STAR Forward Spin Physics at STAR RHIC BNL

STAR Forward Spin Physics at STAR RHIC, BNL Spin-dependent forward particle correlations in p+p collisions at = 200 Ge. V Nikola Poljak University of Zagreb (For the STAR Collaboration)

2 Single Spin Asymmetry • Definition: • dσ↑(↓) – differential cross section of p 0 when incoming proton has spin up(down) Left positive AN: p 0, x. F<0 p 0, x. F>0 more p 0 going p left to polarized beam p Right

Published measurements at STAR: ar. Xiv: hep-ex/0801. 2990 PRL 97, 152302 (2006) accepted for publication in PRL nucl-ex/0602011 RUN 6 • Large transverse single-spin asymmetries at large x. F • x. F dependence matches Sivers effect expectations qualitatively • p. T dependence at fixed x. F not consistent with 1/p. T expectation of p. QCD-based calculations At this energy the cross-section is consistent with NLO p. QCD (run 2 + run 3) 3

4 Developments in theory and experiment • new phenomenological analyses within a generalized parton model can explain both Sivers moments in semi-inclusive deep inelastic scattering and many features of p↑+p → p +X. (PRD 77 051502(R)) • ~20% of the COMPASS transversely polarized proton data has been analyzed and reported. COMPASS finds non-zero Collins moments and Sivers moments compatible with zero, although expected Sivers moments are small in the x, Q 2 range of their experiment. (Levorato, for COMPASS; Ferrara, 2008). • Expectations that the Collins effect is suppressed in p↑+p → p +X (PRD 71 014002) were found incorrect due to a sign error (ar. Xiv: 0804. 3047) The need remains to separate Collins and Sivers effects in p↑+p → p +X

5 Separating Sivers and Collins effects Sivers mechanism: asymmetry in the forward jet or γ production SP Collins mechanism: asymmetry in the forward jet fragmentation SP k. T, q p p p Sensitive to proton spin – parton transverse motion correlations p Sensitive to transversity Sq To discriminate between the two effects we need to go beyond π0 detection to jet-like events k. T, π

Detectors – from FPD to FMS Run 8 and beyond: FMS Runs 3 -6: FPD, FPD++ • FMS will provide full azimuthal coverage for range 2. 5 h 4. 0 • Inclusive p 0 cross x -p plane for inclusive sections • broad acceptance in F T g, p , w, K 0, … production in p+p and • AN 0 for inclusive p 0 productiond(p)+Au • broad acceptance for g-p 0 and p 0 -p 0 from forward jet pairs uses 3 different settings of modular detectors 20 x more single uses a acceptance than previous detectors monolithic detector

Small Cell PSU Type Forward Meson Spectrometer (FMS) New FMS Calorimeter Lead Glass From FNAL E 831 804 cells of 5. 8 cm 60 cm Schott F 2 lead glass QT board 224 of 476 Cockcroft-Walton HV bases with computer control through USB. Designed/built in house for FEU-84. Readout of 1264 channels of Designed and built at Penn State University FMS provided by QT boards. Students prepare cells at test Lab at BNL Each board has • 32 analog inputs • 5 -bit TDC / channel • Five FPGA for data and trigger • Operates at 9. 38 MHz and higher harmonics • Produces 32 bits for each RHIC crossing for trigger • 12 -bit ADC / channel Designed and built at UC Berkeley/SSL

8 Calibration the calibration methodologies employed for the FPD have been successfully adapted to the FMS Offline calibration done cell -by-cell included energy corrections Event selection done with: minbias condition Hightower ADC threshold (400/200 cts. for small/large cells) <0. 7 (small) ; <1. 0 (large cells) fiducial volume cut (0. 5 cell)

Details of data analysis - calibration Minimal run-by-run dependence in mass peak observed LED system : critical calibration tool MIT (LED optics) UC Berkeley/SSL (flasher boards) Texas / Protovino / BNL (assembly) SULI program (Stony Brook students) / BNL (control electronics) Calorimeter stable at level of ~1%.

Association analysis – energy corrections • comparison of generated quantities to reconstructed GEANT simulations • We consider UNCORRECTED Eliminating energy dependence in p 0 mass peak gives the correct average neutral pion energy 10

11 Distributions comparison • Full PYTHIA/GEANT simulations have adequate statistics to reach moderate x. F at large p. T • Cell mass resolution in data is reasonable, given run-6 FPD performance • Simulations have somewhat better resolution than data DATA SIMULATION Present understanding sufficient; further investigations to be done

12 Summary & goals • FMS - a new device, with many more channels (1264 detectors compared to 98 for north/south FPD modules). • FMS has 20 x more acceptance than the previous modular detectors • the FMS involves the large cells, not used in the FPD • methodologies used in FPD successfully adapted to FMS • intercompare reconstructed PYTHIA+GSTAR events against reconstructed data • verify spin information from STAR local polarimeter • extract transverse single spin asymmetries from FMS from data for p↑+ p -> p 0 + X as a point of contact with previous work • extract transverse single spin asymmetries from FMS from data for p↑+ p -> “jet-like” + X final state.

BBC polarization time dependence samples Correlation of multiplicity topology in beam-beam counter (BBC) with polarization direction turns out to be good polarimeter for s = 200 Ge. V see J. Kiryluk (STAR) Ar. Xiv: hepex/0501072 v 1 Polarization For run-8 data, analysis of BBC asymmetries, using effective analyzing powers from run-6, is effective quality assurance for the FMS analysis Flat line fits shown Every run for which there is FMS data, also has BBC data. Relative polarimetry consistent with CNI Star preliminary

First look at analysis results 14 STAR preliminary 75% of run-8 data Octant subdivision of FMS for inclusive p 0 spin sorting. stat. errors only • AN comparable to prior measurements • Azimuthal variation appears to be as expected • Systematic errors being evaluated • First estimate stot. 1. 2 sstat.

First look at “jet-like” events 15 Event selection done with: • • >15 cells with energy > 0. 4 Ge. V in the event (no single pions in the event sample) cone radius = 0. 5 (eta-phi space) “Jet-like” p. T > 1 Ge. V/c ; x. F > 0. 2 2 perimeter fiducial volume cut (small/large cells) The agreement between data and simulation looks convincing

16 Conclusions • FMS is complete and in place. Commissioned and operated in run-8. It has 20 x the acceptance of FPD • Reconstruction and calibration procedures successfully ported from FPD to FMS • Calibration is mostly complete and data shows good agreement with the simulated sample of events • Inclusive p 0 AN(x. F) from FMS is comparable to FPD precision measurements • analysis of jet-like events is under way

17 Outlook • Complete analysis of “jet-like” events • Determine AN(p. T) for p↑+ p -> p 0 + X • Determine AN for final state that contains p 0 pairs • Determine AN for final states with heavier mesons • Run-9 - Go beyond p 0 detection to direct photons + jet final state AN THANK YOU

18 BACKUP

Possible mechanisms • Sivers effect [Phys. Rev. D 41, 83 (1990); 43, 261 (1991)]: Flavor dependent correlation between the proton spin (Sp), proton momentum (Pp) and transverse momentum (k. T) of the unpolarized partons inside. The unpolarized parton distribution function fq(x, k. T) is modified to: • Collins effect [Nucl. Phys. B 396, 161 (1993)]: Correlation between the quark spin (sq), quark momentum (pq) and transverse momentum (k. T) of the pion. The fragmentation function of transversely polarized quark q takes the form:

20 How can the p 0 cross section depend on the proton transversity? 1. Proton quark scattering is insensitive to transverse spin. However, the quark retains its initial spin after a hard scattering, and the quark π0 fragmentation can have azimuthal dependence on the transverse spin of the quark. This process is referred to as the Collins Effect. [Nucl. Phys. B 396, 161 (1993)] 2. A quark inside a proton may have orbital angular momentum that is correlated to the spin of the proton. If two quarks with opposite transverse momentum contribute different scattering amplitudes to the same final state, a case can be made where the proton quark scattering is sensitive to the transverse spin of the proton. This process is referred to as the Sivers Effect. [Phys. Rev. D 41, 83 (1990); 43, 261 (1991)]

21 Collins Effect sq = Spin of the struck quark pq = Momentum of the struck quark k. Tπ = Transverse momentum of the neutral pion y The spin of the scattered quark is correlated with the spin of the proton SP x p sq The fragmentation of the quark to p 0 has sq dependence z p + p p 0 + X pq Spin of the proton affects the scattering angle through the spin of the large x quark k. Tπ P (any polarization) π0 π0

22 Sivers Effect Sp = Spin of the proton Pp = Momentum of the proton k. Tq = Transverse momentum of the quark inside the proton y Quark transverse momentum is correlated with the spin of the proton SP x Pp z p + p p 0 + X k. Tq Quark Parton Distribution Function has k. Tq dependence pq P (any polarization) π0 Spin of the proton affects the scattering angle through the quark transverse momentum π0

23 Background fitting df/dx = S / [σ(2π)½] exp[-(x-μ)2/2σ2] + B[β 2 / (t 1 - t 2)](x - x 0)exp[-β(x - x 0)]; Tuned 2 -γ fit, especially for Large cells. Reduced from 6 to 5 parameters by fixing Eγγ S: Gaussian peak integral, μ: Gaussian peak centroid, σ: Gaussian width, B: integral of background function for |xi - μ| < 3σ, x. P: background peak position, β: background exponential falloff parameter. S and B are spin dependent

Energy-dependent corrections p 0 peak position depends on the energy • Linear correction extracted from p 0 peak position and being applied to photon energies • works for both p 0 s and ηs, and significantly decreases shift from zero in d. Egg = Esimu - Ereco. 24

25 Resolution smearing • A data-driven model is applied to introduce irresolution to the simulation • This smearing is taken from the individual detector performance, as measured from high-tower associated invariant mass • Applying this to the full PYTHIA/GSTAR simulations of the small cells results in a better match between simulation and data

26 Simulation and search algorithm for p 0 p 0 in FMS and its engineering protoype • Without Z vertex information in the calculation above, it is possible to find events where the p 0 pair originated at a significant distance from the origin • One source of such events are decays KS→p 0 p 0 (31% branching fraction) • Plot shows the mass distribution for displaced vertices above 100 cm from the BBC vertex. A pronounced KS mass bump is visible