TRANSVERSE SPIN EFFECTS IN COMPASS

COmmon Muon and Proton Apparatus for Structure and Spectroscopy NA58 Czech Republic, Finland, France, Germany, India, Israel, Italy, Japan, Poland, Portugal, Russia Bielefeld, Bochum, Bonn, Burdwan, Calcutta, CERN, Dubna, Erlangen, Freiburg, Heidelberg, Helsinki, Lisbon, Mainz, Miyazaky, Moscow, Munich, Nagoya, Prague, Protvino, Saclay, Tel Aviv, Torino, Trieste, Warsaw 28 Institutes, ~230 physicists

LHC SPS luminosity: ~5 . 1032 cm-2 s-1 beam intensity: 2.108 m+/spill (4.8s/16.2s) beam momentum: 160 GeV/c longitudinally polarised muon beam longitudinally or transversely polarised target calorimetry particle identification LHC SPS N

The Spectrometer for the Muon Programme Trigger-hodoscopes μ Filter ECal & HCal 50 m SM1 SM2 TWO STAGE SPECTROMETER: RICH MWPC 0.003 < x < 0.5 10-3 < Q2 < 10 (GeV/c)2 Straws 6LiD Target Gems 160 GeV μ Drift chambers Micromegas LAT, PID Silicon SciFi

Target Polarization reversed every week Polarized Target New COMPASS target magnet: 180 mrad geometrical acceptance excellent field homogeneity To match larger acceptance: new microwave cavity 3 target cells: reduction of false asymmetries Target material: NH3 high polarisation very long relaxation time (∼ 4000 h) magnetic field rotation without polarisation loss  Polarisation of NH3 in 2007: -92%, +88%, -83% Target Polarization reversed every week

2007 Transverse data taking statistics 2007 Compass Data taking Begin of run: 18 May 2007 End of run: 11 November 2007 Split between transverse and longitudinal target polarization: - m on tape for transverse (40.0 x1012) - m on tape for longitudinal (41.5 x1012) For the extraction of the asymmetries (this analysis) only a fraction of the full statistic available has been used : nearly 20% of the whole data sample available.

DIS Event Selection DIS cuts: Q2>1 (GeV/c)2 0.1<y<0.9 W>5 GeV/c2

Total statistic for this analysis Hadron Selection All hadrons Track Length<10 X0 Energy Deposit in HCALs>Thr. ( 4 GeV HCal1 and 5 GeV Hcal2 ) Only 1 HCAL fired pT>0.1 GeV/c z>0.2 Positive hadrons Negative hadrons Total statistic for this analysis 5703231 4505088

Data quality checks Data taking stability is needed: A dedicated set of quality checks have been developed and applied to fulfill this condition Different estimators have been considered: the detector profiles stability the number of primary vertices per event the number of tracks per primary vertex beam particles per primary vertex the number of K0 per primary vertex the reconstructed mass of the K0 meson stability of many kinematical variables: ( zvtx, Em’, fm’, xBj Q2,y, W, Ehad , fhadLab , qhadLab , fhadGNS , qhadGNS , pt)

Mean of kinematical quantities

Collins and Sivers angles Azimuthal modulations Collins and Sivers angles C = h - S’  S = h - S S ’ azimuthal angle of spin vector of fragmenting quark (S’ = p - S) h azimuthal angle of hadron momentum

SIDIS azimuthal asymmetries All possible 8 azimuthal asymmetries extracted at once. ... Sivers Collins 6 further modulations M. Diehl, S. Sapeta, Eur.Phys.J C41 (2005) 515-533 hep-ph/0503023

Asymmetry Extraction Splitting middle cell into two parts two couples of cells with opposite polarization two independent values for the asymmetries per period Extraction: 2D Binned Maximum Log-Likelihood Fit: eight by eight grid in fh and fS; in each bin of the matrix one expects Nj counts :

Asymmetry Extraction - II Separation of acceptance and spin dependent modulations: Coupling of two cells (u,d) with opposite polarization () and two periods (p1,p2) with opposite target polarization: Reasonable assumption: 4 · 64 = 256 nonlinear equations 1 + 8 + 3 · 64 = 201 fit parameter, Poisson distribution to account for low statistics: Tests for systematic errors: For false asymmetries: combination of cells with same polarization Comparison of 5 estimators for asymmetry extraction included the one used in previous analysis (deuteron data) For this analysis: overall systematic error is 30% and 50% of the statistical error for Collins and Sivers respectively

Collins asymmetry – proton data systematic errors ~ 0.3 sstat at small x, the asymmetries are compatible with zero in the valence region the asymmetries are different from zero, of opposite sign for positive and negative hadrons, and have the same strength and sign as HERMES

Compass proton data comparison with M. Anselmino et al. predictions

Collins Final on Deuteron

Sivers asymmetry – proton data systematic errors ~ 0.5 sstat the measured symmetries are small, compatible with zero

Sivers Final on Deuteron from COMPASS

Sivers asymmetry– proton data comparison with the most recent predictions from M. Anselmino et al. arXiv:0805.2677

Results: Sivers asymmetry comparison with predictions from S.Arnold, A.V.Efremov, K.Goeke, M.Schlegel and P.Schweitzer, arXiv:0805.2137

Transverse L polarization

Data Selection Secondary vertex downstream of primary vertex. PT > 23 MeV/c to exclude e+e− pairs Proton and pion momenta > 1 GeV/c Q2 > 1 (GeV/c)2 0.1 < y < 0.9 Use of RICH (2007 data) Λ decay distance DΛ > 7 σD Collinearity < 10 mrad

Results with proton target ~60% higher statistics with respect deuteron data (after) Systematic errors have been estimated to be smaller than statistical errors from false polarization. No dependence on x.

(Old) Results on Deuteron Analysis for 2002-2004 deuteron data (and no RICH ID) Only statistical errors are shown (systematic effects not larger than the statistical errors). Small tendency for Λ, but not significant for deuteron target.

Summary First preliminary results of COMPASS 2007 proton transverse run: Collins Asymmetry: different from zero, comparable to HERMES agreement with predictions of Anselmino et al Sivers Asymmetry: small and compatible with zero within present statistical errors Accessing Collins from L:

Thank You

The 3rd Twist-2 structure function three quark distribution functions (DF) are necessary to describe the structure of the nucleon at LO q(x) f1q (x) unpolarised DF quark with momentum xP in a nucleon well known – unpolarised DIS vector charge Dq(x) g1q(x) helicity DF quark with spin parallel to the nucleon spin in a longitudinally polarised nucleon known – polarised DIS  axial charge DTq(x) = q↑↑(x) - q↑↓(x) h1q(x), transversity DF quark with spin parallel to the nucleon spin in a transversely polarised nucleon  tensor charge largely unknown ALL 3 OF EQUAL IMPORTANCE

Misura di ΔTq(x) Chiral-odd: requires another chiral-odd partner Inclusive DIS ppl+l-X lpl’hX impossible direct measurement ΣΔTq(x) ·ΔTq(x) convolution with spin dependent fragment. func. lpl’h1h2X lpl’ΛX SIDIS (e.g. COMPASS and HERMES ) Hard scattering (e.g. RHIC) Drell-Yan Single Spin Asym (e.g. p↑p→πX) Hard scattering (e.g. GSI)

Bakker, Leader, Trueman, PRD 70 (04) Transversity DF q=uv, dv, qsea quark with spin parallel to the nucleon spin in a transversely polarised nucleon DTq(x) = q↑↑(x) - q↑↓(x) h1q(x), dq(x), dTq(x) Properties: probes the relativistic nature of quark dynamics no contribution from the gluons  simple Q2 evolution Positivity: Soffer bound…………….. first moments: tensor charge………. sum rule for transverse spin in Parton Model framework………… it is related to GPD’s is chiral-odd: decouples from inclusive DIS Soffer, PRL 74 (1995) Bakker, Leader, Trueman, PRD 70 (04)

SIVERS Mechanism The Sivers DF is probably the most famous between TMDs… gives a measure of the correlation between the transverse momentum and the transverse spin Requires final/initial state interactions of the struck quark with the spectator system and the interference between different helicity Fock states to survive time-reversal invariance Time-reversal invariance implies: …to be checked In SIDIS:

Global Fits

Global Fit

First Extraction of DTq HERMES, COMPASSBELLE

Global Analysis Stefano Melis DIS

Sivers PDF

2007 Transverse data taking statistics 2007 Compass Data taking Begin of run: 18 May 2007 End of run: 11 November 2007 Split between transverse and longitudinal target polarization: - m on tape for transverse (40.0 x1012) - m on tape for longitudinal (41.5 x1012) For the extraction of the asymmetries (this analysis) only a fraction of the full statistic available has been used : nearly 20% of the whole data sample available. Data taking Period Target Polarization Target Polarization Week 25 - + - Week 30 + - + Week 41 Week 26 Week 31 Week 42a Week 27 Week 39 Week 28 Week 40 Week 43