Deep Virtual Exclusive Scattering with the Electron Ion Collider: Detector Design

and Physics Goals Charles Hyde

Old Dominion University

INT 17-3 25-29 Sept 2017

1

Imaging Quarks & Gluons in the Nucleon and in Nuclei

DVCS: ep—> ep𝛾: ef2, gluons

Deep Virtual 𝜙:

Gluons dominate with 10-20% s-quark interference at modest x

Strong Sudakov corrections for Q2 < 10 GeV2 (Goloskokov & Kroll)

J/Psi: Gluons (intrinsic charm at high-x?)

Pseudo Scaler mesons: Higher twist DA (instanton effects) and Nucleon GPD for Q2< 10 GeV2

𝝆, ω-meson, flavor sensitivity, mechanism unclear at modest Q2

strong violation of SCHC in JLab, HERMES data.

2

JLab 12 GeVExtensive program on proton, x>0.2, Q2>2,

Unpolarized (Halls A,B,C) and Longitudinally polarized (Hall B)

Deuteron

D(e,e’Vn)p, V = 𝛾, vector-meson…

D(e,e’VpS)n: p(pS) > 100 MeV/c

Coherent D, 4He (ALERT Detector)

Transversely polarized targets still challenging (CLAS12 HDice , SOLiD TCS on transversely NH3)

3

Electron Ion Colliders

EIC≈ [s

JLabs

HERA]

1/2

Luminosity ≈ 1034

/cm2/s

High statistics possible on multiple beam-species & polarization states in ≤ 10 years

L & T polarized light ion beams

Access to full spin structure of GPDs

Ion beam has momentum PA

= ZP0

and Rigidity K= PA

/Z = P0

Spectator (p=0 in rest frame) has forward momentum ZP0/A

Tag active neutron ≈”on-shell” by spectator proton tagging in polarized D, 3He

Requires good acceptance and resolution for protons at ≈ 0.5 or 0.33 of beam rigidity

Daughter nucleus Rigidity relative to beam: K/K0

= (A’/Z’)(Z/A)

|ΔK|/K0 ≥ 1% for A≤ 100 (positive for ΔZ = –1, negative for ΔZ=0, ΔA = –1.

Veto nuclear break-up,

Tag spectators, evaporation n,p, d , residual nuclei

4

3"12%GeV

8"100(400)%GeV

8%GeV

Detector Requirements: DVCS on the Proton(also 𝜋0 and η)

Exclusivity:p(e,e’𝛾p) triple coincidence (or N* ➙ Nπ veto)veto neutron in ZDC or proton with p/p0 ≲ M/M*

Imaging:

t = Δ2 resolution requires dispersive focus at Roman Pots. Measure Δ=(p’-p) [better resolution than Δ=(k-k’-q’) ].

Full proton detection acceptance to “Beam-Stay-Clear (BSC)” limit of ∼10× beam rms emittance

Introduction Motivation

Deeply Virtual Compton Scattering (DVCS): �⇤ p ! � p

Handbag diagram

High Q2

Perturbative QCD

Non-perturbativeGPDs

Bjorken limit:

Q2 = �q2 ! 1⌫ ! 1

�xB =

Q2

2M⌫fixed

GPDs accesible through DVCS only at Q2 ! 1Actual value of Q2

must be tested and established by experimentCarlos Munoz Camacho (IPN-Orsay) New DVCS results from Hall A INT-17-3 3 / 23

𝜉 ≈ xB/(2-xB)

5

Deeply Virtual Compton Scattering on the Proton: Transverse Imaging vs xB

0

10

20

0.001 0.01 0.1 1

Q2 (

GeV

2 )

xB

0.1

1

10

102

103

104

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6

dσ

DV

CS/d

t (p

b/G

eV

2)

|t| (GeV2)

γ∗ + p → γ + p

20 GeV on 250 GeV

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6

5 GeV on 100 GeV

∫Ldt = 10 fb-1

|t| (GeV2)

dσ

DV

CS/d

t (p

b/G

eV

2)

γ∗ + p → γ + p

0.004 < xB < 0.0063

10 < Q2/GeV2 < 17.8

0.1 < xB < 0.16

10 < Q2/GeV2 < 17.8 0

0.2

0.4

0.6

0.8

1

0

0.01

0.02

1.4 1.6 1.8

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6

xB F

(xB, b

T)

(fm

-2)

bT (fm)

0

0.1

0.2

0.3

0.4

0.5

0.6

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6

0

0.005

0.01

1.4 1.6 1.8

xB F

(xB, b

T)

(fm

-2)

bT (fm)

Figure 2.21: Top: The DVCS cross-section in two bins of x and Q2. The error bars reflectstatistical and assumed systematic uncertainties, but not the overall normalization uncertaintyfrom the luminosity measurement. For the left panels the assumed luminosity is 10 fb�1 for|t| < 1GeV2 and 100 fb�1 for |t| > 1GeV2. Bottom: The distribution of partons in impactparameter b

T

obtained from the DVCS cross-section. The bands represent the parametric errorsin the fit of d�DV CS

/dt and the uncertainty from di↵erent extrapolations to the regions ofunmeasured (very low and very high) t, as specified in Sec. 3.6 of [2].

partons is “smeared” around the measuredvalue of ⇠ = x/(2 � x), whereas the vari-able b

T

is legitimately interpreted as a trans-verse parton position [85]. The bottom pan-els of Figure 2.21 show that precise imagesare obtained in a wide range of b

T

, includ-ing the large b

T

region where a characteris-tic dependence on b

T

and x due to virtual

pion fluctuations is predicted as discussed inSec. 2.4.1. We emphasize that a broad accep-tance in t is essential to achieve this accuracy.If, for instance, the measured region of |t|starts at (300MeV)2 instead of (175MeV)2,the associated extrapolation uncertainty ex-ceeds 50% for b

T

> 1.5 fm with the modelused here.

53

y=0.7 @ (10�100 GeV2)

ImpacttoDVCSphysicsmeasurementsoflimitedacceptance(II)• Letslookathowtheuncertaintychangesontheparton distribution withdifferentacceptancecuts

11

[fm]Tb0 0.2 0.4 0.6 0.8 1 1.2

]-2) [

fmt,b B

F(x

Bx

0

0.2

0.4

0.6

0.8

1

1.2

-1DVCS - 20 GeV x 250 GeV - 10 fb

0.4<|Pt|(GeV)<1.3

[fm]Tb0 0.2 0.4 0.6 0.8 1 1.2

]-2) [

fmt,b B

F(x

Bx

0

0.2

0.4

0.6

0.8

1

1.2

1.4

-1DVCS - 20 GeV x 250 GeV - 10 fb

0.44<|Pt|(GeV)<1.3

[fm]Tb0 0.2 0.4 0.6 0.8 1 1.2

]-2) [

fmt,b B

F(x

Bx

0

0.2

0.4

0.6

0.8

1

1.2

1.4

-1DVCS - 20 GeV x 250 GeV - 10 fb

0.18 <|Pt|(GeV)<1.0

[fm]Tb0 0.2 0.4 0.6 0.8 1 1.2

]-2) [

fmt,b B

F(x

Bx

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

-1DVCS - 20 GeV x 250 GeV - 10 fb

0.18 <|Pt|(GeV)<0.8

[fm]Tb0 0.2 0.4 0.6 0.8 1 1.2

]-2) [

fmt,b B

F(x

Bx

0

0.2

0.4

0.6

0.8

1

1.2

-1DVCS - 20 GeV x 250 GeV - 10 fb

0.18<|Pt|(GeV)<1.3

0.18<|PT|<1.3 GeV

10<Q2<18 GeV2

0.004 < xB < 0.006

10 fb–1

ImpacttoDVCSphysicsmeasurementsoflimitedacceptance(II)• Letslookathowtheuncertaintychangesontheparton distribution withdifferentacceptancecuts

11

[fm]Tb0 0.2 0.4 0.6 0.8 1 1.2

]-2

) [fm

t,b B

F(x

Bx

0

0.2

0.4

0.6

0.8

1

1.2

-1DVCS - 20 GeV x 250 GeV - 10 fb

0.4<|Pt|(GeV)<1.3

[fm]Tb0 0.2 0.4 0.6 0.8 1 1.2

]-2

) [fm

t,b B

F(x

Bx0

0.2

0.4

0.6

0.8

1

1.2

1.4

-1DVCS - 20 GeV x 250 GeV - 10 fb

0.44<|Pt|(GeV)<1.3

[fm]Tb0 0.2 0.4 0.6 0.8 1 1.2

]-2

) [fm

t,b B

F(x

Bx

0

0.2

0.4

0.6

0.8

1

1.2

1.4

-1DVCS - 20 GeV x 250 GeV - 10 fb

0.18 <|Pt|(GeV)<1.0

[fm]Tb0 0.2 0.4 0.6 0.8 1 1.2

]-2

) [fm

t,b B

F(x

Bx

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

-1DVCS - 20 GeV x 250 GeV - 10 fb

0.18 <|Pt|(GeV)<0.8

[fm]Tb0 0.2 0.4 0.6 0.8 1 1.2

]-2

) [fm

t,b B

F(x

Bx

0

0.2

0.4

0.6

0.8

1

1.2

-1DVCS - 20 GeV x 250 GeV - 10 fb

0.18<|Pt|(GeV)<1.3

Tagging the recoil protons over the full momentum range is essential for precision imaging

6

eRHIC Interaction Region Optics

7

• Spectator protons in RP1,2

• DVCS Recoil protons in RP3,4

JLEIC Full Acceptance Detector

e/𝜋/K/p PID in central region

Full acceptance in forward region, tracking detector design still required.

8

IPe

ions

forward'e'detectionCompton'

polarimetry'dispersion'suppressor/geometric'matchspectrometers

forward'ion'detection

ZDCHigh-Dispersion

Focus (DVES protons)

3He, d spectator

proton tagging

e

EIC vs HERA

9

8 July 2016Charles Hyde EIC UG 8

Acceptance for p’ in DDIS

Acceptance in diffractive peak (XL>~.98) ZEUS: ~2% JLEIC: ~100%

JLEIC ZEUS Leading Proton Spectrometer

Region 2 (Hi. Res)

Region 1

31

t

XL

1

Pt

0.2

0.994 1XL

(GeV)

(GeV2)

0

1

2.5

2 T•m Dipole

20 T•m Dipole

/DVES

Tagging essential for exclusivity

Deep Virtual Vector Meson Production

p(e,e’V)X:

Reconstruct t=Δ2 = (k-k’-p+–p–)2 from charged particle final states: ρ➙𝜋+𝜋–, 𝜙➙K+K–

ω➙𝜋+𝜋–𝜋0 constraint on ω mass refines ω momentum resolution.

Tagging/Veto required for exclusivity

10

DVES on Nuclei

Precision charge densities measured in 1970s ☛

“Neutron Skin” of heavy nuclei has implications for nuclear equation of state & neutron star structure.

p/n ≅ u-quark / d-quark

𝝆,ω: DVES amplitude has charge weight eu∓ed.

Gluon profiles of nuclei from J/𝛹 and 𝜙

VOLUME 38, +UMBER 4 24 JaNUxRv 1977

to + 0.05/o by the field maps of the SP900 spec-trometer. Scattering angles were checked to beaccurate to +0.05'. The incident beam currentwas measured by ferrite monitors and a Faradaycup. The scattered electrons were detected usingthe standard focal-plane equipment. ' Special at-tention was paid to long-term stability which wasfound to be better than a 2%. The overall detec-tion efficiency was obtained by normalizing theangular distribution measured to the Stanford Lin-ear Accelerator Center' (SLAC) and the Universi-ty of Mainz' measurements of "'Pb cross sec-tions at 1.7 fm ', where both sets of data closelyagree. The normalization has been determinedto +3%; it was verified by measuring "C crosssections" at low momentum transfers.The target of 217+ 2 mg/cm' '"Pb (99.14/o) was

held between two aluminum foils. Rater circulat-ing between the aluminum foils cooled the target,and allowed the use of an average beam intensityof 20 pA necessary to measure cross sectionsdown to 10 '0 mb/sr. Aluminum and oxygen con-tributions were separated by recoil energy dif-ference. Background was absent.The experimental results are shown in Fig. 1,

together with previous 502-Me V data taken atSLAC.' The data now span 12 decades.The data analysis has been performed accord-

ing to Sick." The density is expanded on a basisof a sum of Gaussians, the amplitudes of whichare fitted to the data. The limitation to full mod-el independence comes in through the use of Gauss-ians of finite width. This restricts the ampli-tudes of unmeasured high-frequency Four ier com-ponents of p(r). According to present theoreticalunderstanding the amplitudes of such componentsare expected to be severely limited; this is dueto the Schrodinger equation that strongly couplessecond derivatives of nucleon wave functions toknown energy eigenvalues. The width parameterused, y = 1.388 fm, allows one to reproduce a num-ber of theoretical 2 'Pb densities" "with lessthan 0.1% deviation and therefore provides enoughflexibility to reproduce any fine structure in p(r)occurring in presently existing theoretical densi-ties.The error bars on the resulting density are

hence expected to include a realistic estimatefor the completeness error (due to the finite q „).In order to get the most reliable estimate for

p(r), we have included in our analysis all dataconcerning electromagnetic information on ' Pb.The result presented here is based on the mostrecent data published by different laboratories

lo(a)

lo

IO

~ ~

+ SACLAY 76~ STANFORD 69

10

lO'

10+

Eb)g lo~

lo

lO'

-10

R; Q;0 I 0 0038450.7 0.0097241,6 0.0330932. 1 0.0001 202.7 0,0831 073.5 0.08 08694.2 0. 1399575. 1 0.2608926.0 0.3360136.6 0.0336377.6 0.0187298.7 0.000020

t

tt

+20 (b)I' +

~ 't It, ~+n'l l +

0z0I- -20bJCl

+SACLAY~STANFORD- iMAINZ~ DARN STADT

I I

55 l

I

35I

2.5I

l.5 2 3(I (f'17) ')

FIG. 1. (a) Cross sections at E, =502 MeV as a func-tion of effective momentum transfer. The parameters(Ref. 12) of the fit are also given: y=1.388. (b) Devia-tion between fit and data used; the curve shows the dif-ference to the fit of Ref. 7.

[Fig. 1(b)]. This includes the present electron-scattering data (34 points, q= 1.7-3.7 fm '),SLAC data' (87 points, 0.5—2.7 fm '), the Univer-sity of Mainz data' (17 points, 0.6 —1.8 fm '), andthe Technical University of Darmstadt data" (12points, 0.3—0.8 fm '). We have also taken intoaccount the five muonic x-ray transition ener-gies" "that provide additional information onp(r) Howeve. r, for the present fit, we have dis-carded the 289-Me& data points measured recent-ly at the University of Mainz' between 1.8 and 2.3fm '. These points strongly disagree (Fig. 1)with both the present and SLAC data. (The dis-crepancy observed can probably be assigned to adifference in energy calibration. The steep dif-fraction minimum causes a strong energy depen-dence in the "C cross sections" relative to whichthe University of Mainz Pb data' have been nor-

11

Deeply Virtual Vector Meson Production on Nuclei

High resolution reconstruction of |t| from e.g. (e,e’ K+K–) kinematics.

Coherent nuclear recoil is unresolvable: lost in 10σ-BSC.

Excitation of bound-states will wash out minima of coherent scattering.

Doubly-magic nuclei, 𝛾-decay energies are large

EXCLUSIVE DIFFRACTIVE PROCESSES IN ELECTRON- . . . PHYSICAL REVIEW C 87, 024913 (2013)

10-1

10-2

10-3

1

ep no saturationep saturation (bSat)eAu no saturationeAu saturation (bSat)

Experimental Cuts:|η(Kdecay)| < 4p(Kdecay) > 1 GeV/c

2.2

2.0

1.8

1.6

1.4

1.2

1

0.8

0.6

0.4

0.2

0

no saturationsaturation (bSat)

Q2 (GeV2) Q2 (GeV2)1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10

(1/A

4/3 )

σ(eA

u)/σ

(ep)

(1/A

4/3 )

dσ(e

h →

e' h

' φ) /d

Q2 (

nb/G

eV2 )

coherent eventsLdt = 10/A fb-1

x < 0.01

φ → K Ke p/Au → e p /Au φ

(a) (b)

´ ´ ´

∫

FIG. 5. (Color online) (a) Cross sections for φ production differential in Q2 for ep and eAu collisions for both bSat and bNonSat dipolemodels. The cross sections are scaled by 1/A4/3. (b) Ratio of eA to ep cross sections for both models.

Also, the difference is larger for φ mesons. The reason forthis is that the wave-function overlap between the φ mesonand virtual photon allows for larger dipoles than that for J/ψ(see Fig. 3). Therefore, φ production can probe further into thedense gluon regime and exhibits larger differences betweenbSat and bNonSat.

1. Probing the spatial gluon distribution

In Fig. 6 we show the differential cross section with respectto t , dσ/dt , for both J/ψ- and φ-meson production, againfor both dipole models. We assume a conservative t resolutionof 5%, which should be achievable by future EIC detectors.The statistical error bars shown correspond to an integratedluminosity of 10 fb−1. As can be seen, the coherent crosssection clearly exhibits the typical diffractive pattern. Alsodepicted in Fig. 6 is the incoherent cross section, which isproportional to the lumpiness of the nucleus. Experimentally

the sum of the coherent and incoherent parts of the crosssection is measured. Through the detection of emitted neutrons(e.g., by zero-degree calorimeters) from the nuclear breakupin the incoherent case it should be experimentally feasible todisentangle the two contributions unambiguously.

The coherent distributions in Fig. 6 can be used to obtaininformation about the gluon distribution in impact-parameterspace through a Fourier transform. In Eq. (20), the first momentof the diffractive amplitude is a Fourier transform of thedipole cross section averaged over nucleon configurations,times the wave-function overlap between the vector mesonand virtual photon. This represents a transformation fromcoordinate space to momentum space !. The coherent crosssection dσcoherent/dt is proportional to the absolute square ofthis amplitude. Following Ref. [34], we can regain the impact-parameter dependence by performing a Fourier transform onthe amplitude. The amplitude can be obtained by taking thesquare root of the cross section. In order to maintain the

|t | (GeV2 |) t | (GeV2)0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18

J/ φψ

|η(edecay)| < 4p(edecay) > 1 GeV/cδt/t = 5%

∫Ldt = 10/A fb-1

1 < Q2 < 10 GeV2, x < 0.01 ∫Ldt = 10/A fb-1

1 < Q2 < 10 GeV2, x < 0.01

|η(Kdecay)| < 4p(Kdecay) > 1 GeV/cδt/t = 5%

104

103

102

10

1

10-1

10-2

105

104

103

102

10

1

10-1

10-2

coherent - no saturationincoherent - no saturationcoherent - saturation (bSat)incoherent - saturation (bSat)

coherent - no saturationincoherent - no saturationcoherent - saturation (bSat)incoherent - saturation (bSat)

)b()a(

dσ(e

Au

→ e

' Au'

J/ψ

) /dt (

nb/G

eV2 )

dσ(e

Au

→ e

' Au'

φ) /d

t (nb

/GeV

2 )

FIG. 6. (Color online) Differential distributions with respect to t for exclusive J/ψ (a) and φ (b) for coherent and incoherent events. BothbSat and bNonSat models are shown.

024913-712

208Pb(e,e’) & DVESIf bound-excited states are not resolved, they smooth out the diffraction pattern.3–(2.6MeV), 5–(3.2 MeV), 2+(4.1MeV), 4+(4.3MeV)

In DVES@EIC, 𝛾-cascade boosted (×40 JLEIC, ×100 eRHIC)

High Resolution (PbWO4) forward EMCal can veto (~50% acceptance) E

𝛾

> 100 MeV.

Backgrounds?

q (GeV/c) 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8

2 |F

(q)|

-510

-410

-310

-210

-110

1

10

210

310

410 Form FactorC12

O16

Ca48

Pb208

He4

E,pG

Fourier-Bessel Charge Form Factors

EXCITATION OF LOW LYING NATURAL PARITY LEVELS IN. . . 2299

too =J =2+ 208

IO ~=

t-lO ~I-O

bb I04

to &=

Q I I I)

I I I I

jI I I I

)/ I I I

)/ I I I

)II

could either come from an unresolved level or froma real transverse current in this transition. Theclosest level seen in (p,p') is more than 25 keVseparated from this state. Since the resolution inthis experiment was typically 30 keV, contributionsfrom this level cannot be excluded even though theyseem unlikely.Because of the small effect, it was impossible to

develop a reasonable model for these transverse con-tributions. For that reason, we have fitted only theforward scattering data where the effects from thetransverse components should be considerablysmaller than the quoted error bars. The data deter-mine up to Ã= 8 coefficients, in the FB expansion.

C. The 6+ state at 4.424 MeV

The data from the 6+ state were handled in thesame way as for the 2+ state, since there was no in-dication for any transverse current in this transi-tion. The data determine up to %=7 coefficients,in the FB expansion.

D. The 8+ state at 4.610MeV

lO 7=—

)0-8 0.5 2.5I.O

data do not have enough sensitivity to allow a mea-sure of ground-state correlations. It is suggestedthat such experiments be done at 180' where thecontributions from the irrotational, incompressiblepart are considerably smaller.

B. The 4+ state at 4.323 MeV

Combining the data from 90' to 160' in one fitand neglecting the transverse current J",gave unsa-tisfactory fits with much larger g than in fittingthe 90 or the 160' data separately. This indicatesthe presence of transverse cross sections, which

l.5qEFF fm ~

FIG. 3. Cross section for the even spin natural paritystates in Pb divided by (the cross section for a unit-point charge). Data and best fit for the 4+ level arescaled down a factor of 0.03, for the 6+ level a factor of0.001, and for the 8+ level a factor of 0.00003.

The data from the 8+ state again indicate thepresence of a small transverse current. This is notsurprising since our measurements on 10+ states"show a very strong transverse current that can bedescribed by quenched single particle currents witha quenching factor of 0.65. The currents observedin this 8+ state are considerably weaker, and weused our data mainly to correct the 90' data for anytransverse contributions. This was done assumingquenched currents from the proton component,~(lh9&2, lhii~2 '), the quite dominant neutronv(1ji&&q, 3&i~q '), and the neutron v(ii»&z,lii3/2 ') component, which mostly cancels thecurrent contributions from the proton configura-tion.In all cases of the positive parity states, it was

sensible because of the smallness of the transversecurrent to recalculate the data to the maximum in-cident energy. These recalculated data are definedas

dodQ

der(Em, „,q,rr)DWBAdQ,„der(E,„~,q,rr)DWBA

(19)These recalculated data are shown together with

the best fit in Fig. 3. The fact that for the 2+ and

13

Concluding Comment:

An arbitrary JLEIC Run-Plan (similar for eRHIC)

• Neither time- nor priority-ordered.• 2 run periods per year• 15 years for ‘base program’

• Luminosity is important, even for “low luminosity physics”

14

Species e/A Energy/u Ion Pol Run Periodsep 10 x 100 L & T 2

5 x 100 L & T 410 x 40 1

e d 10 x 50 L & T 4e 3He 10 x 75? L & T 3e 4He 10 x 50 1e 9Be 10 x 40 1e 12C 10 x 50 1e 40Ca 10 x 50 1e 48Ca 10 x 40 1e 120Sn 10 x 40 1e 208Pb 10 x 40 1e 238U 10 x 40 1

Positrons 8Total 30

15