Salvatore Fazio BNL for the BNL EIC Science Task Force

Salvatore Fazio BNL

for the BNL EIC Science Task Force

XX International Conference on Deep-Inelastic Scattering and Related SubjectsUniversity of Bonn, Germany

March 26-30, 2012

GPDs at an EIC

p p

gg*

Plan of the talk The eRHIC accelerator and an EIC detector

capabilities for exclusive diffraction

GPDs and DVCS

Bethe-Heitler subtraction

|t|-differential cross sections

Charge and spin asymmetries

Imaging with an EIC (the impact!)

J/ y

Summary

2March. 29, 2012 S. Fazio: DIS2012 - Bonn, Germany

See also: D. Mueller’s andM. Diehl’s talks

S. Fazio: DIS2012 - Bonn, Germany 3March. 29, 2012

The eRHIC collider

5 - 20 GeV electrons on 100-250 (130) GeV protons (nuclei). Polarization of electrons and protons (nuclei)

Lumi: 0.62 x 1033 cm-2s-1 (stage 1–ep pol)

3.1 x 1033 cm-2s-1 (stage 1–ep unp)

9.7 x 1033 cm-2s-1 (stage 2-ep)

Important for exclusive DIS:• Dedicated forward instrumentation• High tracker coverage • Very High lumi!

eSTAR

ePHEN

IX

6 pass 2.5 GeV ERL

Coherent

e-cooler

Beam-dump

Polarized e-gun

eRHIC detector

EIC/eRHIC EIC detector

An electron ring will be built at the RHIC facility

The current experiments can be upgraded for ap(A) physics

A new dedicated detector will be built

S. Fazio: DIS2012 - Bonn, Germany 4

Important for exclusive diffraction:• Hermetic Central Tracking Detector (Si pixels)• Good em calorimeter resolution with fine granularity (fibres)• Very forward calorimetry• Roman pots from the early beginning (and with excellent acceptance)

The EIC detectorep/N

FORW

ARDREAR

General properties:• Hermetic• Asymmetric

March. 29, 2012


20x250 GeV 5x50 GeV

Accepted in“Roman Pot”(example) at s=20mPlots from J-H Lee

L = 27.77 pb-1

55 events (DVCS + BH)

Roman Pots at HERA

Quadrupoles acceptance

10 s from the beam-pipe

• high t acceptance mainly limited by magnet aperture‐• low t acceptance limited by beam envelop (~10σ)‐• t resolution limited by‐

– beam angular divergence ~100μrad for small t– uncertainties in vertex (x,y,z) and transport– ~<5-10% resolution in t (RP at STAR)

Direct |t| measurement @ eRHIC


DVCS (g): H, E, H, E VM ( , , ): r w f H E Info on quark flavors via

PS mesons ( , )p h : H E ~

~ ~

~

quantum number of final state selects different GPDs:

Accessing the GPDs

March. 29, 2012


€

ALU ∝ y F1 t( )H ξ,ξ, t,Q2( ) −

t

4M 2 F2 t( )E ξ,ξ, t,Q2( ) + ...

⎡ ⎣ ⎢

⎤ ⎦ ⎥

March. 29, 2012

€

φ = φN − φl

€

ϕ = ΦT − φN

Angle btw the production and scattering planes

Angle btw the scattering plane and the transverse pol. vector

Accessing the GPDs

€

AC =dσ + − dσ −

dσ + +dσ − ∝ Re ADVCS( ) Requires a positron beam at eRHIC

€

AUT ∝−t

4M 2 F2 t( )H ξ,ξ, t,Q2( ) − F1 t( )E ξ,ξ, t,Q2

( ) + ...[ ]

Dominated by H

sin(FT-fN) governed by E and H

€

dσ

dt~ A0 H

2x, t,Q2

( ) −t

4 M p2 E 2 x, t,Q2

( ) ⎡

⎣ ⎢

⎤

⎦ ⎥ Dominated by H

slightly dependent on E

slightly dependent on E


p p

γγ*

Deeply Virtual Compton ScatteringVM (ρ, ω, φ, J/ψ, Υ) DVCS (γ)

Q2Q2 + M2Scale:

DVCS properties:• Similar to VM production, but γ instead of VM in the final state

• Very clean experimental signature

• Not affected by VM wave-function uncertainty

• Hard scale provided by Q2

• Sensitive to both quarks and gluons

IPp p

Vγ*

March. 29, 2012

A GOLDEN MEASUREMENT!


Scanning the phase space…

March. 29, 2012

EIC will provide sufficient lumi to bin in multi-dimensions

wide x and Q2 range needed to extract GPDs

EIC lumi: ~10 fb-1/year @ stage 1 – 5x100~10 fb-1/month @ stage 2 – 20x250

5 X 100

… we can do a fine binning in Q2 and W… and even in |t|

20 X 250

DVCS |t|-slope @ HERA


DVCS phase-space

Stage 1

Stage 2


MC simulation

GPDs, evolved at NLO by an indipendent code which provides tables of CFF - at LO, the CFFs are just a convolution of GPDs:

provide the real and imaginary parts of Compton form factors (CFFs), used to calculate cross sections for DVCS and DVCS-BH interference.

-> The B slope is allowed to be costant or to vary with Q2:

Proton dissociation (ep → eγY) can be includedOther non-GPD based models are implemented like FFS, DD

€

H (ξ,Q2, t) =ei

2

1 − x ξ − iε± ξ →−ξ{ }

⎡

⎣ ⎢

⎤

⎦ ⎥H i x,ξ,Q2, t( )dx

−1

1

∫u,d,s

∑

The code MILOU is Based on a GPDs convolution by: A. Freund and M. McDermott [All ref.s in: http://durpdg.dur.ac.uk/hepdata/dvcs.html]

Written by E. Perez, L Schoeffel, L. Favart [arXiv:hep-ph/0411389v1]

March. 29, 2012

€

dσ

dxdyd | t | dφdϕ=

α 3xB y

16π 2Q2 1+ε 2

I

e3

€

φ = φN − φl

€

ϕ = ΦT − φN

€

ε ≡ 2xmN

Q

€

I BH

2=

e6

x 2y 2(1+ε 2)2 Δ2P1(φ)P2(φ)c0

BH + cnBH cos nφ( ) + s1

BH sin φ( )n =1

2

∑ ⎧ ⎨ ⎩

⎫ ⎬ ⎭

I DVCS

2=

e6

y 2Q2 c0DVCS + cn

DVCS cos nφ( ) + snDVCS sin nφ( )[ ]

n =1

2

∑ ⎧ ⎨ ⎩

⎫ ⎬ ⎭

I2

=±e6

xy 3Δ2P1(φ)P2(φ)c0

I + cnI cos nφ( ) + s1

I sin φ( )n =1

3

∑ ⎧ ⎨ ⎩

⎫ ⎬ ⎭

€

dσ

d t= exp B Q2

( )t( )

http://durpdg.dur.ac.uk/hepdata/dvcs.html


BHDVCS

p p

ee

Wrong-sign sample: a negative track match to the second candidate

sample: no tracks matching to the second candidate

e sample: a track match to the second candidate

(DVCS+BH)

(BH+ bkgd)

(bkgd)

DVCS - BH

March. 29, 2012

BH is “precisely” known but to certain level! Uncertainty on proton form factor uncertainty on BH xsec ~ 3% (at LO)


BH rejection

March. 29, 2012

Eel

EγEel

Eel

Eel

Eγ

Eγ

Eγ

1. BH electron has very low energy (often below 1 GeV)2. Photon for BH (ISR) goes often forward (trough the

beam pipe)Important: em Cal must discriminate clusters above noise down to 1 GeV


BH rejection

In DVCS most of the photon are less “rear”Than the electrons:(θel-θg) > 0 rejects most of the BH

BH and DVCSBH dominated


BH subtraction will be not an issue for y<0.6

Stage 2

BH fraction


BH subtraction will be relevant in stage 1, at large y, depending on the x-Q2 bin

BUT…

Stage 1

BH fraction

Stage 1-2 overlapping:x-sec. measurements in stage 2 at low-y can cross-check the BH subtrac. made in stage 1


Stage 1: 5 X 100 GeV ~10 fb-1 (~ a year)

Stage 2: 20 X 250 GeV ~100 fb-1 (~ a year)

Data simulation & selection

March. 29, 2012

0.01 < |t| < 0.85 GeV2

(Low-|t| sample)• Very high statistics• Systematics will dominate!• Within Roman pots acceptance

1.0 < |t| < 2.0 GeV2

(Large-|t| sample)• Xsec goes down exponentially• requires a year of data taking @

20x250 GeV (stage 2)• Main detector can be used in

measuring |t|

Acceptance criteria• for Roman pots: 0.03< |t| < 0.88 GeV2

• for |t| > 1GeV2 detect recoil proton in main detector

• 0.01 < y < 0.85 GeV2

• h < 5

BH rejection criteria (applied to x-sec. measurements)• y < 0.6• (θel-θg) > 0 • Eel>1GeV2; Eel>1GeV2

Events smeared for expected resolution in t, Q2, x

Systematic uncertainty assumed to be ~5% (having in mind experience from HERA)

Overall systematic uncertainty from luminosity measurement not taken into account


(~ 10 months EIC)

€

dσ

d t=

#evt

Δbin • A • L

€

~ e−bt

b=5.6Specifications:

• Statistical error down to 1%• It uses smeared t values (5%

momentum resol.)• |t|-binning –> 3 * resolution (or higher)

dσ/d|t|- Stage 1

Smearing of momenta and energies accordig to…


Asymmetries

Plots from D. Mueller


€

q x,r b ,μ 2

( ) =1

4πd t

0

∞

∫ J0

r b t( )H x,η = 0, t,μ 2

( )

Imaging

Errors are extrapolated for:|t| 0 ; |t| > 1.5 GeV2

…FIT…Plots from D. Mueller


See also: D. Mueller’s talk for details M. Diehl’s talk for an overview

Plots from D. Mueller

Imaging

A global fit over all mock data was done, based on the GPDs-based model:

[K. Kumerički, D Müller, K. Passek-Kumerički 2007]

Known values q(x), g(x) are assumed for Hq, Hg (at =0, t=0 forward limits Eq, Eg are unknown)

Excellent reconstruction of Hsea, Hsea and good reconstruction of Hg (from dσ/dt)


g*p -> J/ y p

pseudo-data generated using a version of Pythia tuned to J/y data from HERA

wave function uncert. (non-relativistic approximation)

mass provides hard scale• Sensitive to gluons• Both photo- and electro-production

can be computed

Plots from E. Aschenauer

J/ψ


Summary

A lot of experience carried over from HERA

Simulation shows how an eRHIC can much improve our knowledge of GPDs

a.r.o. HERA

A fine binning of x-sec and symmetries will be possible, uncertainties mostly

dominated by systematics

Large potential for an accurate 2+1D imaging of the polarized and unpolarized

quarks and gluons inside the hadrons (and nuclei!)

March. 29, 2012


Back up

March. 29, 2012


Q2 dependence for the W slope not clear within the uncertainties!

ZEUS: JHEP05(2009)108H1: Phys.Lett.B659:796-806,2008

Fit: σ ~ Wδ

DVCS @ HERA

€

t ~ PTγ2 +PTe

2 ⎛ ⎝ ⎜ ⎞

⎠ ⎟2

t measured indirectly:

€

Fit :dσ

dt∝ e−b |t |

No evidence for W dependence of bby roman pots!

March. 29, 2012


qg vs Eg

20 X 250

March. 29, 2012


qg vs Eg

10 X 100

March. 29, 2012


BH

DVCS

BH

DVCS

• BH dominates at large y. • DVCS dropes with y


Effect of the cuts

The effect of the cut for the 20x250 conf. is that BH never exceeds 70% of the sample

BH fraction - overall


for the 5x100 conf. is that BH can be a problem at large y and large t, depending on the bin

BH fraction - overall


t-xsec (ep -> gp)

March. 29, 2012

Selection criteria:

0.01 < y < 0.6 θγ < 2x10-2 rad θel< 2x10-2 rad Eγ > 1 GeV Eel > 1 GeV

b=5.6 GeV-2

10 x-bins ; 5 Q2-bins€

~ e−bt

HERA eRHIC


(~ 3.5 weeks EIC)

€

dσ

d t=

#evt

Δbin • A • L

€

~ e−bt

b=5.6

dσ/d|t|- Stage 2


(~ 1 year EIC)

€

dσ

d t=

#evt

Δbin • A • L

€

~ e−bt

b=5.6• Resolution: uniform assumed ~20%

• Large-|t| -> |t| is reconstructed using the main detector

dσ/d|t|- Stage 2


• Electron and photon clusters are ofter very close!

• can affect the phi distribution reconstruction!

Crucial!Our em cal must distinguish two clusters within the order of 1 deg

More studies on-going…


How does the nuclear environment modify parton-parton correlations? How do nucleon properties change in the nuclear medium?

(Bethe-Heitler)

Nuclear GPDs ≠ GPDs of free nucleon Enhancement of effect when leaving forward limit?

caused by transverse motion of partons in nuclei?

important role of mesonic degrees of freedom? manifest in strong increase of real part of τDVCS

with atomic mass number A?

DVCS in coherent region: new insights into ‘generalized EMC effect’?

DVCS on nuclear targets

MC simulation for DVCS on nuclei coming soon thanks to an updated version of MILOU code

March. 29, 2012

Documents

Salvatore Fazio BNL for the BNL EIC Science Task Force