measured with the polarized hy-drogen beam (a) population change from state |2〉 to state|3〉 induced by the MFT unit (b) same as (a) with subsequentpopulation change from state |1〉 to state |3〉 induced by theWFT unit

MFT 3-4SFT 2-6

0

2

4

6

8

mI = -1m

I = 0

(a)

mI = +1

0

2

4

6

8(b)

MFT 3-4WFT 1-4 2-3

mI = +1 m

I = 0

mI = -1

mI = +1

mI = 0

mI = -1

mI = +1

mI = 0

mI = -1

0 1000 2000 3000

0

1

2

3

4 (c)

PMT

sign

al [a

u]

Magnetic field in the spinfilter [au]

WFT 1-4SFT 2-6

0 1000 2000 30000

2

4

6

8WFT 1-4SFT 3-5

(d)

FIG 26 Lyman-α spectra measured with the polarized deu-terium beam (a) and (b) vector polarization resulting fromsubsequent transitions MFT (3 rarr 4) and SFT (2 rarr 6) andWFT (1 rarr 4 2 rarr 3) respectively (c) and (d) tensor polar-ization resulting from subsequent transitions WFT (1 rarr 4)and SFT (2 rarr 6) and SFT (3 rarr 5) respectively

The polarization values for the hydrogen and the deu-terium beam derived from the Lyman-α peak-strengthratios with application of the necessary corrections aresummarized in Table IV

The vector polarization for hydrogen of the first linereflects the population of state |1〉 and state |2〉 accordingto the Eqs (3) and (13) The value of 091 deduced fromthe calculated transmission values is confirmed by themeasured one

TABLE IV The vector polarization pz of the hydrogen beamand the vector and the tensor polarization pzz of the deu-terium beam from the ABS measured with the Lamb-shiftpolarimeter

populated pz pzz

state(s)

Hydrogen |1〉 +089plusmn 001 -

|3〉 minus096plusmn 001 -

Deuterium |1〉+ |6〉 +088plusmn 001 +088plusmn 003

|3〉+ |4〉 minus091plusmn 001 +085plusmn 002

|3〉+ |6〉 +0005plusmn 0003 +090plusmn 001

|2〉+ |5〉 +0005plusmn 0003 minus171plusmn 003

VIII CONCLUSIONS AND OUTLOOK

In this paper we present the detailed description ofthe major components of the atomic beam source (ABS)for the polarized internal gas target of the magnet spec-trometer ANKE in COSY-Julich The ABS was builtfor the purpose of extending the physics program ofANKE from unpolarized and single-polarized investiga-tions with stored beams towards double-polarized exper-iments1 thus facilitating nuclear reaction studies involv-

ing ~p~p ~p~d ~d~p and ~d~d initial statesThe mechanical design took into account that at

ANKE the source has to be mounted vertically and trans-versely movable together with the transverse motion ofthe spectrometer magnet D2 The design of the system ofsextupole magnets took advantage of the developments inthe field of rare-earth permanent magnets (NdFeB) Ded-icated tools and methods were developed to determineand to optimize the source parameters ie intensity de-gree of dissociation and polarization Special emphasiswas put on the measurements of the spatial distributionsof the atomic and molecular beam near the focus wherethe feeding tube of the storage cell is located The ABShas been used in a number of investigations at ANKEthe commissioning effort to prepare the target for the usewith polarized H is described in Ref62 Performed stud-ies of the deuteron-charge exchange reaction are summa-rized in Ref6364 studies in near-threshold pion produc-tion are reported about in Ref65The ABS resides at the ANKE target position for a few

months per year only thus during the remaining timeit is used for other studies It had been observed thatthe nuclear polarization in recombined hydrogen is par-tially retained after recombination66 as well as evidencefor nuclear tensor polarization in recombined deuteriummolecules67 In order to investigate this recombinationprocess in more detail a special setup has been developedin the framework of an ISTC project68 and the recombi-nation process for different cell-wall coatings and differ-ent polarizations of the injected hydrogen or deuterium

17

atoms as function of cell-wall temperature strength ofthe magnetic holding field and gas pressure in the cell ispresently investigated69ndash71

Appendix A Preparation of Discharge Tubes and Nozzles

1 Tube Treatment

One end of the discharge tube is machined at a 45

angle while the other is kept flat Both ends are thenremelted and the tubes are tempered at 150 C Thetubes are further treated according to the procedure de-scribed in Ref72 which includes successive cleaning withacetone methanol distilled water and subsequent rins-ing by a 21 acid mixture of concentrated HF (40) andHCl (32) for 5 min The tubes are then flushed bydistilled water and dried

2 Nozzle Treatment

The nozzles are cleaned in an ultrasonic bath oftrichlorethylene acetone methanol and finally distilledwater all at 50 C Anodizing takes place in sulfuric acidto form a thin layer of Al2O3 as described in Ref72 Af-terwards they are immersed in distilled water for 30minat 95 C

ACKNOWLEDGMENTS

The authors want to thank OWB Schult Institut furKernphysik (IKP) Julich who initiated the polarizationprogram of ANKE Thanks go to the design office themechanical workshop and especially to WR Ermer allIKP Valuable advice was received from the PINTEX col-laboration at IUCF from the target group at HERMESespecially NKoch and from DToporkov BINP Novosi-birsk The support by VKoptev PNPI Gatchina whoregrettably passed away in January 2012 is gratefullyacknowledged Thanks go also to R Poprawe and col-leagues Fraunhofer-Institut fur Lasertechnik Aachenwhere the encapsulations of the magnets were laser-welded

1AKacharava F Rathmann and CWilkin Spin Physics fromCOSY to FAIR COSY Experiment Proposal No 152 (2005)Available under httparXivnucl-ex0511028

2WHaeberli in Proc 2nd Int Symp on Polarization Phenomenaof Nucleons Karlsruhe 1965 Eds P Huber and H SchopperExperientia Supplementum 12 64 (Birkhauser Verlag 1966)

3E Steffens and WHaeberli Rep Progr Phys 66 1887 (2003)4SBarsov et al Nucl Instr and Meth A 462 364 (2001)5RMaier Nucl Instr and Meth A 390 1 (1997)6KGrigoryev et al Proc 14th International Workshop on Po-larized Sources Targets and Polarimetry (PSTP 2011) 12-16 September 2011 StPetersburg Russia eds KGrigoryevPKravtsov and AVasilyev ISBN 978-5-86763-282-3 61 (2011)

7REngels et al Rev Sci Instrum 74 4607 (2003)8REngels et al Rev Sci Instrum 76 053305 (2005)9TWise et al Nucl Instr and Meth A 336 410 (1993)

10WADezarn et al Nucl Instr and Meth A 362 36 (1995)11T Rinckel et al Nucl Instr and Meth A 439 117 (2000)12F Stock et al Nucl Instr and Meth A 343 334 (1994)13ANass et al Nucl Instr and Meth A 505 633 (2003)14VDerenchuk et al Proc Conf Polarized Ion Sources and Polar-ized Gas Targets Madison WI 1993 Eds LWAnderson andWHaeberli AIP Conf Proc 293 72 (American Institute ofPhysics 1994)

15HOkamura et al see Ref14 p 8416KHatanaka et al Nucl Instr and Meth A 384 575 (1997)17Manufacturer Schiffer Metall- amp Vakuumtechnik 52428 JulichGermany

18Single-stage type RGS120 refrigerating capacity 120W at 80Kand 20W at 30K Leybold Vacuum GmbH 50968 Koln Ger-many

19Mini UHV gate valve series 010 VAT Germany GmbH 85630Grasbrunn Germany

20Type F3 fomblin oil Pfeiffer Vacuum GmbH 35614 Asslar Ger-many

21Model HU 1 Leybold Vacuum GmbH 50968 Koln Germany22Manufacturer SK Industriemodell GmbH 52531 Ubach-Palenberg Germany

23Type PFG 600 RF with automatic matchbox PFM 1500 A-INDHuttinger Elektronik GmbH 79110 Freiburg Germany

24WKorsch PhD Thesis Philipps Universitat Marburg (1990)25F Stock et al Int Workshop on Polarized Beams and PolarizedGas Targets Koeln Germany 1995 Eds HPaetz gen Schieckand L Sydow (World Scientific Publ Co 1996) p 260

26The first number denotes the outer diameter and the second onethe wall thickness

27Type Duran 8330 equivalent to Corning 7740 (Pyrex) SchottAG 55122 Mainz Germany

28Ultra-Kryomat RUL 80-D Lauda DrRWobser GmbH 97912Lauda-Konigshofen Germany

29ODU-Kontakt GmbH 84444 Muhldorf Germany30Handbook of Chemistry and Physics Ed RCEast (The Chem-ical Rubber Co 1973) p E-10

31NKoch and E Steffens Rev Sci Instrum 70 1631 (1999)32AVassiliev et al Petersburg Nuclear Physics Institute ReportNP-32-1997 No 2175 (1997)

33BLorentz Diploma Thesis Ruprecht-Karls-Universitat Heidel-berg (1993)

34ANass and E Steffens Nucl Instr and Meth A 598 653(2009)

35WHaeberli Ann Rev Nucl Sci 17 373 (1967)36AVassiliev et al Rev Sci Instr 71 3331 (2000)37WKubischta Proc Workshop on Polarized Gas Targetsfor Storage Rings Heidelberg 23-26 September 1991 EdsHGGaul E Steffens and K Zapfe (Max-Planck-Institut furKernphysik Heidelberg)

38The labeling of the hyperfine states as |1〉 = |mj = +12 mI =+12〉 |2〉 = | + 12minus12〉 |3〉 = | minus 12+12〉 and |4〉 =|minus12minus12〉 for hydrogen and |1〉 = |+12+1〉 |2〉 = |+12 0〉|3〉 = | + 12minus1〉 |4〉 = | minus 12minus1〉 |5〉 = | minus 12 0〉 and

7

angle covered by the collimator aperture The factor 14reflects the assumption that the four substates in theatomic beam from the nozzle are equally populated Forq(H2) = 1mbar ls or 27middot1019H2 moleculess one expectsIin(H) sim 1 middot 1017H atomssAs described in the subsequent section the rf tran-

sition units are used to change the relative occupationnumbers of the states The trajectory code allows oneto simulate this change by assigning a microeff of one of thestates to the atoms before they pass a magnet As an ex-ample the medium-field transition unit (MFT) behindmagnet No 3 (see Fig 1) brings H atoms from state |2〉into state |3〉 This is simulated by assigning microeff(|2〉) gt 0to the atoms in the magnets 1minus3 and microeff(|3〉) lt 0 in themagnets 4 minus 6 where they get deflected from the beamaxis This results in a small value Tr(|2〉) = 0017 Fromthis value and the above value Tr(|1〉) = 042 the vectorpolarization is expected as

pz =Tr(|1〉)minus Tr(|2〉)Tr(|1〉) + Tr(|2〉) = 091 (3)

under the assumption of 100 efficiency of the transitionunitThe design and the properties of the permanent sex-

tupole magnets39 were discussed in an earlier paper36To achieve the pole-tip field values of sim15T each mag-net was produced from 24 segments employing three dif-

FIG 6 Projection of the 3-dimensional trajectories of hydro-gen atoms in hyperfine states |1〉 and |2〉 (effective magneticmoment microeff gt 0)) from the nozzle (empty = 2mm Tn = 60K)to the storage cell calculated for the magnet arrangement ofTable II and pole-tip fields of 15 T The positions and lat-eral dimensions of the six magnets and the feeding tube areindicated (in red)

ferent types of NdFeB compounds The expected pole-tip values (Table II) and the precise radial dependenceB(r) sim r2 within the magnet apertures were confirmedFor the first time the predicted high multipole compo-nents40 up to a 102-pole structure very near to the aper-ture surface could be measured36After the field measurements the magnets were encap-

sulated to prevent diffusion of hydrogen into the magnetmaterial which might deteriorate the magnetic proper-ties and to avoid the pumping of gas from the sinteredmagnet bodies The housings were made from thin stain-less steel cans of 02mm thickness for the conical andcylindrical walls within the magnet apertures and 03mmfor the front and end covers During the final welding toclose the housings with magnets installed the local tem-perature of the magnet material had to be kept belowthe Curie temperature of 60 C This was achieved bywelding with the use of a pulsed 15Hz NdYAG laserdelivering 11 J in a pulse of 2ms41 Overlapping weldspots of sim03mm diameter set around the adjacent cir-cular 02mm thick weld lips allowed one to finish thehousings with leak rates sim 10minus10mbar ls Inside thehousings the magnets were fixed to suppress axial androtational movements caused by the force of the adja-cent magnets Finally the free slits within the housingswere filled by sim20mbar krypton to enable leak tests bymass spectroscopy

G Radio Frequency Transition Units

The ABS is equipped with three types of transitionunits a weak field a medium field and a strong field rftransition unit (WFT MFT and SFT units) Togetherwith the selecting properties of the sextupole magnetsthey enable one to achieve all vector and tensor polar-izations of the atomic hydrogen and deuterium gas inthe storage cell In all three units transitions betweenthe hyperfine states split according to the Breit-Rabi di-agram by a static magnetic field (see eg Ref35) areinduced by the magnetic component (Brf) of an rf fieldleading to changes in the population of the states Thestatic field Bstat consists of two parallel components ahomogeneous field Bhom and a superimposed weaker gra-dient field Bgrad both orthogonal to the beam directionThe field gradient along the beam direction is requiredto satisfy the condition of adiabatic passage3542The assemblies of the WFT and the MFT units are

similar43 The layouts follow those of the units devel-oped for the HERMES experiment44 In both units therf field is produced by a coil with the axis along the beamdirection and consequently Brf orthogonal to Bstat TheMFT unit is shown in Fig 7 Figure 8 schematicallyshows one of the grooved aluminum frames with thewindings producing the gradient field A WFT unitis operated in a weak magnetic field Bstat le10G for hy-drogen and le5G for deuterium where the total atomicspin F is a good quantum number In hydrogen the

8

F = 1 levels |1〉 |2〉 and |3〉 with magnetic quantumnumbers mF = +1 0 and minus1 respectively can be re-garded as equally spaced In deuterium the same holdsfor the four F = 32 levels |1〉 |2〉 |3〉 and |4〉) withmF = +32 +12 minus12 and minus32 respectively andfor the two F = 12 levels |5〉 and |6〉 with mF = minus12and +12 respectively The magnetic component of therf dipole field induces transitions between each pair ofneighboring mF states with ∆mF = plusmn1 |∆mF| = 2transitions are forbidden The interchange of the popu-lation between the states |1〉 and |3〉 in hydrogen eg iscaused by a two-quantum transition via the intermedi-ate state |2〉 In the classical description of the adiabaticpassage method42 the population change should not de-pend on the sign of the magnetic field gradient relativeto the beam direction An exact quantum-mechanicaltreatment4546 however indicates that cleaner popula-tion changes from state |1〉 to |3〉 in hydrogen and fromstate |1〉 to |4〉 in deuterium are obtained with a nega-tive field gradient ie a Brf field decreasing in the beamdirection Deviations from adiabaticity are discussed inRef4547

The MFT unit is operated at higher values of Bstatwhere the differences in the energy spacings of pairs of hy-

FIG 7 Three-quarter-section view of the MFT unit with thesupport structure (1 self-supporting rf coil with spacers 2pick-up loop 3 Al tubes defining the length of the transition-inducing rf field 4 Cu cavity 5 coil around the pole shoe(6) providing the static field Bstat 7 slit between pole shoeand cavity wall housing the gradient-field coil 8 componentsof the static magnet yoke also serving as shielding againstexternal fields 9 cavity-positioning element 10 Cu padscooled by means of water-carrying tubes The cavity withthe rf coil and the pick-up loop can be taken out from thesurrounding components

z

Bstatic

transition

region

FIG 8 Arrangement of the windings producing the staticgradient field Bgrad shown in the left-hand side of the figureIn all transition units the field lies in the direction of the statichomogeneous field the field gradient dBdz lies in the beamdirection which defines the z axis In z direction the tran-sition reagion (indicated by the dashed lines) is confined tothe range of constant gradient by the Al tubes in orthogonaldirection by the beam diameter

perfine states with ∆mF = plusmn1 allow one to select singletransitions Originally developed for an polarized alkaliion source48 the MFT unit now is a standard compo-nent in polarized hydrogen and deuterium sources as dis-cussed eg in Ref49 Appropriate choice ofBhom Bgradand the rf frequency allows one to induce selected tran-sitions |1〉 harr |2〉 and |2〉 harr |3〉 in hydrogen or |1〉 harr |2〉|2〉 harr |3〉 and |3〉 harr |4〉 in deuterium Furthermore thechoice of the field gradient allows one to achieve consecu-tive transitions As an example a negative field gradientin the MFT unit behind the first set of magnets ie aB field decreasing in beam direction at a fixed rf fre-quency leads to the sequence of the transitions |3〉 rarr |4〉|2〉 rarr |3〉 and finally |1〉 rarr |2〉 in deuterium leaving thestate |1〉 empty

The SFT unit is used to induce transitions betweenstates in the upper and lower hyperfine multiplet in hy-drogen and deuterium Contrary to the historical nameindicating a strong magnetic field the SFT unit is op-erated with magnetic fields comparable to those used inthe MFT unit The transition frequencies are comparablewith those of the hyperfine splitting (1420MHz for hy-drogen and 327MHz for deuterium) and thus are muchhigher than those in the WFT and MFT units The rffield in a SFT unit is produced by a twin-line resonatorinside a Cu box tuned to the λ4 resonance50 The SFTunit51 is shown in Fig 9 Again the layout follows that ofthe unit used in the HERMES experiment44 Two vari-able capacitors at the free ends of the conducting rodsfed by the rf power with a relative phase shift of 180 allow one to tune the device

9

FIG 9 Three-quarter-section view of the rf cavity of theSFT unit for deuterium (1 the two resonant-field creatingconductors 2 the adjustible capacitor plates 3 Cu cavity)The inner dimensions of the cavity are 56mm along Bstat36mm orthogonal to it and 36mm height The cross sectionof the conductors is 14times 4mm2

H Slow Control System

Industrial components providing reliable and long-term support were selected for the control system of thewhole setup consisting of the ABS and the diagnosticstools the storage cell positioning system the Lamb-shiftpolarimeter and the supply system of a calibrated flowof unpolarized molecular gas The interlock system hasbeen implemented on the basis of SIEMENS SIMATICS7-300 family of programmable logic controllers In orderto unify the interfacing to the control computer all front-end equipment is connected via the PROFIBUS DP field-bus The process control software was implemented us-ing the Windows-based WinCC toolkit from SIEMENSThe system controls the operation of the pumps and thevalves It reads the pressure gauges and controls theregeneration cycles of the cryopumps Via a control net-work the temperature of the nozzle is stabilized withinplusmn05K Furthermore all power-supply units rf genera-tors and amplifiers are set and controlled The wholevariety of components to be controlled the logical struc-ture of the control and interlock system and a separatedevice for parameter studies are described in Ref52

III STUDIES OF THE FREE HYDROGEN JET

A Atomic beam profile near the nozzle

A novel device has been used to measure the profile ofan atomic beam via the deposition of recombination heaton thin wires in a two-dimensional grid5354 Atoms stuckon the surface of gold-plated tungsten wires of 5microm diam-eter recombine and are reemitted as molecules The re-combination heat (4476 eV per hydrogen molecule) leadsto a change of temperature and thus resistance alongeach wire The measurement of the resistance changes of

FIG 10 Two-dimensional profile of the atomic hydrogenbeam 10mm from the nozzle deduced from recombinationheating of gold-plated tungsten wires of 5microm in a 8times 8 wiregrid

all the wires in the grid allows one to deduce the centerand the profile of the beam Figure 10 shows the beamprofile resulting with a 8times8 wire grid positioned betweenskimmer and collimator performed as a first proof of themethod Later such a device has been used to comparemeasured and calculated beam profiles along the beamaxis between nozzle and skimmer34

B Degree of dissociation of the free atomic jet

The dissociation of the primary molecules is achievedby the interaction of the electrons and the hydrogen ordeuterium molecules in the plasma of the dissociatorThe degree of dissociation of the beam from the nozzledepends on the rf power applied to maintain the plasmathe primary molecular gas flow into the dissociator andthe temperature of the nozzle and the lower end of thedischarge tube These dependencies have been studiedbefore installation of the sextupole magnets with a setupcontaining a crossed-beam quadrupole mass spectrome-ter5556

α =ρa

ρa + 2 middot ρm (4)

The admixture of molecules in an atomic beam is de-scribed by the degree of dissociation where ρa and ρmare the densities of atomic and molecular hydrogen ordeuterium in the beam Other authors (eg Ref13) usethe atomic and molecular intensities Ia and Im in the def-inition of the degree of dissociation (αI) in Eq (4) Thetwo definitions of are related by

ImIa

=vmva

middot 1minus α

2α=

1minus αI

2αI (5)

10

This quantity was determined with the quadrupolemass spectrometer (QMS) in a conventional way as

α =Slowasta

Slowasta + 2 kv kion kdetSm

(6)

Here Slowasta = SaminusδSm denotes the atomic signal corrected

for dissociative ionization The parameter δ = 00141was obtained following the method described in Ref31The coefficient kv = vmva accounting for the difference

in atom and molecule velocity was chosen as 1radic2 under

the assumption of thermalization of the beam emergingfrom the nozzle Furthermore kion = 064 57 accounts forthe differences in ionization cross section for atomic andmolecular hydrogen and kdet = 084 for the detectionprobability55 As an example of the parameter stud-ies Fig 11 shows the deduced dependencies on the rfpower for a set of primary molecular hydrogen gas flowsFor typical flow values q(H2) le 10mbar ls a saturationvalue around 08 was obtained

0 50 100 150 200 250 300 350 400

00

02

04

06

08

10

Deg

ree

of d

isso

ciat

ion

()

Dissociator rf power [W]

q=07 mbar ls q=08 mbar ls q=10 mbar ls q=15 mbar ls

FIG 11 Degree of dissociation α of the free hydrogen jet asfunction of the applied rf power for different primary molec-ular hydrogen flows and a nozzle temperature of 70K

IV BEAM INTENSITY

The intensity of the polarized beam from the ABS to-gether with the layout of the storage cell determines theareal density of the target gas The intensity of the beamhas been measured with the use of a compression-tubesetup5859 shown in Fig 12 to optimize the ABS opera-tion parameters The measurements were performed at a300mm distance from the compression-tube entrance tothe last magnet and an inner tube diameter of 100mmas set in the tracking calculations The length of thecompression tube of 100mm was made equal to that ofthe foreseen feeding tube of the storage cell The narrowtube around the compression tube on a support based onthe lower flange separates the volume around the tube

FIG 12 Side view of the compression-tube setup made fromstandard ultra-high-vacuum components with a partial cutalong the axis (1 compression tube 2 support of thecompression tube based on the lower flange 3 narrow tubearound 1 closing the upper volume and allowing axial shiftsof the tube by the support 4 compression volume 5 hot-cathode pressure gauge 6 xy manipulator 7 z manipulator8 glass viewport 9 electromagnetic valve for gas inlet

from the compression volume The xy manipulator servesfor centering the tubes and for intensity-profile measure-ments The construction allows axial shifts of the com-pression tube by the z manipulator and the use of tubesof different diametersThe intensity of the beam entering the compression

volume through the compression tube is measured viathe pressure in the compression volume It is determinedby the equilibrium between the incoming beam intensityIin and the outgoing intensity Iout Under the assump-tion of a pure atomic beam and complete recombinationin the compression volume

Iin(atomss) = 2 middot Iout(moleculess)

= 2 middot∆P middot Ctube

= 2 middot∆P middot 103 middot 1020 middot d3

l

radic

T

M (7)

Here ∆P is the difference between the pressure measuredin the compreesion volume and that in the ABS cham-ber V The conductance of the compression tube Ctubeis determined by the inner diameter d of the tube itslength l the gas temperature T and the molar massM of the gas (given in cm and K respectively)60 The

11

CG

RVC200

H2

F

EV

EV

EV

EV

EV

MP

V1V2

EVR116

HC

V0

EV EVA016UDV136

HCCPLC

PS

220V AC

220V AC

Interface

24V DC

FIG 13 Source of calibrated molecular gas flow (V0 com-pression chamber V1 gas-storage chamber feeding V0 viathe needle valve UDV136a V2 chamber of calibrated volumeused to determine that of V1) The pressure in V1 is mea-sured by the capacitance gauge CG and is kept constant bythe dosing valve EVR116 with the gauge controller RVC200The whole setup including the evacuation elements can beoperated manually or by the programmable logic controllerPLC either within the ABS control system52 or as a separatesystem

a All the valves and the gauge controller are supplied by PfeifferVacuum GmbH Dndash35614 Asslar Germany (manufacturerBalzers AG Liechtenstein)

factor 2 takes into account that the same pressure is mea-sured in the hot-cathode gauge for 2 middotIin (H atomss) and1 middot Iin (H2 moleculess) For d = 10mm l = 100mmT = 290K and M = 2 for hydrogen pressure differences∆P on the order of 10minus4mbar are expected for atomichydrogen beam intensities in the order of 1017 atomssThe relation between Iin and ∆P for hydrogen has beendetermined experimentally with the use of a source ofcalibrated molecular hydrogen gas flow5859 depicted inFig 13 The measured dependence with a linear fit isshown in Fig 14 The calibration curve allows one to de-termine absolute values of Iin of hydrogen and deuteriumbeams The calibration for deuterium was deduced fromthe one for hydrogen by scaling with a factor 1

radic2 ac-

cording to Eq (7)

The dependences of Iin on the dissociator-operationparameters primary molecular hydrogen flow q(H2) noz-zle temperature Tn and dissociator power Pdiss have beenstudied to find the optimum values They are shown inthe Figs 15 16 and 17 respectively for different nozzle-orifice diameters The figures show that for the hydrogenbeam (states |1〉 and |2〉) with the standard operation pa-rameters qH2

= 11 mbar ls Tn = 70K Pdiss = 350Wand with a nozzle-orifice diameter of 23mm an inten-sity of Iin(H) = (75 plusmn 02) middot 1016 particless is achievedquite close to the earlier estimate from Eq (2) Besidesthe dominant atomic component of H atoms this value

04 06 08 10 12 14 16 18 20 2201

02

03

04

05

06

07

08

09

10

11

12

Gas

flow

from

the

stor

age

volu

me

[1017

ats

]

Pressure in compression volume [10-4 mbar]

FIG 14 Calibration curve for hydrogen used to deduce fromthe measured pressures the intensities of the hydrogen anddeuterium beam injected into the compression tube

00 05 10 15 20 2500

10

20

30

40

50

60

70

80 D = 20 mm D = 23 mm D = 25 mm

Bea

m in

tens

ity [1

016 a

ts]

Primary gas flow [mbar ls]

FIG 15 Intensity of the hydrogen beam (states |1〉 and |2〉)injected into the compression tube as function of the primarymolecular gas flow for different nozzle diameters D (nozzletemperature 60K dissociator power 300W)

includes small admixtures of H atoms in state |3〉 andmolecular hydrogen The first kind can be estimatedwith the use of the calculated transmissions (Sec II F)as 0017084 asymp 2 The amount of the second admix-ture has been measured as described below

For the deuterium beam (states |1〉 |2〉 and |3〉) theoptimization procedure gave an intensity of Iin(D) =(39 plusmn 02) middot 1016 particless achieved with q(D2) = 09mbar ls Tn = 65K and Pdiss = 300W slightly lowerthan those for hydrogen

12

50 60 70 80 90 100 110

40

50

60

70

80 D = 20 mm D = 23 mm D = 25 mm

Bea

m in

tens

ity [1

016 a

ts]

Nozzle temperature [K]

FIG 16 Intensity of the hydrogen beam (states |1〉 and |2〉)injected into the compression tube as function of the nozzletemperature for different nozzle diameters D (primary molec-ular gas flow 1mbar ls dissociator power 300W)

100 200 300 400 50010

20

30

40

50

60

70

80 D = 20 mm

Bea

m in

tens

ity [1

016 a

ts]

Dissociator power [W]

FIG 17 Intensity of the hydrogen beam (states |1〉 and |2〉)into the compression tube as function of the dissociator powerfor a nozzle diameter of 2 mm (nozzle temperature 60K pri-mary molecular gas flow 1mbar ls)

V HYDROGEN BEAM PROFILES

Beam profiles were measured at various positions atvarious positions behind the last sextupole magnet withthe use of

bull a compression tube of reduced dimensions (5mmdiameter)

bull a crossed-beam quadrupole mass spectrometer andbull a supplementary method of reduction of MoO3 by

hydrogen

A Measurements with the compression tube

For the determination of the beam dimensions at twopositions 300mm and 337mm behind the last magnetthe compression tube setup (Fig 12) was used makinguse of the possibility of axial movement by the z ma-nipulator and of that to install a narrower and shortercompression tube of 5mm diameter and 50mm lengthto enhance the spatial resolution The xy manipula-tor provided a lateral displacement of the compressiontube by plusmn10 mm in x and y direction The center co-ordinates of the geometrical axis of the source had beendetermined with the use of a bi-directional laser cen-tered inside the bore of the central support plate (seeFig 1) The relative intensity distributions in the xzand yz planes given by the measured pressure in thecompression volume are shown in Fig 18 Fits by Gaus-sian distributions to the data yield full widths at halfmaximum Γx = (642plusmn 009)mm Γy = (699plusmn 006)mmfor the distributions measured at z = 300mm and Γx =(627plusmn 008)mm Γy = (658plusmn 008)mm at 337mm

00

02

04

06

08

10

12

00

02

04

06

08

10

12

0 5 10 15 20 25

00

02

04

06

08

10

12

14

0 5 10 15 20 25

00

02

04

06

08

10

12

14

Pre

ssur

e in

the

com

pres

sion

tube

[au

]

Z = 300 mm

(a) (b)

(c) (d)

Z = 337 mm

X [mm]

Y [mm]

FIG 18 Cross sections of the beam profile in the mid-planemeasured with compression tube of 5mm diameter and 50mmlength Measurements in the xz-plane (a c) and yz-plane (bd) performed at two different positions z = 300mm (a b)and z = 337mm (c d) behind the last sextupole magnet ofthe ABS The shaded area represents position and dimensionsof the compression tube used in intensity measurements

The center of gravity of the measured profile definedas

rc =

sum

ij

radic

x2i + y2j middot P (xi yj)

sum

ij

P (xi yj) (8)

where xi and yj give the position of the compression-tubeaxis and P (xi yj) is the pressure measured in the com-

13

pression volume The resulting rc shows a deviation of012 mm from the geometrical axis of the source Further-more the data measured with the narrow compressiontube of 25mm radius can be used to derive the fractionof the beam entering the compression tube of 5mm ra-dius used in the intensity measurement of Sec IV Theratio

η =

rile25 mmsum

0P (xi yj)

rile10 mmsum

0P (xi yj)

(9)

where ri is the distance of the compression-tube axis tothe beam axis yields η asymp 07

B Measurements with the QMS

The beam-profile studies of Sec VA were extendedwith a setup utilizing a crossed-beam quadrupole massspectrometer (QMS) in the setup of Fig 19 Contraryto the measurements with the compression tube thosewith the QMS allow to separate the atomic and molec-ular fractions in the beam A 2mm diameter aperturewas installed at the entrance of the sensitive volume of

FIG 19 Setup for the measurements of the beam profilewith the QMS (1) xy-table enabling two-dimensional dis-placement of the entrance window of the QMS against thegeometrical axis of the ABS (2) the QMS (3) manually op-erated beam shutter The beam dump is an axially mountedcryo pump

the QMS to improve the resolution compared with thatachieved by the compression tube of 5mm diameter usedin measurements of the preceding section The layout ofthe setup presented in Fig 19 shows that in the presentcase the profile could not be measured at a distance ofz = 300mm to the last magnet Instead measurementswere performed at z = 567mm and with installation ofan extension tube at z = 697mm The xy manipulatorenabled displacements of the aperture axis from the geo-metrical axis of the source in any direction within limitsset by the bore diameter of the xy manipulatorThe first measured distribution of the atomic hydro-

gen (Fig 20) showed a distinct deviation from azimuthalsymmetry indicating an insufficient relative alignment ofnozzle and skimmer The three threaded rods support-ing the dissociator with the nozzle via the three-legged

FIG 20 Two-dimensional distribution of the atomic hydro-gen component of the beam at z = 567mm before the nozzle-to-skimmer adjustment showing a disinct deviation from az-imuthal symmetry

plate (label 2 in Fig 1) allow one to vary the position ofthe nozzle relative to that of the skimmer while the sourceis running This possibility has been used to find a nozzleposition which results in an azimuthally symmetric distri-bution The achieved symmetric distribution is shown inFig 21 and profiles of the atomic hadrogen component inthe beam measured in x and y direction at z = 567mmand z = 697mm are presented in Fig 22 Fits by Gaus-sian distributions to the data yield full widths at halfmaximum Γx = (736plusmn 043)mm Γy = (668plusmn 080)mmfor the distributions measured at z = 567mm and Γx =(669plusmn 022)mm Γy = (638plusmn 027)mm at 697mm

C Reduction of MoO3 by hydrogen

In addition to the compression tube and the QMS tech-nique a supplementary attempt was made to determinethe beam profile by exposing molybdenium trioxide (a

14

FIG 21 The distribution corresponding to that of Fig 20 af-ter nozzle-to-skimmer adjustment resulting in azimuthal sym-metry

0

20

40

60

80

0

20

40

60

80

2 4 6 8 10 12 14 16 18 20 22

0

20

40

60

80

6 8 10 12 14 16 18 20 22 24

0

20

40

60

80

QM

S si

gnal

s [a

u]

(a)

Z = 567 mm

Z = 697 mm

(c)

(b)

X [mm]

(d)

Y [mm]

FIG 22 Profiles of the atomic hydrogen component inthe beam measured with the QMS 567mm and 697mmbehind the last magnet

yellowish powder) on a glass plate to the beam The prin-ciple of this method is based on the reduction of MoO3 toa lower oxide of blue colour It first was used in the ex-periment to measure the magnetic moment of the hydro-gen atom by splitting of the beam in an inhomogeneousmagnetic field 61

This method is much simpler than the time-consumingmeasurements described in Secs VA and VB It givesqualitative results as presented in Fig 23 A quantitativeanalysis however requires development of the measuringtechnique (eg preparation of glass plates study of theoptimum exposure time digital image processing)

FIG 23 Photo of the glass plate covered with molybdenumtrioxide MoO3 exposed to the atomic hydrogen beam

D Summary of the profile measurements

Table III summarizes results of the measurements ofthe ABS beam profile with the compression-tube and theQMS setup The larger errors of the widths measuredwith the QMS are due to the lack of measurements withthe dissociator switched off and the necessity to estimatethe background signal from the existing data Within theerrors the measured widths do not show a dependenceon the distance from the last magnet This facilitatesto position the feeding tube of the storage cell in a widerange of a distances to the last magnet The average val-ues Γx = (638 plusmn 060)mm and Γy = (684 plusmn 033)mmagree within the errors and yield a common width ofΓxy = (673 plusmn 029)mm The two-dimensional Gaus-sian distribution of this width allows one to estimate thefraction η of the beam intensity injected into the com-pression tube or a feeding tube For a tube of 10mmdiameter η = 078plusmn 003 comparable with η asymp 07 givenin Sec VA

TABLE III Dimensions (FWHM) of the atomic hydrogenbeam measured with the compression tube (CT) and thecrossed-beam quadrupole mass spectrometer (QMS) at dis-tances z to the last magnet along perpendicular directions xand y

z[mm] Γx[mm] Γy[mm]

CT 300 642 plusmn 009 699plusmn 006

CT 337 627 plusmn 008 658plusmn 008

QMS 567 736 plusmn 043 668plusmn 080

QMS 697 669 plusmn 022 638plusmn 027

15

VI DEGREE OF DISSOCIATION

Besides the intensity of the atomic beam it is impor-tant to determine the molecular fraction in the beamMolecules injected into the feeding tube reduce the po-larization of the target gas

A Measurements with crossed-beam QMS

In addition to the data on the profile of the atomichydrogen beam (Sec VB) data on the distributions ofmolecular hydrogen in the beam were taken too at thepositions z = 567mm and 697mm behind the last mag-net The relation between the degree of dissociation andthe QMS signals by the atomic and molecular beam com-ponent was given above by Eq (6) The coefficient kv =vmva however is chosen here under the assumptionthat the average velocity of the atoms is determined bythe nozzle temperature of 65K and that of the moleculesby scattering and recombination on the ABS chamberwalls at 290K This yields kv =

radic

2 middot 65290 = 067 ingood agreement with Ref13 where this coefficient wasdetermined by the measured velocity distributions undersimilar conditionsThe measured profiles of the atomic fraction (identical

to those of Fig 21) those of the molecular fraction andthose of the degree of dissociation deduced from Eq (6)are collected in Fig 24As it is seen from the figure the distribution of the

degree of dissociation shows a dip around the central linedue to the higher density of molecular hydrogen originat-ing from the nozzle The mean value in an aperture of

0

20

40

60

80

100

120

2 4 6 8 10 12 14 16 18 20 22

0

20

40

60

80

100

120

6 8 10 12 14 16 18 20 22 24

QM

S si

gnal

s [a

u]

(a)

(b)

(c)

00

02

04

06

08

10

12

Deg

ree

of d

isso

ciat

ion

()

X [mm]

Y [mm]

(d)

00

02

04

06

08

10

12

Z = 567 mm

Z = 697 mm

FIG 24 Spatial distributions of H1 (bull)H2 () and degree ofdissociation () averaged over 3mm wide bands in the xz andyz planes respectively (here the z-axis is the geometrical axisof the ABS)

10mm diameter results as α = 095 plusmn 004

B Measurements with the Lamb-shift polarimeter

A cup in the quench chamber of the Lamb-shift po-larimeter (LSP) described in Ref8 allows one to mea-sure the currents Icup(H1) and Icup(H2) of the H+

1 andH+

2 ions extracted from the ionizer and separated bythe Wien filter with the cesium evaporation and the spinfilter switched off The relation between the degree ofdissociation α and the measured currents is

α =Icup(H1)minus r1

r2Icup(H2)

Icup(H1)minus r1r2Icup(H2) + 2kv

r2Icup(H2)

(10)

Among the three coefficients kv = 067 as for the mea-surement with the QMS For the electron energy of about100keV the ratio r1 of dissociative to non-dissociativeionization of H2 is8

r1 =σ(H2 rarr 2H+

1 )

σ(H2 rarr H+2 )

= 0095plusmn 0008 (11)

and the ratio between the ionization cross sections is857

r2 =σion(H2)

σion(H1)= 17plusmn 01 (12)

At the standard operation parameters of the source(Sec IV) the measured currents are Icup(H1) = (125 plusmn05) nA and Icup(H2) = (64 plusmn 01) nA yielding α =(096plusmn004) in excellent agreement with the value result-ing from the measurements with the QMS (Sec VIA)

VII BEAM POLARIZATION

The Lamb-shift polarimeter was designed built andtested at Universitat zu Koln7 It was used to measureand to optimize the polarization of the atomic hydrogenand deuterium beams delivered by the ABS Details arefound in Ref7The vector polarization pz for hydrogen is defined by

the relative hyperfine-state occupation numbers N(mI)

pz =N(+ 1

2 )minusN(minus 12 )

N(+ 12 ) +N(minus 1

2 ) (13)

for deuterium

pz =N(+1)minusN(minus1)

N(+1) +N(0) +N(minus1) (14)

Deuterium tensor polarization pzz is given by

pzz =N(+1) +N(minus1)minus 2 middotN(0)

N(+1) +N(0) +N(minus1) (15)

16

These polarizations can be derived from the measuredLyman-α peak strengths S by application of a number ofcorrection factors78

Typical Lyman α spectra measured with the polarizedhydrogen and deuterium beam from the ABS are shownin the Figs 25 and 26

MFT 2-3

200 300 400 5000

40

80

120

160

Magnetic field in the spinfilter [au]

PM

T si

gnal

[au

]

mI = -12

(a)

mI = +12

100 200 300 400 5000

40

80

120

160(b)

MFT 2-3WFT 1-3

mI = +12

mI = -12

FIG 25 Lyman-α spectra measured with the polarized hy-drogen beam (a) population change from state |2〉 to state|3〉 induced by the MFT unit (b) same as (a) with subsequentpopulation change from state |1〉 to state |3〉 induced by theWFT unit

MFT 3-4SFT 2-6

0

2

4

6

8

mI = -1m

I = 0

(a)

mI = +1

0

2

4

6

8(b)

MFT 3-4WFT 1-4 2-3

mI = +1 m

I = 0

mI = -1

mI = +1

mI = 0

mI = -1

mI = +1

mI = 0

mI = -1

0 1000 2000 3000

0

1

2

3

4 (c)

PMT

sign

al [a

u]

Magnetic field in the spinfilter [au]

WFT 1-4SFT 2-6

0 1000 2000 30000

2

4

6

8WFT 1-4SFT 3-5

(d)

FIG 26 Lyman-α spectra measured with the polarized deu-terium beam (a) and (b) vector polarization resulting fromsubsequent transitions MFT (3 rarr 4) and SFT (2 rarr 6) andWFT (1 rarr 4 2 rarr 3) respectively (c) and (d) tensor polar-ization resulting from subsequent transitions WFT (1 rarr 4)and SFT (2 rarr 6) and SFT (3 rarr 5) respectively

The polarization values for the hydrogen and the deu-terium beam derived from the Lyman-α peak-strengthratios with application of the necessary corrections aresummarized in Table IV

The vector polarization for hydrogen of the first linereflects the population of state |1〉 and state |2〉 accordingto the Eqs (3) and (13) The value of 091 deduced fromthe calculated transmission values is confirmed by themeasured one

TABLE IV The vector polarization pz of the hydrogen beamand the vector and the tensor polarization pzz of the deu-terium beam from the ABS measured with the Lamb-shiftpolarimeter

populated pz pzz

state(s)

Hydrogen |1〉 +089plusmn 001 -

|3〉 minus096plusmn 001 -

Deuterium |1〉+ |6〉 +088plusmn 001 +088plusmn 003

|3〉+ |4〉 minus091plusmn 001 +085plusmn 002

|3〉+ |6〉 +0005plusmn 0003 +090plusmn 001

|2〉+ |5〉 +0005plusmn 0003 minus171plusmn 003

VIII CONCLUSIONS AND OUTLOOK

In this paper we present the detailed description ofthe major components of the atomic beam source (ABS)for the polarized internal gas target of the magnet spec-trometer ANKE in COSY-Julich The ABS was builtfor the purpose of extending the physics program ofANKE from unpolarized and single-polarized investiga-tions with stored beams towards double-polarized exper-iments1 thus facilitating nuclear reaction studies involv-

ing ~p~p ~p~d ~d~p and ~d~d initial statesThe mechanical design took into account that at

ANKE the source has to be mounted vertically and trans-versely movable together with the transverse motion ofthe spectrometer magnet D2 The design of the system ofsextupole magnets took advantage of the developments inthe field of rare-earth permanent magnets (NdFeB) Ded-icated tools and methods were developed to determineand to optimize the source parameters ie intensity de-gree of dissociation and polarization Special emphasiswas put on the measurements of the spatial distributionsof the atomic and molecular beam near the focus wherethe feeding tube of the storage cell is located The ABShas been used in a number of investigations at ANKEthe commissioning effort to prepare the target for the usewith polarized H is described in Ref62 Performed stud-ies of the deuteron-charge exchange reaction are summa-rized in Ref6364 studies in near-threshold pion produc-tion are reported about in Ref65The ABS resides at the ANKE target position for a few

months per year only thus during the remaining timeit is used for other studies It had been observed thatthe nuclear polarization in recombined hydrogen is par-tially retained after recombination66 as well as evidencefor nuclear tensor polarization in recombined deuteriummolecules67 In order to investigate this recombinationprocess in more detail a special setup has been developedin the framework of an ISTC project68 and the recombi-nation process for different cell-wall coatings and differ-ent polarizations of the injected hydrogen or deuterium

17

atoms as function of cell-wall temperature strength ofthe magnetic holding field and gas pressure in the cell ispresently investigated69ndash71

Appendix A Preparation of Discharge Tubes and Nozzles

1 Tube Treatment

One end of the discharge tube is machined at a 45

angle while the other is kept flat Both ends are thenremelted and the tubes are tempered at 150 C Thetubes are further treated according to the procedure de-scribed in Ref72 which includes successive cleaning withacetone methanol distilled water and subsequent rins-ing by a 21 acid mixture of concentrated HF (40) andHCl (32) for 5 min The tubes are then flushed bydistilled water and dried

2 Nozzle Treatment

The nozzles are cleaned in an ultrasonic bath oftrichlorethylene acetone methanol and finally distilledwater all at 50 C Anodizing takes place in sulfuric acidto form a thin layer of Al2O3 as described in Ref72 Af-terwards they are immersed in distilled water for 30minat 95 C

ACKNOWLEDGMENTS

The authors want to thank OWB Schult Institut furKernphysik (IKP) Julich who initiated the polarizationprogram of ANKE Thanks go to the design office themechanical workshop and especially to WR Ermer allIKP Valuable advice was received from the PINTEX col-laboration at IUCF from the target group at HERMESespecially NKoch and from DToporkov BINP Novosi-birsk The support by VKoptev PNPI Gatchina whoregrettably passed away in January 2012 is gratefullyacknowledged Thanks go also to R Poprawe and col-leagues Fraunhofer-Institut fur Lasertechnik Aachenwhere the encapsulations of the magnets were laser-welded

1AKacharava F Rathmann and CWilkin Spin Physics fromCOSY to FAIR COSY Experiment Proposal No 152 (2005)Available under httparXivnucl-ex0511028

2WHaeberli in Proc 2nd Int Symp on Polarization Phenomenaof Nucleons Karlsruhe 1965 Eds P Huber and H SchopperExperientia Supplementum 12 64 (Birkhauser Verlag 1966)

3E Steffens and WHaeberli Rep Progr Phys 66 1887 (2003)4SBarsov et al Nucl Instr and Meth A 462 364 (2001)5RMaier Nucl Instr and Meth A 390 1 (1997)6KGrigoryev et al Proc 14th International Workshop on Po-larized Sources Targets and Polarimetry (PSTP 2011) 12-16 September 2011 StPetersburg Russia eds KGrigoryevPKravtsov and AVasilyev ISBN 978-5-86763-282-3 61 (2011)

7REngels et al Rev Sci Instrum 74 4607 (2003)8REngels et al Rev Sci Instrum 76 053305 (2005)9TWise et al Nucl Instr and Meth A 336 410 (1993)

10WADezarn et al Nucl Instr and Meth A 362 36 (1995)11T Rinckel et al Nucl Instr and Meth A 439 117 (2000)12F Stock et al Nucl Instr and Meth A 343 334 (1994)13ANass et al Nucl Instr and Meth A 505 633 (2003)14VDerenchuk et al Proc Conf Polarized Ion Sources and Polar-ized Gas Targets Madison WI 1993 Eds LWAnderson andWHaeberli AIP Conf Proc 293 72 (American Institute ofPhysics 1994)

15HOkamura et al see Ref14 p 8416KHatanaka et al Nucl Instr and Meth A 384 575 (1997)17Manufacturer Schiffer Metall- amp Vakuumtechnik 52428 JulichGermany

18Single-stage type RGS120 refrigerating capacity 120W at 80Kand 20W at 30K Leybold Vacuum GmbH 50968 Koln Ger-many

19Mini UHV gate valve series 010 VAT Germany GmbH 85630Grasbrunn Germany

20Type F3 fomblin oil Pfeiffer Vacuum GmbH 35614 Asslar Ger-many

21Model HU 1 Leybold Vacuum GmbH 50968 Koln Germany22Manufacturer SK Industriemodell GmbH 52531 Ubach-Palenberg Germany

23Type PFG 600 RF with automatic matchbox PFM 1500 A-INDHuttinger Elektronik GmbH 79110 Freiburg Germany

24WKorsch PhD Thesis Philipps Universitat Marburg (1990)25F Stock et al Int Workshop on Polarized Beams and PolarizedGas Targets Koeln Germany 1995 Eds HPaetz gen Schieckand L Sydow (World Scientific Publ Co 1996) p 260

26The first number denotes the outer diameter and the second onethe wall thickness

27Type Duran 8330 equivalent to Corning 7740 (Pyrex) SchottAG 55122 Mainz Germany

28Ultra-Kryomat RUL 80-D Lauda DrRWobser GmbH 97912Lauda-Konigshofen Germany

29ODU-Kontakt GmbH 84444 Muhldorf Germany30Handbook of Chemistry and Physics Ed RCEast (The Chem-ical Rubber Co 1973) p E-10

31NKoch and E Steffens Rev Sci Instrum 70 1631 (1999)32AVassiliev et al Petersburg Nuclear Physics Institute ReportNP-32-1997 No 2175 (1997)

33BLorentz Diploma Thesis Ruprecht-Karls-Universitat Heidel-berg (1993)

34ANass and E Steffens Nucl Instr and Meth A 598 653(2009)

35WHaeberli Ann Rev Nucl Sci 17 373 (1967)36AVassiliev et al Rev Sci Instr 71 3331 (2000)37WKubischta Proc Workshop on Polarized Gas Targetsfor Storage Rings Heidelberg 23-26 September 1991 EdsHGGaul E Steffens and K Zapfe (Max-Planck-Institut furKernphysik Heidelberg)

38The labeling of the hyperfine states as |1〉 = |mj = +12 mI =+12〉 |2〉 = | + 12minus12〉 |3〉 = | minus 12+12〉 and |4〉 =|minus12minus12〉 for hydrogen and |1〉 = |+12+1〉 |2〉 = |+12 0〉|3〉 = | + 12minus1〉 |4〉 = | minus 12minus1〉 |5〉 = | minus 12 0〉 and

8

F = 1 levels |1〉 |2〉 and |3〉 with magnetic quantumnumbers mF = +1 0 and minus1 respectively can be re-garded as equally spaced In deuterium the same holdsfor the four F = 32 levels |1〉 |2〉 |3〉 and |4〉) withmF = +32 +12 minus12 and minus32 respectively andfor the two F = 12 levels |5〉 and |6〉 with mF = minus12and +12 respectively The magnetic component of therf dipole field induces transitions between each pair ofneighboring mF states with ∆mF = plusmn1 |∆mF| = 2transitions are forbidden The interchange of the popu-lation between the states |1〉 and |3〉 in hydrogen eg iscaused by a two-quantum transition via the intermedi-ate state |2〉 In the classical description of the adiabaticpassage method42 the population change should not de-pend on the sign of the magnetic field gradient relativeto the beam direction An exact quantum-mechanicaltreatment4546 however indicates that cleaner popula-tion changes from state |1〉 to |3〉 in hydrogen and fromstate |1〉 to |4〉 in deuterium are obtained with a nega-tive field gradient ie a Brf field decreasing in the beamdirection Deviations from adiabaticity are discussed inRef4547

The MFT unit is operated at higher values of Bstatwhere the differences in the energy spacings of pairs of hy-

FIG 7 Three-quarter-section view of the MFT unit with thesupport structure (1 self-supporting rf coil with spacers 2pick-up loop 3 Al tubes defining the length of the transition-inducing rf field 4 Cu cavity 5 coil around the pole shoe(6) providing the static field Bstat 7 slit between pole shoeand cavity wall housing the gradient-field coil 8 componentsof the static magnet yoke also serving as shielding againstexternal fields 9 cavity-positioning element 10 Cu padscooled by means of water-carrying tubes The cavity withthe rf coil and the pick-up loop can be taken out from thesurrounding components

z

Bstatic

transition

region

FIG 8 Arrangement of the windings producing the staticgradient field Bgrad shown in the left-hand side of the figureIn all transition units the field lies in the direction of the statichomogeneous field the field gradient dBdz lies in the beamdirection which defines the z axis In z direction the tran-sition reagion (indicated by the dashed lines) is confined tothe range of constant gradient by the Al tubes in orthogonaldirection by the beam diameter

perfine states with ∆mF = plusmn1 allow one to select singletransitions Originally developed for an polarized alkaliion source48 the MFT unit now is a standard compo-nent in polarized hydrogen and deuterium sources as dis-cussed eg in Ref49 Appropriate choice ofBhom Bgradand the rf frequency allows one to induce selected tran-sitions |1〉 harr |2〉 and |2〉 harr |3〉 in hydrogen or |1〉 harr |2〉|2〉 harr |3〉 and |3〉 harr |4〉 in deuterium Furthermore thechoice of the field gradient allows one to achieve consecu-tive transitions As an example a negative field gradientin the MFT unit behind the first set of magnets ie aB field decreasing in beam direction at a fixed rf fre-quency leads to the sequence of the transitions |3〉 rarr |4〉|2〉 rarr |3〉 and finally |1〉 rarr |2〉 in deuterium leaving thestate |1〉 empty

The SFT unit is used to induce transitions betweenstates in the upper and lower hyperfine multiplet in hy-drogen and deuterium Contrary to the historical nameindicating a strong magnetic field the SFT unit is op-erated with magnetic fields comparable to those used inthe MFT unit The transition frequencies are comparablewith those of the hyperfine splitting (1420MHz for hy-drogen and 327MHz for deuterium) and thus are muchhigher than those in the WFT and MFT units The rffield in a SFT unit is produced by a twin-line resonatorinside a Cu box tuned to the λ4 resonance50 The SFTunit51 is shown in Fig 9 Again the layout follows that ofthe unit used in the HERMES experiment44 Two vari-able capacitors at the free ends of the conducting rodsfed by the rf power with a relative phase shift of 180 allow one to tune the device

9

FIG 9 Three-quarter-section view of the rf cavity of theSFT unit for deuterium (1 the two resonant-field creatingconductors 2 the adjustible capacitor plates 3 Cu cavity)The inner dimensions of the cavity are 56mm along Bstat36mm orthogonal to it and 36mm height The cross sectionof the conductors is 14times 4mm2

H Slow Control System

Industrial components providing reliable and long-term support were selected for the control system of thewhole setup consisting of the ABS and the diagnosticstools the storage cell positioning system the Lamb-shiftpolarimeter and the supply system of a calibrated flowof unpolarized molecular gas The interlock system hasbeen implemented on the basis of SIEMENS SIMATICS7-300 family of programmable logic controllers In orderto unify the interfacing to the control computer all front-end equipment is connected via the PROFIBUS DP field-bus The process control software was implemented us-ing the Windows-based WinCC toolkit from SIEMENSThe system controls the operation of the pumps and thevalves It reads the pressure gauges and controls theregeneration cycles of the cryopumps Via a control net-work the temperature of the nozzle is stabilized withinplusmn05K Furthermore all power-supply units rf genera-tors and amplifiers are set and controlled The wholevariety of components to be controlled the logical struc-ture of the control and interlock system and a separatedevice for parameter studies are described in Ref52

III STUDIES OF THE FREE HYDROGEN JET

A Atomic beam profile near the nozzle

A novel device has been used to measure the profile ofan atomic beam via the deposition of recombination heaton thin wires in a two-dimensional grid5354 Atoms stuckon the surface of gold-plated tungsten wires of 5microm diam-eter recombine and are reemitted as molecules The re-combination heat (4476 eV per hydrogen molecule) leadsto a change of temperature and thus resistance alongeach wire The measurement of the resistance changes of

FIG 10 Two-dimensional profile of the atomic hydrogenbeam 10mm from the nozzle deduced from recombinationheating of gold-plated tungsten wires of 5microm in a 8times 8 wiregrid

all the wires in the grid allows one to deduce the centerand the profile of the beam Figure 10 shows the beamprofile resulting with a 8times8 wire grid positioned betweenskimmer and collimator performed as a first proof of themethod Later such a device has been used to comparemeasured and calculated beam profiles along the beamaxis between nozzle and skimmer34

B Degree of dissociation of the free atomic jet

The dissociation of the primary molecules is achievedby the interaction of the electrons and the hydrogen ordeuterium molecules in the plasma of the dissociatorThe degree of dissociation of the beam from the nozzledepends on the rf power applied to maintain the plasmathe primary molecular gas flow into the dissociator andthe temperature of the nozzle and the lower end of thedischarge tube These dependencies have been studiedbefore installation of the sextupole magnets with a setupcontaining a crossed-beam quadrupole mass spectrome-ter5556

α =ρa

ρa + 2 middot ρm (4)

The admixture of molecules in an atomic beam is de-scribed by the degree of dissociation where ρa and ρmare the densities of atomic and molecular hydrogen ordeuterium in the beam Other authors (eg Ref13) usethe atomic and molecular intensities Ia and Im in the def-inition of the degree of dissociation (αI) in Eq (4) Thetwo definitions of are related by

ImIa

=vmva

middot 1minus α

2α=

1minus αI

2αI (5)

10

This quantity was determined with the quadrupolemass spectrometer (QMS) in a conventional way as

α =Slowasta

Slowasta + 2 kv kion kdetSm

(6)

Here Slowasta = SaminusδSm denotes the atomic signal corrected

for dissociative ionization The parameter δ = 00141was obtained following the method described in Ref31The coefficient kv = vmva accounting for the difference

in atom and molecule velocity was chosen as 1radic2 under

the assumption of thermalization of the beam emergingfrom the nozzle Furthermore kion = 064 57 accounts forthe differences in ionization cross section for atomic andmolecular hydrogen and kdet = 084 for the detectionprobability55 As an example of the parameter stud-ies Fig 11 shows the deduced dependencies on the rfpower for a set of primary molecular hydrogen gas flowsFor typical flow values q(H2) le 10mbar ls a saturationvalue around 08 was obtained

0 50 100 150 200 250 300 350 400

00

02

04

06

08

10

Deg

ree

of d

isso

ciat

ion

()

Dissociator rf power [W]

q=07 mbar ls q=08 mbar ls q=10 mbar ls q=15 mbar ls

FIG 11 Degree of dissociation α of the free hydrogen jet asfunction of the applied rf power for different primary molec-ular hydrogen flows and a nozzle temperature of 70K

IV BEAM INTENSITY

The intensity of the polarized beam from the ABS to-gether with the layout of the storage cell determines theareal density of the target gas The intensity of the beamhas been measured with the use of a compression-tubesetup5859 shown in Fig 12 to optimize the ABS opera-tion parameters The measurements were performed at a300mm distance from the compression-tube entrance tothe last magnet and an inner tube diameter of 100mmas set in the tracking calculations The length of thecompression tube of 100mm was made equal to that ofthe foreseen feeding tube of the storage cell The narrowtube around the compression tube on a support based onthe lower flange separates the volume around the tube

FIG 12 Side view of the compression-tube setup made fromstandard ultra-high-vacuum components with a partial cutalong the axis (1 compression tube 2 support of thecompression tube based on the lower flange 3 narrow tubearound 1 closing the upper volume and allowing axial shiftsof the tube by the support 4 compression volume 5 hot-cathode pressure gauge 6 xy manipulator 7 z manipulator8 glass viewport 9 electromagnetic valve for gas inlet

from the compression volume The xy manipulator servesfor centering the tubes and for intensity-profile measure-ments The construction allows axial shifts of the com-pression tube by the z manipulator and the use of tubesof different diametersThe intensity of the beam entering the compression

volume through the compression tube is measured viathe pressure in the compression volume It is determinedby the equilibrium between the incoming beam intensityIin and the outgoing intensity Iout Under the assump-tion of a pure atomic beam and complete recombinationin the compression volume

Iin(atomss) = 2 middot Iout(moleculess)

= 2 middot∆P middot Ctube

= 2 middot∆P middot 103 middot 1020 middot d3

l

radic

T

M (7)

Here ∆P is the difference between the pressure measuredin the compreesion volume and that in the ABS cham-ber V The conductance of the compression tube Ctubeis determined by the inner diameter d of the tube itslength l the gas temperature T and the molar massM of the gas (given in cm and K respectively)60 The

11

CG

RVC200

H2

F

EV

EV

EV

EV

EV

MP

V1V2

EVR116

HC

V0

EV EVA016UDV136

HCCPLC

PS

220V AC

220V AC

Interface

24V DC

FIG 13 Source of calibrated molecular gas flow (V0 com-pression chamber V1 gas-storage chamber feeding V0 viathe needle valve UDV136a V2 chamber of calibrated volumeused to determine that of V1) The pressure in V1 is mea-sured by the capacitance gauge CG and is kept constant bythe dosing valve EVR116 with the gauge controller RVC200The whole setup including the evacuation elements can beoperated manually or by the programmable logic controllerPLC either within the ABS control system52 or as a separatesystem

a All the valves and the gauge controller are supplied by PfeifferVacuum GmbH Dndash35614 Asslar Germany (manufacturerBalzers AG Liechtenstein)

factor 2 takes into account that the same pressure is mea-sured in the hot-cathode gauge for 2 middotIin (H atomss) and1 middot Iin (H2 moleculess) For d = 10mm l = 100mmT = 290K and M = 2 for hydrogen pressure differences∆P on the order of 10minus4mbar are expected for atomichydrogen beam intensities in the order of 1017 atomssThe relation between Iin and ∆P for hydrogen has beendetermined experimentally with the use of a source ofcalibrated molecular hydrogen gas flow5859 depicted inFig 13 The measured dependence with a linear fit isshown in Fig 14 The calibration curve allows one to de-termine absolute values of Iin of hydrogen and deuteriumbeams The calibration for deuterium was deduced fromthe one for hydrogen by scaling with a factor 1

radic2 ac-

cording to Eq (7)

The dependences of Iin on the dissociator-operationparameters primary molecular hydrogen flow q(H2) noz-zle temperature Tn and dissociator power Pdiss have beenstudied to find the optimum values They are shown inthe Figs 15 16 and 17 respectively for different nozzle-orifice diameters The figures show that for the hydrogenbeam (states |1〉 and |2〉) with the standard operation pa-rameters qH2

= 11 mbar ls Tn = 70K Pdiss = 350Wand with a nozzle-orifice diameter of 23mm an inten-sity of Iin(H) = (75 plusmn 02) middot 1016 particless is achievedquite close to the earlier estimate from Eq (2) Besidesthe dominant atomic component of H atoms this value

04 06 08 10 12 14 16 18 20 2201

02

03

04

05

06

07

08

09

10

11

12

Gas

flow

from

the

stor

age

volu

me

[1017

ats

]

Pressure in compression volume [10-4 mbar]

FIG 14 Calibration curve for hydrogen used to deduce fromthe measured pressures the intensities of the hydrogen anddeuterium beam injected into the compression tube

00 05 10 15 20 2500

10

20

30

40

50

60

70

80 D = 20 mm D = 23 mm D = 25 mm

Bea

m in

tens

ity [1

016 a

ts]

Primary gas flow [mbar ls]

FIG 15 Intensity of the hydrogen beam (states |1〉 and |2〉)injected into the compression tube as function of the primarymolecular gas flow for different nozzle diameters D (nozzletemperature 60K dissociator power 300W)

includes small admixtures of H atoms in state |3〉 andmolecular hydrogen The first kind can be estimatedwith the use of the calculated transmissions (Sec II F)as 0017084 asymp 2 The amount of the second admix-ture has been measured as described below

For the deuterium beam (states |1〉 |2〉 and |3〉) theoptimization procedure gave an intensity of Iin(D) =(39 plusmn 02) middot 1016 particless achieved with q(D2) = 09mbar ls Tn = 65K and Pdiss = 300W slightly lowerthan those for hydrogen

12

50 60 70 80 90 100 110

40

50

60

70

80 D = 20 mm D = 23 mm D = 25 mm

Bea

m in

tens

ity [1

016 a

ts]

Nozzle temperature [K]

FIG 16 Intensity of the hydrogen beam (states |1〉 and |2〉)injected into the compression tube as function of the nozzletemperature for different nozzle diameters D (primary molec-ular gas flow 1mbar ls dissociator power 300W)

100 200 300 400 50010

20

30

40

50

60

70

80 D = 20 mm

Bea

m in

tens

ity [1

016 a

ts]

Dissociator power [W]

FIG 17 Intensity of the hydrogen beam (states |1〉 and |2〉)into the compression tube as function of the dissociator powerfor a nozzle diameter of 2 mm (nozzle temperature 60K pri-mary molecular gas flow 1mbar ls)

V HYDROGEN BEAM PROFILES

Beam profiles were measured at various positions atvarious positions behind the last sextupole magnet withthe use of

bull a compression tube of reduced dimensions (5mmdiameter)

bull a crossed-beam quadrupole mass spectrometer andbull a supplementary method of reduction of MoO3 by

hydrogen

A Measurements with the compression tube

For the determination of the beam dimensions at twopositions 300mm and 337mm behind the last magnetthe compression tube setup (Fig 12) was used makinguse of the possibility of axial movement by the z ma-nipulator and of that to install a narrower and shortercompression tube of 5mm diameter and 50mm lengthto enhance the spatial resolution The xy manipula-tor provided a lateral displacement of the compressiontube by plusmn10 mm in x and y direction The center co-ordinates of the geometrical axis of the source had beendetermined with the use of a bi-directional laser cen-tered inside the bore of the central support plate (seeFig 1) The relative intensity distributions in the xzand yz planes given by the measured pressure in thecompression volume are shown in Fig 18 Fits by Gaus-sian distributions to the data yield full widths at halfmaximum Γx = (642plusmn 009)mm Γy = (699plusmn 006)mmfor the distributions measured at z = 300mm and Γx =(627plusmn 008)mm Γy = (658plusmn 008)mm at 337mm

00

02

04

06

08

10

12

00

02

04

06

08

10

12

0 5 10 15 20 25

00

02

04

06

08

10

12

14

0 5 10 15 20 25

00

02

04

06

08

10

12

14

Pre

ssur

e in

the

com

pres

sion

tube

[au

]

Z = 300 mm

(a) (b)

(c) (d)

Z = 337 mm

X [mm]

Y [mm]

FIG 18 Cross sections of the beam profile in the mid-planemeasured with compression tube of 5mm diameter and 50mmlength Measurements in the xz-plane (a c) and yz-plane (bd) performed at two different positions z = 300mm (a b)and z = 337mm (c d) behind the last sextupole magnet ofthe ABS The shaded area represents position and dimensionsof the compression tube used in intensity measurements

The center of gravity of the measured profile definedas

rc =

sum

ij

radic

x2i + y2j middot P (xi yj)

sum

ij

P (xi yj) (8)

where xi and yj give the position of the compression-tubeaxis and P (xi yj) is the pressure measured in the com-

13

pression volume The resulting rc shows a deviation of012 mm from the geometrical axis of the source Further-more the data measured with the narrow compressiontube of 25mm radius can be used to derive the fractionof the beam entering the compression tube of 5mm ra-dius used in the intensity measurement of Sec IV Theratio

η =

rile25 mmsum

0P (xi yj)

rile10 mmsum

0P (xi yj)

(9)

where ri is the distance of the compression-tube axis tothe beam axis yields η asymp 07

B Measurements with the QMS

The beam-profile studies of Sec VA were extendedwith a setup utilizing a crossed-beam quadrupole massspectrometer (QMS) in the setup of Fig 19 Contraryto the measurements with the compression tube thosewith the QMS allow to separate the atomic and molec-ular fractions in the beam A 2mm diameter aperturewas installed at the entrance of the sensitive volume of

FIG 19 Setup for the measurements of the beam profilewith the QMS (1) xy-table enabling two-dimensional dis-placement of the entrance window of the QMS against thegeometrical axis of the ABS (2) the QMS (3) manually op-erated beam shutter The beam dump is an axially mountedcryo pump

the QMS to improve the resolution compared with thatachieved by the compression tube of 5mm diameter usedin measurements of the preceding section The layout ofthe setup presented in Fig 19 shows that in the presentcase the profile could not be measured at a distance ofz = 300mm to the last magnet Instead measurementswere performed at z = 567mm and with installation ofan extension tube at z = 697mm The xy manipulatorenabled displacements of the aperture axis from the geo-metrical axis of the source in any direction within limitsset by the bore diameter of the xy manipulatorThe first measured distribution of the atomic hydro-

gen (Fig 20) showed a distinct deviation from azimuthalsymmetry indicating an insufficient relative alignment ofnozzle and skimmer The three threaded rods support-ing the dissociator with the nozzle via the three-legged

FIG 20 Two-dimensional distribution of the atomic hydro-gen component of the beam at z = 567mm before the nozzle-to-skimmer adjustment showing a disinct deviation from az-imuthal symmetry

plate (label 2 in Fig 1) allow one to vary the position ofthe nozzle relative to that of the skimmer while the sourceis running This possibility has been used to find a nozzleposition which results in an azimuthally symmetric distri-bution The achieved symmetric distribution is shown inFig 21 and profiles of the atomic hadrogen component inthe beam measured in x and y direction at z = 567mmand z = 697mm are presented in Fig 22 Fits by Gaus-sian distributions to the data yield full widths at halfmaximum Γx = (736plusmn 043)mm Γy = (668plusmn 080)mmfor the distributions measured at z = 567mm and Γx =(669plusmn 022)mm Γy = (638plusmn 027)mm at 697mm

C Reduction of MoO3 by hydrogen

In addition to the compression tube and the QMS tech-nique a supplementary attempt was made to determinethe beam profile by exposing molybdenium trioxide (a

14

FIG 21 The distribution corresponding to that of Fig 20 af-ter nozzle-to-skimmer adjustment resulting in azimuthal sym-metry

0

20

40

60

80

0

20

40

60

80

2 4 6 8 10 12 14 16 18 20 22

0

20

40

60

80

6 8 10 12 14 16 18 20 22 24

0

20

40

60

80

QM

S si

gnal

s [a

u]

(a)

Z = 567 mm

Z = 697 mm

(c)

(b)

X [mm]

(d)

Y [mm]

FIG 22 Profiles of the atomic hydrogen component inthe beam measured with the QMS 567mm and 697mmbehind the last magnet

yellowish powder) on a glass plate to the beam The prin-ciple of this method is based on the reduction of MoO3 toa lower oxide of blue colour It first was used in the ex-periment to measure the magnetic moment of the hydro-gen atom by splitting of the beam in an inhomogeneousmagnetic field 61

This method is much simpler than the time-consumingmeasurements described in Secs VA and VB It givesqualitative results as presented in Fig 23 A quantitativeanalysis however requires development of the measuringtechnique (eg preparation of glass plates study of theoptimum exposure time digital image processing)

FIG 23 Photo of the glass plate covered with molybdenumtrioxide MoO3 exposed to the atomic hydrogen beam

D Summary of the profile measurements

Table III summarizes results of the measurements ofthe ABS beam profile with the compression-tube and theQMS setup The larger errors of the widths measuredwith the QMS are due to the lack of measurements withthe dissociator switched off and the necessity to estimatethe background signal from the existing data Within theerrors the measured widths do not show a dependenceon the distance from the last magnet This facilitatesto position the feeding tube of the storage cell in a widerange of a distances to the last magnet The average val-ues Γx = (638 plusmn 060)mm and Γy = (684 plusmn 033)mmagree within the errors and yield a common width ofΓxy = (673 plusmn 029)mm The two-dimensional Gaus-sian distribution of this width allows one to estimate thefraction η of the beam intensity injected into the com-pression tube or a feeding tube For a tube of 10mmdiameter η = 078plusmn 003 comparable with η asymp 07 givenin Sec VA

TABLE III Dimensions (FWHM) of the atomic hydrogenbeam measured with the compression tube (CT) and thecrossed-beam quadrupole mass spectrometer (QMS) at dis-tances z to the last magnet along perpendicular directions xand y

z[mm] Γx[mm] Γy[mm]

CT 300 642 plusmn 009 699plusmn 006

CT 337 627 plusmn 008 658plusmn 008

QMS 567 736 plusmn 043 668plusmn 080

QMS 697 669 plusmn 022 638plusmn 027

15

VI DEGREE OF DISSOCIATION

Besides the intensity of the atomic beam it is impor-tant to determine the molecular fraction in the beamMolecules injected into the feeding tube reduce the po-larization of the target gas

A Measurements with crossed-beam QMS

In addition to the data on the profile of the atomichydrogen beam (Sec VB) data on the distributions ofmolecular hydrogen in the beam were taken too at thepositions z = 567mm and 697mm behind the last mag-net The relation between the degree of dissociation andthe QMS signals by the atomic and molecular beam com-ponent was given above by Eq (6) The coefficient kv =vmva however is chosen here under the assumptionthat the average velocity of the atoms is determined bythe nozzle temperature of 65K and that of the moleculesby scattering and recombination on the ABS chamberwalls at 290K This yields kv =

radic

2 middot 65290 = 067 ingood agreement with Ref13 where this coefficient wasdetermined by the measured velocity distributions undersimilar conditionsThe measured profiles of the atomic fraction (identical

to those of Fig 21) those of the molecular fraction andthose of the degree of dissociation deduced from Eq (6)are collected in Fig 24As it is seen from the figure the distribution of the

degree of dissociation shows a dip around the central linedue to the higher density of molecular hydrogen originat-ing from the nozzle The mean value in an aperture of

0

20

40

60

80

100

120

2 4 6 8 10 12 14 16 18 20 22

0

20

40

60

80

100

120

6 8 10 12 14 16 18 20 22 24

QM

S si

gnal

s [a

u]

(a)

(b)

(c)

00

02

04

06

08

10

12

Deg

ree

of d

isso

ciat

ion

()

X [mm]

Y [mm]

(d)

00

02

04

06

08

10

12

Z = 567 mm

Z = 697 mm

FIG 24 Spatial distributions of H1 (bull)H2 () and degree ofdissociation () averaged over 3mm wide bands in the xz andyz planes respectively (here the z-axis is the geometrical axisof the ABS)

10mm diameter results as α = 095 plusmn 004

B Measurements with the Lamb-shift polarimeter

A cup in the quench chamber of the Lamb-shift po-larimeter (LSP) described in Ref8 allows one to mea-sure the currents Icup(H1) and Icup(H2) of the H+

1 andH+

2 ions extracted from the ionizer and separated bythe Wien filter with the cesium evaporation and the spinfilter switched off The relation between the degree ofdissociation α and the measured currents is

α =Icup(H1)minus r1

r2Icup(H2)

Icup(H1)minus r1r2Icup(H2) + 2kv

r2Icup(H2)

(10)

Among the three coefficients kv = 067 as for the mea-surement with the QMS For the electron energy of about100keV the ratio r1 of dissociative to non-dissociativeionization of H2 is8

r1 =σ(H2 rarr 2H+

1 )

σ(H2 rarr H+2 )

= 0095plusmn 0008 (11)

and the ratio between the ionization cross sections is857

r2 =σion(H2)

σion(H1)= 17plusmn 01 (12)

At the standard operation parameters of the source(Sec IV) the measured currents are Icup(H1) = (125 plusmn05) nA and Icup(H2) = (64 plusmn 01) nA yielding α =(096plusmn004) in excellent agreement with the value result-ing from the measurements with the QMS (Sec VIA)

VII BEAM POLARIZATION

The Lamb-shift polarimeter was designed built andtested at Universitat zu Koln7 It was used to measureand to optimize the polarization of the atomic hydrogenand deuterium beams delivered by the ABS Details arefound in Ref7The vector polarization pz for hydrogen is defined by

the relative hyperfine-state occupation numbers N(mI)

pz =N(+ 1

2 )minusN(minus 12 )

N(+ 12 ) +N(minus 1

2 ) (13)

for deuterium

pz =N(+1)minusN(minus1)

N(+1) +N(0) +N(minus1) (14)

Deuterium tensor polarization pzz is given by

pzz =N(+1) +N(minus1)minus 2 middotN(0)

N(+1) +N(0) +N(minus1) (15)

16

These polarizations can be derived from the measuredLyman-α peak strengths S by application of a number ofcorrection factors78

Typical Lyman α spectra measured with the polarizedhydrogen and deuterium beam from the ABS are shownin the Figs 25 and 26

MFT 2-3

200 300 400 5000

40

80

120

160

Magnetic field in the spinfilter [au]

PM

T si

gnal

[au

]

mI = -12

(a)

mI = +12

100 200 300 400 5000

40

80

120

160(b)

MFT 2-3WFT 1-3

mI = +12

mI = -12

FIG 25 Lyman-α spectra measured with the polarized hy-drogen beam (a) population change from state |2〉 to state|3〉 induced by the MFT unit (b) same as (a) with subsequentpopulation change from state |1〉 to state |3〉 induced by theWFT unit

MFT 3-4SFT 2-6

0

2

4

6

8

mI = -1m

I = 0

(a)

mI = +1

0

2

4

6

8(b)

MFT 3-4WFT 1-4 2-3

mI = +1 m

I = 0

mI = -1

mI = +1

mI = 0

mI = -1

mI = +1

mI = 0

mI = -1

0 1000 2000 3000

0

1

2

3

4 (c)

PMT

sign

al [a

u]

Magnetic field in the spinfilter [au]

WFT 1-4SFT 2-6

0 1000 2000 30000

2

4

6

8WFT 1-4SFT 3-5

(d)

FIG 26 Lyman-α spectra measured with the polarized deu-terium beam (a) and (b) vector polarization resulting fromsubsequent transitions MFT (3 rarr 4) and SFT (2 rarr 6) andWFT (1 rarr 4 2 rarr 3) respectively (c) and (d) tensor polar-ization resulting from subsequent transitions WFT (1 rarr 4)and SFT (2 rarr 6) and SFT (3 rarr 5) respectively

The polarization values for the hydrogen and the deu-terium beam derived from the Lyman-α peak-strengthratios with application of the necessary corrections aresummarized in Table IV

The vector polarization for hydrogen of the first linereflects the population of state |1〉 and state |2〉 accordingto the Eqs (3) and (13) The value of 091 deduced fromthe calculated transmission values is confirmed by themeasured one

TABLE IV The vector polarization pz of the hydrogen beamand the vector and the tensor polarization pzz of the deu-terium beam from the ABS measured with the Lamb-shiftpolarimeter

populated pz pzz

state(s)

Hydrogen |1〉 +089plusmn 001 -

|3〉 minus096plusmn 001 -

Deuterium |1〉+ |6〉 +088plusmn 001 +088plusmn 003

|3〉+ |4〉 minus091plusmn 001 +085plusmn 002

|3〉+ |6〉 +0005plusmn 0003 +090plusmn 001

|2〉+ |5〉 +0005plusmn 0003 minus171plusmn 003

VIII CONCLUSIONS AND OUTLOOK

In this paper we present the detailed description ofthe major components of the atomic beam source (ABS)for the polarized internal gas target of the magnet spec-trometer ANKE in COSY-Julich The ABS was builtfor the purpose of extending the physics program ofANKE from unpolarized and single-polarized investiga-tions with stored beams towards double-polarized exper-iments1 thus facilitating nuclear reaction studies involv-

ing ~p~p ~p~d ~d~p and ~d~d initial statesThe mechanical design took into account that at

ANKE the source has to be mounted vertically and trans-versely movable together with the transverse motion ofthe spectrometer magnet D2 The design of the system ofsextupole magnets took advantage of the developments inthe field of rare-earth permanent magnets (NdFeB) Ded-icated tools and methods were developed to determineand to optimize the source parameters ie intensity de-gree of dissociation and polarization Special emphasiswas put on the measurements of the spatial distributionsof the atomic and molecular beam near the focus wherethe feeding tube of the storage cell is located The ABShas been used in a number of investigations at ANKEthe commissioning effort to prepare the target for the usewith polarized H is described in Ref62 Performed stud-ies of the deuteron-charge exchange reaction are summa-rized in Ref6364 studies in near-threshold pion produc-tion are reported about in Ref65The ABS resides at the ANKE target position for a few

months per year only thus during the remaining timeit is used for other studies It had been observed thatthe nuclear polarization in recombined hydrogen is par-tially retained after recombination66 as well as evidencefor nuclear tensor polarization in recombined deuteriummolecules67 In order to investigate this recombinationprocess in more detail a special setup has been developedin the framework of an ISTC project68 and the recombi-nation process for different cell-wall coatings and differ-ent polarizations of the injected hydrogen or deuterium

17

atoms as function of cell-wall temperature strength ofthe magnetic holding field and gas pressure in the cell ispresently investigated69ndash71

Appendix A Preparation of Discharge Tubes and Nozzles

1 Tube Treatment

One end of the discharge tube is machined at a 45

angle while the other is kept flat Both ends are thenremelted and the tubes are tempered at 150 C Thetubes are further treated according to the procedure de-scribed in Ref72 which includes successive cleaning withacetone methanol distilled water and subsequent rins-ing by a 21 acid mixture of concentrated HF (40) andHCl (32) for 5 min The tubes are then flushed bydistilled water and dried

2 Nozzle Treatment

The nozzles are cleaned in an ultrasonic bath oftrichlorethylene acetone methanol and finally distilledwater all at 50 C Anodizing takes place in sulfuric acidto form a thin layer of Al2O3 as described in Ref72 Af-terwards they are immersed in distilled water for 30minat 95 C

ACKNOWLEDGMENTS

The authors want to thank OWB Schult Institut furKernphysik (IKP) Julich who initiated the polarizationprogram of ANKE Thanks go to the design office themechanical workshop and especially to WR Ermer allIKP Valuable advice was received from the PINTEX col-laboration at IUCF from the target group at HERMESespecially NKoch and from DToporkov BINP Novosi-birsk The support by VKoptev PNPI Gatchina whoregrettably passed away in January 2012 is gratefullyacknowledged Thanks go also to R Poprawe and col-leagues Fraunhofer-Institut fur Lasertechnik Aachenwhere the encapsulations of the magnets were laser-welded

1AKacharava F Rathmann and CWilkin Spin Physics fromCOSY to FAIR COSY Experiment Proposal No 152 (2005)Available under httparXivnucl-ex0511028

2WHaeberli in Proc 2nd Int Symp on Polarization Phenomenaof Nucleons Karlsruhe 1965 Eds P Huber and H SchopperExperientia Supplementum 12 64 (Birkhauser Verlag 1966)

3E Steffens and WHaeberli Rep Progr Phys 66 1887 (2003)4SBarsov et al Nucl Instr and Meth A 462 364 (2001)5RMaier Nucl Instr and Meth A 390 1 (1997)6KGrigoryev et al Proc 14th International Workshop on Po-larized Sources Targets and Polarimetry (PSTP 2011) 12-16 September 2011 StPetersburg Russia eds KGrigoryevPKravtsov and AVasilyev ISBN 978-5-86763-282-3 61 (2011)

7REngels et al Rev Sci Instrum 74 4607 (2003)8REngels et al Rev Sci Instrum 76 053305 (2005)9TWise et al Nucl Instr and Meth A 336 410 (1993)

10WADezarn et al Nucl Instr and Meth A 362 36 (1995)11T Rinckel et al Nucl Instr and Meth A 439 117 (2000)12F Stock et al Nucl Instr and Meth A 343 334 (1994)13ANass et al Nucl Instr and Meth A 505 633 (2003)14VDerenchuk et al Proc Conf Polarized Ion Sources and Polar-ized Gas Targets Madison WI 1993 Eds LWAnderson andWHaeberli AIP Conf Proc 293 72 (American Institute ofPhysics 1994)

15HOkamura et al see Ref14 p 8416KHatanaka et al Nucl Instr and Meth A 384 575 (1997)17Manufacturer Schiffer Metall- amp Vakuumtechnik 52428 JulichGermany

18Single-stage type RGS120 refrigerating capacity 120W at 80Kand 20W at 30K Leybold Vacuum GmbH 50968 Koln Ger-many

19Mini UHV gate valve series 010 VAT Germany GmbH 85630Grasbrunn Germany

20Type F3 fomblin oil Pfeiffer Vacuum GmbH 35614 Asslar Ger-many

21Model HU 1 Leybold Vacuum GmbH 50968 Koln Germany22Manufacturer SK Industriemodell GmbH 52531 Ubach-Palenberg Germany

23Type PFG 600 RF with automatic matchbox PFM 1500 A-INDHuttinger Elektronik GmbH 79110 Freiburg Germany

24WKorsch PhD Thesis Philipps Universitat Marburg (1990)25F Stock et al Int Workshop on Polarized Beams and PolarizedGas Targets Koeln Germany 1995 Eds HPaetz gen Schieckand L Sydow (World Scientific Publ Co 1996) p 260

26The first number denotes the outer diameter and the second onethe wall thickness

27Type Duran 8330 equivalent to Corning 7740 (Pyrex) SchottAG 55122 Mainz Germany

28Ultra-Kryomat RUL 80-D Lauda DrRWobser GmbH 97912Lauda-Konigshofen Germany

29ODU-Kontakt GmbH 84444 Muhldorf Germany30Handbook of Chemistry and Physics Ed RCEast (The Chem-ical Rubber Co 1973) p E-10

31NKoch and E Steffens Rev Sci Instrum 70 1631 (1999)32AVassiliev et al Petersburg Nuclear Physics Institute ReportNP-32-1997 No 2175 (1997)

33BLorentz Diploma Thesis Ruprecht-Karls-Universitat Heidel-berg (1993)

34ANass and E Steffens Nucl Instr and Meth A 598 653(2009)

35WHaeberli Ann Rev Nucl Sci 17 373 (1967)36AVassiliev et al Rev Sci Instr 71 3331 (2000)37WKubischta Proc Workshop on Polarized Gas Targetsfor Storage Rings Heidelberg 23-26 September 1991 EdsHGGaul E Steffens and K Zapfe (Max-Planck-Institut furKernphysik Heidelberg)

38The labeling of the hyperfine states as |1〉 = |mj = +12 mI =+12〉 |2〉 = | + 12minus12〉 |3〉 = | minus 12+12〉 and |4〉 =|minus12minus12〉 for hydrogen and |1〉 = |+12+1〉 |2〉 = |+12 0〉|3〉 = | + 12minus1〉 |4〉 = | minus 12minus1〉 |5〉 = | minus 12 0〉 and

9

FIG 9 Three-quarter-section view of the rf cavity of theSFT unit for deuterium (1 the two resonant-field creatingconductors 2 the adjustible capacitor plates 3 Cu cavity)The inner dimensions of the cavity are 56mm along Bstat36mm orthogonal to it and 36mm height The cross sectionof the conductors is 14times 4mm2

H Slow Control System

Industrial components providing reliable and long-term support were selected for the control system of thewhole setup consisting of the ABS and the diagnosticstools the storage cell positioning system the Lamb-shiftpolarimeter and the supply system of a calibrated flowof unpolarized molecular gas The interlock system hasbeen implemented on the basis of SIEMENS SIMATICS7-300 family of programmable logic controllers In orderto unify the interfacing to the control computer all front-end equipment is connected via the PROFIBUS DP field-bus The process control software was implemented us-ing the Windows-based WinCC toolkit from SIEMENSThe system controls the operation of the pumps and thevalves It reads the pressure gauges and controls theregeneration cycles of the cryopumps Via a control net-work the temperature of the nozzle is stabilized withinplusmn05K Furthermore all power-supply units rf genera-tors and amplifiers are set and controlled The wholevariety of components to be controlled the logical struc-ture of the control and interlock system and a separatedevice for parameter studies are described in Ref52

III STUDIES OF THE FREE HYDROGEN JET

A Atomic beam profile near the nozzle

A novel device has been used to measure the profile ofan atomic beam via the deposition of recombination heaton thin wires in a two-dimensional grid5354 Atoms stuckon the surface of gold-plated tungsten wires of 5microm diam-eter recombine and are reemitted as molecules The re-combination heat (4476 eV per hydrogen molecule) leadsto a change of temperature and thus resistance alongeach wire The measurement of the resistance changes of

FIG 10 Two-dimensional profile of the atomic hydrogenbeam 10mm from the nozzle deduced from recombinationheating of gold-plated tungsten wires of 5microm in a 8times 8 wiregrid

all the wires in the grid allows one to deduce the centerand the profile of the beam Figure 10 shows the beamprofile resulting with a 8times8 wire grid positioned betweenskimmer and collimator performed as a first proof of themethod Later such a device has been used to comparemeasured and calculated beam profiles along the beamaxis between nozzle and skimmer34

B Degree of dissociation of the free atomic jet

The dissociation of the primary molecules is achievedby the interaction of the electrons and the hydrogen ordeuterium molecules in the plasma of the dissociatorThe degree of dissociation of the beam from the nozzledepends on the rf power applied to maintain the plasmathe primary molecular gas flow into the dissociator andthe temperature of the nozzle and the lower end of thedischarge tube These dependencies have been studiedbefore installation of the sextupole magnets with a setupcontaining a crossed-beam quadrupole mass spectrome-ter5556

α =ρa

ρa + 2 middot ρm (4)

The admixture of molecules in an atomic beam is de-scribed by the degree of dissociation where ρa and ρmare the densities of atomic and molecular hydrogen ordeuterium in the beam Other authors (eg Ref13) usethe atomic and molecular intensities Ia and Im in the def-inition of the degree of dissociation (αI) in Eq (4) Thetwo definitions of are related by

ImIa

=vmva

middot 1minus α

2α=

1minus αI

2αI (5)

10

This quantity was determined with the quadrupolemass spectrometer (QMS) in a conventional way as

α =Slowasta

Slowasta + 2 kv kion kdetSm

(6)

Here Slowasta = SaminusδSm denotes the atomic signal corrected

for dissociative ionization The parameter δ = 00141was obtained following the method described in Ref31The coefficient kv = vmva accounting for the difference

in atom and molecule velocity was chosen as 1radic2 under

the assumption of thermalization of the beam emergingfrom the nozzle Furthermore kion = 064 57 accounts forthe differences in ionization cross section for atomic andmolecular hydrogen and kdet = 084 for the detectionprobability55 As an example of the parameter stud-ies Fig 11 shows the deduced dependencies on the rfpower for a set of primary molecular hydrogen gas flowsFor typical flow values q(H2) le 10mbar ls a saturationvalue around 08 was obtained

0 50 100 150 200 250 300 350 400

00

02

04

06

08

10

Deg

ree

of d

isso

ciat

ion

()

Dissociator rf power [W]

q=07 mbar ls q=08 mbar ls q=10 mbar ls q=15 mbar ls

FIG 11 Degree of dissociation α of the free hydrogen jet asfunction of the applied rf power for different primary molec-ular hydrogen flows and a nozzle temperature of 70K

IV BEAM INTENSITY

The intensity of the polarized beam from the ABS to-gether with the layout of the storage cell determines theareal density of the target gas The intensity of the beamhas been measured with the use of a compression-tubesetup5859 shown in Fig 12 to optimize the ABS opera-tion parameters The measurements were performed at a300mm distance from the compression-tube entrance tothe last magnet and an inner tube diameter of 100mmas set in the tracking calculations The length of thecompression tube of 100mm was made equal to that ofthe foreseen feeding tube of the storage cell The narrowtube around the compression tube on a support based onthe lower flange separates the volume around the tube

FIG 12 Side view of the compression-tube setup made fromstandard ultra-high-vacuum components with a partial cutalong the axis (1 compression tube 2 support of thecompression tube based on the lower flange 3 narrow tubearound 1 closing the upper volume and allowing axial shiftsof the tube by the support 4 compression volume 5 hot-cathode pressure gauge 6 xy manipulator 7 z manipulator8 glass viewport 9 electromagnetic valve for gas inlet

from the compression volume The xy manipulator servesfor centering the tubes and for intensity-profile measure-ments The construction allows axial shifts of the com-pression tube by the z manipulator and the use of tubesof different diametersThe intensity of the beam entering the compression

volume through the compression tube is measured viathe pressure in the compression volume It is determinedby the equilibrium between the incoming beam intensityIin and the outgoing intensity Iout Under the assump-tion of a pure atomic beam and complete recombinationin the compression volume

Iin(atomss) = 2 middot Iout(moleculess)

= 2 middot∆P middot Ctube

= 2 middot∆P middot 103 middot 1020 middot d3

l

radic

T

M (7)

Here ∆P is the difference between the pressure measuredin the compreesion volume and that in the ABS cham-ber V The conductance of the compression tube Ctubeis determined by the inner diameter d of the tube itslength l the gas temperature T and the molar massM of the gas (given in cm and K respectively)60 The

11

CG

RVC200

H2

F

EV

EV

EV

EV

EV

MP

V1V2

EVR116

HC

V0

EV EVA016UDV136

HCCPLC

PS

220V AC

220V AC

Interface

24V DC

FIG 13 Source of calibrated molecular gas flow (V0 com-pression chamber V1 gas-storage chamber feeding V0 viathe needle valve UDV136a V2 chamber of calibrated volumeused to determine that of V1) The pressure in V1 is mea-sured by the capacitance gauge CG and is kept constant bythe dosing valve EVR116 with the gauge controller RVC200The whole setup including the evacuation elements can beoperated manually or by the programmable logic controllerPLC either within the ABS control system52 or as a separatesystem

a All the valves and the gauge controller are supplied by PfeifferVacuum GmbH Dndash35614 Asslar Germany (manufacturerBalzers AG Liechtenstein)

factor 2 takes into account that the same pressure is mea-sured in the hot-cathode gauge for 2 middotIin (H atomss) and1 middot Iin (H2 moleculess) For d = 10mm l = 100mmT = 290K and M = 2 for hydrogen pressure differences∆P on the order of 10minus4mbar are expected for atomichydrogen beam intensities in the order of 1017 atomssThe relation between Iin and ∆P for hydrogen has beendetermined experimentally with the use of a source ofcalibrated molecular hydrogen gas flow5859 depicted inFig 13 The measured dependence with a linear fit isshown in Fig 14 The calibration curve allows one to de-termine absolute values of Iin of hydrogen and deuteriumbeams The calibration for deuterium was deduced fromthe one for hydrogen by scaling with a factor 1

radic2 ac-

cording to Eq (7)

The dependences of Iin on the dissociator-operationparameters primary molecular hydrogen flow q(H2) noz-zle temperature Tn and dissociator power Pdiss have beenstudied to find the optimum values They are shown inthe Figs 15 16 and 17 respectively for different nozzle-orifice diameters The figures show that for the hydrogenbeam (states |1〉 and |2〉) with the standard operation pa-rameters qH2

= 11 mbar ls Tn = 70K Pdiss = 350Wand with a nozzle-orifice diameter of 23mm an inten-sity of Iin(H) = (75 plusmn 02) middot 1016 particless is achievedquite close to the earlier estimate from Eq (2) Besidesthe dominant atomic component of H atoms this value

04 06 08 10 12 14 16 18 20 2201

02

03

04

05

06

07

08

09

10

11

12

Gas

flow

from

the

stor

age

volu

me

[1017

ats

]

Pressure in compression volume [10-4 mbar]

FIG 14 Calibration curve for hydrogen used to deduce fromthe measured pressures the intensities of the hydrogen anddeuterium beam injected into the compression tube

00 05 10 15 20 2500

10

20

30

40

50

60

70

80 D = 20 mm D = 23 mm D = 25 mm

Bea

m in

tens

ity [1

016 a

ts]

Primary gas flow [mbar ls]

FIG 15 Intensity of the hydrogen beam (states |1〉 and |2〉)injected into the compression tube as function of the primarymolecular gas flow for different nozzle diameters D (nozzletemperature 60K dissociator power 300W)

includes small admixtures of H atoms in state |3〉 andmolecular hydrogen The first kind can be estimatedwith the use of the calculated transmissions (Sec II F)as 0017084 asymp 2 The amount of the second admix-ture has been measured as described below

For the deuterium beam (states |1〉 |2〉 and |3〉) theoptimization procedure gave an intensity of Iin(D) =(39 plusmn 02) middot 1016 particless achieved with q(D2) = 09mbar ls Tn = 65K and Pdiss = 300W slightly lowerthan those for hydrogen

12

50 60 70 80 90 100 110

40

50

60

70

80 D = 20 mm D = 23 mm D = 25 mm

Bea

m in

tens

ity [1

016 a

ts]

Nozzle temperature [K]

FIG 16 Intensity of the hydrogen beam (states |1〉 and |2〉)injected into the compression tube as function of the nozzletemperature for different nozzle diameters D (primary molec-ular gas flow 1mbar ls dissociator power 300W)

100 200 300 400 50010

20

30

40

50

60

70

80 D = 20 mm

Bea

m in

tens

ity [1

016 a

ts]

Dissociator power [W]

FIG 17 Intensity of the hydrogen beam (states |1〉 and |2〉)into the compression tube as function of the dissociator powerfor a nozzle diameter of 2 mm (nozzle temperature 60K pri-mary molecular gas flow 1mbar ls)

V HYDROGEN BEAM PROFILES

Beam profiles were measured at various positions atvarious positions behind the last sextupole magnet withthe use of

bull a compression tube of reduced dimensions (5mmdiameter)

bull a crossed-beam quadrupole mass spectrometer andbull a supplementary method of reduction of MoO3 by

hydrogen

A Measurements with the compression tube

For the determination of the beam dimensions at twopositions 300mm and 337mm behind the last magnetthe compression tube setup (Fig 12) was used makinguse of the possibility of axial movement by the z ma-nipulator and of that to install a narrower and shortercompression tube of 5mm diameter and 50mm lengthto enhance the spatial resolution The xy manipula-tor provided a lateral displacement of the compressiontube by plusmn10 mm in x and y direction The center co-ordinates of the geometrical axis of the source had beendetermined with the use of a bi-directional laser cen-tered inside the bore of the central support plate (seeFig 1) The relative intensity distributions in the xzand yz planes given by the measured pressure in thecompression volume are shown in Fig 18 Fits by Gaus-sian distributions to the data yield full widths at halfmaximum Γx = (642plusmn 009)mm Γy = (699plusmn 006)mmfor the distributions measured at z = 300mm and Γx =(627plusmn 008)mm Γy = (658plusmn 008)mm at 337mm

00

02

04

06

08

10

12

00

02

04

06

08

10

12

0 5 10 15 20 25

00

02

04

06

08

10

12

14

0 5 10 15 20 25

00

02

04

06

08

10

12

14

Pre

ssur

e in

the

com

pres

sion

tube

[au

]

Z = 300 mm

(a) (b)

(c) (d)

Z = 337 mm

X [mm]

Y [mm]

FIG 18 Cross sections of the beam profile in the mid-planemeasured with compression tube of 5mm diameter and 50mmlength Measurements in the xz-plane (a c) and yz-plane (bd) performed at two different positions z = 300mm (a b)and z = 337mm (c d) behind the last sextupole magnet ofthe ABS The shaded area represents position and dimensionsof the compression tube used in intensity measurements

The center of gravity of the measured profile definedas

rc =

sum

ij

radic

x2i + y2j middot P (xi yj)

sum

ij

P (xi yj) (8)

where xi and yj give the position of the compression-tubeaxis and P (xi yj) is the pressure measured in the com-

13

pression volume The resulting rc shows a deviation of012 mm from the geometrical axis of the source Further-more the data measured with the narrow compressiontube of 25mm radius can be used to derive the fractionof the beam entering the compression tube of 5mm ra-dius used in the intensity measurement of Sec IV Theratio

η =

rile25 mmsum

0P (xi yj)

rile10 mmsum

0P (xi yj)

(9)

where ri is the distance of the compression-tube axis tothe beam axis yields η asymp 07

B Measurements with the QMS

The beam-profile studies of Sec VA were extendedwith a setup utilizing a crossed-beam quadrupole massspectrometer (QMS) in the setup of Fig 19 Contraryto the measurements with the compression tube thosewith the QMS allow to separate the atomic and molec-ular fractions in the beam A 2mm diameter aperturewas installed at the entrance of the sensitive volume of

FIG 19 Setup for the measurements of the beam profilewith the QMS (1) xy-table enabling two-dimensional dis-placement of the entrance window of the QMS against thegeometrical axis of the ABS (2) the QMS (3) manually op-erated beam shutter The beam dump is an axially mountedcryo pump

the QMS to improve the resolution compared with thatachieved by the compression tube of 5mm diameter usedin measurements of the preceding section The layout ofthe setup presented in Fig 19 shows that in the presentcase the profile could not be measured at a distance ofz = 300mm to the last magnet Instead measurementswere performed at z = 567mm and with installation ofan extension tube at z = 697mm The xy manipulatorenabled displacements of the aperture axis from the geo-metrical axis of the source in any direction within limitsset by the bore diameter of the xy manipulatorThe first measured distribution of the atomic hydro-

gen (Fig 20) showed a distinct deviation from azimuthalsymmetry indicating an insufficient relative alignment ofnozzle and skimmer The three threaded rods support-ing the dissociator with the nozzle via the three-legged

FIG 20 Two-dimensional distribution of the atomic hydro-gen component of the beam at z = 567mm before the nozzle-to-skimmer adjustment showing a disinct deviation from az-imuthal symmetry

plate (label 2 in Fig 1) allow one to vary the position ofthe nozzle relative to that of the skimmer while the sourceis running This possibility has been used to find a nozzleposition which results in an azimuthally symmetric distri-bution The achieved symmetric distribution is shown inFig 21 and profiles of the atomic hadrogen component inthe beam measured in x and y direction at z = 567mmand z = 697mm are presented in Fig 22 Fits by Gaus-sian distributions to the data yield full widths at halfmaximum Γx = (736plusmn 043)mm Γy = (668plusmn 080)mmfor the distributions measured at z = 567mm and Γx =(669plusmn 022)mm Γy = (638plusmn 027)mm at 697mm

C Reduction of MoO3 by hydrogen

In addition to the compression tube and the QMS tech-nique a supplementary attempt was made to determinethe beam profile by exposing molybdenium trioxide (a

14

FIG 21 The distribution corresponding to that of Fig 20 af-ter nozzle-to-skimmer adjustment resulting in azimuthal sym-metry

0

20

40

60

80

0

20

40

60

80

2 4 6 8 10 12 14 16 18 20 22

0

20

40

60

80

6 8 10 12 14 16 18 20 22 24

0

20

40

60

80

QM

S si

gnal

s [a

u]

(a)

Z = 567 mm

Z = 697 mm

(c)

(b)

X [mm]

(d)

Y [mm]

FIG 22 Profiles of the atomic hydrogen component inthe beam measured with the QMS 567mm and 697mmbehind the last magnet

yellowish powder) on a glass plate to the beam The prin-ciple of this method is based on the reduction of MoO3 toa lower oxide of blue colour It first was used in the ex-periment to measure the magnetic moment of the hydro-gen atom by splitting of the beam in an inhomogeneousmagnetic field 61

This method is much simpler than the time-consumingmeasurements described in Secs VA and VB It givesqualitative results as presented in Fig 23 A quantitativeanalysis however requires development of the measuringtechnique (eg preparation of glass plates study of theoptimum exposure time digital image processing)

FIG 23 Photo of the glass plate covered with molybdenumtrioxide MoO3 exposed to the atomic hydrogen beam

D Summary of the profile measurements

Table III summarizes results of the measurements ofthe ABS beam profile with the compression-tube and theQMS setup The larger errors of the widths measuredwith the QMS are due to the lack of measurements withthe dissociator switched off and the necessity to estimatethe background signal from the existing data Within theerrors the measured widths do not show a dependenceon the distance from the last magnet This facilitatesto position the feeding tube of the storage cell in a widerange of a distances to the last magnet The average val-ues Γx = (638 plusmn 060)mm and Γy = (684 plusmn 033)mmagree within the errors and yield a common width ofΓxy = (673 plusmn 029)mm The two-dimensional Gaus-sian distribution of this width allows one to estimate thefraction η of the beam intensity injected into the com-pression tube or a feeding tube For a tube of 10mmdiameter η = 078plusmn 003 comparable with η asymp 07 givenin Sec VA

TABLE III Dimensions (FWHM) of the atomic hydrogenbeam measured with the compression tube (CT) and thecrossed-beam quadrupole mass spectrometer (QMS) at dis-tances z to the last magnet along perpendicular directions xand y

z[mm] Γx[mm] Γy[mm]

CT 300 642 plusmn 009 699plusmn 006

CT 337 627 plusmn 008 658plusmn 008

QMS 567 736 plusmn 043 668plusmn 080

QMS 697 669 plusmn 022 638plusmn 027

15

VI DEGREE OF DISSOCIATION

Besides the intensity of the atomic beam it is impor-tant to determine the molecular fraction in the beamMolecules injected into the feeding tube reduce the po-larization of the target gas

A Measurements with crossed-beam QMS

In addition to the data on the profile of the atomichydrogen beam (Sec VB) data on the distributions ofmolecular hydrogen in the beam were taken too at thepositions z = 567mm and 697mm behind the last mag-net The relation between the degree of dissociation andthe QMS signals by the atomic and molecular beam com-ponent was given above by Eq (6) The coefficient kv =vmva however is chosen here under the assumptionthat the average velocity of the atoms is determined bythe nozzle temperature of 65K and that of the moleculesby scattering and recombination on the ABS chamberwalls at 290K This yields kv =

radic

2 middot 65290 = 067 ingood agreement with Ref13 where this coefficient wasdetermined by the measured velocity distributions undersimilar conditionsThe measured profiles of the atomic fraction (identical

to those of Fig 21) those of the molecular fraction andthose of the degree of dissociation deduced from Eq (6)are collected in Fig 24As it is seen from the figure the distribution of the

degree of dissociation shows a dip around the central linedue to the higher density of molecular hydrogen originat-ing from the nozzle The mean value in an aperture of

0

20

40

60

80

100

120

2 4 6 8 10 12 14 16 18 20 22

0

20

40

60

80

100

120

6 8 10 12 14 16 18 20 22 24

QM

S si

gnal

s [a

u]

(a)

(b)

(c)

00

02

04

06

08

10

12

Deg

ree

of d

isso

ciat

ion

()

X [mm]

Y [mm]

(d)

00

02

04

06

08

10

12

Z = 567 mm

Z = 697 mm

FIG 24 Spatial distributions of H1 (bull)H2 () and degree ofdissociation () averaged over 3mm wide bands in the xz andyz planes respectively (here the z-axis is the geometrical axisof the ABS)

10mm diameter results as α = 095 plusmn 004

B Measurements with the Lamb-shift polarimeter

A cup in the quench chamber of the Lamb-shift po-larimeter (LSP) described in Ref8 allows one to mea-sure the currents Icup(H1) and Icup(H2) of the H+

1 andH+

2 ions extracted from the ionizer and separated bythe Wien filter with the cesium evaporation and the spinfilter switched off The relation between the degree ofdissociation α and the measured currents is

α =Icup(H1)minus r1

r2Icup(H2)

Icup(H1)minus r1r2Icup(H2) + 2kv

r2Icup(H2)

(10)

Among the three coefficients kv = 067 as for the mea-surement with the QMS For the electron energy of about100keV the ratio r1 of dissociative to non-dissociativeionization of H2 is8

r1 =σ(H2 rarr 2H+

1 )

σ(H2 rarr H+2 )

= 0095plusmn 0008 (11)

and the ratio between the ionization cross sections is857

r2 =σion(H2)

σion(H1)= 17plusmn 01 (12)

At the standard operation parameters of the source(Sec IV) the measured currents are Icup(H1) = (125 plusmn05) nA and Icup(H2) = (64 plusmn 01) nA yielding α =(096plusmn004) in excellent agreement with the value result-ing from the measurements with the QMS (Sec VIA)

VII BEAM POLARIZATION

The Lamb-shift polarimeter was designed built andtested at Universitat zu Koln7 It was used to measureand to optimize the polarization of the atomic hydrogenand deuterium beams delivered by the ABS Details arefound in Ref7The vector polarization pz for hydrogen is defined by

the relative hyperfine-state occupation numbers N(mI)

pz =N(+ 1

2 )minusN(minus 12 )

N(+ 12 ) +N(minus 1

2 ) (13)

for deuterium

pz =N(+1)minusN(minus1)

N(+1) +N(0) +N(minus1) (14)

Deuterium tensor polarization pzz is given by

pzz =N(+1) +N(minus1)minus 2 middotN(0)

N(+1) +N(0) +N(minus1) (15)

16

These polarizations can be derived from the measuredLyman-α peak strengths S by application of a number ofcorrection factors78

Typical Lyman α spectra measured with the polarizedhydrogen and deuterium beam from the ABS are shownin the Figs 25 and 26

MFT 2-3

200 300 400 5000

40

80

120

160

Magnetic field in the spinfilter [au]

PM

T si

gnal

[au

]

mI = -12

(a)

mI = +12

100 200 300 400 5000

40

80

120

160(b)

MFT 2-3WFT 1-3

mI = +12

mI = -12

FIG 25 Lyman-α spectra measured with the polarized hy-drogen beam (a) population change from state |2〉 to state|3〉 induced by the MFT unit (b) same as (a) with subsequentpopulation change from state |1〉 to state |3〉 induced by theWFT unit

MFT 3-4SFT 2-6

0

2

4

6

8

mI = -1m

I = 0

(a)

mI = +1

0

2

4

6

8(b)

MFT 3-4WFT 1-4 2-3

mI = +1 m

I = 0

mI = -1

mI = +1

mI = 0

mI = -1

mI = +1

mI = 0

mI = -1

0 1000 2000 3000

0

1

2

3

4 (c)

PMT

sign

al [a

u]

Magnetic field in the spinfilter [au]

WFT 1-4SFT 2-6

0 1000 2000 30000

2

4

6

8WFT 1-4SFT 3-5

(d)

FIG 26 Lyman-α spectra measured with the polarized deu-terium beam (a) and (b) vector polarization resulting fromsubsequent transitions MFT (3 rarr 4) and SFT (2 rarr 6) andWFT (1 rarr 4 2 rarr 3) respectively (c) and (d) tensor polar-ization resulting from subsequent transitions WFT (1 rarr 4)and SFT (2 rarr 6) and SFT (3 rarr 5) respectively

The polarization values for the hydrogen and the deu-terium beam derived from the Lyman-α peak-strengthratios with application of the necessary corrections aresummarized in Table IV

The vector polarization for hydrogen of the first linereflects the population of state |1〉 and state |2〉 accordingto the Eqs (3) and (13) The value of 091 deduced fromthe calculated transmission values is confirmed by themeasured one

TABLE IV The vector polarization pz of the hydrogen beamand the vector and the tensor polarization pzz of the deu-terium beam from the ABS measured with the Lamb-shiftpolarimeter

populated pz pzz

state(s)

Hydrogen |1〉 +089plusmn 001 -

|3〉 minus096plusmn 001 -

Deuterium |1〉+ |6〉 +088plusmn 001 +088plusmn 003

|3〉+ |4〉 minus091plusmn 001 +085plusmn 002

|3〉+ |6〉 +0005plusmn 0003 +090plusmn 001

|2〉+ |5〉 +0005plusmn 0003 minus171plusmn 003

VIII CONCLUSIONS AND OUTLOOK

In this paper we present the detailed description ofthe major components of the atomic beam source (ABS)for the polarized internal gas target of the magnet spec-trometer ANKE in COSY-Julich The ABS was builtfor the purpose of extending the physics program ofANKE from unpolarized and single-polarized investiga-tions with stored beams towards double-polarized exper-iments1 thus facilitating nuclear reaction studies involv-

ing ~p~p ~p~d ~d~p and ~d~d initial statesThe mechanical design took into account that at

ANKE the source has to be mounted vertically and trans-versely movable together with the transverse motion ofthe spectrometer magnet D2 The design of the system ofsextupole magnets took advantage of the developments inthe field of rare-earth permanent magnets (NdFeB) Ded-icated tools and methods were developed to determineand to optimize the source parameters ie intensity de-gree of dissociation and polarization Special emphasiswas put on the measurements of the spatial distributionsof the atomic and molecular beam near the focus wherethe feeding tube of the storage cell is located The ABShas been used in a number of investigations at ANKEthe commissioning effort to prepare the target for the usewith polarized H is described in Ref62 Performed stud-ies of the deuteron-charge exchange reaction are summa-rized in Ref6364 studies in near-threshold pion produc-tion are reported about in Ref65The ABS resides at the ANKE target position for a few

months per year only thus during the remaining timeit is used for other studies It had been observed thatthe nuclear polarization in recombined hydrogen is par-tially retained after recombination66 as well as evidencefor nuclear tensor polarization in recombined deuteriummolecules67 In order to investigate this recombinationprocess in more detail a special setup has been developedin the framework of an ISTC project68 and the recombi-nation process for different cell-wall coatings and differ-ent polarizations of the injected hydrogen or deuterium

17

atoms as function of cell-wall temperature strength ofthe magnetic holding field and gas pressure in the cell ispresently investigated69ndash71

Appendix A Preparation of Discharge Tubes and Nozzles

1 Tube Treatment

One end of the discharge tube is machined at a 45

angle while the other is kept flat Both ends are thenremelted and the tubes are tempered at 150 C Thetubes are further treated according to the procedure de-scribed in Ref72 which includes successive cleaning withacetone methanol distilled water and subsequent rins-ing by a 21 acid mixture of concentrated HF (40) andHCl (32) for 5 min The tubes are then flushed bydistilled water and dried

2 Nozzle Treatment

The nozzles are cleaned in an ultrasonic bath oftrichlorethylene acetone methanol and finally distilledwater all at 50 C Anodizing takes place in sulfuric acidto form a thin layer of Al2O3 as described in Ref72 Af-terwards they are immersed in distilled water for 30minat 95 C

ACKNOWLEDGMENTS

The authors want to thank OWB Schult Institut furKernphysik (IKP) Julich who initiated the polarizationprogram of ANKE Thanks go to the design office themechanical workshop and especially to WR Ermer allIKP Valuable advice was received from the PINTEX col-laboration at IUCF from the target group at HERMESespecially NKoch and from DToporkov BINP Novosi-birsk The support by VKoptev PNPI Gatchina whoregrettably passed away in January 2012 is gratefullyacknowledged Thanks go also to R Poprawe and col-leagues Fraunhofer-Institut fur Lasertechnik Aachenwhere the encapsulations of the magnets were laser-welded

1AKacharava F Rathmann and CWilkin Spin Physics fromCOSY to FAIR COSY Experiment Proposal No 152 (2005)Available under httparXivnucl-ex0511028

2WHaeberli in Proc 2nd Int Symp on Polarization Phenomenaof Nucleons Karlsruhe 1965 Eds P Huber and H SchopperExperientia Supplementum 12 64 (Birkhauser Verlag 1966)

3E Steffens and WHaeberli Rep Progr Phys 66 1887 (2003)4SBarsov et al Nucl Instr and Meth A 462 364 (2001)5RMaier Nucl Instr and Meth A 390 1 (1997)6KGrigoryev et al Proc 14th International Workshop on Po-larized Sources Targets and Polarimetry (PSTP 2011) 12-16 September 2011 StPetersburg Russia eds KGrigoryevPKravtsov and AVasilyev ISBN 978-5-86763-282-3 61 (2011)

7REngels et al Rev Sci Instrum 74 4607 (2003)8REngels et al Rev Sci Instrum 76 053305 (2005)9TWise et al Nucl Instr and Meth A 336 410 (1993)

10WADezarn et al Nucl Instr and Meth A 362 36 (1995)11T Rinckel et al Nucl Instr and Meth A 439 117 (2000)12F Stock et al Nucl Instr and Meth A 343 334 (1994)13ANass et al Nucl Instr and Meth A 505 633 (2003)14VDerenchuk et al Proc Conf Polarized Ion Sources and Polar-ized Gas Targets Madison WI 1993 Eds LWAnderson andWHaeberli AIP Conf Proc 293 72 (American Institute ofPhysics 1994)

15HOkamura et al see Ref14 p 8416KHatanaka et al Nucl Instr and Meth A 384 575 (1997)17Manufacturer Schiffer Metall- amp Vakuumtechnik 52428 JulichGermany

18Single-stage type RGS120 refrigerating capacity 120W at 80Kand 20W at 30K Leybold Vacuum GmbH 50968 Koln Ger-many

19Mini UHV gate valve series 010 VAT Germany GmbH 85630Grasbrunn Germany

20Type F3 fomblin oil Pfeiffer Vacuum GmbH 35614 Asslar Ger-many

21Model HU 1 Leybold Vacuum GmbH 50968 Koln Germany22Manufacturer SK Industriemodell GmbH 52531 Ubach-Palenberg Germany

23Type PFG 600 RF with automatic matchbox PFM 1500 A-INDHuttinger Elektronik GmbH 79110 Freiburg Germany

24WKorsch PhD Thesis Philipps Universitat Marburg (1990)25F Stock et al Int Workshop on Polarized Beams and PolarizedGas Targets Koeln Germany 1995 Eds HPaetz gen Schieckand L Sydow (World Scientific Publ Co 1996) p 260

26The first number denotes the outer diameter and the second onethe wall thickness

27Type Duran 8330 equivalent to Corning 7740 (Pyrex) SchottAG 55122 Mainz Germany

28Ultra-Kryomat RUL 80-D Lauda DrRWobser GmbH 97912Lauda-Konigshofen Germany

29ODU-Kontakt GmbH 84444 Muhldorf Germany30Handbook of Chemistry and Physics Ed RCEast (The Chem-ical Rubber Co 1973) p E-10

31NKoch and E Steffens Rev Sci Instrum 70 1631 (1999)32AVassiliev et al Petersburg Nuclear Physics Institute ReportNP-32-1997 No 2175 (1997)

33BLorentz Diploma Thesis Ruprecht-Karls-Universitat Heidel-berg (1993)

34ANass and E Steffens Nucl Instr and Meth A 598 653(2009)

35WHaeberli Ann Rev Nucl Sci 17 373 (1967)36AVassiliev et al Rev Sci Instr 71 3331 (2000)37WKubischta Proc Workshop on Polarized Gas Targetsfor Storage Rings Heidelberg 23-26 September 1991 EdsHGGaul E Steffens and K Zapfe (Max-Planck-Institut furKernphysik Heidelberg)

38The labeling of the hyperfine states as |1〉 = |mj = +12 mI =+12〉 |2〉 = | + 12minus12〉 |3〉 = | minus 12+12〉 and |4〉 =|minus12minus12〉 for hydrogen and |1〉 = |+12+1〉 |2〉 = |+12 0〉|3〉 = | + 12minus1〉 |4〉 = | minus 12minus1〉 |5〉 = | minus 12 0〉 and

10

This quantity was determined with the quadrupolemass spectrometer (QMS) in a conventional way as

α =Slowasta

Slowasta + 2 kv kion kdetSm

(6)

Here Slowasta = SaminusδSm denotes the atomic signal corrected

for dissociative ionization The parameter δ = 00141was obtained following the method described in Ref31The coefficient kv = vmva accounting for the difference

in atom and molecule velocity was chosen as 1radic2 under

the assumption of thermalization of the beam emergingfrom the nozzle Furthermore kion = 064 57 accounts forthe differences in ionization cross section for atomic andmolecular hydrogen and kdet = 084 for the detectionprobability55 As an example of the parameter stud-ies Fig 11 shows the deduced dependencies on the rfpower for a set of primary molecular hydrogen gas flowsFor typical flow values q(H2) le 10mbar ls a saturationvalue around 08 was obtained

0 50 100 150 200 250 300 350 400

00

02

04

06

08

10

Deg

ree

of d

isso

ciat

ion

()

Dissociator rf power [W]

q=07 mbar ls q=08 mbar ls q=10 mbar ls q=15 mbar ls

FIG 11 Degree of dissociation α of the free hydrogen jet asfunction of the applied rf power for different primary molec-ular hydrogen flows and a nozzle temperature of 70K

IV BEAM INTENSITY

The intensity of the polarized beam from the ABS to-gether with the layout of the storage cell determines theareal density of the target gas The intensity of the beamhas been measured with the use of a compression-tubesetup5859 shown in Fig 12 to optimize the ABS opera-tion parameters The measurements were performed at a300mm distance from the compression-tube entrance tothe last magnet and an inner tube diameter of 100mmas set in the tracking calculations The length of thecompression tube of 100mm was made equal to that ofthe foreseen feeding tube of the storage cell The narrowtube around the compression tube on a support based onthe lower flange separates the volume around the tube

FIG 12 Side view of the compression-tube setup made fromstandard ultra-high-vacuum components with a partial cutalong the axis (1 compression tube 2 support of thecompression tube based on the lower flange 3 narrow tubearound 1 closing the upper volume and allowing axial shiftsof the tube by the support 4 compression volume 5 hot-cathode pressure gauge 6 xy manipulator 7 z manipulator8 glass viewport 9 electromagnetic valve for gas inlet

from the compression volume The xy manipulator servesfor centering the tubes and for intensity-profile measure-ments The construction allows axial shifts of the com-pression tube by the z manipulator and the use of tubesof different diametersThe intensity of the beam entering the compression

volume through the compression tube is measured viathe pressure in the compression volume It is determinedby the equilibrium between the incoming beam intensityIin and the outgoing intensity Iout Under the assump-tion of a pure atomic beam and complete recombinationin the compression volume

Iin(atomss) = 2 middot Iout(moleculess)

= 2 middot∆P middot Ctube

= 2 middot∆P middot 103 middot 1020 middot d3

l

radic

T

M (7)

Here ∆P is the difference between the pressure measuredin the compreesion volume and that in the ABS cham-ber V The conductance of the compression tube Ctubeis determined by the inner diameter d of the tube itslength l the gas temperature T and the molar massM of the gas (given in cm and K respectively)60 The

11

CG

RVC200

H2

F

EV

EV

EV

EV

EV

MP

V1V2

EVR116

HC

V0

EV EVA016UDV136

HCCPLC

PS

220V AC

220V AC

Interface

24V DC

FIG 13 Source of calibrated molecular gas flow (V0 com-pression chamber V1 gas-storage chamber feeding V0 viathe needle valve UDV136a V2 chamber of calibrated volumeused to determine that of V1) The pressure in V1 is mea-sured by the capacitance gauge CG and is kept constant bythe dosing valve EVR116 with the gauge controller RVC200The whole setup including the evacuation elements can beoperated manually or by the programmable logic controllerPLC either within the ABS control system52 or as a separatesystem

a All the valves and the gauge controller are supplied by PfeifferVacuum GmbH Dndash35614 Asslar Germany (manufacturerBalzers AG Liechtenstein)

factor 2 takes into account that the same pressure is mea-sured in the hot-cathode gauge for 2 middotIin (H atomss) and1 middot Iin (H2 moleculess) For d = 10mm l = 100mmT = 290K and M = 2 for hydrogen pressure differences∆P on the order of 10minus4mbar are expected for atomichydrogen beam intensities in the order of 1017 atomssThe relation between Iin and ∆P for hydrogen has beendetermined experimentally with the use of a source ofcalibrated molecular hydrogen gas flow5859 depicted inFig 13 The measured dependence with a linear fit isshown in Fig 14 The calibration curve allows one to de-termine absolute values of Iin of hydrogen and deuteriumbeams The calibration for deuterium was deduced fromthe one for hydrogen by scaling with a factor 1

radic2 ac-

cording to Eq (7)

The dependences of Iin on the dissociator-operationparameters primary molecular hydrogen flow q(H2) noz-zle temperature Tn and dissociator power Pdiss have beenstudied to find the optimum values They are shown inthe Figs 15 16 and 17 respectively for different nozzle-orifice diameters The figures show that for the hydrogenbeam (states |1〉 and |2〉) with the standard operation pa-rameters qH2

= 11 mbar ls Tn = 70K Pdiss = 350Wand with a nozzle-orifice diameter of 23mm an inten-sity of Iin(H) = (75 plusmn 02) middot 1016 particless is achievedquite close to the earlier estimate from Eq (2) Besidesthe dominant atomic component of H atoms this value

04 06 08 10 12 14 16 18 20 2201

02

03

04

05

06

07

08

09

10

11

12

Gas

flow

from

the

stor

age

volu

me

[1017

ats

]

Pressure in compression volume [10-4 mbar]

FIG 14 Calibration curve for hydrogen used to deduce fromthe measured pressures the intensities of the hydrogen anddeuterium beam injected into the compression tube

00 05 10 15 20 2500

10

20

30

40

50

60

70

80 D = 20 mm D = 23 mm D = 25 mm

Bea

m in

tens

ity [1

016 a

ts]

Primary gas flow [mbar ls]

FIG 15 Intensity of the hydrogen beam (states |1〉 and |2〉)injected into the compression tube as function of the primarymolecular gas flow for different nozzle diameters D (nozzletemperature 60K dissociator power 300W)

includes small admixtures of H atoms in state |3〉 andmolecular hydrogen The first kind can be estimatedwith the use of the calculated transmissions (Sec II F)as 0017084 asymp 2 The amount of the second admix-ture has been measured as described below

For the deuterium beam (states |1〉 |2〉 and |3〉) theoptimization procedure gave an intensity of Iin(D) =(39 plusmn 02) middot 1016 particless achieved with q(D2) = 09mbar ls Tn = 65K and Pdiss = 300W slightly lowerthan those for hydrogen

12

50 60 70 80 90 100 110

40

50

60

70

80 D = 20 mm D = 23 mm D = 25 mm

Bea

m in

tens

ity [1

016 a

ts]

Nozzle temperature [K]

FIG 16 Intensity of the hydrogen beam (states |1〉 and |2〉)injected into the compression tube as function of the nozzletemperature for different nozzle diameters D (primary molec-ular gas flow 1mbar ls dissociator power 300W)

100 200 300 400 50010

20

30

40

50

60

70

80 D = 20 mm

Bea

m in

tens

ity [1

016 a

ts]

Dissociator power [W]

FIG 17 Intensity of the hydrogen beam (states |1〉 and |2〉)into the compression tube as function of the dissociator powerfor a nozzle diameter of 2 mm (nozzle temperature 60K pri-mary molecular gas flow 1mbar ls)

V HYDROGEN BEAM PROFILES

Beam profiles were measured at various positions atvarious positions behind the last sextupole magnet withthe use of

bull a compression tube of reduced dimensions (5mmdiameter)

bull a crossed-beam quadrupole mass spectrometer andbull a supplementary method of reduction of MoO3 by

hydrogen

A Measurements with the compression tube

For the determination of the beam dimensions at twopositions 300mm and 337mm behind the last magnetthe compression tube setup (Fig 12) was used makinguse of the possibility of axial movement by the z ma-nipulator and of that to install a narrower and shortercompression tube of 5mm diameter and 50mm lengthto enhance the spatial resolution The xy manipula-tor provided a lateral displacement of the compressiontube by plusmn10 mm in x and y direction The center co-ordinates of the geometrical axis of the source had beendetermined with the use of a bi-directional laser cen-tered inside the bore of the central support plate (seeFig 1) The relative intensity distributions in the xzand yz planes given by the measured pressure in thecompression volume are shown in Fig 18 Fits by Gaus-sian distributions to the data yield full widths at halfmaximum Γx = (642plusmn 009)mm Γy = (699plusmn 006)mmfor the distributions measured at z = 300mm and Γx =(627plusmn 008)mm Γy = (658plusmn 008)mm at 337mm

00

02

04

06

08

10

12

00

02

04

06

08

10

12

0 5 10 15 20 25

00

02

04

06

08

10

12

14

0 5 10 15 20 25

00

02

04

06

08

10

12

14

Pre

ssur

e in

the

com

pres

sion

tube

[au

]

Z = 300 mm

(a) (b)

(c) (d)

Z = 337 mm

X [mm]

Y [mm]

FIG 18 Cross sections of the beam profile in the mid-planemeasured with compression tube of 5mm diameter and 50mmlength Measurements in the xz-plane (a c) and yz-plane (bd) performed at two different positions z = 300mm (a b)and z = 337mm (c d) behind the last sextupole magnet ofthe ABS The shaded area represents position and dimensionsof the compression tube used in intensity measurements

The center of gravity of the measured profile definedas

rc =

sum

ij

radic

x2i + y2j middot P (xi yj)

sum

ij

P (xi yj) (8)

where xi and yj give the position of the compression-tubeaxis and P (xi yj) is the pressure measured in the com-

13

pression volume The resulting rc shows a deviation of012 mm from the geometrical axis of the source Further-more the data measured with the narrow compressiontube of 25mm radius can be used to derive the fractionof the beam entering the compression tube of 5mm ra-dius used in the intensity measurement of Sec IV Theratio

η =

rile25 mmsum

0P (xi yj)

rile10 mmsum

0P (xi yj)

(9)

where ri is the distance of the compression-tube axis tothe beam axis yields η asymp 07

B Measurements with the QMS

The beam-profile studies of Sec VA were extendedwith a setup utilizing a crossed-beam quadrupole massspectrometer (QMS) in the setup of Fig 19 Contraryto the measurements with the compression tube thosewith the QMS allow to separate the atomic and molec-ular fractions in the beam A 2mm diameter aperturewas installed at the entrance of the sensitive volume of

FIG 19 Setup for the measurements of the beam profilewith the QMS (1) xy-table enabling two-dimensional dis-placement of the entrance window of the QMS against thegeometrical axis of the ABS (2) the QMS (3) manually op-erated beam shutter The beam dump is an axially mountedcryo pump

the QMS to improve the resolution compared with thatachieved by the compression tube of 5mm diameter usedin measurements of the preceding section The layout ofthe setup presented in Fig 19 shows that in the presentcase the profile could not be measured at a distance ofz = 300mm to the last magnet Instead measurementswere performed at z = 567mm and with installation ofan extension tube at z = 697mm The xy manipulatorenabled displacements of the aperture axis from the geo-metrical axis of the source in any direction within limitsset by the bore diameter of the xy manipulatorThe first measured distribution of the atomic hydro-

gen (Fig 20) showed a distinct deviation from azimuthalsymmetry indicating an insufficient relative alignment ofnozzle and skimmer The three threaded rods support-ing the dissociator with the nozzle via the three-legged

FIG 20 Two-dimensional distribution of the atomic hydro-gen component of the beam at z = 567mm before the nozzle-to-skimmer adjustment showing a disinct deviation from az-imuthal symmetry

plate (label 2 in Fig 1) allow one to vary the position ofthe nozzle relative to that of the skimmer while the sourceis running This possibility has been used to find a nozzleposition which results in an azimuthally symmetric distri-bution The achieved symmetric distribution is shown inFig 21 and profiles of the atomic hadrogen component inthe beam measured in x and y direction at z = 567mmand z = 697mm are presented in Fig 22 Fits by Gaus-sian distributions to the data yield full widths at halfmaximum Γx = (736plusmn 043)mm Γy = (668plusmn 080)mmfor the distributions measured at z = 567mm and Γx =(669plusmn 022)mm Γy = (638plusmn 027)mm at 697mm

C Reduction of MoO3 by hydrogen

In addition to the compression tube and the QMS tech-nique a supplementary attempt was made to determinethe beam profile by exposing molybdenium trioxide (a

14

FIG 21 The distribution corresponding to that of Fig 20 af-ter nozzle-to-skimmer adjustment resulting in azimuthal sym-metry

0

20

40

60

80

0

20

40

60

80

2 4 6 8 10 12 14 16 18 20 22

0

20

40

60

80

6 8 10 12 14 16 18 20 22 24

0

20

40

60

80

QM

S si

gnal

s [a

u]

(a)

Z = 567 mm

Z = 697 mm

(c)

(b)

X [mm]

(d)

Y [mm]

FIG 22 Profiles of the atomic hydrogen component inthe beam measured with the QMS 567mm and 697mmbehind the last magnet

yellowish powder) on a glass plate to the beam The prin-ciple of this method is based on the reduction of MoO3 toa lower oxide of blue colour It first was used in the ex-periment to measure the magnetic moment of the hydro-gen atom by splitting of the beam in an inhomogeneousmagnetic field 61

This method is much simpler than the time-consumingmeasurements described in Secs VA and VB It givesqualitative results as presented in Fig 23 A quantitativeanalysis however requires development of the measuringtechnique (eg preparation of glass plates study of theoptimum exposure time digital image processing)

FIG 23 Photo of the glass plate covered with molybdenumtrioxide MoO3 exposed to the atomic hydrogen beam

D Summary of the profile measurements

Table III summarizes results of the measurements ofthe ABS beam profile with the compression-tube and theQMS setup The larger errors of the widths measuredwith the QMS are due to the lack of measurements withthe dissociator switched off and the necessity to estimatethe background signal from the existing data Within theerrors the measured widths do not show a dependenceon the distance from the last magnet This facilitatesto position the feeding tube of the storage cell in a widerange of a distances to the last magnet The average val-ues Γx = (638 plusmn 060)mm and Γy = (684 plusmn 033)mmagree within the errors and yield a common width ofΓxy = (673 plusmn 029)mm The two-dimensional Gaus-sian distribution of this width allows one to estimate thefraction η of the beam intensity injected into the com-pression tube or a feeding tube For a tube of 10mmdiameter η = 078plusmn 003 comparable with η asymp 07 givenin Sec VA

TABLE III Dimensions (FWHM) of the atomic hydrogenbeam measured with the compression tube (CT) and thecrossed-beam quadrupole mass spectrometer (QMS) at dis-tances z to the last magnet along perpendicular directions xand y

z[mm] Γx[mm] Γy[mm]

CT 300 642 plusmn 009 699plusmn 006

CT 337 627 plusmn 008 658plusmn 008

QMS 567 736 plusmn 043 668plusmn 080

QMS 697 669 plusmn 022 638plusmn 027

15

VI DEGREE OF DISSOCIATION

Besides the intensity of the atomic beam it is impor-tant to determine the molecular fraction in the beamMolecules injected into the feeding tube reduce the po-larization of the target gas

A Measurements with crossed-beam QMS

In addition to the data on the profile of the atomichydrogen beam (Sec VB) data on the distributions ofmolecular hydrogen in the beam were taken too at thepositions z = 567mm and 697mm behind the last mag-net The relation between the degree of dissociation andthe QMS signals by the atomic and molecular beam com-ponent was given above by Eq (6) The coefficient kv =vmva however is chosen here under the assumptionthat the average velocity of the atoms is determined bythe nozzle temperature of 65K and that of the moleculesby scattering and recombination on the ABS chamberwalls at 290K This yields kv =

radic

2 middot 65290 = 067 ingood agreement with Ref13 where this coefficient wasdetermined by the measured velocity distributions undersimilar conditionsThe measured profiles of the atomic fraction (identical

to those of Fig 21) those of the molecular fraction andthose of the degree of dissociation deduced from Eq (6)are collected in Fig 24As it is seen from the figure the distribution of the

degree of dissociation shows a dip around the central linedue to the higher density of molecular hydrogen originat-ing from the nozzle The mean value in an aperture of

0

20

40

60

80

100

120

2 4 6 8 10 12 14 16 18 20 22

0

20

40

60

80

100

120

6 8 10 12 14 16 18 20 22 24

QM

S si

gnal

s [a

u]

(a)

(b)

(c)

00

02

04

06

08

10

12

Deg

ree

of d

isso

ciat

ion

()

X [mm]

Y [mm]

(d)

00

02

04

06

08

10

12

Z = 567 mm

Z = 697 mm

FIG 24 Spatial distributions of H1 (bull)H2 () and degree ofdissociation () averaged over 3mm wide bands in the xz andyz planes respectively (here the z-axis is the geometrical axisof the ABS)

10mm diameter results as α = 095 plusmn 004

B Measurements with the Lamb-shift polarimeter

A cup in the quench chamber of the Lamb-shift po-larimeter (LSP) described in Ref8 allows one to mea-sure the currents Icup(H1) and Icup(H2) of the H+

1 andH+

2 ions extracted from the ionizer and separated bythe Wien filter with the cesium evaporation and the spinfilter switched off The relation between the degree ofdissociation α and the measured currents is

α =Icup(H1)minus r1

r2Icup(H2)

Icup(H1)minus r1r2Icup(H2) + 2kv

r2Icup(H2)

(10)

Among the three coefficients kv = 067 as for the mea-surement with the QMS For the electron energy of about100keV the ratio r1 of dissociative to non-dissociativeionization of H2 is8

r1 =σ(H2 rarr 2H+

1 )

σ(H2 rarr H+2 )

= 0095plusmn 0008 (11)

and the ratio between the ionization cross sections is857

r2 =σion(H2)

σion(H1)= 17plusmn 01 (12)

At the standard operation parameters of the source(Sec IV) the measured currents are Icup(H1) = (125 plusmn05) nA and Icup(H2) = (64 plusmn 01) nA yielding α =(096plusmn004) in excellent agreement with the value result-ing from the measurements with the QMS (Sec VIA)

VII BEAM POLARIZATION

The Lamb-shift polarimeter was designed built andtested at Universitat zu Koln7 It was used to measureand to optimize the polarization of the atomic hydrogenand deuterium beams delivered by the ABS Details arefound in Ref7The vector polarization pz for hydrogen is defined by

the relative hyperfine-state occupation numbers N(mI)

pz =N(+ 1

2 )minusN(minus 12 )

N(+ 12 ) +N(minus 1

2 ) (13)

for deuterium

pz =N(+1)minusN(minus1)

N(+1) +N(0) +N(minus1) (14)

Deuterium tensor polarization pzz is given by

pzz =N(+1) +N(minus1)minus 2 middotN(0)

N(+1) +N(0) +N(minus1) (15)

16

These polarizations can be derived from the measuredLyman-α peak strengths S by application of a number ofcorrection factors78

Typical Lyman α spectra measured with the polarizedhydrogen and deuterium beam from the ABS are shownin the Figs 25 and 26

MFT 2-3

200 300 400 5000

40

80

120

160

Magnetic field in the spinfilter [au]

PM

T si

gnal

[au

]

mI = -12

(a)

mI = +12

100 200 300 400 5000

40

80

120

160(b)

MFT 2-3WFT 1-3

mI = +12

mI = -12

FIG 25 Lyman-α spectra measured with the polarized hy-drogen beam (a) population change from state |2〉 to state|3〉 induced by the MFT unit (b) same as (a) with subsequentpopulation change from state |1〉 to state |3〉 induced by theWFT unit

MFT 3-4SFT 2-6

0

2

4

6

8

mI = -1m

I = 0

(a)

mI = +1

0

2

4

6

8(b)

MFT 3-4WFT 1-4 2-3

mI = +1 m

I = 0

mI = -1

mI = +1

mI = 0

mI = -1

mI = +1

mI = 0

mI = -1

0 1000 2000 3000

0

1

2

3

4 (c)

PMT

sign

al [a

u]

Magnetic field in the spinfilter [au]

WFT 1-4SFT 2-6

0 1000 2000 30000

2

4

6

8WFT 1-4SFT 3-5

(d)

FIG 26 Lyman-α spectra measured with the polarized deu-terium beam (a) and (b) vector polarization resulting fromsubsequent transitions MFT (3 rarr 4) and SFT (2 rarr 6) andWFT (1 rarr 4 2 rarr 3) respectively (c) and (d) tensor polar-ization resulting from subsequent transitions WFT (1 rarr 4)and SFT (2 rarr 6) and SFT (3 rarr 5) respectively

The polarization values for the hydrogen and the deu-terium beam derived from the Lyman-α peak-strengthratios with application of the necessary corrections aresummarized in Table IV

The vector polarization for hydrogen of the first linereflects the population of state |1〉 and state |2〉 accordingto the Eqs (3) and (13) The value of 091 deduced fromthe calculated transmission values is confirmed by themeasured one

TABLE IV The vector polarization pz of the hydrogen beamand the vector and the tensor polarization pzz of the deu-terium beam from the ABS measured with the Lamb-shiftpolarimeter

populated pz pzz

state(s)

Hydrogen |1〉 +089plusmn 001 -

|3〉 minus096plusmn 001 -

Deuterium |1〉+ |6〉 +088plusmn 001 +088plusmn 003

|3〉+ |4〉 minus091plusmn 001 +085plusmn 002

|3〉+ |6〉 +0005plusmn 0003 +090plusmn 001

|2〉+ |5〉 +0005plusmn 0003 minus171plusmn 003

VIII CONCLUSIONS AND OUTLOOK

In this paper we present the detailed description ofthe major components of the atomic beam source (ABS)for the polarized internal gas target of the magnet spec-trometer ANKE in COSY-Julich The ABS was builtfor the purpose of extending the physics program ofANKE from unpolarized and single-polarized investiga-tions with stored beams towards double-polarized exper-iments1 thus facilitating nuclear reaction studies involv-

ing ~p~p ~p~d ~d~p and ~d~d initial statesThe mechanical design took into account that at

ANKE the source has to be mounted vertically and trans-versely movable together with the transverse motion ofthe spectrometer magnet D2 The design of the system ofsextupole magnets took advantage of the developments inthe field of rare-earth permanent magnets (NdFeB) Ded-icated tools and methods were developed to determineand to optimize the source parameters ie intensity de-gree of dissociation and polarization Special emphasiswas put on the measurements of the spatial distributionsof the atomic and molecular beam near the focus wherethe feeding tube of the storage cell is located The ABShas been used in a number of investigations at ANKEthe commissioning effort to prepare the target for the usewith polarized H is described in Ref62 Performed stud-ies of the deuteron-charge exchange reaction are summa-rized in Ref6364 studies in near-threshold pion produc-tion are reported about in Ref65The ABS resides at the ANKE target position for a few

months per year only thus during the remaining timeit is used for other studies It had been observed thatthe nuclear polarization in recombined hydrogen is par-tially retained after recombination66 as well as evidencefor nuclear tensor polarization in recombined deuteriummolecules67 In order to investigate this recombinationprocess in more detail a special setup has been developedin the framework of an ISTC project68 and the recombi-nation process for different cell-wall coatings and differ-ent polarizations of the injected hydrogen or deuterium

17

atoms as function of cell-wall temperature strength ofthe magnetic holding field and gas pressure in the cell ispresently investigated69ndash71

Appendix A Preparation of Discharge Tubes and Nozzles

1 Tube Treatment

One end of the discharge tube is machined at a 45

angle while the other is kept flat Both ends are thenremelted and the tubes are tempered at 150 C Thetubes are further treated according to the procedure de-scribed in Ref72 which includes successive cleaning withacetone methanol distilled water and subsequent rins-ing by a 21 acid mixture of concentrated HF (40) andHCl (32) for 5 min The tubes are then flushed bydistilled water and dried

2 Nozzle Treatment

The nozzles are cleaned in an ultrasonic bath oftrichlorethylene acetone methanol and finally distilledwater all at 50 C Anodizing takes place in sulfuric acidto form a thin layer of Al2O3 as described in Ref72 Af-terwards they are immersed in distilled water for 30minat 95 C

ACKNOWLEDGMENTS

The authors want to thank OWB Schult Institut furKernphysik (IKP) Julich who initiated the polarizationprogram of ANKE Thanks go to the design office themechanical workshop and especially to WR Ermer allIKP Valuable advice was received from the PINTEX col-laboration at IUCF from the target group at HERMESespecially NKoch and from DToporkov BINP Novosi-birsk The support by VKoptev PNPI Gatchina whoregrettably passed away in January 2012 is gratefullyacknowledged Thanks go also to R Poprawe and col-leagues Fraunhofer-Institut fur Lasertechnik Aachenwhere the encapsulations of the magnets were laser-welded

1AKacharava F Rathmann and CWilkin Spin Physics fromCOSY to FAIR COSY Experiment Proposal No 152 (2005)Available under httparXivnucl-ex0511028

2WHaeberli in Proc 2nd Int Symp on Polarization Phenomenaof Nucleons Karlsruhe 1965 Eds P Huber and H SchopperExperientia Supplementum 12 64 (Birkhauser Verlag 1966)

3E Steffens and WHaeberli Rep Progr Phys 66 1887 (2003)4SBarsov et al Nucl Instr and Meth A 462 364 (2001)5RMaier Nucl Instr and Meth A 390 1 (1997)6KGrigoryev et al Proc 14th International Workshop on Po-larized Sources Targets and Polarimetry (PSTP 2011) 12-16 September 2011 StPetersburg Russia eds KGrigoryevPKravtsov and AVasilyev ISBN 978-5-86763-282-3 61 (2011)

7REngels et al Rev Sci Instrum 74 4607 (2003)8REngels et al Rev Sci Instrum 76 053305 (2005)9TWise et al Nucl Instr and Meth A 336 410 (1993)

10WADezarn et al Nucl Instr and Meth A 362 36 (1995)11T Rinckel et al Nucl Instr and Meth A 439 117 (2000)12F Stock et al Nucl Instr and Meth A 343 334 (1994)13ANass et al Nucl Instr and Meth A 505 633 (2003)14VDerenchuk et al Proc Conf Polarized Ion Sources and Polar-ized Gas Targets Madison WI 1993 Eds LWAnderson andWHaeberli AIP Conf Proc 293 72 (American Institute ofPhysics 1994)

15HOkamura et al see Ref14 p 8416KHatanaka et al Nucl Instr and Meth A 384 575 (1997)17Manufacturer Schiffer Metall- amp Vakuumtechnik 52428 JulichGermany

18Single-stage type RGS120 refrigerating capacity 120W at 80Kand 20W at 30K Leybold Vacuum GmbH 50968 Koln Ger-many

19Mini UHV gate valve series 010 VAT Germany GmbH 85630Grasbrunn Germany

20Type F3 fomblin oil Pfeiffer Vacuum GmbH 35614 Asslar Ger-many

21Model HU 1 Leybold Vacuum GmbH 50968 Koln Germany22Manufacturer SK Industriemodell GmbH 52531 Ubach-Palenberg Germany

23Type PFG 600 RF with automatic matchbox PFM 1500 A-INDHuttinger Elektronik GmbH 79110 Freiburg Germany

24WKorsch PhD Thesis Philipps Universitat Marburg (1990)25F Stock et al Int Workshop on Polarized Beams and PolarizedGas Targets Koeln Germany 1995 Eds HPaetz gen Schieckand L Sydow (World Scientific Publ Co 1996) p 260

26The first number denotes the outer diameter and the second onethe wall thickness

27Type Duran 8330 equivalent to Corning 7740 (Pyrex) SchottAG 55122 Mainz Germany

28Ultra-Kryomat RUL 80-D Lauda DrRWobser GmbH 97912Lauda-Konigshofen Germany

29ODU-Kontakt GmbH 84444 Muhldorf Germany30Handbook of Chemistry and Physics Ed RCEast (The Chem-ical Rubber Co 1973) p E-10

31NKoch and E Steffens Rev Sci Instrum 70 1631 (1999)32AVassiliev et al Petersburg Nuclear Physics Institute ReportNP-32-1997 No 2175 (1997)

33BLorentz Diploma Thesis Ruprecht-Karls-Universitat Heidel-berg (1993)

34ANass and E Steffens Nucl Instr and Meth A 598 653(2009)

35WHaeberli Ann Rev Nucl Sci 17 373 (1967)36AVassiliev et al Rev Sci Instr 71 3331 (2000)37WKubischta Proc Workshop on Polarized Gas Targetsfor Storage Rings Heidelberg 23-26 September 1991 EdsHGGaul E Steffens and K Zapfe (Max-Planck-Institut furKernphysik Heidelberg)

38The labeling of the hyperfine states as |1〉 = |mj = +12 mI =+12〉 |2〉 = | + 12minus12〉 |3〉 = | minus 12+12〉 and |4〉 =|minus12minus12〉 for hydrogen and |1〉 = |+12+1〉 |2〉 = |+12 0〉|3〉 = | + 12minus1〉 |4〉 = | minus 12minus1〉 |5〉 = | minus 12 0〉 and

11

CG

RVC200

H2

F

EV

EV

EV

EV

EV

MP

V1V2

EVR116

HC

V0

EV EVA016UDV136

HCCPLC

PS

220V AC

220V AC

Interface

24V DC

FIG 13 Source of calibrated molecular gas flow (V0 com-pression chamber V1 gas-storage chamber feeding V0 viathe needle valve UDV136a V2 chamber of calibrated volumeused to determine that of V1) The pressure in V1 is mea-sured by the capacitance gauge CG and is kept constant bythe dosing valve EVR116 with the gauge controller RVC200The whole setup including the evacuation elements can beoperated manually or by the programmable logic controllerPLC either within the ABS control system52 or as a separatesystem

a All the valves and the gauge controller are supplied by PfeifferVacuum GmbH Dndash35614 Asslar Germany (manufacturerBalzers AG Liechtenstein)

factor 2 takes into account that the same pressure is mea-sured in the hot-cathode gauge for 2 middotIin (H atomss) and1 middot Iin (H2 moleculess) For d = 10mm l = 100mmT = 290K and M = 2 for hydrogen pressure differences∆P on the order of 10minus4mbar are expected for atomichydrogen beam intensities in the order of 1017 atomssThe relation between Iin and ∆P for hydrogen has beendetermined experimentally with the use of a source ofcalibrated molecular hydrogen gas flow5859 depicted inFig 13 The measured dependence with a linear fit isshown in Fig 14 The calibration curve allows one to de-termine absolute values of Iin of hydrogen and deuteriumbeams The calibration for deuterium was deduced fromthe one for hydrogen by scaling with a factor 1

radic2 ac-

cording to Eq (7)

The dependences of Iin on the dissociator-operationparameters primary molecular hydrogen flow q(H2) noz-zle temperature Tn and dissociator power Pdiss have beenstudied to find the optimum values They are shown inthe Figs 15 16 and 17 respectively for different nozzle-orifice diameters The figures show that for the hydrogenbeam (states |1〉 and |2〉) with the standard operation pa-rameters qH2

= 11 mbar ls Tn = 70K Pdiss = 350Wand with a nozzle-orifice diameter of 23mm an inten-sity of Iin(H) = (75 plusmn 02) middot 1016 particless is achievedquite close to the earlier estimate from Eq (2) Besidesthe dominant atomic component of H atoms this value

04 06 08 10 12 14 16 18 20 2201

02

03

04

05

06

07

08

09

10

11

12

Gas

flow

from

the

stor

age

volu

me

[1017

ats

]

Pressure in compression volume [10-4 mbar]

FIG 14 Calibration curve for hydrogen used to deduce fromthe measured pressures the intensities of the hydrogen anddeuterium beam injected into the compression tube

00 05 10 15 20 2500

10

20

30

40

50

60

70

80 D = 20 mm D = 23 mm D = 25 mm

Bea

m in

tens

ity [1

016 a

ts]

Primary gas flow [mbar ls]

FIG 15 Intensity of the hydrogen beam (states |1〉 and |2〉)injected into the compression tube as function of the primarymolecular gas flow for different nozzle diameters D (nozzletemperature 60K dissociator power 300W)

includes small admixtures of H atoms in state |3〉 andmolecular hydrogen The first kind can be estimatedwith the use of the calculated transmissions (Sec II F)as 0017084 asymp 2 The amount of the second admix-ture has been measured as described below

For the deuterium beam (states |1〉 |2〉 and |3〉) theoptimization procedure gave an intensity of Iin(D) =(39 plusmn 02) middot 1016 particless achieved with q(D2) = 09mbar ls Tn = 65K and Pdiss = 300W slightly lowerthan those for hydrogen

12

50 60 70 80 90 100 110

40

50

60

70

80 D = 20 mm D = 23 mm D = 25 mm

Bea

m in

tens

ity [1

016 a

ts]

Nozzle temperature [K]

FIG 16 Intensity of the hydrogen beam (states |1〉 and |2〉)injected into the compression tube as function of the nozzletemperature for different nozzle diameters D (primary molec-ular gas flow 1mbar ls dissociator power 300W)

100 200 300 400 50010

20

30

40

50

60

70

80 D = 20 mm

Bea

m in

tens

ity [1

016 a

ts]

Dissociator power [W]

FIG 17 Intensity of the hydrogen beam (states |1〉 and |2〉)into the compression tube as function of the dissociator powerfor a nozzle diameter of 2 mm (nozzle temperature 60K pri-mary molecular gas flow 1mbar ls)

V HYDROGEN BEAM PROFILES

Beam profiles were measured at various positions atvarious positions behind the last sextupole magnet withthe use of

bull a compression tube of reduced dimensions (5mmdiameter)

bull a crossed-beam quadrupole mass spectrometer andbull a supplementary method of reduction of MoO3 by

hydrogen

A Measurements with the compression tube

For the determination of the beam dimensions at twopositions 300mm and 337mm behind the last magnetthe compression tube setup (Fig 12) was used makinguse of the possibility of axial movement by the z ma-nipulator and of that to install a narrower and shortercompression tube of 5mm diameter and 50mm lengthto enhance the spatial resolution The xy manipula-tor provided a lateral displacement of the compressiontube by plusmn10 mm in x and y direction The center co-ordinates of the geometrical axis of the source had beendetermined with the use of a bi-directional laser cen-tered inside the bore of the central support plate (seeFig 1) The relative intensity distributions in the xzand yz planes given by the measured pressure in thecompression volume are shown in Fig 18 Fits by Gaus-sian distributions to the data yield full widths at halfmaximum Γx = (642plusmn 009)mm Γy = (699plusmn 006)mmfor the distributions measured at z = 300mm and Γx =(627plusmn 008)mm Γy = (658plusmn 008)mm at 337mm

00

02

04

06

08

10

12

00

02

04

06

08

10

12

0 5 10 15 20 25

00

02

04

06

08

10

12

14

0 5 10 15 20 25

00

02

04

06

08

10

12

14

Pre

ssur

e in

the

com

pres

sion

tube

[au

]

Z = 300 mm

(a) (b)

(c) (d)

Z = 337 mm

X [mm]

Y [mm]

FIG 18 Cross sections of the beam profile in the mid-planemeasured with compression tube of 5mm diameter and 50mmlength Measurements in the xz-plane (a c) and yz-plane (bd) performed at two different positions z = 300mm (a b)and z = 337mm (c d) behind the last sextupole magnet ofthe ABS The shaded area represents position and dimensionsof the compression tube used in intensity measurements

The center of gravity of the measured profile definedas

rc =

sum

ij

radic

x2i + y2j middot P (xi yj)

sum

ij

P (xi yj) (8)

where xi and yj give the position of the compression-tubeaxis and P (xi yj) is the pressure measured in the com-

13

pression volume The resulting rc shows a deviation of012 mm from the geometrical axis of the source Further-more the data measured with the narrow compressiontube of 25mm radius can be used to derive the fractionof the beam entering the compression tube of 5mm ra-dius used in the intensity measurement of Sec IV Theratio

η =

rile25 mmsum

0P (xi yj)

rile10 mmsum

0P (xi yj)

(9)

where ri is the distance of the compression-tube axis tothe beam axis yields η asymp 07

B Measurements with the QMS

The beam-profile studies of Sec VA were extendedwith a setup utilizing a crossed-beam quadrupole massspectrometer (QMS) in the setup of Fig 19 Contraryto the measurements with the compression tube thosewith the QMS allow to separate the atomic and molec-ular fractions in the beam A 2mm diameter aperturewas installed at the entrance of the sensitive volume of

FIG 19 Setup for the measurements of the beam profilewith the QMS (1) xy-table enabling two-dimensional dis-placement of the entrance window of the QMS against thegeometrical axis of the ABS (2) the QMS (3) manually op-erated beam shutter The beam dump is an axially mountedcryo pump

the QMS to improve the resolution compared with thatachieved by the compression tube of 5mm diameter usedin measurements of the preceding section The layout ofthe setup presented in Fig 19 shows that in the presentcase the profile could not be measured at a distance ofz = 300mm to the last magnet Instead measurementswere performed at z = 567mm and with installation ofan extension tube at z = 697mm The xy manipulatorenabled displacements of the aperture axis from the geo-metrical axis of the source in any direction within limitsset by the bore diameter of the xy manipulatorThe first measured distribution of the atomic hydro-

gen (Fig 20) showed a distinct deviation from azimuthalsymmetry indicating an insufficient relative alignment ofnozzle and skimmer The three threaded rods support-ing the dissociator with the nozzle via the three-legged

FIG 20 Two-dimensional distribution of the atomic hydro-gen component of the beam at z = 567mm before the nozzle-to-skimmer adjustment showing a disinct deviation from az-imuthal symmetry

plate (label 2 in Fig 1) allow one to vary the position ofthe nozzle relative to that of the skimmer while the sourceis running This possibility has been used to find a nozzleposition which results in an azimuthally symmetric distri-bution The achieved symmetric distribution is shown inFig 21 and profiles of the atomic hadrogen component inthe beam measured in x and y direction at z = 567mmand z = 697mm are presented in Fig 22 Fits by Gaus-sian distributions to the data yield full widths at halfmaximum Γx = (736plusmn 043)mm Γy = (668plusmn 080)mmfor the distributions measured at z = 567mm and Γx =(669plusmn 022)mm Γy = (638plusmn 027)mm at 697mm

C Reduction of MoO3 by hydrogen

In addition to the compression tube and the QMS tech-nique a supplementary attempt was made to determinethe beam profile by exposing molybdenium trioxide (a

14

FIG 21 The distribution corresponding to that of Fig 20 af-ter nozzle-to-skimmer adjustment resulting in azimuthal sym-metry

0

20

40

60

80

0

20

40

60

80

2 4 6 8 10 12 14 16 18 20 22

0

20

40

60

80

6 8 10 12 14 16 18 20 22 24

0

20

40

60

80

QM

S si

gnal

s [a

u]

(a)

Z = 567 mm

Z = 697 mm

(c)

(b)

X [mm]

(d)

Y [mm]

FIG 22 Profiles of the atomic hydrogen component inthe beam measured with the QMS 567mm and 697mmbehind the last magnet

yellowish powder) on a glass plate to the beam The prin-ciple of this method is based on the reduction of MoO3 toa lower oxide of blue colour It first was used in the ex-periment to measure the magnetic moment of the hydro-gen atom by splitting of the beam in an inhomogeneousmagnetic field 61

This method is much simpler than the time-consumingmeasurements described in Secs VA and VB It givesqualitative results as presented in Fig 23 A quantitativeanalysis however requires development of the measuringtechnique (eg preparation of glass plates study of theoptimum exposure time digital image processing)

FIG 23 Photo of the glass plate covered with molybdenumtrioxide MoO3 exposed to the atomic hydrogen beam

D Summary of the profile measurements

Table III summarizes results of the measurements ofthe ABS beam profile with the compression-tube and theQMS setup The larger errors of the widths measuredwith the QMS are due to the lack of measurements withthe dissociator switched off and the necessity to estimatethe background signal from the existing data Within theerrors the measured widths do not show a dependenceon the distance from the last magnet This facilitatesto position the feeding tube of the storage cell in a widerange of a distances to the last magnet The average val-ues Γx = (638 plusmn 060)mm and Γy = (684 plusmn 033)mmagree within the errors and yield a common width ofΓxy = (673 plusmn 029)mm The two-dimensional Gaus-sian distribution of this width allows one to estimate thefraction η of the beam intensity injected into the com-pression tube or a feeding tube For a tube of 10mmdiameter η = 078plusmn 003 comparable with η asymp 07 givenin Sec VA

TABLE III Dimensions (FWHM) of the atomic hydrogenbeam measured with the compression tube (CT) and thecrossed-beam quadrupole mass spectrometer (QMS) at dis-tances z to the last magnet along perpendicular directions xand y

z[mm] Γx[mm] Γy[mm]

CT 300 642 plusmn 009 699plusmn 006

CT 337 627 plusmn 008 658plusmn 008

QMS 567 736 plusmn 043 668plusmn 080

QMS 697 669 plusmn 022 638plusmn 027

15

VI DEGREE OF DISSOCIATION

Besides the intensity of the atomic beam it is impor-tant to determine the molecular fraction in the beamMolecules injected into the feeding tube reduce the po-larization of the target gas

A Measurements with crossed-beam QMS

In addition to the data on the profile of the atomichydrogen beam (Sec VB) data on the distributions ofmolecular hydrogen in the beam were taken too at thepositions z = 567mm and 697mm behind the last mag-net The relation between the degree of dissociation andthe QMS signals by the atomic and molecular beam com-ponent was given above by Eq (6) The coefficient kv =vmva however is chosen here under the assumptionthat the average velocity of the atoms is determined bythe nozzle temperature of 65K and that of the moleculesby scattering and recombination on the ABS chamberwalls at 290K This yields kv =

radic

2 middot 65290 = 067 ingood agreement with Ref13 where this coefficient wasdetermined by the measured velocity distributions undersimilar conditionsThe measured profiles of the atomic fraction (identical

to those of Fig 21) those of the molecular fraction andthose of the degree of dissociation deduced from Eq (6)are collected in Fig 24As it is seen from the figure the distribution of the

degree of dissociation shows a dip around the central linedue to the higher density of molecular hydrogen originat-ing from the nozzle The mean value in an aperture of

0

20

40

60

80

100

120

2 4 6 8 10 12 14 16 18 20 22

0

20

40

60

80

100

120

6 8 10 12 14 16 18 20 22 24

QM

S si

gnal

s [a

u]

(a)

(b)

(c)

00

02

04

06

08

10

12

Deg

ree

of d

isso

ciat

ion

()

X [mm]

Y [mm]

(d)

00

02

04

06

08

10

12

Z = 567 mm

Z = 697 mm

FIG 24 Spatial distributions of H1 (bull)H2 () and degree ofdissociation () averaged over 3mm wide bands in the xz andyz planes respectively (here the z-axis is the geometrical axisof the ABS)

10mm diameter results as α = 095 plusmn 004

B Measurements with the Lamb-shift polarimeter

A cup in the quench chamber of the Lamb-shift po-larimeter (LSP) described in Ref8 allows one to mea-sure the currents Icup(H1) and Icup(H2) of the H+

1 andH+

2 ions extracted from the ionizer and separated bythe Wien filter with the cesium evaporation and the spinfilter switched off The relation between the degree ofdissociation α and the measured currents is

α =Icup(H1)minus r1

r2Icup(H2)

Icup(H1)minus r1r2Icup(H2) + 2kv

r2Icup(H2)

(10)

Among the three coefficients kv = 067 as for the mea-surement with the QMS For the electron energy of about100keV the ratio r1 of dissociative to non-dissociativeionization of H2 is8

r1 =σ(H2 rarr 2H+

1 )

σ(H2 rarr H+2 )

= 0095plusmn 0008 (11)

and the ratio between the ionization cross sections is857

r2 =σion(H2)

σion(H1)= 17plusmn 01 (12)

At the standard operation parameters of the source(Sec IV) the measured currents are Icup(H1) = (125 plusmn05) nA and Icup(H2) = (64 plusmn 01) nA yielding α =(096plusmn004) in excellent agreement with the value result-ing from the measurements with the QMS (Sec VIA)

VII BEAM POLARIZATION

The Lamb-shift polarimeter was designed built andtested at Universitat zu Koln7 It was used to measureand to optimize the polarization of the atomic hydrogenand deuterium beams delivered by the ABS Details arefound in Ref7The vector polarization pz for hydrogen is defined by

the relative hyperfine-state occupation numbers N(mI)

pz =N(+ 1

2 )minusN(minus 12 )

N(+ 12 ) +N(minus 1

2 ) (13)

for deuterium

pz =N(+1)minusN(minus1)

N(+1) +N(0) +N(minus1) (14)

Deuterium tensor polarization pzz is given by

pzz =N(+1) +N(minus1)minus 2 middotN(0)

N(+1) +N(0) +N(minus1) (15)

16

These polarizations can be derived from the measuredLyman-α peak strengths S by application of a number ofcorrection factors78

Typical Lyman α spectra measured with the polarizedhydrogen and deuterium beam from the ABS are shownin the Figs 25 and 26

MFT 2-3

200 300 400 5000

40

80

120

160

Magnetic field in the spinfilter [au]

PM

T si

gnal

[au

]

mI = -12

(a)

mI = +12

100 200 300 400 5000

40

80

120

160(b)

MFT 2-3WFT 1-3

mI = +12

mI = -12

FIG 25 Lyman-α spectra measured with the polarized hy-drogen beam (a) population change from state |2〉 to state|3〉 induced by the MFT unit (b) same as (a) with subsequentpopulation change from state |1〉 to state |3〉 induced by theWFT unit

MFT 3-4SFT 2-6

0

2

4

6

8

mI = -1m

I = 0

(a)

mI = +1

0

2

4

6

8(b)

MFT 3-4WFT 1-4 2-3

mI = +1 m

I = 0

mI = -1

mI = +1

mI = 0

mI = -1

mI = +1

mI = 0

mI = -1

0 1000 2000 3000

0

1

2

3

4 (c)

PMT

sign

al [a

u]

Magnetic field in the spinfilter [au]

WFT 1-4SFT 2-6

0 1000 2000 30000

2

4

6

8WFT 1-4SFT 3-5

(d)

FIG 26 Lyman-α spectra measured with the polarized deu-terium beam (a) and (b) vector polarization resulting fromsubsequent transitions MFT (3 rarr 4) and SFT (2 rarr 6) andWFT (1 rarr 4 2 rarr 3) respectively (c) and (d) tensor polar-ization resulting from subsequent transitions WFT (1 rarr 4)and SFT (2 rarr 6) and SFT (3 rarr 5) respectively

The polarization values for the hydrogen and the deu-terium beam derived from the Lyman-α peak-strengthratios with application of the necessary corrections aresummarized in Table IV

The vector polarization for hydrogen of the first linereflects the population of state |1〉 and state |2〉 accordingto the Eqs (3) and (13) The value of 091 deduced fromthe calculated transmission values is confirmed by themeasured one

TABLE IV The vector polarization pz of the hydrogen beamand the vector and the tensor polarization pzz of the deu-terium beam from the ABS measured with the Lamb-shiftpolarimeter

populated pz pzz

state(s)

Hydrogen |1〉 +089plusmn 001 -

|3〉 minus096plusmn 001 -

Deuterium |1〉+ |6〉 +088plusmn 001 +088plusmn 003

|3〉+ |4〉 minus091plusmn 001 +085plusmn 002

|3〉+ |6〉 +0005plusmn 0003 +090plusmn 001

|2〉+ |5〉 +0005plusmn 0003 minus171plusmn 003

VIII CONCLUSIONS AND OUTLOOK

In this paper we present the detailed description ofthe major components of the atomic beam source (ABS)for the polarized internal gas target of the magnet spec-trometer ANKE in COSY-Julich The ABS was builtfor the purpose of extending the physics program ofANKE from unpolarized and single-polarized investiga-tions with stored beams towards double-polarized exper-iments1 thus facilitating nuclear reaction studies involv-

ing ~p~p ~p~d ~d~p and ~d~d initial statesThe mechanical design took into account that at

ANKE the source has to be mounted vertically and trans-versely movable together with the transverse motion ofthe spectrometer magnet D2 The design of the system ofsextupole magnets took advantage of the developments inthe field of rare-earth permanent magnets (NdFeB) Ded-icated tools and methods were developed to determineand to optimize the source parameters ie intensity de-gree of dissociation and polarization Special emphasiswas put on the measurements of the spatial distributionsof the atomic and molecular beam near the focus wherethe feeding tube of the storage cell is located The ABShas been used in a number of investigations at ANKEthe commissioning effort to prepare the target for the usewith polarized H is described in Ref62 Performed stud-ies of the deuteron-charge exchange reaction are summa-rized in Ref6364 studies in near-threshold pion produc-tion are reported about in Ref65The ABS resides at the ANKE target position for a few

months per year only thus during the remaining timeit is used for other studies It had been observed thatthe nuclear polarization in recombined hydrogen is par-tially retained after recombination66 as well as evidencefor nuclear tensor polarization in recombined deuteriummolecules67 In order to investigate this recombinationprocess in more detail a special setup has been developedin the framework of an ISTC project68 and the recombi-nation process for different cell-wall coatings and differ-ent polarizations of the injected hydrogen or deuterium

17

atoms as function of cell-wall temperature strength ofthe magnetic holding field and gas pressure in the cell ispresently investigated69ndash71

Appendix A Preparation of Discharge Tubes and Nozzles

1 Tube Treatment

One end of the discharge tube is machined at a 45

angle while the other is kept flat Both ends are thenremelted and the tubes are tempered at 150 C Thetubes are further treated according to the procedure de-scribed in Ref72 which includes successive cleaning withacetone methanol distilled water and subsequent rins-ing by a 21 acid mixture of concentrated HF (40) andHCl (32) for 5 min The tubes are then flushed bydistilled water and dried

2 Nozzle Treatment

The nozzles are cleaned in an ultrasonic bath oftrichlorethylene acetone methanol and finally distilledwater all at 50 C Anodizing takes place in sulfuric acidto form a thin layer of Al2O3 as described in Ref72 Af-terwards they are immersed in distilled water for 30minat 95 C

ACKNOWLEDGMENTS

The authors want to thank OWB Schult Institut furKernphysik (IKP) Julich who initiated the polarizationprogram of ANKE Thanks go to the design office themechanical workshop and especially to WR Ermer allIKP Valuable advice was received from the PINTEX col-laboration at IUCF from the target group at HERMESespecially NKoch and from DToporkov BINP Novosi-birsk The support by VKoptev PNPI Gatchina whoregrettably passed away in January 2012 is gratefullyacknowledged Thanks go also to R Poprawe and col-leagues Fraunhofer-Institut fur Lasertechnik Aachenwhere the encapsulations of the magnets were laser-welded

1AKacharava F Rathmann and CWilkin Spin Physics fromCOSY to FAIR COSY Experiment Proposal No 152 (2005)Available under httparXivnucl-ex0511028

2WHaeberli in Proc 2nd Int Symp on Polarization Phenomenaof Nucleons Karlsruhe 1965 Eds P Huber and H SchopperExperientia Supplementum 12 64 (Birkhauser Verlag 1966)

3E Steffens and WHaeberli Rep Progr Phys 66 1887 (2003)4SBarsov et al Nucl Instr and Meth A 462 364 (2001)5RMaier Nucl Instr and Meth A 390 1 (1997)6KGrigoryev et al Proc 14th International Workshop on Po-larized Sources Targets and Polarimetry (PSTP 2011) 12-16 September 2011 StPetersburg Russia eds KGrigoryevPKravtsov and AVasilyev ISBN 978-5-86763-282-3 61 (2011)

7REngels et al Rev Sci Instrum 74 4607 (2003)8REngels et al Rev Sci Instrum 76 053305 (2005)9TWise et al Nucl Instr and Meth A 336 410 (1993)

10WADezarn et al Nucl Instr and Meth A 362 36 (1995)11T Rinckel et al Nucl Instr and Meth A 439 117 (2000)12F Stock et al Nucl Instr and Meth A 343 334 (1994)13ANass et al Nucl Instr and Meth A 505 633 (2003)14VDerenchuk et al Proc Conf Polarized Ion Sources and Polar-ized Gas Targets Madison WI 1993 Eds LWAnderson andWHaeberli AIP Conf Proc 293 72 (American Institute ofPhysics 1994)

15HOkamura et al see Ref14 p 8416KHatanaka et al Nucl Instr and Meth A 384 575 (1997)17Manufacturer Schiffer Metall- amp Vakuumtechnik 52428 JulichGermany

18Single-stage type RGS120 refrigerating capacity 120W at 80Kand 20W at 30K Leybold Vacuum GmbH 50968 Koln Ger-many

19Mini UHV gate valve series 010 VAT Germany GmbH 85630Grasbrunn Germany

20Type F3 fomblin oil Pfeiffer Vacuum GmbH 35614 Asslar Ger-many

21Model HU 1 Leybold Vacuum GmbH 50968 Koln Germany22Manufacturer SK Industriemodell GmbH 52531 Ubach-Palenberg Germany

23Type PFG 600 RF with automatic matchbox PFM 1500 A-INDHuttinger Elektronik GmbH 79110 Freiburg Germany

24WKorsch PhD Thesis Philipps Universitat Marburg (1990)25F Stock et al Int Workshop on Polarized Beams and PolarizedGas Targets Koeln Germany 1995 Eds HPaetz gen Schieckand L Sydow (World Scientific Publ Co 1996) p 260

26The first number denotes the outer diameter and the second onethe wall thickness

27Type Duran 8330 equivalent to Corning 7740 (Pyrex) SchottAG 55122 Mainz Germany

28Ultra-Kryomat RUL 80-D Lauda DrRWobser GmbH 97912Lauda-Konigshofen Germany

29ODU-Kontakt GmbH 84444 Muhldorf Germany30Handbook of Chemistry and Physics Ed RCEast (The Chem-ical Rubber Co 1973) p E-10

31NKoch and E Steffens Rev Sci Instrum 70 1631 (1999)32AVassiliev et al Petersburg Nuclear Physics Institute ReportNP-32-1997 No 2175 (1997)

33BLorentz Diploma Thesis Ruprecht-Karls-Universitat Heidel-berg (1993)

34ANass and E Steffens Nucl Instr and Meth A 598 653(2009)

35WHaeberli Ann Rev Nucl Sci 17 373 (1967)36AVassiliev et al Rev Sci Instr 71 3331 (2000)37WKubischta Proc Workshop on Polarized Gas Targetsfor Storage Rings Heidelberg 23-26 September 1991 EdsHGGaul E Steffens and K Zapfe (Max-Planck-Institut furKernphysik Heidelberg)

38The labeling of the hyperfine states as |1〉 = |mj = +12 mI =+12〉 |2〉 = | + 12minus12〉 |3〉 = | minus 12+12〉 and |4〉 =|minus12minus12〉 for hydrogen and |1〉 = |+12+1〉 |2〉 = |+12 0〉|3〉 = | + 12minus1〉 |4〉 = | minus 12minus1〉 |5〉 = | minus 12 0〉 and

12

50 60 70 80 90 100 110

40

50

60

70

80 D = 20 mm D = 23 mm D = 25 mm

Bea

m in

tens

ity [1

016 a

ts]

Nozzle temperature [K]

FIG 16 Intensity of the hydrogen beam (states |1〉 and |2〉)injected into the compression tube as function of the nozzletemperature for different nozzle diameters D (primary molec-ular gas flow 1mbar ls dissociator power 300W)

100 200 300 400 50010

20

30

40

50

60

70

80 D = 20 mm

Bea

m in

tens

ity [1

016 a

ts]

Dissociator power [W]

FIG 17 Intensity of the hydrogen beam (states |1〉 and |2〉)into the compression tube as function of the dissociator powerfor a nozzle diameter of 2 mm (nozzle temperature 60K pri-mary molecular gas flow 1mbar ls)

V HYDROGEN BEAM PROFILES

Beam profiles were measured at various positions atvarious positions behind the last sextupole magnet withthe use of

bull a compression tube of reduced dimensions (5mmdiameter)

bull a crossed-beam quadrupole mass spectrometer andbull a supplementary method of reduction of MoO3 by

hydrogen

A Measurements with the compression tube

For the determination of the beam dimensions at twopositions 300mm and 337mm behind the last magnetthe compression tube setup (Fig 12) was used makinguse of the possibility of axial movement by the z ma-nipulator and of that to install a narrower and shortercompression tube of 5mm diameter and 50mm lengthto enhance the spatial resolution The xy manipula-tor provided a lateral displacement of the compressiontube by plusmn10 mm in x and y direction The center co-ordinates of the geometrical axis of the source had beendetermined with the use of a bi-directional laser cen-tered inside the bore of the central support plate (seeFig 1) The relative intensity distributions in the xzand yz planes given by the measured pressure in thecompression volume are shown in Fig 18 Fits by Gaus-sian distributions to the data yield full widths at halfmaximum Γx = (642plusmn 009)mm Γy = (699plusmn 006)mmfor the distributions measured at z = 300mm and Γx =(627plusmn 008)mm Γy = (658plusmn 008)mm at 337mm

00

02

04

06

08

10

12

00

02

04

06

08

10

12

0 5 10 15 20 25

00

02

04

06

08

10

12

14

0 5 10 15 20 25

00

02

04

06

08

10

12

14

Pre

ssur

e in

the

com

pres

sion

tube

[au

]

Z = 300 mm

(a) (b)

(c) (d)

Z = 337 mm

X [mm]

Y [mm]

FIG 18 Cross sections of the beam profile in the mid-planemeasured with compression tube of 5mm diameter and 50mmlength Measurements in the xz-plane (a c) and yz-plane (bd) performed at two different positions z = 300mm (a b)and z = 337mm (c d) behind the last sextupole magnet ofthe ABS The shaded area represents position and dimensionsof the compression tube used in intensity measurements

The center of gravity of the measured profile definedas

rc =

sum

ij

radic

x2i + y2j middot P (xi yj)

sum

ij

P (xi yj) (8)

where xi and yj give the position of the compression-tubeaxis and P (xi yj) is the pressure measured in the com-

13

pression volume The resulting rc shows a deviation of012 mm from the geometrical axis of the source Further-more the data measured with the narrow compressiontube of 25mm radius can be used to derive the fractionof the beam entering the compression tube of 5mm ra-dius used in the intensity measurement of Sec IV Theratio

η =

rile25 mmsum

0P (xi yj)

rile10 mmsum

0P (xi yj)

(9)

where ri is the distance of the compression-tube axis tothe beam axis yields η asymp 07

B Measurements with the QMS

The beam-profile studies of Sec VA were extendedwith a setup utilizing a crossed-beam quadrupole massspectrometer (QMS) in the setup of Fig 19 Contraryto the measurements with the compression tube thosewith the QMS allow to separate the atomic and molec-ular fractions in the beam A 2mm diameter aperturewas installed at the entrance of the sensitive volume of

FIG 19 Setup for the measurements of the beam profilewith the QMS (1) xy-table enabling two-dimensional dis-placement of the entrance window of the QMS against thegeometrical axis of the ABS (2) the QMS (3) manually op-erated beam shutter The beam dump is an axially mountedcryo pump

the QMS to improve the resolution compared with thatachieved by the compression tube of 5mm diameter usedin measurements of the preceding section The layout ofthe setup presented in Fig 19 shows that in the presentcase the profile could not be measured at a distance ofz = 300mm to the last magnet Instead measurementswere performed at z = 567mm and with installation ofan extension tube at z = 697mm The xy manipulatorenabled displacements of the aperture axis from the geo-metrical axis of the source in any direction within limitsset by the bore diameter of the xy manipulatorThe first measured distribution of the atomic hydro-

gen (Fig 20) showed a distinct deviation from azimuthalsymmetry indicating an insufficient relative alignment ofnozzle and skimmer The three threaded rods support-ing the dissociator with the nozzle via the three-legged

FIG 20 Two-dimensional distribution of the atomic hydro-gen component of the beam at z = 567mm before the nozzle-to-skimmer adjustment showing a disinct deviation from az-imuthal symmetry

plate (label 2 in Fig 1) allow one to vary the position ofthe nozzle relative to that of the skimmer while the sourceis running This possibility has been used to find a nozzleposition which results in an azimuthally symmetric distri-bution The achieved symmetric distribution is shown inFig 21 and profiles of the atomic hadrogen component inthe beam measured in x and y direction at z = 567mmand z = 697mm are presented in Fig 22 Fits by Gaus-sian distributions to the data yield full widths at halfmaximum Γx = (736plusmn 043)mm Γy = (668plusmn 080)mmfor the distributions measured at z = 567mm and Γx =(669plusmn 022)mm Γy = (638plusmn 027)mm at 697mm

C Reduction of MoO3 by hydrogen

In addition to the compression tube and the QMS tech-nique a supplementary attempt was made to determinethe beam profile by exposing molybdenium trioxide (a

14

FIG 21 The distribution corresponding to that of Fig 20 af-ter nozzle-to-skimmer adjustment resulting in azimuthal sym-metry

0

20

40

60

80

0

20

40

60

80

2 4 6 8 10 12 14 16 18 20 22

0

20

40

60

80

6 8 10 12 14 16 18 20 22 24

0

20

40

60

80

QM

S si

gnal

s [a

u]

(a)

Z = 567 mm

Z = 697 mm

(c)

(b)

X [mm]

(d)

Y [mm]

FIG 22 Profiles of the atomic hydrogen component inthe beam measured with the QMS 567mm and 697mmbehind the last magnet

yellowish powder) on a glass plate to the beam The prin-ciple of this method is based on the reduction of MoO3 toa lower oxide of blue colour It first was used in the ex-periment to measure the magnetic moment of the hydro-gen atom by splitting of the beam in an inhomogeneousmagnetic field 61

This method is much simpler than the time-consumingmeasurements described in Secs VA and VB It givesqualitative results as presented in Fig 23 A quantitativeanalysis however requires development of the measuringtechnique (eg preparation of glass plates study of theoptimum exposure time digital image processing)

FIG 23 Photo of the glass plate covered with molybdenumtrioxide MoO3 exposed to the atomic hydrogen beam

D Summary of the profile measurements

Table III summarizes results of the measurements ofthe ABS beam profile with the compression-tube and theQMS setup The larger errors of the widths measuredwith the QMS are due to the lack of measurements withthe dissociator switched off and the necessity to estimatethe background signal from the existing data Within theerrors the measured widths do not show a dependenceon the distance from the last magnet This facilitatesto position the feeding tube of the storage cell in a widerange of a distances to the last magnet The average val-ues Γx = (638 plusmn 060)mm and Γy = (684 plusmn 033)mmagree within the errors and yield a common width ofΓxy = (673 plusmn 029)mm The two-dimensional Gaus-sian distribution of this width allows one to estimate thefraction η of the beam intensity injected into the com-pression tube or a feeding tube For a tube of 10mmdiameter η = 078plusmn 003 comparable with η asymp 07 givenin Sec VA

TABLE III Dimensions (FWHM) of the atomic hydrogenbeam measured with the compression tube (CT) and thecrossed-beam quadrupole mass spectrometer (QMS) at dis-tances z to the last magnet along perpendicular directions xand y

z[mm] Γx[mm] Γy[mm]

CT 300 642 plusmn 009 699plusmn 006

CT 337 627 plusmn 008 658plusmn 008

QMS 567 736 plusmn 043 668plusmn 080

QMS 697 669 plusmn 022 638plusmn 027

15

VI DEGREE OF DISSOCIATION

Besides the intensity of the atomic beam it is impor-tant to determine the molecular fraction in the beamMolecules injected into the feeding tube reduce the po-larization of the target gas

A Measurements with crossed-beam QMS

In addition to the data on the profile of the atomichydrogen beam (Sec VB) data on the distributions ofmolecular hydrogen in the beam were taken too at thepositions z = 567mm and 697mm behind the last mag-net The relation between the degree of dissociation andthe QMS signals by the atomic and molecular beam com-ponent was given above by Eq (6) The coefficient kv =vmva however is chosen here under the assumptionthat the average velocity of the atoms is determined bythe nozzle temperature of 65K and that of the moleculesby scattering and recombination on the ABS chamberwalls at 290K This yields kv =

radic

2 middot 65290 = 067 ingood agreement with Ref13 where this coefficient wasdetermined by the measured velocity distributions undersimilar conditionsThe measured profiles of the atomic fraction (identical

to those of Fig 21) those of the molecular fraction andthose of the degree of dissociation deduced from Eq (6)are collected in Fig 24As it is seen from the figure the distribution of the

degree of dissociation shows a dip around the central linedue to the higher density of molecular hydrogen originat-ing from the nozzle The mean value in an aperture of

0

20

40

60

80

100

120

2 4 6 8 10 12 14 16 18 20 22

0

20

40

60

80

100

120

6 8 10 12 14 16 18 20 22 24

QM

S si

gnal

s [a

u]

(a)

(b)

(c)

00

02

04

06

08

10

12

Deg

ree

of d

isso

ciat

ion

()

X [mm]

Y [mm]

(d)

00

02

04

06

08

10

12

Z = 567 mm

Z = 697 mm

FIG 24 Spatial distributions of H1 (bull)H2 () and degree ofdissociation () averaged over 3mm wide bands in the xz andyz planes respectively (here the z-axis is the geometrical axisof the ABS)

10mm diameter results as α = 095 plusmn 004

B Measurements with the Lamb-shift polarimeter

A cup in the quench chamber of the Lamb-shift po-larimeter (LSP) described in Ref8 allows one to mea-sure the currents Icup(H1) and Icup(H2) of the H+

1 andH+

2 ions extracted from the ionizer and separated bythe Wien filter with the cesium evaporation and the spinfilter switched off The relation between the degree ofdissociation α and the measured currents is

α =Icup(H1)minus r1

r2Icup(H2)

Icup(H1)minus r1r2Icup(H2) + 2kv

r2Icup(H2)

(10)

Among the three coefficients kv = 067 as for the mea-surement with the QMS For the electron energy of about100keV the ratio r1 of dissociative to non-dissociativeionization of H2 is8

r1 =σ(H2 rarr 2H+

1 )

σ(H2 rarr H+2 )

= 0095plusmn 0008 (11)

and the ratio between the ionization cross sections is857

r2 =σion(H2)

σion(H1)= 17plusmn 01 (12)

At the standard operation parameters of the source(Sec IV) the measured currents are Icup(H1) = (125 plusmn05) nA and Icup(H2) = (64 plusmn 01) nA yielding α =(096plusmn004) in excellent agreement with the value result-ing from the measurements with the QMS (Sec VIA)

VII BEAM POLARIZATION

The Lamb-shift polarimeter was designed built andtested at Universitat zu Koln7 It was used to measureand to optimize the polarization of the atomic hydrogenand deuterium beams delivered by the ABS Details arefound in Ref7The vector polarization pz for hydrogen is defined by

the relative hyperfine-state occupation numbers N(mI)

pz =N(+ 1

2 )minusN(minus 12 )

N(+ 12 ) +N(minus 1

2 ) (13)

for deuterium

pz =N(+1)minusN(minus1)

N(+1) +N(0) +N(minus1) (14)

Deuterium tensor polarization pzz is given by

pzz =N(+1) +N(minus1)minus 2 middotN(0)

N(+1) +N(0) +N(minus1) (15)

16

These polarizations can be derived from the measuredLyman-α peak strengths S by application of a number ofcorrection factors78

Typical Lyman α spectra measured with the polarizedhydrogen and deuterium beam from the ABS are shownin the Figs 25 and 26

MFT 2-3

200 300 400 5000

40

80

120

160

Magnetic field in the spinfilter [au]

PM

T si

gnal

[au

]

mI = -12

(a)

mI = +12

100 200 300 400 5000

40

80

120

160(b)

MFT 2-3WFT 1-3

mI = +12

mI = -12

FIG 25 Lyman-α spectra measured with the polarized hy-drogen beam (a) population change from state |2〉 to state|3〉 induced by the MFT unit (b) same as (a) with subsequentpopulation change from state |1〉 to state |3〉 induced by theWFT unit

MFT 3-4SFT 2-6

0

2

4

6

8

mI = -1m

I = 0

(a)

mI = +1

0

2

4

6

8(b)

MFT 3-4WFT 1-4 2-3

mI = +1 m

I = 0

mI = -1

mI = +1

mI = 0

mI = -1

mI = +1

mI = 0

mI = -1

0 1000 2000 3000

0

1

2

3

4 (c)

PMT

sign

al [a

u]

Magnetic field in the spinfilter [au]

WFT 1-4SFT 2-6

0 1000 2000 30000

2

4

6

8WFT 1-4SFT 3-5

(d)

FIG 26 Lyman-α spectra measured with the polarized deu-terium beam (a) and (b) vector polarization resulting fromsubsequent transitions MFT (3 rarr 4) and SFT (2 rarr 6) andWFT (1 rarr 4 2 rarr 3) respectively (c) and (d) tensor polar-ization resulting from subsequent transitions WFT (1 rarr 4)and SFT (2 rarr 6) and SFT (3 rarr 5) respectively

The polarization values for the hydrogen and the deu-terium beam derived from the Lyman-α peak-strengthratios with application of the necessary corrections aresummarized in Table IV

The vector polarization for hydrogen of the first linereflects the population of state |1〉 and state |2〉 accordingto the Eqs (3) and (13) The value of 091 deduced fromthe calculated transmission values is confirmed by themeasured one

TABLE IV The vector polarization pz of the hydrogen beamand the vector and the tensor polarization pzz of the deu-terium beam from the ABS measured with the Lamb-shiftpolarimeter

populated pz pzz

state(s)

Hydrogen |1〉 +089plusmn 001 -

|3〉 minus096plusmn 001 -

Deuterium |1〉+ |6〉 +088plusmn 001 +088plusmn 003

|3〉+ |4〉 minus091plusmn 001 +085plusmn 002

|3〉+ |6〉 +0005plusmn 0003 +090plusmn 001

|2〉+ |5〉 +0005plusmn 0003 minus171plusmn 003

VIII CONCLUSIONS AND OUTLOOK

In this paper we present the detailed description ofthe major components of the atomic beam source (ABS)for the polarized internal gas target of the magnet spec-trometer ANKE in COSY-Julich The ABS was builtfor the purpose of extending the physics program ofANKE from unpolarized and single-polarized investiga-tions with stored beams towards double-polarized exper-iments1 thus facilitating nuclear reaction studies involv-

ing ~p~p ~p~d ~d~p and ~d~d initial statesThe mechanical design took into account that at

ANKE the source has to be mounted vertically and trans-versely movable together with the transverse motion ofthe spectrometer magnet D2 The design of the system ofsextupole magnets took advantage of the developments inthe field of rare-earth permanent magnets (NdFeB) Ded-icated tools and methods were developed to determineand to optimize the source parameters ie intensity de-gree of dissociation and polarization Special emphasiswas put on the measurements of the spatial distributionsof the atomic and molecular beam near the focus wherethe feeding tube of the storage cell is located The ABShas been used in a number of investigations at ANKEthe commissioning effort to prepare the target for the usewith polarized H is described in Ref62 Performed stud-ies of the deuteron-charge exchange reaction are summa-rized in Ref6364 studies in near-threshold pion produc-tion are reported about in Ref65The ABS resides at the ANKE target position for a few

months per year only thus during the remaining timeit is used for other studies It had been observed thatthe nuclear polarization in recombined hydrogen is par-tially retained after recombination66 as well as evidencefor nuclear tensor polarization in recombined deuteriummolecules67 In order to investigate this recombinationprocess in more detail a special setup has been developedin the framework of an ISTC project68 and the recombi-nation process for different cell-wall coatings and differ-ent polarizations of the injected hydrogen or deuterium

17

atoms as function of cell-wall temperature strength ofthe magnetic holding field and gas pressure in the cell ispresently investigated69ndash71

Appendix A Preparation of Discharge Tubes and Nozzles

1 Tube Treatment

One end of the discharge tube is machined at a 45

angle while the other is kept flat Both ends are thenremelted and the tubes are tempered at 150 C Thetubes are further treated according to the procedure de-scribed in Ref72 which includes successive cleaning withacetone methanol distilled water and subsequent rins-ing by a 21 acid mixture of concentrated HF (40) andHCl (32) for 5 min The tubes are then flushed bydistilled water and dried

2 Nozzle Treatment

The nozzles are cleaned in an ultrasonic bath oftrichlorethylene acetone methanol and finally distilledwater all at 50 C Anodizing takes place in sulfuric acidto form a thin layer of Al2O3 as described in Ref72 Af-terwards they are immersed in distilled water for 30minat 95 C

ACKNOWLEDGMENTS

The authors want to thank OWB Schult Institut furKernphysik (IKP) Julich who initiated the polarizationprogram of ANKE Thanks go to the design office themechanical workshop and especially to WR Ermer allIKP Valuable advice was received from the PINTEX col-laboration at IUCF from the target group at HERMESespecially NKoch and from DToporkov BINP Novosi-birsk The support by VKoptev PNPI Gatchina whoregrettably passed away in January 2012 is gratefullyacknowledged Thanks go also to R Poprawe and col-leagues Fraunhofer-Institut fur Lasertechnik Aachenwhere the encapsulations of the magnets were laser-welded

1AKacharava F Rathmann and CWilkin Spin Physics fromCOSY to FAIR COSY Experiment Proposal No 152 (2005)Available under httparXivnucl-ex0511028

2WHaeberli in Proc 2nd Int Symp on Polarization Phenomenaof Nucleons Karlsruhe 1965 Eds P Huber and H SchopperExperientia Supplementum 12 64 (Birkhauser Verlag 1966)

3E Steffens and WHaeberli Rep Progr Phys 66 1887 (2003)4SBarsov et al Nucl Instr and Meth A 462 364 (2001)5RMaier Nucl Instr and Meth A 390 1 (1997)6KGrigoryev et al Proc 14th International Workshop on Po-larized Sources Targets and Polarimetry (PSTP 2011) 12-16 September 2011 StPetersburg Russia eds KGrigoryevPKravtsov and AVasilyev ISBN 978-5-86763-282-3 61 (2011)

7REngels et al Rev Sci Instrum 74 4607 (2003)8REngels et al Rev Sci Instrum 76 053305 (2005)9TWise et al Nucl Instr and Meth A 336 410 (1993)

10WADezarn et al Nucl Instr and Meth A 362 36 (1995)11T Rinckel et al Nucl Instr and Meth A 439 117 (2000)12F Stock et al Nucl Instr and Meth A 343 334 (1994)13ANass et al Nucl Instr and Meth A 505 633 (2003)14VDerenchuk et al Proc Conf Polarized Ion Sources and Polar-ized Gas Targets Madison WI 1993 Eds LWAnderson andWHaeberli AIP Conf Proc 293 72 (American Institute ofPhysics 1994)

15HOkamura et al see Ref14 p 8416KHatanaka et al Nucl Instr and Meth A 384 575 (1997)17Manufacturer Schiffer Metall- amp Vakuumtechnik 52428 JulichGermany

18Single-stage type RGS120 refrigerating capacity 120W at 80Kand 20W at 30K Leybold Vacuum GmbH 50968 Koln Ger-many

19Mini UHV gate valve series 010 VAT Germany GmbH 85630Grasbrunn Germany

20Type F3 fomblin oil Pfeiffer Vacuum GmbH 35614 Asslar Ger-many

21Model HU 1 Leybold Vacuum GmbH 50968 Koln Germany22Manufacturer SK Industriemodell GmbH 52531 Ubach-Palenberg Germany

23Type PFG 600 RF with automatic matchbox PFM 1500 A-INDHuttinger Elektronik GmbH 79110 Freiburg Germany

24WKorsch PhD Thesis Philipps Universitat Marburg (1990)25F Stock et al Int Workshop on Polarized Beams and PolarizedGas Targets Koeln Germany 1995 Eds HPaetz gen Schieckand L Sydow (World Scientific Publ Co 1996) p 260

26The first number denotes the outer diameter and the second onethe wall thickness

27Type Duran 8330 equivalent to Corning 7740 (Pyrex) SchottAG 55122 Mainz Germany

28Ultra-Kryomat RUL 80-D Lauda DrRWobser GmbH 97912Lauda-Konigshofen Germany

29ODU-Kontakt GmbH 84444 Muhldorf Germany30Handbook of Chemistry and Physics Ed RCEast (The Chem-ical Rubber Co 1973) p E-10

31NKoch and E Steffens Rev Sci Instrum 70 1631 (1999)32AVassiliev et al Petersburg Nuclear Physics Institute ReportNP-32-1997 No 2175 (1997)

33BLorentz Diploma Thesis Ruprecht-Karls-Universitat Heidel-berg (1993)

34ANass and E Steffens Nucl Instr and Meth A 598 653(2009)

35WHaeberli Ann Rev Nucl Sci 17 373 (1967)36AVassiliev et al Rev Sci Instr 71 3331 (2000)37WKubischta Proc Workshop on Polarized Gas Targetsfor Storage Rings Heidelberg 23-26 September 1991 EdsHGGaul E Steffens and K Zapfe (Max-Planck-Institut furKernphysik Heidelberg)

38The labeling of the hyperfine states as |1〉 = |mj = +12 mI =+12〉 |2〉 = | + 12minus12〉 |3〉 = | minus 12+12〉 and |4〉 =|minus12minus12〉 for hydrogen and |1〉 = |+12+1〉 |2〉 = |+12 0〉|3〉 = | + 12minus1〉 |4〉 = | minus 12minus1〉 |5〉 = | minus 12 0〉 and

13

pression volume The resulting rc shows a deviation of012 mm from the geometrical axis of the source Further-more the data measured with the narrow compressiontube of 25mm radius can be used to derive the fractionof the beam entering the compression tube of 5mm ra-dius used in the intensity measurement of Sec IV Theratio

η =

rile25 mmsum

0P (xi yj)

rile10 mmsum

0P (xi yj)

(9)

where ri is the distance of the compression-tube axis tothe beam axis yields η asymp 07

B Measurements with the QMS

The beam-profile studies of Sec VA were extendedwith a setup utilizing a crossed-beam quadrupole massspectrometer (QMS) in the setup of Fig 19 Contraryto the measurements with the compression tube thosewith the QMS allow to separate the atomic and molec-ular fractions in the beam A 2mm diameter aperturewas installed at the entrance of the sensitive volume of

FIG 19 Setup for the measurements of the beam profilewith the QMS (1) xy-table enabling two-dimensional dis-placement of the entrance window of the QMS against thegeometrical axis of the ABS (2) the QMS (3) manually op-erated beam shutter The beam dump is an axially mountedcryo pump

the QMS to improve the resolution compared with thatachieved by the compression tube of 5mm diameter usedin measurements of the preceding section The layout ofthe setup presented in Fig 19 shows that in the presentcase the profile could not be measured at a distance ofz = 300mm to the last magnet Instead measurementswere performed at z = 567mm and with installation ofan extension tube at z = 697mm The xy manipulatorenabled displacements of the aperture axis from the geo-metrical axis of the source in any direction within limitsset by the bore diameter of the xy manipulatorThe first measured distribution of the atomic hydro-

gen (Fig 20) showed a distinct deviation from azimuthalsymmetry indicating an insufficient relative alignment ofnozzle and skimmer The three threaded rods support-ing the dissociator with the nozzle via the three-legged

FIG 20 Two-dimensional distribution of the atomic hydro-gen component of the beam at z = 567mm before the nozzle-to-skimmer adjustment showing a disinct deviation from az-imuthal symmetry

plate (label 2 in Fig 1) allow one to vary the position ofthe nozzle relative to that of the skimmer while the sourceis running This possibility has been used to find a nozzleposition which results in an azimuthally symmetric distri-bution The achieved symmetric distribution is shown inFig 21 and profiles of the atomic hadrogen component inthe beam measured in x and y direction at z = 567mmand z = 697mm are presented in Fig 22 Fits by Gaus-sian distributions to the data yield full widths at halfmaximum Γx = (736plusmn 043)mm Γy = (668plusmn 080)mmfor the distributions measured at z = 567mm and Γx =(669plusmn 022)mm Γy = (638plusmn 027)mm at 697mm

C Reduction of MoO3 by hydrogen

In addition to the compression tube and the QMS tech-nique a supplementary attempt was made to determinethe beam profile by exposing molybdenium trioxide (a

14

FIG 21 The distribution corresponding to that of Fig 20 af-ter nozzle-to-skimmer adjustment resulting in azimuthal sym-metry

0

20

40

60

80

0

20

40

60

80

2 4 6 8 10 12 14 16 18 20 22

0

20

40

60

80

6 8 10 12 14 16 18 20 22 24

0

20

40

60

80

QM

S si

gnal

s [a

u]

(a)

Z = 567 mm

Z = 697 mm

(c)

(b)

X [mm]

(d)

Y [mm]

FIG 22 Profiles of the atomic hydrogen component inthe beam measured with the QMS 567mm and 697mmbehind the last magnet

yellowish powder) on a glass plate to the beam The prin-ciple of this method is based on the reduction of MoO3 toa lower oxide of blue colour It first was used in the ex-periment to measure the magnetic moment of the hydro-gen atom by splitting of the beam in an inhomogeneousmagnetic field 61

This method is much simpler than the time-consumingmeasurements described in Secs VA and VB It givesqualitative results as presented in Fig 23 A quantitativeanalysis however requires development of the measuringtechnique (eg preparation of glass plates study of theoptimum exposure time digital image processing)

FIG 23 Photo of the glass plate covered with molybdenumtrioxide MoO3 exposed to the atomic hydrogen beam

D Summary of the profile measurements

Table III summarizes results of the measurements ofthe ABS beam profile with the compression-tube and theQMS setup The larger errors of the widths measuredwith the QMS are due to the lack of measurements withthe dissociator switched off and the necessity to estimatethe background signal from the existing data Within theerrors the measured widths do not show a dependenceon the distance from the last magnet This facilitatesto position the feeding tube of the storage cell in a widerange of a distances to the last magnet The average val-ues Γx = (638 plusmn 060)mm and Γy = (684 plusmn 033)mmagree within the errors and yield a common width ofΓxy = (673 plusmn 029)mm The two-dimensional Gaus-sian distribution of this width allows one to estimate thefraction η of the beam intensity injected into the com-pression tube or a feeding tube For a tube of 10mmdiameter η = 078plusmn 003 comparable with η asymp 07 givenin Sec VA

TABLE III Dimensions (FWHM) of the atomic hydrogenbeam measured with the compression tube (CT) and thecrossed-beam quadrupole mass spectrometer (QMS) at dis-tances z to the last magnet along perpendicular directions xand y

z[mm] Γx[mm] Γy[mm]

CT 300 642 plusmn 009 699plusmn 006

CT 337 627 plusmn 008 658plusmn 008

QMS 567 736 plusmn 043 668plusmn 080

QMS 697 669 plusmn 022 638plusmn 027

15

VI DEGREE OF DISSOCIATION

Besides the intensity of the atomic beam it is impor-tant to determine the molecular fraction in the beamMolecules injected into the feeding tube reduce the po-larization of the target gas

A Measurements with crossed-beam QMS

In addition to the data on the profile of the atomichydrogen beam (Sec VB) data on the distributions ofmolecular hydrogen in the beam were taken too at thepositions z = 567mm and 697mm behind the last mag-net The relation between the degree of dissociation andthe QMS signals by the atomic and molecular beam com-ponent was given above by Eq (6) The coefficient kv =vmva however is chosen here under the assumptionthat the average velocity of the atoms is determined bythe nozzle temperature of 65K and that of the moleculesby scattering and recombination on the ABS chamberwalls at 290K This yields kv =

radic

2 middot 65290 = 067 ingood agreement with Ref13 where this coefficient wasdetermined by the measured velocity distributions undersimilar conditionsThe measured profiles of the atomic fraction (identical

to those of Fig 21) those of the molecular fraction andthose of the degree of dissociation deduced from Eq (6)are collected in Fig 24As it is seen from the figure the distribution of the

degree of dissociation shows a dip around the central linedue to the higher density of molecular hydrogen originat-ing from the nozzle The mean value in an aperture of

0

20

40

60

80

100

120

2 4 6 8 10 12 14 16 18 20 22

0

20

40

60

80

100

120

6 8 10 12 14 16 18 20 22 24

QM

S si

gnal

s [a

u]

(a)

(b)

(c)

00

02

04

06

08

10

12

Deg

ree

of d

isso

ciat

ion

()

X [mm]

Y [mm]

(d)

00

02

04

06

08

10

12

Z = 567 mm

Z = 697 mm

FIG 24 Spatial distributions of H1 (bull)H2 () and degree ofdissociation () averaged over 3mm wide bands in the xz andyz planes respectively (here the z-axis is the geometrical axisof the ABS)

10mm diameter results as α = 095 plusmn 004

B Measurements with the Lamb-shift polarimeter

A cup in the quench chamber of the Lamb-shift po-larimeter (LSP) described in Ref8 allows one to mea-sure the currents Icup(H1) and Icup(H2) of the H+

1 andH+

2 ions extracted from the ionizer and separated bythe Wien filter with the cesium evaporation and the spinfilter switched off The relation between the degree ofdissociation α and the measured currents is

α =Icup(H1)minus r1

r2Icup(H2)

Icup(H1)minus r1r2Icup(H2) + 2kv

r2Icup(H2)

(10)

Among the three coefficients kv = 067 as for the mea-surement with the QMS For the electron energy of about100keV the ratio r1 of dissociative to non-dissociativeionization of H2 is8

r1 =σ(H2 rarr 2H+

1 )

σ(H2 rarr H+2 )

= 0095plusmn 0008 (11)

and the ratio between the ionization cross sections is857

r2 =σion(H2)

σion(H1)= 17plusmn 01 (12)

At the standard operation parameters of the source(Sec IV) the measured currents are Icup(H1) = (125 plusmn05) nA and Icup(H2) = (64 plusmn 01) nA yielding α =(096plusmn004) in excellent agreement with the value result-ing from the measurements with the QMS (Sec VIA)

VII BEAM POLARIZATION

The Lamb-shift polarimeter was designed built andtested at Universitat zu Koln7 It was used to measureand to optimize the polarization of the atomic hydrogenand deuterium beams delivered by the ABS Details arefound in Ref7The vector polarization pz for hydrogen is defined by

the relative hyperfine-state occupation numbers N(mI)

pz =N(+ 1

2 )minusN(minus 12 )

N(+ 12 ) +N(minus 1

2 ) (13)

for deuterium

pz =N(+1)minusN(minus1)

N(+1) +N(0) +N(minus1) (14)

Deuterium tensor polarization pzz is given by

pzz =N(+1) +N(minus1)minus 2 middotN(0)

N(+1) +N(0) +N(minus1) (15)

16

These polarizations can be derived from the measuredLyman-α peak strengths S by application of a number ofcorrection factors78

Typical Lyman α spectra measured with the polarizedhydrogen and deuterium beam from the ABS are shownin the Figs 25 and 26

MFT 2-3

200 300 400 5000

40

80

120

160

Magnetic field in the spinfilter [au]

PM

T si

gnal

[au

]

mI = -12

(a)

mI = +12

100 200 300 400 5000

40

80

120

160(b)

MFT 2-3WFT 1-3

mI = +12

mI = -12

FIG 25 Lyman-α spectra measured with the polarized hy-drogen beam (a) population change from state |2〉 to state|3〉 induced by the MFT unit (b) same as (a) with subsequentpopulation change from state |1〉 to state |3〉 induced by theWFT unit

MFT 3-4SFT 2-6

0

2

4

6

8

mI = -1m

I = 0

(a)

mI = +1

0

2

4

6

8(b)

MFT 3-4WFT 1-4 2-3

mI = +1 m

I = 0

mI = -1

mI = +1

mI = 0

mI = -1

mI = +1

mI = 0

mI = -1

0 1000 2000 3000

0

1

2

3

4 (c)

PMT

sign

al [a

u]

Magnetic field in the spinfilter [au]

WFT 1-4SFT 2-6

0 1000 2000 30000

2

4

6

8WFT 1-4SFT 3-5

(d)

FIG 26 Lyman-α spectra measured with the polarized deu-terium beam (a) and (b) vector polarization resulting fromsubsequent transitions MFT (3 rarr 4) and SFT (2 rarr 6) andWFT (1 rarr 4 2 rarr 3) respectively (c) and (d) tensor polar-ization resulting from subsequent transitions WFT (1 rarr 4)and SFT (2 rarr 6) and SFT (3 rarr 5) respectively

The polarization values for the hydrogen and the deu-terium beam derived from the Lyman-α peak-strengthratios with application of the necessary corrections aresummarized in Table IV

The vector polarization for hydrogen of the first linereflects the population of state |1〉 and state |2〉 accordingto the Eqs (3) and (13) The value of 091 deduced fromthe calculated transmission values is confirmed by themeasured one

TABLE IV The vector polarization pz of the hydrogen beamand the vector and the tensor polarization pzz of the deu-terium beam from the ABS measured with the Lamb-shiftpolarimeter

populated pz pzz

state(s)

Hydrogen |1〉 +089plusmn 001 -

|3〉 minus096plusmn 001 -

Deuterium |1〉+ |6〉 +088plusmn 001 +088plusmn 003

|3〉+ |4〉 minus091plusmn 001 +085plusmn 002

|3〉+ |6〉 +0005plusmn 0003 +090plusmn 001

|2〉+ |5〉 +0005plusmn 0003 minus171plusmn 003

VIII CONCLUSIONS AND OUTLOOK

In this paper we present the detailed description ofthe major components of the atomic beam source (ABS)for the polarized internal gas target of the magnet spec-trometer ANKE in COSY-Julich The ABS was builtfor the purpose of extending the physics program ofANKE from unpolarized and single-polarized investiga-tions with stored beams towards double-polarized exper-iments1 thus facilitating nuclear reaction studies involv-

ing ~p~p ~p~d ~d~p and ~d~d initial statesThe mechanical design took into account that at

ANKE the source has to be mounted vertically and trans-versely movable together with the transverse motion ofthe spectrometer magnet D2 The design of the system ofsextupole magnets took advantage of the developments inthe field of rare-earth permanent magnets (NdFeB) Ded-icated tools and methods were developed to determineand to optimize the source parameters ie intensity de-gree of dissociation and polarization Special emphasiswas put on the measurements of the spatial distributionsof the atomic and molecular beam near the focus wherethe feeding tube of the storage cell is located The ABShas been used in a number of investigations at ANKEthe commissioning effort to prepare the target for the usewith polarized H is described in Ref62 Performed stud-ies of the deuteron-charge exchange reaction are summa-rized in Ref6364 studies in near-threshold pion produc-tion are reported about in Ref65The ABS resides at the ANKE target position for a few

months per year only thus during the remaining timeit is used for other studies It had been observed thatthe nuclear polarization in recombined hydrogen is par-tially retained after recombination66 as well as evidencefor nuclear tensor polarization in recombined deuteriummolecules67 In order to investigate this recombinationprocess in more detail a special setup has been developedin the framework of an ISTC project68 and the recombi-nation process for different cell-wall coatings and differ-ent polarizations of the injected hydrogen or deuterium

17

atoms as function of cell-wall temperature strength ofthe magnetic holding field and gas pressure in the cell ispresently investigated69ndash71

Appendix A Preparation of Discharge Tubes and Nozzles

1 Tube Treatment

One end of the discharge tube is machined at a 45

angle while the other is kept flat Both ends are thenremelted and the tubes are tempered at 150 C Thetubes are further treated according to the procedure de-scribed in Ref72 which includes successive cleaning withacetone methanol distilled water and subsequent rins-ing by a 21 acid mixture of concentrated HF (40) andHCl (32) for 5 min The tubes are then flushed bydistilled water and dried

2 Nozzle Treatment

The nozzles are cleaned in an ultrasonic bath oftrichlorethylene acetone methanol and finally distilledwater all at 50 C Anodizing takes place in sulfuric acidto form a thin layer of Al2O3 as described in Ref72 Af-terwards they are immersed in distilled water for 30minat 95 C

ACKNOWLEDGMENTS

The authors want to thank OWB Schult Institut furKernphysik (IKP) Julich who initiated the polarizationprogram of ANKE Thanks go to the design office themechanical workshop and especially to WR Ermer allIKP Valuable advice was received from the PINTEX col-laboration at IUCF from the target group at HERMESespecially NKoch and from DToporkov BINP Novosi-birsk The support by VKoptev PNPI Gatchina whoregrettably passed away in January 2012 is gratefullyacknowledged Thanks go also to R Poprawe and col-leagues Fraunhofer-Institut fur Lasertechnik Aachenwhere the encapsulations of the magnets were laser-welded

1AKacharava F Rathmann and CWilkin Spin Physics fromCOSY to FAIR COSY Experiment Proposal No 152 (2005)Available under httparXivnucl-ex0511028

2WHaeberli in Proc 2nd Int Symp on Polarization Phenomenaof Nucleons Karlsruhe 1965 Eds P Huber and H SchopperExperientia Supplementum 12 64 (Birkhauser Verlag 1966)

3E Steffens and WHaeberli Rep Progr Phys 66 1887 (2003)4SBarsov et al Nucl Instr and Meth A 462 364 (2001)5RMaier Nucl Instr and Meth A 390 1 (1997)6KGrigoryev et al Proc 14th International Workshop on Po-larized Sources Targets and Polarimetry (PSTP 2011) 12-16 September 2011 StPetersburg Russia eds KGrigoryevPKravtsov and AVasilyev ISBN 978-5-86763-282-3 61 (2011)

7REngels et al Rev Sci Instrum 74 4607 (2003)8REngels et al Rev Sci Instrum 76 053305 (2005)9TWise et al Nucl Instr and Meth A 336 410 (1993)

10WADezarn et al Nucl Instr and Meth A 362 36 (1995)11T Rinckel et al Nucl Instr and Meth A 439 117 (2000)12F Stock et al Nucl Instr and Meth A 343 334 (1994)13ANass et al Nucl Instr and Meth A 505 633 (2003)14VDerenchuk et al Proc Conf Polarized Ion Sources and Polar-ized Gas Targets Madison WI 1993 Eds LWAnderson andWHaeberli AIP Conf Proc 293 72 (American Institute ofPhysics 1994)

15HOkamura et al see Ref14 p 8416KHatanaka et al Nucl Instr and Meth A 384 575 (1997)17Manufacturer Schiffer Metall- amp Vakuumtechnik 52428 JulichGermany

18Single-stage type RGS120 refrigerating capacity 120W at 80Kand 20W at 30K Leybold Vacuum GmbH 50968 Koln Ger-many

19Mini UHV gate valve series 010 VAT Germany GmbH 85630Grasbrunn Germany

20Type F3 fomblin oil Pfeiffer Vacuum GmbH 35614 Asslar Ger-many

21Model HU 1 Leybold Vacuum GmbH 50968 Koln Germany22Manufacturer SK Industriemodell GmbH 52531 Ubach-Palenberg Germany

23Type PFG 600 RF with automatic matchbox PFM 1500 A-INDHuttinger Elektronik GmbH 79110 Freiburg Germany

24WKorsch PhD Thesis Philipps Universitat Marburg (1990)25F Stock et al Int Workshop on Polarized Beams and PolarizedGas Targets Koeln Germany 1995 Eds HPaetz gen Schieckand L Sydow (World Scientific Publ Co 1996) p 260

26The first number denotes the outer diameter and the second onethe wall thickness

27Type Duran 8330 equivalent to Corning 7740 (Pyrex) SchottAG 55122 Mainz Germany

28Ultra-Kryomat RUL 80-D Lauda DrRWobser GmbH 97912Lauda-Konigshofen Germany

29ODU-Kontakt GmbH 84444 Muhldorf Germany30Handbook of Chemistry and Physics Ed RCEast (The Chem-ical Rubber Co 1973) p E-10

31NKoch and E Steffens Rev Sci Instrum 70 1631 (1999)32AVassiliev et al Petersburg Nuclear Physics Institute ReportNP-32-1997 No 2175 (1997)

33BLorentz Diploma Thesis Ruprecht-Karls-Universitat Heidel-berg (1993)

34ANass and E Steffens Nucl Instr and Meth A 598 653(2009)

35WHaeberli Ann Rev Nucl Sci 17 373 (1967)36AVassiliev et al Rev Sci Instr 71 3331 (2000)37WKubischta Proc Workshop on Polarized Gas Targetsfor Storage Rings Heidelberg 23-26 September 1991 EdsHGGaul E Steffens and K Zapfe (Max-Planck-Institut furKernphysik Heidelberg)

38The labeling of the hyperfine states as |1〉 = |mj = +12 mI =+12〉 |2〉 = | + 12minus12〉 |3〉 = | minus 12+12〉 and |4〉 =|minus12minus12〉 for hydrogen and |1〉 = |+12+1〉 |2〉 = |+12 0〉|3〉 = | + 12minus1〉 |4〉 = | minus 12minus1〉 |5〉 = | minus 12 0〉 and

14

FIG 21 The distribution corresponding to that of Fig 20 af-ter nozzle-to-skimmer adjustment resulting in azimuthal sym-metry

0

20

40

60

80

0

20

40

60

80

2 4 6 8 10 12 14 16 18 20 22

0

20

40

60

80

6 8 10 12 14 16 18 20 22 24

0

20

40

60

80

QM

S si

gnal

s [a

u]

(a)

Z = 567 mm

Z = 697 mm

(c)

(b)

X [mm]

(d)

Y [mm]

FIG 22 Profiles of the atomic hydrogen component inthe beam measured with the QMS 567mm and 697mmbehind the last magnet

yellowish powder) on a glass plate to the beam The prin-ciple of this method is based on the reduction of MoO3 toa lower oxide of blue colour It first was used in the ex-periment to measure the magnetic moment of the hydro-gen atom by splitting of the beam in an inhomogeneousmagnetic field 61

This method is much simpler than the time-consumingmeasurements described in Secs VA and VB It givesqualitative results as presented in Fig 23 A quantitativeanalysis however requires development of the measuringtechnique (eg preparation of glass plates study of theoptimum exposure time digital image processing)

FIG 23 Photo of the glass plate covered with molybdenumtrioxide MoO3 exposed to the atomic hydrogen beam

D Summary of the profile measurements

Table III summarizes results of the measurements ofthe ABS beam profile with the compression-tube and theQMS setup The larger errors of the widths measuredwith the QMS are due to the lack of measurements withthe dissociator switched off and the necessity to estimatethe background signal from the existing data Within theerrors the measured widths do not show a dependenceon the distance from the last magnet This facilitatesto position the feeding tube of the storage cell in a widerange of a distances to the last magnet The average val-ues Γx = (638 plusmn 060)mm and Γy = (684 plusmn 033)mmagree within the errors and yield a common width ofΓxy = (673 plusmn 029)mm The two-dimensional Gaus-sian distribution of this width allows one to estimate thefraction η of the beam intensity injected into the com-pression tube or a feeding tube For a tube of 10mmdiameter η = 078plusmn 003 comparable with η asymp 07 givenin Sec VA

TABLE III Dimensions (FWHM) of the atomic hydrogenbeam measured with the compression tube (CT) and thecrossed-beam quadrupole mass spectrometer (QMS) at dis-tances z to the last magnet along perpendicular directions xand y

z[mm] Γx[mm] Γy[mm]

CT 300 642 plusmn 009 699plusmn 006

CT 337 627 plusmn 008 658plusmn 008

QMS 567 736 plusmn 043 668plusmn 080

QMS 697 669 plusmn 022 638plusmn 027

15

VI DEGREE OF DISSOCIATION

Besides the intensity of the atomic beam it is impor-tant to determine the molecular fraction in the beamMolecules injected into the feeding tube reduce the po-larization of the target gas

A Measurements with crossed-beam QMS

In addition to the data on the profile of the atomichydrogen beam (Sec VB) data on the distributions ofmolecular hydrogen in the beam were taken too at thepositions z = 567mm and 697mm behind the last mag-net The relation between the degree of dissociation andthe QMS signals by the atomic and molecular beam com-ponent was given above by Eq (6) The coefficient kv =vmva however is chosen here under the assumptionthat the average velocity of the atoms is determined bythe nozzle temperature of 65K and that of the moleculesby scattering and recombination on the ABS chamberwalls at 290K This yields kv =

radic

2 middot 65290 = 067 ingood agreement with Ref13 where this coefficient wasdetermined by the measured velocity distributions undersimilar conditionsThe measured profiles of the atomic fraction (identical

to those of Fig 21) those of the molecular fraction andthose of the degree of dissociation deduced from Eq (6)are collected in Fig 24As it is seen from the figure the distribution of the

degree of dissociation shows a dip around the central linedue to the higher density of molecular hydrogen originat-ing from the nozzle The mean value in an aperture of

0

20

40

60

80

100

120

2 4 6 8 10 12 14 16 18 20 22

0

20

40

60

80

100

120

6 8 10 12 14 16 18 20 22 24

QM

S si

gnal

s [a

u]

(a)

(b)

(c)

00

02

04

06

08

10

12

Deg

ree

of d

isso

ciat

ion

()

X [mm]

Y [mm]

(d)

00

02

04

06

08

10

12

Z = 567 mm

Z = 697 mm

FIG 24 Spatial distributions of H1 (bull)H2 () and degree ofdissociation () averaged over 3mm wide bands in the xz andyz planes respectively (here the z-axis is the geometrical axisof the ABS)

10mm diameter results as α = 095 plusmn 004

B Measurements with the Lamb-shift polarimeter

A cup in the quench chamber of the Lamb-shift po-larimeter (LSP) described in Ref8 allows one to mea-sure the currents Icup(H1) and Icup(H2) of the H+

1 andH+

2 ions extracted from the ionizer and separated bythe Wien filter with the cesium evaporation and the spinfilter switched off The relation between the degree ofdissociation α and the measured currents is

α =Icup(H1)minus r1

r2Icup(H2)

Icup(H1)minus r1r2Icup(H2) + 2kv

r2Icup(H2)

(10)

Among the three coefficients kv = 067 as for the mea-surement with the QMS For the electron energy of about100keV the ratio r1 of dissociative to non-dissociativeionization of H2 is8

r1 =σ(H2 rarr 2H+

1 )

σ(H2 rarr H+2 )

= 0095plusmn 0008 (11)

and the ratio between the ionization cross sections is857

r2 =σion(H2)

σion(H1)= 17plusmn 01 (12)

At the standard operation parameters of the source(Sec IV) the measured currents are Icup(H1) = (125 plusmn05) nA and Icup(H2) = (64 plusmn 01) nA yielding α =(096plusmn004) in excellent agreement with the value result-ing from the measurements with the QMS (Sec VIA)

VII BEAM POLARIZATION

The Lamb-shift polarimeter was designed built andtested at Universitat zu Koln7 It was used to measureand to optimize the polarization of the atomic hydrogenand deuterium beams delivered by the ABS Details arefound in Ref7The vector polarization pz for hydrogen is defined by

the relative hyperfine-state occupation numbers N(mI)

pz =N(+ 1

2 )minusN(minus 12 )

N(+ 12 ) +N(minus 1

2 ) (13)

for deuterium

pz =N(+1)minusN(minus1)

N(+1) +N(0) +N(minus1) (14)

Deuterium tensor polarization pzz is given by

pzz =N(+1) +N(minus1)minus 2 middotN(0)

N(+1) +N(0) +N(minus1) (15)

16

These polarizations can be derived from the measuredLyman-α peak strengths S by application of a number ofcorrection factors78

Typical Lyman α spectra measured with the polarizedhydrogen and deuterium beam from the ABS are shownin the Figs 25 and 26

MFT 2-3

200 300 400 5000

40

80

120

160

Magnetic field in the spinfilter [au]

PM

T si

gnal

[au

]

mI = -12

(a)

mI = +12

100 200 300 400 5000

40

80

120

160(b)

MFT 2-3WFT 1-3

mI = +12

mI = -12

FIG 25 Lyman-α spectra measured with the polarized hy-drogen beam (a) population change from state |2〉 to state|3〉 induced by the MFT unit (b) same as (a) with subsequentpopulation change from state |1〉 to state |3〉 induced by theWFT unit

MFT 3-4SFT 2-6

0

2

4

6

8

mI = -1m

I = 0

(a)

mI = +1

0

2

4

6

8(b)

MFT 3-4WFT 1-4 2-3

mI = +1 m

I = 0

mI = -1

mI = +1

mI = 0

mI = -1

mI = +1

mI = 0

mI = -1

0 1000 2000 3000

0

1

2

3

4 (c)

PMT

sign

al [a

u]

Magnetic field in the spinfilter [au]

WFT 1-4SFT 2-6

0 1000 2000 30000

2

4

6

8WFT 1-4SFT 3-5

(d)

FIG 26 Lyman-α spectra measured with the polarized deu-terium beam (a) and (b) vector polarization resulting fromsubsequent transitions MFT (3 rarr 4) and SFT (2 rarr 6) andWFT (1 rarr 4 2 rarr 3) respectively (c) and (d) tensor polar-ization resulting from subsequent transitions WFT (1 rarr 4)and SFT (2 rarr 6) and SFT (3 rarr 5) respectively

The polarization values for the hydrogen and the deu-terium beam derived from the Lyman-α peak-strengthratios with application of the necessary corrections aresummarized in Table IV

The vector polarization for hydrogen of the first linereflects the population of state |1〉 and state |2〉 accordingto the Eqs (3) and (13) The value of 091 deduced fromthe calculated transmission values is confirmed by themeasured one

TABLE IV The vector polarization pz of the hydrogen beamand the vector and the tensor polarization pzz of the deu-terium beam from the ABS measured with the Lamb-shiftpolarimeter

populated pz pzz

state(s)

Hydrogen |1〉 +089plusmn 001 -

|3〉 minus096plusmn 001 -

Deuterium |1〉+ |6〉 +088plusmn 001 +088plusmn 003

|3〉+ |4〉 minus091plusmn 001 +085plusmn 002

|3〉+ |6〉 +0005plusmn 0003 +090plusmn 001

|2〉+ |5〉 +0005plusmn 0003 minus171plusmn 003

VIII CONCLUSIONS AND OUTLOOK

In this paper we present the detailed description ofthe major components of the atomic beam source (ABS)for the polarized internal gas target of the magnet spec-trometer ANKE in COSY-Julich The ABS was builtfor the purpose of extending the physics program ofANKE from unpolarized and single-polarized investiga-tions with stored beams towards double-polarized exper-iments1 thus facilitating nuclear reaction studies involv-

ing ~p~p ~p~d ~d~p and ~d~d initial statesThe mechanical design took into account that at

ANKE the source has to be mounted vertically and trans-versely movable together with the transverse motion ofthe spectrometer magnet D2 The design of the system ofsextupole magnets took advantage of the developments inthe field of rare-earth permanent magnets (NdFeB) Ded-icated tools and methods were developed to determineand to optimize the source parameters ie intensity de-gree of dissociation and polarization Special emphasiswas put on the measurements of the spatial distributionsof the atomic and molecular beam near the focus wherethe feeding tube of the storage cell is located The ABShas been used in a number of investigations at ANKEthe commissioning effort to prepare the target for the usewith polarized H is described in Ref62 Performed stud-ies of the deuteron-charge exchange reaction are summa-rized in Ref6364 studies in near-threshold pion produc-tion are reported about in Ref65The ABS resides at the ANKE target position for a few

months per year only thus during the remaining timeit is used for other studies It had been observed thatthe nuclear polarization in recombined hydrogen is par-tially retained after recombination66 as well as evidencefor nuclear tensor polarization in recombined deuteriummolecules67 In order to investigate this recombinationprocess in more detail a special setup has been developedin the framework of an ISTC project68 and the recombi-nation process for different cell-wall coatings and differ-ent polarizations of the injected hydrogen or deuterium

17

atoms as function of cell-wall temperature strength ofthe magnetic holding field and gas pressure in the cell ispresently investigated69ndash71

Appendix A Preparation of Discharge Tubes and Nozzles

1 Tube Treatment

One end of the discharge tube is machined at a 45

angle while the other is kept flat Both ends are thenremelted and the tubes are tempered at 150 C Thetubes are further treated according to the procedure de-scribed in Ref72 which includes successive cleaning withacetone methanol distilled water and subsequent rins-ing by a 21 acid mixture of concentrated HF (40) andHCl (32) for 5 min The tubes are then flushed bydistilled water and dried

2 Nozzle Treatment

The nozzles are cleaned in an ultrasonic bath oftrichlorethylene acetone methanol and finally distilledwater all at 50 C Anodizing takes place in sulfuric acidto form a thin layer of Al2O3 as described in Ref72 Af-terwards they are immersed in distilled water for 30minat 95 C

ACKNOWLEDGMENTS

The authors want to thank OWB Schult Institut furKernphysik (IKP) Julich who initiated the polarizationprogram of ANKE Thanks go to the design office themechanical workshop and especially to WR Ermer allIKP Valuable advice was received from the PINTEX col-laboration at IUCF from the target group at HERMESespecially NKoch and from DToporkov BINP Novosi-birsk The support by VKoptev PNPI Gatchina whoregrettably passed away in January 2012 is gratefullyacknowledged Thanks go also to R Poprawe and col-leagues Fraunhofer-Institut fur Lasertechnik Aachenwhere the encapsulations of the magnets were laser-welded

1AKacharava F Rathmann and CWilkin Spin Physics fromCOSY to FAIR COSY Experiment Proposal No 152 (2005)Available under httparXivnucl-ex0511028

2WHaeberli in Proc 2nd Int Symp on Polarization Phenomenaof Nucleons Karlsruhe 1965 Eds P Huber and H SchopperExperientia Supplementum 12 64 (Birkhauser Verlag 1966)

3E Steffens and WHaeberli Rep Progr Phys 66 1887 (2003)4SBarsov et al Nucl Instr and Meth A 462 364 (2001)5RMaier Nucl Instr and Meth A 390 1 (1997)6KGrigoryev et al Proc 14th International Workshop on Po-larized Sources Targets and Polarimetry (PSTP 2011) 12-16 September 2011 StPetersburg Russia eds KGrigoryevPKravtsov and AVasilyev ISBN 978-5-86763-282-3 61 (2011)

7REngels et al Rev Sci Instrum 74 4607 (2003)8REngels et al Rev Sci Instrum 76 053305 (2005)9TWise et al Nucl Instr and Meth A 336 410 (1993)

10WADezarn et al Nucl Instr and Meth A 362 36 (1995)11T Rinckel et al Nucl Instr and Meth A 439 117 (2000)12F Stock et al Nucl Instr and Meth A 343 334 (1994)13ANass et al Nucl Instr and Meth A 505 633 (2003)14VDerenchuk et al Proc Conf Polarized Ion Sources and Polar-ized Gas Targets Madison WI 1993 Eds LWAnderson andWHaeberli AIP Conf Proc 293 72 (American Institute ofPhysics 1994)

15HOkamura et al see Ref14 p 8416KHatanaka et al Nucl Instr and Meth A 384 575 (1997)17Manufacturer Schiffer Metall- amp Vakuumtechnik 52428 JulichGermany

18Single-stage type RGS120 refrigerating capacity 120W at 80Kand 20W at 30K Leybold Vacuum GmbH 50968 Koln Ger-many

19Mini UHV gate valve series 010 VAT Germany GmbH 85630Grasbrunn Germany

20Type F3 fomblin oil Pfeiffer Vacuum GmbH 35614 Asslar Ger-many

21Model HU 1 Leybold Vacuum GmbH 50968 Koln Germany22Manufacturer SK Industriemodell GmbH 52531 Ubach-Palenberg Germany

23Type PFG 600 RF with automatic matchbox PFM 1500 A-INDHuttinger Elektronik GmbH 79110 Freiburg Germany

24WKorsch PhD Thesis Philipps Universitat Marburg (1990)25F Stock et al Int Workshop on Polarized Beams and PolarizedGas Targets Koeln Germany 1995 Eds HPaetz gen Schieckand L Sydow (World Scientific Publ Co 1996) p 260

26The first number denotes the outer diameter and the second onethe wall thickness

27Type Duran 8330 equivalent to Corning 7740 (Pyrex) SchottAG 55122 Mainz Germany

28Ultra-Kryomat RUL 80-D Lauda DrRWobser GmbH 97912Lauda-Konigshofen Germany

29ODU-Kontakt GmbH 84444 Muhldorf Germany30Handbook of Chemistry and Physics Ed RCEast (The Chem-ical Rubber Co 1973) p E-10

31NKoch and E Steffens Rev Sci Instrum 70 1631 (1999)32AVassiliev et al Petersburg Nuclear Physics Institute ReportNP-32-1997 No 2175 (1997)

33BLorentz Diploma Thesis Ruprecht-Karls-Universitat Heidel-berg (1993)

34ANass and E Steffens Nucl Instr and Meth A 598 653(2009)

35WHaeberli Ann Rev Nucl Sci 17 373 (1967)36AVassiliev et al Rev Sci Instr 71 3331 (2000)37WKubischta Proc Workshop on Polarized Gas Targetsfor Storage Rings Heidelberg 23-26 September 1991 EdsHGGaul E Steffens and K Zapfe (Max-Planck-Institut furKernphysik Heidelberg)

38The labeling of the hyperfine states as |1〉 = |mj = +12 mI =+12〉 |2〉 = | + 12minus12〉 |3〉 = | minus 12+12〉 and |4〉 =|minus12minus12〉 for hydrogen and |1〉 = |+12+1〉 |2〉 = |+12 0〉|3〉 = | + 12minus1〉 |4〉 = | minus 12minus1〉 |5〉 = | minus 12 0〉 and

15

VI DEGREE OF DISSOCIATION

Besides the intensity of the atomic beam it is impor-tant to determine the molecular fraction in the beamMolecules injected into the feeding tube reduce the po-larization of the target gas

A Measurements with crossed-beam QMS

In addition to the data on the profile of the atomichydrogen beam (Sec VB) data on the distributions ofmolecular hydrogen in the beam were taken too at thepositions z = 567mm and 697mm behind the last mag-net The relation between the degree of dissociation andthe QMS signals by the atomic and molecular beam com-ponent was given above by Eq (6) The coefficient kv =vmva however is chosen here under the assumptionthat the average velocity of the atoms is determined bythe nozzle temperature of 65K and that of the moleculesby scattering and recombination on the ABS chamberwalls at 290K This yields kv =

radic

2 middot 65290 = 067 ingood agreement with Ref13 where this coefficient wasdetermined by the measured velocity distributions undersimilar conditionsThe measured profiles of the atomic fraction (identical

to those of Fig 21) those of the molecular fraction andthose of the degree of dissociation deduced from Eq (6)are collected in Fig 24As it is seen from the figure the distribution of the

degree of dissociation shows a dip around the central linedue to the higher density of molecular hydrogen originat-ing from the nozzle The mean value in an aperture of

0

20

40

60

80

100

120

2 4 6 8 10 12 14 16 18 20 22

0

20

40

60

80

100

120

6 8 10 12 14 16 18 20 22 24

QM

S si

gnal

s [a

u]

(a)

(b)

(c)

00

02

04

06

08

10

12

Deg

ree

of d

isso

ciat

ion

()

X [mm]

Y [mm]

(d)

00

02

04

06

08

10

12

Z = 567 mm

Z = 697 mm

FIG 24 Spatial distributions of H1 (bull)H2 () and degree ofdissociation () averaged over 3mm wide bands in the xz andyz planes respectively (here the z-axis is the geometrical axisof the ABS)

10mm diameter results as α = 095 plusmn 004

B Measurements with the Lamb-shift polarimeter

A cup in the quench chamber of the Lamb-shift po-larimeter (LSP) described in Ref8 allows one to mea-sure the currents Icup(H1) and Icup(H2) of the H+

1 andH+

2 ions extracted from the ionizer and separated bythe Wien filter with the cesium evaporation and the spinfilter switched off The relation between the degree ofdissociation α and the measured currents is

α =Icup(H1)minus r1

r2Icup(H2)

Icup(H1)minus r1r2Icup(H2) + 2kv

r2Icup(H2)

(10)

Among the three coefficients kv = 067 as for the mea-surement with the QMS For the electron energy of about100keV the ratio r1 of dissociative to non-dissociativeionization of H2 is8

r1 =σ(H2 rarr 2H+

1 )

σ(H2 rarr H+2 )

= 0095plusmn 0008 (11)

and the ratio between the ionization cross sections is857

r2 =σion(H2)

σion(H1)= 17plusmn 01 (12)

At the standard operation parameters of the source(Sec IV) the measured currents are Icup(H1) = (125 plusmn05) nA and Icup(H2) = (64 plusmn 01) nA yielding α =(096plusmn004) in excellent agreement with the value result-ing from the measurements with the QMS (Sec VIA)

VII BEAM POLARIZATION

The Lamb-shift polarimeter was designed built andtested at Universitat zu Koln7 It was used to measureand to optimize the polarization of the atomic hydrogenand deuterium beams delivered by the ABS Details arefound in Ref7The vector polarization pz for hydrogen is defined by

the relative hyperfine-state occupation numbers N(mI)

pz =N(+ 1

2 )minusN(minus 12 )

N(+ 12 ) +N(minus 1

2 ) (13)

for deuterium

pz =N(+1)minusN(minus1)

N(+1) +N(0) +N(minus1) (14)

Deuterium tensor polarization pzz is given by

pzz =N(+1) +N(minus1)minus 2 middotN(0)

N(+1) +N(0) +N(minus1) (15)

16

These polarizations can be derived from the measuredLyman-α peak strengths S by application of a number ofcorrection factors78

Typical Lyman α spectra measured with the polarizedhydrogen and deuterium beam from the ABS are shownin the Figs 25 and 26

MFT 2-3

200 300 400 5000

40

80

120

160

Magnetic field in the spinfilter [au]

PM

T si

gnal

[au

]

mI = -12

(a)

mI = +12

100 200 300 400 5000

40

80

120

160(b)

MFT 2-3WFT 1-3

mI = +12

mI = -12

FIG 25 Lyman-α spectra measured with the polarized hy-drogen beam (a) population change from state |2〉 to state|3〉 induced by the MFT unit (b) same as (a) with subsequentpopulation change from state |1〉 to state |3〉 induced by theWFT unit

MFT 3-4SFT 2-6

0

2

4

6

8

mI = -1m

I = 0

(a)

mI = +1

0

2

4

6

8(b)

MFT 3-4WFT 1-4 2-3

mI = +1 m

I = 0

mI = -1

mI = +1

mI = 0

mI = -1

mI = +1

mI = 0

mI = -1

0 1000 2000 3000

0

1

2

3

4 (c)

PMT

sign

al [a

u]

Magnetic field in the spinfilter [au]

WFT 1-4SFT 2-6

0 1000 2000 30000

2

4

6

8WFT 1-4SFT 3-5

(d)

FIG 26 Lyman-α spectra measured with the polarized deu-terium beam (a) and (b) vector polarization resulting fromsubsequent transitions MFT (3 rarr 4) and SFT (2 rarr 6) andWFT (1 rarr 4 2 rarr 3) respectively (c) and (d) tensor polar-ization resulting from subsequent transitions WFT (1 rarr 4)and SFT (2 rarr 6) and SFT (3 rarr 5) respectively

The polarization values for the hydrogen and the deu-terium beam derived from the Lyman-α peak-strengthratios with application of the necessary corrections aresummarized in Table IV

The vector polarization for hydrogen of the first linereflects the population of state |1〉 and state |2〉 accordingto the Eqs (3) and (13) The value of 091 deduced fromthe calculated transmission values is confirmed by themeasured one

TABLE IV The vector polarization pz of the hydrogen beamand the vector and the tensor polarization pzz of the deu-terium beam from the ABS measured with the Lamb-shiftpolarimeter

populated pz pzz

state(s)

Hydrogen |1〉 +089plusmn 001 -

|3〉 minus096plusmn 001 -

Deuterium |1〉+ |6〉 +088plusmn 001 +088plusmn 003

|3〉+ |4〉 minus091plusmn 001 +085plusmn 002

|3〉+ |6〉 +0005plusmn 0003 +090plusmn 001

|2〉+ |5〉 +0005plusmn 0003 minus171plusmn 003

VIII CONCLUSIONS AND OUTLOOK

In this paper we present the detailed description ofthe major components of the atomic beam source (ABS)for the polarized internal gas target of the magnet spec-trometer ANKE in COSY-Julich The ABS was builtfor the purpose of extending the physics program ofANKE from unpolarized and single-polarized investiga-tions with stored beams towards double-polarized exper-iments1 thus facilitating nuclear reaction studies involv-

ing ~p~p ~p~d ~d~p and ~d~d initial statesThe mechanical design took into account that at

ANKE the source has to be mounted vertically and trans-versely movable together with the transverse motion ofthe spectrometer magnet D2 The design of the system ofsextupole magnets took advantage of the developments inthe field of rare-earth permanent magnets (NdFeB) Ded-icated tools and methods were developed to determineand to optimize the source parameters ie intensity de-gree of dissociation and polarization Special emphasiswas put on the measurements of the spatial distributionsof the atomic and molecular beam near the focus wherethe feeding tube of the storage cell is located The ABShas been used in a number of investigations at ANKEthe commissioning effort to prepare the target for the usewith polarized H is described in Ref62 Performed stud-ies of the deuteron-charge exchange reaction are summa-rized in Ref6364 studies in near-threshold pion produc-tion are reported about in Ref65The ABS resides at the ANKE target position for a few

months per year only thus during the remaining timeit is used for other studies It had been observed thatthe nuclear polarization in recombined hydrogen is par-tially retained after recombination66 as well as evidencefor nuclear tensor polarization in recombined deuteriummolecules67 In order to investigate this recombinationprocess in more detail a special setup has been developedin the framework of an ISTC project68 and the recombi-nation process for different cell-wall coatings and differ-ent polarizations of the injected hydrogen or deuterium

17

atoms as function of cell-wall temperature strength ofthe magnetic holding field and gas pressure in the cell ispresently investigated69ndash71

Appendix A Preparation of Discharge Tubes and Nozzles

1 Tube Treatment

One end of the discharge tube is machined at a 45

angle while the other is kept flat Both ends are thenremelted and the tubes are tempered at 150 C Thetubes are further treated according to the procedure de-scribed in Ref72 which includes successive cleaning withacetone methanol distilled water and subsequent rins-ing by a 21 acid mixture of concentrated HF (40) andHCl (32) for 5 min The tubes are then flushed bydistilled water and dried

2 Nozzle Treatment

The nozzles are cleaned in an ultrasonic bath oftrichlorethylene acetone methanol and finally distilledwater all at 50 C Anodizing takes place in sulfuric acidto form a thin layer of Al2O3 as described in Ref72 Af-terwards they are immersed in distilled water for 30minat 95 C

ACKNOWLEDGMENTS

The authors want to thank OWB Schult Institut furKernphysik (IKP) Julich who initiated the polarizationprogram of ANKE Thanks go to the design office themechanical workshop and especially to WR Ermer allIKP Valuable advice was received from the PINTEX col-laboration at IUCF from the target group at HERMESespecially NKoch and from DToporkov BINP Novosi-birsk The support by VKoptev PNPI Gatchina whoregrettably passed away in January 2012 is gratefullyacknowledged Thanks go also to R Poprawe and col-leagues Fraunhofer-Institut fur Lasertechnik Aachenwhere the encapsulations of the magnets were laser-welded

1AKacharava F Rathmann and CWilkin Spin Physics fromCOSY to FAIR COSY Experiment Proposal No 152 (2005)Available under httparXivnucl-ex0511028

2WHaeberli in Proc 2nd Int Symp on Polarization Phenomenaof Nucleons Karlsruhe 1965 Eds P Huber and H SchopperExperientia Supplementum 12 64 (Birkhauser Verlag 1966)

3E Steffens and WHaeberli Rep Progr Phys 66 1887 (2003)4SBarsov et al Nucl Instr and Meth A 462 364 (2001)5RMaier Nucl Instr and Meth A 390 1 (1997)6KGrigoryev et al Proc 14th International Workshop on Po-larized Sources Targets and Polarimetry (PSTP 2011) 12-16 September 2011 StPetersburg Russia eds KGrigoryevPKravtsov and AVasilyev ISBN 978-5-86763-282-3 61 (2011)

7REngels et al Rev Sci Instrum 74 4607 (2003)8REngels et al Rev Sci Instrum 76 053305 (2005)9TWise et al Nucl Instr and Meth A 336 410 (1993)

10WADezarn et al Nucl Instr and Meth A 362 36 (1995)11T Rinckel et al Nucl Instr and Meth A 439 117 (2000)12F Stock et al Nucl Instr and Meth A 343 334 (1994)13ANass et al Nucl Instr and Meth A 505 633 (2003)14VDerenchuk et al Proc Conf Polarized Ion Sources and Polar-ized Gas Targets Madison WI 1993 Eds LWAnderson andWHaeberli AIP Conf Proc 293 72 (American Institute ofPhysics 1994)

15HOkamura et al see Ref14 p 8416KHatanaka et al Nucl Instr and Meth A 384 575 (1997)17Manufacturer Schiffer Metall- amp Vakuumtechnik 52428 JulichGermany

18Single-stage type RGS120 refrigerating capacity 120W at 80Kand 20W at 30K Leybold Vacuum GmbH 50968 Koln Ger-many

19Mini UHV gate valve series 010 VAT Germany GmbH 85630Grasbrunn Germany

20Type F3 fomblin oil Pfeiffer Vacuum GmbH 35614 Asslar Ger-many

21Model HU 1 Leybold Vacuum GmbH 50968 Koln Germany22Manufacturer SK Industriemodell GmbH 52531 Ubach-Palenberg Germany

23Type PFG 600 RF with automatic matchbox PFM 1500 A-INDHuttinger Elektronik GmbH 79110 Freiburg Germany

24WKorsch PhD Thesis Philipps Universitat Marburg (1990)25F Stock et al Int Workshop on Polarized Beams and PolarizedGas Targets Koeln Germany 1995 Eds HPaetz gen Schieckand L Sydow (World Scientific Publ Co 1996) p 260

26The first number denotes the outer diameter and the second onethe wall thickness

27Type Duran 8330 equivalent to Corning 7740 (Pyrex) SchottAG 55122 Mainz Germany

28Ultra-Kryomat RUL 80-D Lauda DrRWobser GmbH 97912Lauda-Konigshofen Germany

29ODU-Kontakt GmbH 84444 Muhldorf Germany30Handbook of Chemistry and Physics Ed RCEast (The Chem-ical Rubber Co 1973) p E-10

31NKoch and E Steffens Rev Sci Instrum 70 1631 (1999)32AVassiliev et al Petersburg Nuclear Physics Institute ReportNP-32-1997 No 2175 (1997)

33BLorentz Diploma Thesis Ruprecht-Karls-Universitat Heidel-berg (1993)

34ANass and E Steffens Nucl Instr and Meth A 598 653(2009)

35WHaeberli Ann Rev Nucl Sci 17 373 (1967)36AVassiliev et al Rev Sci Instr 71 3331 (2000)37WKubischta Proc Workshop on Polarized Gas Targetsfor Storage Rings Heidelberg 23-26 September 1991 EdsHGGaul E Steffens and K Zapfe (Max-Planck-Institut furKernphysik Heidelberg)

38The labeling of the hyperfine states as |1〉 = |mj = +12 mI =+12〉 |2〉 = | + 12minus12〉 |3〉 = | minus 12+12〉 and |4〉 =|minus12minus12〉 for hydrogen and |1〉 = |+12+1〉 |2〉 = |+12 0〉|3〉 = | + 12minus1〉 |4〉 = | minus 12minus1〉 |5〉 = | minus 12 0〉 and

16

These polarizations can be derived from the measuredLyman-α peak strengths S by application of a number ofcorrection factors78

Typical Lyman α spectra measured with the polarizedhydrogen and deuterium beam from the ABS are shownin the Figs 25 and 26

MFT 2-3

200 300 400 5000

40

80

120

160

Magnetic field in the spinfilter [au]

PM

T si

gnal

[au

]

mI = -12

(a)

mI = +12

100 200 300 400 5000

40

80

120

160(b)

MFT 2-3WFT 1-3

mI = +12

mI = -12

FIG 25 Lyman-α spectra measured with the polarized hy-drogen beam (a) population change from state |2〉 to state|3〉 induced by the MFT unit (b) same as (a) with subsequentpopulation change from state |1〉 to state |3〉 induced by theWFT unit

MFT 3-4SFT 2-6

0

2

4

6

8

mI = -1m

I = 0

(a)

mI = +1

0

2

4

6

8(b)

MFT 3-4WFT 1-4 2-3

mI = +1 m

I = 0

mI = -1

mI = +1

mI = 0

mI = -1

mI = +1

mI = 0

mI = -1

0 1000 2000 3000

0

1

2

3

4 (c)

PMT

sign

al [a

u]

Magnetic field in the spinfilter [au]

WFT 1-4SFT 2-6

0 1000 2000 30000

2

4

6

8WFT 1-4SFT 3-5

(d)

FIG 26 Lyman-α spectra measured with the polarized deu-terium beam (a) and (b) vector polarization resulting fromsubsequent transitions MFT (3 rarr 4) and SFT (2 rarr 6) andWFT (1 rarr 4 2 rarr 3) respectively (c) and (d) tensor polar-ization resulting from subsequent transitions WFT (1 rarr 4)and SFT (2 rarr 6) and SFT (3 rarr 5) respectively

The polarization values for the hydrogen and the deu-terium beam derived from the Lyman-α peak-strengthratios with application of the necessary corrections aresummarized in Table IV

The vector polarization for hydrogen of the first linereflects the population of state |1〉 and state |2〉 accordingto the Eqs (3) and (13) The value of 091 deduced fromthe calculated transmission values is confirmed by themeasured one

TABLE IV The vector polarization pz of the hydrogen beamand the vector and the tensor polarization pzz of the deu-terium beam from the ABS measured with the Lamb-shiftpolarimeter

populated pz pzz

state(s)

Hydrogen |1〉 +089plusmn 001 -

|3〉 minus096plusmn 001 -

Deuterium |1〉+ |6〉 +088plusmn 001 +088plusmn 003

|3〉+ |4〉 minus091plusmn 001 +085plusmn 002

|3〉+ |6〉 +0005plusmn 0003 +090plusmn 001

|2〉+ |5〉 +0005plusmn 0003 minus171plusmn 003

VIII CONCLUSIONS AND OUTLOOK

In this paper we present the detailed description ofthe major components of the atomic beam source (ABS)for the polarized internal gas target of the magnet spec-trometer ANKE in COSY-Julich The ABS was builtfor the purpose of extending the physics program ofANKE from unpolarized and single-polarized investiga-tions with stored beams towards double-polarized exper-iments1 thus facilitating nuclear reaction studies involv-

ing ~p~p ~p~d ~d~p and ~d~d initial statesThe mechanical design took into account that at

ANKE the source has to be mounted vertically and trans-versely movable together with the transverse motion ofthe spectrometer magnet D2 The design of the system ofsextupole magnets took advantage of the developments inthe field of rare-earth permanent magnets (NdFeB) Ded-icated tools and methods were developed to determineand to optimize the source parameters ie intensity de-gree of dissociation and polarization Special emphasiswas put on the measurements of the spatial distributionsof the atomic and molecular beam near the focus wherethe feeding tube of the storage cell is located The ABShas been used in a number of investigations at ANKEthe commissioning effort to prepare the target for the usewith polarized H is described in Ref62 Performed stud-ies of the deuteron-charge exchange reaction are summa-rized in Ref6364 studies in near-threshold pion produc-tion are reported about in Ref65The ABS resides at the ANKE target position for a few

months per year only thus during the remaining timeit is used for other studies It had been observed thatthe nuclear polarization in recombined hydrogen is par-tially retained after recombination66 as well as evidencefor nuclear tensor polarization in recombined deuteriummolecules67 In order to investigate this recombinationprocess in more detail a special setup has been developedin the framework of an ISTC project68 and the recombi-nation process for different cell-wall coatings and differ-ent polarizations of the injected hydrogen or deuterium

17

atoms as function of cell-wall temperature strength ofthe magnetic holding field and gas pressure in the cell ispresently investigated69ndash71

Appendix A Preparation of Discharge Tubes and Nozzles

1 Tube Treatment

One end of the discharge tube is machined at a 45

angle while the other is kept flat Both ends are thenremelted and the tubes are tempered at 150 C Thetubes are further treated according to the procedure de-scribed in Ref72 which includes successive cleaning withacetone methanol distilled water and subsequent rins-ing by a 21 acid mixture of concentrated HF (40) andHCl (32) for 5 min The tubes are then flushed bydistilled water and dried

2 Nozzle Treatment

The nozzles are cleaned in an ultrasonic bath oftrichlorethylene acetone methanol and finally distilledwater all at 50 C Anodizing takes place in sulfuric acidto form a thin layer of Al2O3 as described in Ref72 Af-terwards they are immersed in distilled water for 30minat 95 C

ACKNOWLEDGMENTS

The authors want to thank OWB Schult Institut furKernphysik (IKP) Julich who initiated the polarizationprogram of ANKE Thanks go to the design office themechanical workshop and especially to WR Ermer allIKP Valuable advice was received from the PINTEX col-laboration at IUCF from the target group at HERMESespecially NKoch and from DToporkov BINP Novosi-birsk The support by VKoptev PNPI Gatchina whoregrettably passed away in January 2012 is gratefullyacknowledged Thanks go also to R Poprawe and col-leagues Fraunhofer-Institut fur Lasertechnik Aachenwhere the encapsulations of the magnets were laser-welded

1AKacharava F Rathmann and CWilkin Spin Physics fromCOSY to FAIR COSY Experiment Proposal No 152 (2005)Available under httparXivnucl-ex0511028

2WHaeberli in Proc 2nd Int Symp on Polarization Phenomenaof Nucleons Karlsruhe 1965 Eds P Huber and H SchopperExperientia Supplementum 12 64 (Birkhauser Verlag 1966)

3E Steffens and WHaeberli Rep Progr Phys 66 1887 (2003)4SBarsov et al Nucl Instr and Meth A 462 364 (2001)5RMaier Nucl Instr and Meth A 390 1 (1997)6KGrigoryev et al Proc 14th International Workshop on Po-larized Sources Targets and Polarimetry (PSTP 2011) 12-16 September 2011 StPetersburg Russia eds KGrigoryevPKravtsov and AVasilyev ISBN 978-5-86763-282-3 61 (2011)

7REngels et al Rev Sci Instrum 74 4607 (2003)8REngels et al Rev Sci Instrum 76 053305 (2005)9TWise et al Nucl Instr and Meth A 336 410 (1993)

10WADezarn et al Nucl Instr and Meth A 362 36 (1995)11T Rinckel et al Nucl Instr and Meth A 439 117 (2000)12F Stock et al Nucl Instr and Meth A 343 334 (1994)13ANass et al Nucl Instr and Meth A 505 633 (2003)14VDerenchuk et al Proc Conf Polarized Ion Sources and Polar-ized Gas Targets Madison WI 1993 Eds LWAnderson andWHaeberli AIP Conf Proc 293 72 (American Institute ofPhysics 1994)

15HOkamura et al see Ref14 p 8416KHatanaka et al Nucl Instr and Meth A 384 575 (1997)17Manufacturer Schiffer Metall- amp Vakuumtechnik 52428 JulichGermany

18Single-stage type RGS120 refrigerating capacity 120W at 80Kand 20W at 30K Leybold Vacuum GmbH 50968 Koln Ger-many

19Mini UHV gate valve series 010 VAT Germany GmbH 85630Grasbrunn Germany

20Type F3 fomblin oil Pfeiffer Vacuum GmbH 35614 Asslar Ger-many

21Model HU 1 Leybold Vacuum GmbH 50968 Koln Germany22Manufacturer SK Industriemodell GmbH 52531 Ubach-Palenberg Germany

23Type PFG 600 RF with automatic matchbox PFM 1500 A-INDHuttinger Elektronik GmbH 79110 Freiburg Germany

24WKorsch PhD Thesis Philipps Universitat Marburg (1990)25F Stock et al Int Workshop on Polarized Beams and PolarizedGas Targets Koeln Germany 1995 Eds HPaetz gen Schieckand L Sydow (World Scientific Publ Co 1996) p 260

26The first number denotes the outer diameter and the second onethe wall thickness

27Type Duran 8330 equivalent to Corning 7740 (Pyrex) SchottAG 55122 Mainz Germany

28Ultra-Kryomat RUL 80-D Lauda DrRWobser GmbH 97912Lauda-Konigshofen Germany

29ODU-Kontakt GmbH 84444 Muhldorf Germany30Handbook of Chemistry and Physics Ed RCEast (The Chem-ical Rubber Co 1973) p E-10

31NKoch and E Steffens Rev Sci Instrum 70 1631 (1999)32AVassiliev et al Petersburg Nuclear Physics Institute ReportNP-32-1997 No 2175 (1997)

33BLorentz Diploma Thesis Ruprecht-Karls-Universitat Heidel-berg (1993)

34ANass and E Steffens Nucl Instr and Meth A 598 653(2009)

35WHaeberli Ann Rev Nucl Sci 17 373 (1967)36AVassiliev et al Rev Sci Instr 71 3331 (2000)37WKubischta Proc Workshop on Polarized Gas Targetsfor Storage Rings Heidelberg 23-26 September 1991 EdsHGGaul E Steffens and K Zapfe (Max-Planck-Institut furKernphysik Heidelberg)

38The labeling of the hyperfine states as |1〉 = |mj = +12 mI =+12〉 |2〉 = | + 12minus12〉 |3〉 = | minus 12+12〉 and |4〉 =|minus12minus12〉 for hydrogen and |1〉 = |+12+1〉 |2〉 = |+12 0〉|3〉 = | + 12minus1〉 |4〉 = | minus 12minus1〉 |5〉 = | minus 12 0〉 and

17

atoms as function of cell-wall temperature strength ofthe magnetic holding field and gas pressure in the cell ispresently investigated69ndash71

Appendix A Preparation of Discharge Tubes and Nozzles

1 Tube Treatment

One end of the discharge tube is machined at a 45

angle while the other is kept flat Both ends are thenremelted and the tubes are tempered at 150 C Thetubes are further treated according to the procedure de-scribed in Ref72 which includes successive cleaning withacetone methanol distilled water and subsequent rins-ing by a 21 acid mixture of concentrated HF (40) andHCl (32) for 5 min The tubes are then flushed bydistilled water and dried

2 Nozzle Treatment

The nozzles are cleaned in an ultrasonic bath oftrichlorethylene acetone methanol and finally distilledwater all at 50 C Anodizing takes place in sulfuric acidto form a thin layer of Al2O3 as described in Ref72 Af-terwards they are immersed in distilled water for 30minat 95 C

ACKNOWLEDGMENTS

The authors want to thank OWB Schult Institut furKernphysik (IKP) Julich who initiated the polarizationprogram of ANKE Thanks go to the design office themechanical workshop and especially to WR Ermer allIKP Valuable advice was received from the PINTEX col-laboration at IUCF from the target group at HERMESespecially NKoch and from DToporkov BINP Novosi-birsk The support by VKoptev PNPI Gatchina whoregrettably passed away in January 2012 is gratefullyacknowledged Thanks go also to R Poprawe and col-leagues Fraunhofer-Institut fur Lasertechnik Aachenwhere the encapsulations of the magnets were laser-welded

1AKacharava F Rathmann and CWilkin Spin Physics fromCOSY to FAIR COSY Experiment Proposal No 152 (2005)Available under httparXivnucl-ex0511028

2WHaeberli in Proc 2nd Int Symp on Polarization Phenomenaof Nucleons Karlsruhe 1965 Eds P Huber and H SchopperExperientia Supplementum 12 64 (Birkhauser Verlag 1966)

3E Steffens and WHaeberli Rep Progr Phys 66 1887 (2003)4SBarsov et al Nucl Instr and Meth A 462 364 (2001)5RMaier Nucl Instr and Meth A 390 1 (1997)6KGrigoryev et al Proc 14th International Workshop on Po-larized Sources Targets and Polarimetry (PSTP 2011) 12-16 September 2011 StPetersburg Russia eds KGrigoryevPKravtsov and AVasilyev ISBN 978-5-86763-282-3 61 (2011)

7REngels et al Rev Sci Instrum 74 4607 (2003)8REngels et al Rev Sci Instrum 76 053305 (2005)9TWise et al Nucl Instr and Meth A 336 410 (1993)

10WADezarn et al Nucl Instr and Meth A 362 36 (1995)11T Rinckel et al Nucl Instr and Meth A 439 117 (2000)12F Stock et al Nucl Instr and Meth A 343 334 (1994)13ANass et al Nucl Instr and Meth A 505 633 (2003)14VDerenchuk et al Proc Conf Polarized Ion Sources and Polar-ized Gas Targets Madison WI 1993 Eds LWAnderson andWHaeberli AIP Conf Proc 293 72 (American Institute ofPhysics 1994)

15HOkamura et al see Ref14 p 8416KHatanaka et al Nucl Instr and Meth A 384 575 (1997)17Manufacturer Schiffer Metall- amp Vakuumtechnik 52428 JulichGermany

18Single-stage type RGS120 refrigerating capacity 120W at 80Kand 20W at 30K Leybold Vacuum GmbH 50968 Koln Ger-many

19Mini UHV gate valve series 010 VAT Germany GmbH 85630Grasbrunn Germany

20Type F3 fomblin oil Pfeiffer Vacuum GmbH 35614 Asslar Ger-many

21Model HU 1 Leybold Vacuum GmbH 50968 Koln Germany22Manufacturer SK Industriemodell GmbH 52531 Ubach-Palenberg Germany

23Type PFG 600 RF with automatic matchbox PFM 1500 A-INDHuttinger Elektronik GmbH 79110 Freiburg Germany

24WKorsch PhD Thesis Philipps Universitat Marburg (1990)25F Stock et al Int Workshop on Polarized Beams and PolarizedGas Targets Koeln Germany 1995 Eds HPaetz gen Schieckand L Sydow (World Scientific Publ Co 1996) p 260

26The first number denotes the outer diameter and the second onethe wall thickness

27Type Duran 8330 equivalent to Corning 7740 (Pyrex) SchottAG 55122 Mainz Germany

28Ultra-Kryomat RUL 80-D Lauda DrRWobser GmbH 97912Lauda-Konigshofen Germany

29ODU-Kontakt GmbH 84444 Muhldorf Germany30Handbook of Chemistry and Physics Ed RCEast (The Chem-ical Rubber Co 1973) p E-10

31NKoch and E Steffens Rev Sci Instrum 70 1631 (1999)32AVassiliev et al Petersburg Nuclear Physics Institute ReportNP-32-1997 No 2175 (1997)

33BLorentz Diploma Thesis Ruprecht-Karls-Universitat Heidel-berg (1993)

34ANass and E Steffens Nucl Instr and Meth A 598 653(2009)

35WHaeberli Ann Rev Nucl Sci 17 373 (1967)36AVassiliev et al Rev Sci Instr 71 3331 (2000)37WKubischta Proc Workshop on Polarized Gas Targetsfor Storage Rings Heidelberg 23-26 September 1991 EdsHGGaul E Steffens and K Zapfe (Max-Planck-Institut furKernphysik Heidelberg)

38The labeling of the hyperfine states as |1〉 = |mj = +12 mI =+12〉 |2〉 = | + 12minus12〉 |3〉 = | minus 12+12〉 and |4〉 =|minus12minus12〉 for hydrogen and |1〉 = |+12+1〉 |2〉 = |+12 0〉|3〉 = | + 12minus1〉 |4〉 = | minus 12minus1〉 |5〉 = | minus 12 0〉 and

18

|6〉 = | minus 12+1〉 for deuterium follows that of Ref3539Produced from VACODYM 510HR 383HR and 400HR by Vacu-umschmelze GmbH 63412 Hanau Germany

40KHalbach Nucl Instr and Meth 169 1 (1980)41Welding performed at Fraunhofer-Institut fur Lasertechnik52074 Aachen Germany

42AAbragam and JM Winter Phys Rev Lett 1 374 (1958)43S Lorenz Diploma Thesis Friedrich-Alexander-UniversitatErlangen-Nurnberg (1999)

44H-GGaul and E Steffens Nucl Instr and Meth A 316 297(1992)

45SOh Nucl Instr and Meth 82 189 (1970)46HPaetz gen Schieck Nucl Instr and Meth A 587 213 (2008)47RJ Philpott Nucl Instr and Meth A 259 317 (1987)48H Jansch et al Hyperfine Interactions 22 253 (1985)49ADRoberts et al Nucl Instr and Meth A 322 6 (1992)50MCapiluppi et al httptheorjinrru~spin2012talkss6Steffenspdf(to be published in Physics of ElementaryParticles and Atomic Nuclei JINR Russiahttppepanjinrrupepanengabout)

51Manufactured by St Petersburg Nuclear Physics Institute188300 Gatchina Russia

52HKleines et al Nucl Instr Meth A 560 503 (2006)53AVassiliev et al Petersburg Nuclear Physics Institute ReportEP-46-1998 No 2260 (1998)

54AVassiliev et al Proc Int Workshop Polarized Sourcesand Targets Erlangen Germany September 29 -October 21999 Eds AGute S Lorenz E Steffens (Universitat Erlangen-Nurnberg 1999) p 200

55MMikirtytchiants Diploma Thesis St Petersburg State Tech-nical University (1999)

56MMikirtytchiants et al see Ref54 p 47857YKKim et al Electron-impact cross section database 2002httppysicsnistgovPhysRefDataIonization

58MNekipelov Diploma Thesis St Petersburg State TechnicalUniversity (1999)

59MNekipelov et al see Ref54 p 48660ARoth Vacuum Technology (Elsevier Amsterdam 1996)61TEPhipps and JBTaylor Phys Rev 29 309 (1927)62MMikirtychyants et al J Phys Conf Ser 295 012148 (2011)63DMchedlishvili et al J Phys Conf Ser 295 012099 (2011)64FRathmann J Phys Conf Ser 295 012006 (2011)65SDymov (for the ANKE collaboration) J Phys Conf Ser 295012095 (2011)

66TWise et al Phys Rev Lett 87 042701 (2001)67JFJ van denBrand et al PhysRev Lett 78 1235 (1997)68International Science and Technology Center Project No 186169Work now financed by Deutsche Forschungsgemeinschaft project436 RUS 11397701

70REngels et al Proc 13th Int Workshop on Polarized SourcesTargets and Polarimetry Ferrara Italy September 7-11 2009Eds G Ciullo MContalbrigo P Lenisa (World Scientific 2011)p 215

71REngels et al J Phys Conf Ser 295 012161 (2011)72NKoch PhD Thesis Friedrich-Alexander-Universitat Erlangen-Nurnberg (1999)