Prompt Photons in Photoproduction and Deep Inelastic Scattering at HERA

1. Prompt Photons in Photoproduction and Deep Inelastic Scattering at HERA Eric Brownson

2. Outline • Introduction to ep scattering • HERA I, luminosity upgrade to HERA II and ZEUS • Kinematics • HERA I MC generation and usage • Verification via Deeply Virtual Compton Scattering • Prompt photons with jet in photoproduction • Event Sample and Cuts • Cross Sections • HERA II MC generation and usage • New Tools Developed • Verification via Deeply Virtual Compton Scattering • Prompt Photons in Deep Inelastic Scattering • Event Sample and Cuts • Cross Section

3. Lepton Proton Collisions Particle Scattering • A smaller wavelength means greater resolution • Wavelength of probe: l = h/Q • Q: Probe’s momentum • HERA Collisions • Ee=27.5 GeV , Ep=920 GeV • HERA provides ep collisions with CMS Energy ~ 320 GeV • Provides γor W/Z as probes • Deep Inelastic Scattering (DIS): Q2 < 40,000 GeV2 • Currently possible to probe to .001 fm (Proton is 1 fm)

4. Quark-Parton Model • Hadrons: particles that interact strongly • Bound states of structure-less particles (quarks) • Quark-parton model • Quark properties: mass, electric charge, spin • Quarks treated as point-like, non-interacting

5. Quark-Parton Model • Proton contains only valence quarks • Partons considered point-like particles • Structure functions describing individual particles’ momenta distribution depend only on xBJ • XBJ is the fraction of the proton’s momentum carried by the valence parton • No Q dependence (Bjorken scaling) • fi(x) Parton density functions (PDF’s) • Must be experimentally determined

6. QCD and Colored Gluons • Problems with Quark-Parton Model • Statistics for fermion D++ • D++ comprised of 3 u quarks • Violation of exclusion principle under QPM • Valence quarks don’t carry proton’s entire momentum • By experiment • Single quarks never observed • Quantum Chromodynamics: Theory with color quantum number • D++ quark composition: uRuBuG • Mediator of color force  gluon • Gluons carry roughly half proton momentum • Gluons carry color Observed particles “colorless” • Isolated quarks not observed  Confinement

7. Deep Inelastic Scattering • Photon has high 4-momentum • Photon is highly virtual • Deep Inelastic Scattering (Q2>1) • Scattered e- observed • Cross section has a (1/Q4) dependence • Only proton PDFs contribute

8. Photoproduction Direct: Resolved: • Photon carries very little 4-momentum • Photon is almost real • Photoproduction (Q2~0) • Scattered e- not seen (escapes down beampipe) • Direct Photoproduction: Photon couples to a parton • Resolved Photoproduction: Photon fluctuates into partonic state xg = fraction of exchange photon’s momentum involved in the collision

9. Prompt Photons Background: ISR Prompt Photon: • g is produced in the hard scatter  Carries information about the struck parton • No Hadronisation correction • Sensitive to both quark and gluon densities • Non-Prompt Background: • Initial State Radiation (ISR): Photon is radiated from e- before interaction • Final State Radiation (FSR): Photon is radiated from e- after interaction • Radiative events: Photon is radiated after the interaction • Neutral mesons: Photon originates from a decay of a hadron FSR Radiative

10. Perturbative QCD for Prompt Photons in PHP • Arrange diagrams by increasing powers of as LO: A = A0 + A1as + A2as2+ … Compton Processaem2 NLO: Resolved initial photonasaem Resolved final photonasaem asaem2 asaem2 NNLO: incomplete + … Box diagramas2aem2 Double resolvedas2aem2

11. Jets and Hadronization Struck Parton  Jet • Colored parton produced in the interaction  “Parton Level” • Colorless hadrons form via hadronization  “Hadron Level” • Collimated “spray” of particles  “Jets” • Particle showers observed as energy deposits in detectors  “Detector Level”

12. Prompt Photons + Jet in Photoproduction • Presence of a jet: • More sensitivity to underlying partonic processes • Introduces some hadronisation • Smaller hadronisation correction than dijets • Photoproduction (Q2 ≈ 0): • Exchange photon is real • Large cross section • No additional Pt given to the g+jet system by e± • The g+jet will be back to back  Well separated • No ISR/FSR background • Resolved contribution: • g hadronic structure • Constrain gluon distribution • NLO calculation available

13. NLO QCD Prompt g with Jet in PHP Calculations • K.Krawczyk & A.Zembrzuski (KZ): • GRV parametrisation: • photon structure function • proton structure function • fragmentation function • Fontanaz, Guillet & Heinrich (FGH): • MRST proton structure function • AFG photon structure function • BFG fragmentation function • A.Lipatov & N.Zotov (LZ): • Kt-factorization approach • Unintegrated quark/gluon densities using Kimber-Martin-Ryskin prescription

14. Prompt Photons in DIS • Presence of a jet (Not Used): • More sensitivity to underlying partonic processes • Decrease s by a factor of ~6.5 • Deep Inelastic Scattering (Q2 » 0): • Exchange photon is highly virtual • Only proton PDFs contribute • Q2 energy scale • Smaller cross section than PHP • ISR/FSR background contribution e± ISR FSR

15. HERA Description • 920 GeV Protons • 27.5 GeV e- or e+ • CMS Energy 320 GeV • Equivalent to 50 TeV fixed target • 220 bunches • Not all filled • 96 ns crossing time • Currents: • ~90mA protons • ~40mA positrons • Instantaneous Luminosity: • 1.8x1031cm-2s-1 ZEUS DESY Hamburg, Germany H1 & ZEUS are general purpose detectors

16. HERA Luminosity • Total HERA I Integrated Luminosity from ’93  ’00: ~193 pb-1 • Total HERA II Integrated Luminosity from ’02  ’07 : ~561 pb-1 HERA II HERA I

17. ZEUS Detector Proton (920 GeV) Electron (27.5 GeV)

18. Central Tracking Detector e p View Along Beam Pipe Side View • Cylindrical Drift Chamber inside 1.43 T Solenoid • Region of good acceptance: -1.75 < h < 1.75 • Good track resolution • Measures event vertex • Vertex Resolution • Transverse (x-y): 1mm • Longitudinal (z): 4mm Pseudorapidity

19. Uranium-Scintillator Calorimeter h = 1.1 q = 36.7o h = 0.0 q = 90.0o h = -0.75 q = 129.1o h = 3.0 q = 5.7o h = -3.0 q = 174.3o Hadronic (HAC) Cells Electromagnetic (EMC) Cells • Depleted Uranium and Scintillator • 99.8% Solid Angle Coverage Near Hermetic • Energy Resolution (single particle test beam) • Electromagnetic: • Hadronic: • Compensating • Equal signal from hadrons and g / electron particles of same energy • Measures energy and position of final state particles

20. Barrel Preshower Detector • As a particle moves from the interaction point it passes through dead material in front of the BCAL • This leads to energy loss before measurement • BCAL Presampler measurement • Measured energy is proportional to the number of particles, not the energy of the individual particles  Neutral meson separation

21. First Level Dedicated custom hardware Pipelined without deadtime Global and regional energy sums Isolated m and e+ recognition Track quality information Second Level Commodity Transputers Calorimeter timing cuts E - pz cuts Energy, momentum conservation Vertex information Simple physics filters Third Level Commodity processor farm Full event info available Refined jet and electron finding Advanced physics filters ZEUS Trigger 107 Hz Crossing Rate, 105 Hz Background Rate, 10 Hz Physics Rate

22. Kinematic Variables • Center of Mass Energy of ep system squared • s = (p+k)2 ~ 4EpEe • Center of Mass Energy of gp system squared • W2 = (q+p)2 • Photon Virtuality (4-momentum transfer squared at electron vertex) • q2 = -Q2 = (k-k’)2 • Fraction of Proton’s Momentum carried by struck quark • x = Q2/(2p·q) • Fraction of e’s energy transferred to Proton in Proton’s rest frame • y = (p·q)/(p·k) • Variables are related • Q2 = sxy

23. Kinematic Reconstruction Four measured quantities: EH, gH, E‘e, qe Three reconstruction methods used: Double angle, Electron method, Jaquet-Blondel • Methods have different resolutions over different kinematic regions • Photoproduction scattered electron escapes down beampipe  Jaquet-Blondel gH qe

24. Model Events: PYTHIA • Parton Level • Matrix element + Parton shower • Hadron Level Model • Fragmentation Model • Lund String (Next Slide) • Detector Level • Detector simulation based on GEANT Parton Level Hadron Level Detector Simulation Factorization: Long range interactions below certain scale absorbed into proton’s structure PDFs

25. Lund String Fragmentation

26. CAL Islands & Energy Flow Objects • CAL cells have a finite size  Particles transverse multiple cells • Reconstruct particles by clustering cells with highest energy neighbor  “CAL Islands” • Can be thought of as number of local maxima • CAL Islands are clustered via probability functions of angular separation • Tracks are extrapolated to an impact point with the CAL • CAL Islands and tracks are clustered if the track’s extrapolated path is within 20cm of the CAL island  “Energy Flow Objects”

27. Jet Finding: Longitudinally Invariant KT Algorithm • In ep: kT is transverse momentum with respect to beamline • For every object i and every pair of objects i, j compute • di = E2T,i(distance to beamline in momentum space) • dij = min{E2T,i,E2T,j}[Dh2 + Df2] (distance between objects) • Calculate min{ di , dij } for all objects • If (dij/R2)is the smallest, combine objects i and j into a new object • If di is the smallest, then object i is a jet • Advantages: • No ambiguities (no seed required and no overlapping jets) • kT distributions can be predicted by QCD

28. Isolated g Identification • Photons • Isolated e.m. calorimeter shower • No associated track • Neutral mesons decay into photons • Produce wider shower in the calorimeter (width) • Shower shape variables • Deposit energy at different rates • Barrel Preshower Detector (# of particles) Barrel Preshower Detector Barrel EMC Calorimeter Barrel HAC Calorimeter g p0

29. g Finders • Two methods can be used: • ELEC5 • Modified e- finder • 10 most energetic EMC cells • Energy in cones around it • Calculate & cut on a probibility • Extrapolate tracks to CAL for rejection • Standard method & relatively simple • Cone based isolation • KTCLUS • Modified KTCLUS jet finder • Run KT Algorithm on EFOs • Only jets that are trackless, electromagnetic & small number of CAL islands • Inherent track handling & kT-Cluster based isolation • New use as a g finder

30. Deeply Virtual Compton Scattering • Provides a clean photon sample • DVCS Sample: • Only one track per event (e±) • Two isolated EM deposits (g,e±) • No hadronic activity • 1999-2000 Data sample • Other cuts as in the paper: “Measurement of Deeply Virtual Compton Scattering at HERA” Physics Letters B537 (2003) 46-62 g Detection of DVCS photons: BPRE BCAL emc BCAL hac

31. HERA I: DVCS Data vs. GenDVCS Energy deposited in BPRE in mips • GenDVCS MC • Minimum ionizing particle units (mips) • Counts the number of charged particles passing through it • Improved agreement with additional dead material in MC • Probability for a photon to convert (ge+e-) well reproduced • Validates the modeling of photons by GenDVCS MC

32. HERA I: Prompt g with Jet in PHP • HERA I, 99-00 Data, 77.1 pb-1 • Photoproduction Sample: • 0.2 ≤ YJB ≤ 0.8 • Q2 <1 GeV2 • 2 or more jets from the Kt algorithm: • Photon candidate: EEMC/ETotal ≥ 0.9 -0.7 ≤ hg ≤ 1.1 (BCAL region) 5.0 ≤ Etg ≤ 16.0 GeV No associated track or within 1.0 in h,j Low multiplicity: # of energy flow objects • Associated jet: EEMC/ETotal ≤ 0.9 -1.6 ≤ hjet ≤ 2.4 6.0 ≤ Etjet ≤ 17.0 GeV • (Note: Etg cut ≠ Etjet cut) Jet g

33. HERA I: PHP Prompt g + Jet LO MC • Data will contain prompt g with jet & inclusive Dijet events • Two PYTHIA MC sample must be generated • Signal MC: PYTHIA Prompt photon with jet subprocess only • Background MC: Inclusive PYTHIA Dijet MC without the Prompt photon subprocess • The relative amount of each is unknown • Add the Signal and Background MC according to a fit to BPRE distribution (g response confirmed  DVCS photons) • Require agreement in lowest BPRE bin • Use fractions from fit for analysis

34. Prompt g + Jet Data vs. MC • Fit sum of prompt g MC & background MC to Barrel Preshower Detector (BPRE) • Require agreement in lowest BPRE bin • Determine relative amounts • Large fraction of events with < 1 MIP  high purity • Excess of MC events in 1-3 mip range • No cut used • Tail well described • Examine other variables • DET = ET,Total – ET,(g + jet) • Dr = Distance (in hf) from g to any track • Both are well reproduced by the sum of MCs • Description verified via comparison between DVCS Data and GenDVCS MC

35. Identification of Jet & g by emc fraction • Fraction of energy in BEMC for g & jet • For Jet: EEMC/ETotal ≤ 0.9 • For g: EEMC/ETotal ≥ 0.9 • Well reconstructed • Minor shift towards lower Eemcg/Etotalg • Prevents double counting, • jet & g also identified by: tracking distinction and number of energy flow objects Unique Assignments

36. Prompt g with Jet Data vs. MC • ETg and ETjet well described • hgand hjet shifted forward in data • Indication that higher order diagrams contribute

37. Prompt g with Jet Data vs. MC: xg • xg = fraction of exchange photon’s momentum involved in the collision • PYTHIA has deficiency in Resolved PHP • Indication that higher order diagrams contribute

38. Differential Cross Section & Correction Factors • Only small deviations from bin to bin • High-hg and Low-xg are ~3x higher • Decrease in efficiency in forward region • Isolation from proton remnant N is the number of events in bin of size DY for the integrated luminosity L Number of reconstructed PYTHIA events C= Number of hadron level prompt photon with jet PYTHIA events

39. Prompt g + Jet in PHP

40. Prompt g + Jet in PHP Data vs. LO MC • I use one global fit of dijet background to prompt g signal MC • Published ZEUS fits background ratio in each bin ETg, ETjet, hgand hjet • Agree well except for: high-hg and hjet • high-hg, has low efficiency • hjet, depends on photon in an indirect way • Data above the MC predictions • Particularly at low ET • High order diagrams ZEUS 99-00, This Thesis ZEUS 99-00, Published PYTHIA 6.3 HERWIG 6.5 ETg (GeV) hg ETjet (GeV) hjet Results published: Eur.Phys.J.C49:511-522,2007

41. Prompt g + Jet in PHP Data vs. NLO Calculation ZEUS H1 • LO, HERWIG & PYTHIA: • Underestimate measured cross section, particularly at Low-ETg • NLO, KZ & FGH: • Improved agreement with measured cross section • NLO, LZ: • Improves description for Low Etg & low hg Published ZEUS Result Published H1 Result Published H1 Result ETg (GeV) ETg (GeV) hg Published ZEUS Result hg Eur.Phys.J.C49:511-522,2007

42. Prompt g + Jet in PHP Data vs. NLO Calculation ZEUS H1 • LO, HERWIG & PYTHIA: • Underestimate measured cross section, particularly at high hjet • NLO, KZ & FGH: • Improved agreement with measured cross section • NLO, LZ: • Improves description of high hjet Published ZEUS Result Published H1 Result Published H1 Result ETjet (GeV) ETjet (GeV) hjet Published ZEUS Result hjet Eur.Phys.J.C49:511-522,2007

43. Prompt g + Jet in PHP: xg ZEUS 99-00, This Thesis ZEUS 99-00, Published PYTHIA 6.3 HERWIG 6.5 Published ZEUS Result • Agreement with published ZEUS data • Published ZEUS fits background ratio in each bin • KZ & FGH: • Improvement compared to LO MC, particularly at high Xg (Direct contribution) • LZ: • Improvement for low Xg (Resolved contribution) Eur.Phys.J.C49:511-522,2007

44. Prompt Photons in DIS • Same as photoproduction: • Isolated photon • Confirm description via DVCS study • Hadronic activity • Separation from background (neutral hadrons) • Additional selection and tools: • ELEC5 photon finder also used • Shower shapes • No jet requirement • Scattered electron • Large contribution from ISR & FSR from e± • ARIADNE for background simulation e±

45. Photon Finding by Shower Shapes Combine signal (PYTHIA) & background (ARIADNE) MC to describe fmax & <dz> p0 Energy in most energetic BEMC cell Total energy of cluster fmax= Energy weighted width in z direction fmax <dz> (cells)

46. HERA II: 04-05 DVCS data • Start with cuts as the HERA I, 99-00, DVCS data study • Only one track per event (e±) • Two isolated EM deposits (g,e±) • No hadronic activity • New Requirements: • More restrictive hg requirements • -0.7 ≤ hg ≤ 0.9 • Both ELEC5 and KT algorithms used • BPRE, fmax and <dz> studied • <dz> ≤ 0.65 cells • Provides a comparison between: • KTCLUS Vs. ELEC5 • BPRE vs. fmax vs. <dz>

47. HERA II: DVCS Data vs. GenDVCS (ELEC5) • g selected with ELEC5 photon finder • ETg and hg not well reproduced • Confirm agreement in each region of ETg and hg (Next slide) • Number of cells in g, BPRE, fmax and <dz> are well reproduced

48. HERA II: DVCS Data vs. GenDVCS hg regions (ELEC5) • No disagreement for ranges of ETg(Not shown) • Central regions show more compact photons • Less dead material • Shorter distances BCAL g g Dead material (Exaggerated) Interaction point

49. HERA II: DVCS Data vs. GenDVCS (KTCLUS) • g selected with KTCLUS photon finder • Agrees with what was seen for ELEC5 selection • Number of cells in g, BPRE, fmax and <dz> are well reproduced • <dz> Is slightly broader than ELEC5

50. HERA II DVCS: ELEC5 vs. KTCLUS • Select photons found by ELEC5 and KTCLUS • Compare ELEC5 and KTCLUS reconstruction • KT clusters more cells into photon • ETg is similar • fmax and <dz> show a slight broadening for KT • Good agreement between ELEC5 and KTCLUS reconstruction of HERA II DVCS photons