The HERMES experiment

Gerard van der Steenhoven, 19 September 2004

Search the carriers of proton spin

½ = ½ q + G + Lq

• Three possible sources:– quarks:

valence quarks

sea quarks

– gluons– orbital momentum

• Mathematically:

~ 10% ? ?

EMC (85): q ~ 10%

)( qq

The experimental strategy

• Polarization of the sea quarks

• Polarization of the gluons

• Orbital angular

momentum

• Transversity:

“switch off

the gluons”

Outline of the lecture

1. The origin of proton spin?– Polarization of quarks– Gluon contributions

2. New developments– Generalized parton distributions → DVCS → Lq,g

– Transverse spin → switch off the gluons

3. Other surprises– The HERMES pentaquark signal– Parton energy loss in nuclei

4. Outlook

How to probe the quark polarization?

Measure yield asymmetry:

Polarizeddeep inelasticelectronscattering

1

1

NN

NN

PDPA

BT

Parallel electron & quark spins

Spin-dependent Structure Function

Anti-parallel electron & quark spins

In the Quark-Parton Model:

)()(

1

)(

)( 2

11

11

fff xqe

xFxF

xgA

Why HERMES?• Original purpose (~1990):

– Measure inclusive spin structure

functions g1(x) for proton & neutron

– Measure polarization of u-, d- and

sea-quarks separately: qu,d,sea(x)

• What came out sofar (~2004)?

– Precise data on g1n,p(x), qu,d,sea(x)

– First measurements of G(x), DVCS,

transversity, parton energy loss,…

→ design a reliable multi-purpose detector system !

gluon

Quark-antiquark pair

The HERMES experiment

e+ BEAM

Target area

~1.5

m

Magnet

TRDRICH

Lambda Wheels

Beam Loss Monitor

EM Calorimeter

The HERMES spectrometer

27.6 GeV e+/e-

0.02 < x < 0.6, 1.0 < Q2 < 15 GeV2

p/p ~ 1-2%, < 0.6 mrad

Data taking since 1996

2002 - 20041996-2000

Spin-dependent structure functions• The function g1(x):

• Evaluate the integrals:

• 1999 result:

)( 91

181

61

11 qpp DFdxxg

From hyperon decays

Total spin carried by quarks

0.1 0.2 q

Q2 dependence of F2(x) and g1(x)

),( 2

2

x

QxF++

2 2

→ Gluons contribute to the nucleon spin !

QCD analysis of world data (’03)• Next-to-Leading-Order analysis of -data)(1 xg

Excellent data for x > 0.01

Polarized Parton Densities• First moments:

– input scale

– pol. singlet density:

– pol. gluon density:

(th) 0.070 (exp) 0.133

(stat) 0.169 0.167

q

There must be other sources of angular momentum in the proton

220 GeV 0.4Q

(th) 0.424 (exp) 0.175

(stat) 0.388 0.616

G

Flavour decomposition of spin

• Semi-inclusive deep

inelastic scattering

• Hadron tags flavour of

struck quark

• Derive purity of tag from

unpolarized data

Key issue: role of sea quarks in nucleon spin

24.08.2004

Particle identification• Dual radiator RICH:

C4F10

Aerogel

P

K

K

Detection efficiencies

f

hfff

f

hfffh

z,QdzDx,Qqe

z,QdzDx,Qqex,QA

)()(

)()()(

222

222

21

Hadron asymmetries (measured)

Known quantities(from other data)

Polarized Parton Distribution Functions !

Flavour decomposition: results• The method:

• Conclusion: qsea 0

Strong breaking of flavour symmetry

No significant breaking of flavour symmetry.

Flavour symmetry breaking Unpolarized data: Polarized data:

0AVMD||

q

ΔqADIS

|| q

ΔqAQCDC

|| G

ΔGAPGF

||

Gluon polarization• Photoproduction of high

pT –hadron pairs →

• Contributing diagrams:

• Corresponding asymmetries:

Data and plans for G/G• Asymmetry for high-pT hadron pairs production:

• New high-precision data →

±0.18±0.03

0.100.280.20Δ :SMC G/G

Generalized Parton Distributions• Consider exclusive processes:

– Deeply virtual Compton scatt.– Exclusive vector meson prod.

• Collins et al. proved factorization theorem (1997):

Distribution amplitude(meson) final state

finalquark

initialquark

2

2*.. ),,( ),,( ),(

f

pf

mfmprodexcl dtxHQxcz

Hard scatteringcoefficient (QCD)

Generalized PartonDistribution (GPD)

(NASTY: x = xBj for quarks and x = -xBj for antiquarks → x [-1,1])

’

x+ x-

t L,

The remarkable properties of GPDs

• Integration over x gives Proton Form Factors:

)(),,(~

);(),,( 0,0, xqtxHxqtxH tq

tq

• The forward limit:

• Second moment (X. Ji, PRL 1997):

)(),,(~

)(),,(

)(),,(~

),(),,(

1

1-

2

1

1

1

1-

1

1

1

tGtxEdxtFtxEdx

tGtxHdxtFtxHdx

P

A

qqqt

qq LJdxtxEtxHx ),,(),,( 210

1

121

Dirac

Pauli

Axial vector

Pseudoscalar

GPDs give access to Orbital Angular Momentum of Quarks

Applying the GPD framework• GPDs enter description of different processes:

• Take Fourier transform of leading GPD:

dtetxHbxq tibff ),,(),( 22

1

As Jq = ½q + Lq

information on Jq gives data on Lq.

Spatial distribution of quarks in the perpendicular direction Spatial distribution of quarks in the perpendicular direction

A 3D-view of partons in the proton

A.V. Belitsky, D. Muller, NP A711 (2002) 118c

Form Factor Parton Density Gen. Parton Distribution

Experimental access to GPDs• Exclusive meson electroproduction:

– Vector mesons (0):

– Pseudoscalar mesons ():

• Deeply virtual Compton scattering (DVCS):

),,( and ),,( txEtxH

),,(~

and ),,(~

txEtxH

DVCS Bethe-Heitler

Experimental access to DVCS

• DVCS observables:

– Cross section:

– Beam charge asymmetry:

– Beam spin asymmetry:

– Longitudinal target spin asymmetry:

)(22BH

*DVCSDVCS

*BHDVCSBH ττττ||τ||τdσ

Keydifferences

First DVCS results Beam spin asymmetryB

eam

cha

rge

asym

met

ry

• Three leading order quark distributions:

momentum carried by quarks

longitudinal quark spin,

What is transversity?

transverse quark spin,

• Gluons don’t contribute to h1(x), while dominant in g1(x):

Study nucleon spin while switching off the gluons

• New QCD tests: Q2 evolution h1(x) & (lattice)

• The relevant diagram:– helicity flip of quark & target

– chirally odd process

• Consequences:

– no gluon contributions….

Measuring transversity

+

+ -

-quark flip

target flip

2

1

… & measure single-spin asymmetries:

),(),(

),(),(1),(

shsh

shsh

Ts

hUT

NN

NN

PA

Single – Spin Asymmetries• Sivers effect: AUT driven by

orbital motion

struck quark:

measure L

• Collins effect: AUT driven by

fragmentation

process: measure

transversity

First data on transversity

)()(~)sin( )1(11 zHxh

zM

Ps

‘Sivers’:‘Collins’: )()(~)sin( 1)1(

1 zDxfzM

PTs

p

First evidence for non-zero Collins (h1) and Sivers effects (Lq)

HERMES, hep-ex/0408013

Parton Energy Loss

• Energy loss mechanisms:– hadron-nucleon rescattering

– quark-gluon propagation

(QCD: LPM effect)

• Relevance:– Verification novel QCD effect– Study of Quark-Gluon Plasma

in relativ. heavy-ion collisions.

3/1 AEhadron

3/22 AEparton

DIS on heavy nuclei• Hadron attenuation in 14N, 84Kr:

Search for quark-gluon plasmaData: EPJC 21 (2001) 599

Dashed: X. Wang et al. (2002)[QCD + LPM effect + tune g(x)]

Solid: Accardi et al. (2003)[N incl. Q2 rescaling effects]

Energy loss in hot matter 0 production in Au + Au

collisions at Phenix:

• Adjust energy loss to fit data (cf. cold matter)

x

z

y

mattercoldmatterhot gg

New data on hadron attenuation• Cronin effect:

– enhancement at

high pT2 (rescatt.)

• Attenuation for 0:

Search for quark-gluon plasma

Two-hadron attenuation• Evaluate:

• Partonic energy loss:

R2h →1

• Hadronic energy loss:

R2h ~ (R1h)2 0.5(Kr) - 0.8(N)

DzdNzzNd

AzdNzzNd

h zR)(

),(

)(),(

22

1

212

1

212

)(

Both partonic and hadronic energy loss processes are relevant

The HERMES pentaquark signal

• Quasi-real photoproduction: e+D X

• Decay mode: ps p

27.6 GeV e-beam

deuteron gas target

Invariant mass from identified

decay particles

+ Invariant mass peak

• Background: 3rd order polynomial

• M =1528 2.6 MeV

= 8 2 MeV (dominated by exp. resolution)

• Significance:– naïve:

– realistic:

Gauss + 3rd order

polynom.

7.3/ sigsig NN

7.4/ 22 bcksig NN

Background below the +

• Background: MC sim-ulation + resonances

• M = 1527 2.3 MeV

= 9.2 2 MeV

• Significance:– naïve: 6.1 – Realistic: 4.3

Gaussian + resonances + background fit

Pythia6 background

Mixed event background

additional *+ resonances (not in Pythia6)

Comparison of pentaquark data

• Mean: 1532.52.4 MeV

nK+

0spK

Average of all data:M = 1532.5 2.4 MeVIncludes latest from - JINR (hep-ex/0403044) - LPI (hep-ex/0404003)

Latest HERMES results on +

• Require additional in + mass spectrum

• Impose veto on K* and(1116)

Signal/background improves from 1:3 2 :1

Summary• HERMES results:

– Quark sea → unpolarized– Gluons → polarized // proton– First data on transversity:

Quarks carry orbital momentum?

– First exploration of GPDs– Partonic energy loss seen – Co-discovery pentaquarks

• The future:– End of HERA operations:

summer of 2007