ΔG/G Extraction From High- P t Hadron Pairs at COMPASS Ahmed El Alaoui Nuclear Physics School, Erice, September 2007 On Behalf Of COMPASS Collaboration

Outline Introduction COMPASS Experimental Setup Data Analysis Results Summary and Conclusion

- To access the gluon contribution to the nucleon spin - To understand the role of the Axial Anomaly in the explanation of the spin crisis Nucleon Spin Naive Quark Model Pure valence description of constituent quarks: ∆u = + 4/3 ∆d = - 1/3 ∆Σ = 1 Relativistic Quark Model ∆Σ ≈ 0.75 QCD framework Hyperons β decay constants + SU(3) flavor symmetry ∆Σ = a 0 ≈ 0.60 compatible with the Relativistic QM prediction However, the EMC measured ∆Σ = 0.12 ± 0.09 ± 0.14 SPIN CRISIS a 0 = ΔΣ – (3α s /2π)ΔG A measurement of ∆G is needed

How To Acess ΔG/G Indirect Measurement: QCD analysis: fit to the nucleon spin structure function g 1 (x) Direct Measurement: ∆G/G can be accessed via Photon Gluon Fusion (PGF) process Unfortunately, the limited range in Q 2 does not allow for a precise determination of ∆G

PGF Process Two approaches are used to tag PGF process q = c: - Open Charm D 0, D * decay - Clean signal - Combinatorial background - Low statistics q = u, d, s: - High-p t hadron Pairs - Physical background - High statistics A PGF = a LL x (∆G/G) factorization theorem PGF

How To Acess ΔG/G Direct Measurement: ∆G/G can be accessed via Photon Gluon Fusion (PGF) process Three independent measurements were done at COMPASS - Open Charm - High p t hadron pairs production at Q 2 >1GeV 2 - High p t hadron pairs production at Q 2 <1GeV 2 Indirect Measurement: QCD analysis: fit to the nucleon spin structure function g 1 (x) Unfortunately, the limited range in Q 2 does not allow a for precise determination of ∆G

COMPASS Collaboration COMPASS COmmon Muon and Proton Apparatus for Structure and Spectroscopy 250 Physicists 18 Institutes 12 Countries

LHC SPS luminosity: ~ cm -2 s -1 beam intensity: µ+/spill (4.8s/16.2s) beam momentum: 160 GeV/c Experiment Layout

COMPASS Spectrometer Tracking: SciFi, Silicon, MicroMegas, GEMs, MWPC, Straws PID: RICH, Calorimeters, μ Filters SM1 SM2 RICH E/HCAL1 Muon Wall 1Muon Wall 2 E/HCAL2 Polarized Target 160 GeV μ beam Polarization ~ 80% 50 m long LAS SAS MicroMegas DC

COMPASS Target Two 60 cm long oppositely polarized cells 6 LiD is used as a material dilution factor ~ 0.4 Target Polarization ~ 50% 70 mrad acceptance (180 mrad for 2006 target) [cm] Vertex distribution

High P t Events Selection Primary vertex with at least μ, μ’ and 2 hadrons m inv (h 1,h 2 ) > 1.5 GeV 0.0 < z, x F < 1.0 P t > 0.7 GeV 0.1 1GeV 2 ) E Calo /P > < z 1 +z 2 < 0.95 ΣP t > 2.5 GeV < y < 0.9 (Q 2 <1GeV 2 )

High P t Spin Asymmetry The acceptance is not identical in both cells Asymmetry bias μ B A exp = (N u - N d )/(N u + N d )

High P t Spin Asymmetry Polarisation reversal each 8 hours A exp = (N u - N d )/(N u + N d ) ‘‘‘‘‘ To improve the statistical error, a weighted method is used in the asymmetry calculation: w = fDP B (event-wise weight) A || /D=(A exp - A exp )/2fP T P B D ‘ μ B μ B f Dilution factor P T(B) Target(Beam) polarization D Depolarization factor

∆G/G Extraction at Q 2 <1GeV 2

R i (fraction of the process i), a LL, ∆q, q, q and G are obtained from - Monte Carlo Simulation based on PYTHIA generator and Geant. - pQCD Calculation - pdf in the nucleon from GRSV2000 and GRV98LO parametrization - pdf in the photon from GRS parametrization The polarized pdfs in the photon ∆q and ∆G are not available. Therefore the positivity limit is used to constrain them which leads to 2 extreme scenarios. ∆G/G at Q 2 <1GeV 2 A || /D = R PGF ∆G/G a LL PGF + R qq ∆q/q a LL (∆q/q) qq γ + R qg ∆G/G a LL (∆q/q) qg γ + R gq ∆q/q a LL (∆G/G) gq γ + R gg ∆G/G a LL (∆G/G) gg γ γ γ γ Included as systematic error in the estimation of ∆G/G + R QCDC ∆q/q a LL QCDC qq’ Resolved photon processes γ

Monte Carlo vs. Data (Q 2 <1GeV 2 ) x Bj

Process fractions (Q 2 <1GeV 2 ) Resolved photons processes 50% 32% 12%

∆G/G Result at Q 2 <1GeV data A || /D = ± 0.013(stat.) ± 0.003(syst.) ∆G/G(x g,μ 2 ) = ± 0.058(stat.) ± 0.055(syst.) x g = μ 2 = 3GeV 2 Contribution to Syst. error comes from - False asymmetry - Monte Carlo tuning - Resolved photon process

∆G/G Extraction at Q 2 >1GeV 2

∆G/G at Q 2 >1GeV 2 A || /D = R PGF ∆G/G a LL + R QCDC ∆Q/Q a LL PGF QCDC + R LO ∆Q/Q a LL LO At Q 2 >1GeV 2 analysis, Lepto generator seems to describe the real data much better than PYTHIA. It was then used to estimate the fraction of each process PGF QCDC LO Contribution from resolved photon precesses is negligible in this case

Monte Carlo vs. Data at Q 2 >1GeV 2 34%

∆G/G Result at Q 2 >1GeV Data A || /D = ± 0.080(stat.) ± 0.013(syst.) ∆G/G(x g, μ 2 ) = 0.06 ± 0.31(stat.) ± 0.06(syst.) = Data Analysis is in progress… Contribution to Syst. error comes from - False asymmetry - Monte Carlo tuning μ 2 = 3GeV 2

Results ΔG/G

Summary and Conclusion - The new solenoid, installed in 2006, has an acceptance (180 mrad) three times larger than the previous one. - High-P t asymmetries at Q 2 >1GeV 2 and Q 2 <1GeV 2 were presented - The measured ∆G/G is compatible with zero at x g = Analysis of 2004 data (at Q 2 >1GeV 2 ) is almost finished. It will be released soon Double the statistics obtained in 2004 Access higher value of x g Thank you The most precise measurement up to now

Backup slides

Fit to g 1 (x) using DGLAP evolution equations provides 2 differents solutions : ∆G >0 and ∆G <0. Both solutions describe the data well. The first moment of ∆G obtained from the fit is ∆G(x g ) dx g ∫ ≈ ≡ ∆G

Muon Beam Line