Dileptons with PHENIX Dilepton analysis PHENIX setup with and w/o HBD Key: background subtraction Results Well understood baseline pp w & w/o HBD, dAu A new handle on charm/bottom separation in dAu Dilepton puzzle in AuAu Comment on STAR data HBD analysis HBD performance in central AuAu collisions Status at QM2012 Recent progress Summary Axel Drees, May 22 nd 2013, Trento

Axel Drees PHENIX Setup without HBD background  0  e - e +   e - e + e-e- ee PC1 DC magnetic field & tracking detectors e + e - pairs E/p and RICH 2 e - and e + acceptance are different No background rejection! dilepton S/B < 1: GeV Run-4 AuAu Run-5 pp, CuCu Run8 dAu ee e-e-

Axel Drees PHENIX Setup with HBD background  0  e - e +   e - e + e-e- ee PC1 DC magnetic field & tracking detectors e + e - pairs E/p and RICH 3 background rejection in field free region! S/B improves by GeV Run-9 pp Run-10 AuAu 62.4 GeV Run-10 AuAu ee e-e- HBD field free region PHENIX Setup with HBD

Axel Drees Combinatorial background: e + and e  from different uncorrelated source Need event mixing because of acceptance differences for e + and e  Use like sign pairs to check event mixing Unphysical correlated background Track overlaps in detectors Not reproducible by mixed events: removed from event sample (pair cut) Correlated background: e + and e  from same source but not “signal” “Cross” pairs  “jet” pairs Use Monte Carlo simulation and like sign data to estimate and subtract background X π0π0 π0π0 e+e+ e-e- e+e+ e-e- γ γ π0π0 e-e- γ e+e+ Subtractions dominate systematic uncertainty 4 Key: Understanding the Background Subtraction

Axel Drees Component-by-Component Bkg Subtraction Mixed events Correlated Signal = Data-Mix Cross pairs Simulate cross pairs with decay generator Normalize to like sign data for small mass Jet pairs Simulate with PYTHIA Normalize to like sign data Like Sign Data Unlike Sign Data Unlike sign pairs same simulations normalization from like sign pairs Alternative methode Correct like sign correlated background with mixed pairs Signal: S/B  1 5 pp raw data

/MeV /MeV Key to dilepton analysis is to understand the background /MeV /MeV Correlated background roughly consistent with expectation   : N  AA /N  pp *  AA  pp jet: N coll *R AA *  AA  pp

Alternative Method: Relative Acceptance Correction FG1122 = S = FG12 –  FG1122 Method not usable for AuAu A large number of simulations show that relative acceptance correction can not be controlled to better than a few % PHENIX d+Au run-8

Sys. Uncertainties on Relative Acceptance Relative acceptance correction depends on Physics source, thus mixed events not perfect description Difference in number of electron and positrons Differences in pt spectra of electrons & positrons z-vertex position Variations in active area (within similar run groups) Case study: variation of active area in dAu setup In MC simulation remove randomly drift chamber readout cards (1/80 of acceptance) Axel Drees 8

Dilepton Continuum in p+p Collisions PHENIX Phys. Lett. B 670, 313 (2009) Data and Cocktail of known sources represent pairs with e  and e  PHENIX acceptance Data are efficiency corrected Excellent agreement of data and hadron decay contributions with 30% systematic uncertainties 9 Consistent with PHENIX single electron measurement  c = 567±57±193  b * * PYTHIA, 1 st order, D/B semileptonic decay

p+p Data with the HBD Cocktail updated BR, trans. form factors,  -shape, bremsstrahlung, c/b New Charm/Bottom simulation using cross sections determined in d+Au Like sign contribution subtracted in data and simulation Axel Drees 10 Consistent with known sources!

Axel Drees Dilepton Continuum in d+Au Collisions 11 PHENIX preliminary PYTHIA 3 data sets consistent with cocktail within 20-30% pp, dAu w/o HBD; and pp w HBD Baseline at RHIC energies well understood!

Detailed double differential data Subtract hadron decay contribution Remaining yield dominated by heavy flavor production d+Au Data: p T Dependence Axel Drees 12 Common wisdom: charm bottom (but not quite true!)

What can we learn about Charm/Bottom Axel Drees 13 Any like sign subtraction in data must be accounted for if compared to simulation! Use 2 nd order pQCD Monte-Carlo Significant improvement over 1 st PYTHIA (gg-fusion) Subtle differences compared to “tuned” PYTHIA New features of bottom production not relevant for charm Feed-down: BR (B→e) ~ BR (B→D→e) ~ 10% B 0 B 0 oscillations 60% of pairs unlikesign ee-pairs from bottom production ee-pairs from charm & bottom production in PHENIX acceptance

Charm/Bottom Cross Section from d+Au Data Axel Drees 14 Hadron decay cocktail and like-sign pairs subtracted normalized to double differential data (m vs p T ) Extrapolated heavy flavor cross sections:  cc = 710  62 (stat)  183 (syst)  80 (model)  b  bb = 4.5  0.7 (stat)  1.1 (syst)  0.2 (model)  b DY not considered, bottom might be overestimated

Double Differential View Axel Drees 15 Relative importance of c/b changes with mass and p T c/b discrimination power Low p T low mass → charm High p T low mass → bottom Possible new opportunity for thermal radiation (Au+Au) Mass – p T independent Inverse slope in mass reflects temperature of ~ few 100 MeV Should be distinguishable from c/b Will be effected by heavy flavor modification d+Au at 200 GeV c > b c ~ b c < b

Axel Drees Au+Au Dilepton Continuum Excess 150 <m ee <750 MeV: 4.7 ± 0.4(stat.) ± 1.5(syst.) ± 0.9(model) Charm from PYTHIA filtered by acceptance  cc = N coll × 567±57±193  b Charm “thermalized” filtered by acceptance  cc = N coll × 567±57±193  b Intermediate-mass continuum: consistent with PYTHIA since charm is modified room for thermal radiation hadron decay cocktail tuned to AuAu 16 Need better handle on heavy flavor PHENIX Phys. Rev. C 81 (2010)

Soft Low Mass Dilepton Puzzle m T spectrum of excess dileptons Subtract cocktail Correct for pair acceptance Fit two exponentials in m T –m 0 1 st component T ~ 260 MeV Consistent with thermal photon yield 2 nd “soft” component T ~ 100 MeV Independent of mass More than 50% of yield Soft component also seen at SPS in NA60 and CERES data Axel Drees  11  9 MeV 258  37  10 MeV 300 < m < 750 MeV Soft component eludes theoretical explanation acceptance corrected PHENIX Phys. Rev. C 81 (2010)

Soft Low Mass Dilepton Puzzle m T spectrum of excess dileptons Subtract cocktail Correct for pair acceptance Fit two exponentials in m T –m 0 1 st component T ~ 260 MeV Consistent with thermal photon yield 2 nd “soft” component T ~ 100 MeV Independent of mass More than 50% of yield Soft component also seen at SPS in NA60 and CERES data Axel Drees  11  9 MeV 258  37  10 MeV 300 < m < 750 MeV acceptance corrected PHENIX Phys. Rev. C 81 (2010) CERES Soft component eludes theoretical explanation

A look at STAR p+p Dilepton Data Axel Drees 19 STAR arXiv: charm cross section: STAR  = 920  b  = 710  b* acceptance: STAR  |  |=2 PHENIX cocktail in STAR acceptance PHENIX cocktail and STAR cocktail consistent *consistent with PYTHIA 570  b

A look at STAR pp and AuAu Data Axel Drees 20 Compared to PHENIX cocktail no appreciable enhancement On a different note: comparing PHENIX data to cocktail Not acceptance corrected Soft component emphasized 20%-30% increase of enhancement Ratio STAR data/ PHENIX cocktail in STAR acceptance

Data from PHENIX HBD Upgrade HBD fully operational: Single electron ~ 20 P.E. Conversion rejection ~ 90% Dalitz rejection ~ 80% Improvement of S/B factor 5-10 to published results Axel Drees 21 Window less CF4 Cherenkov detector GEM/CSI photo cathode readout Operated in B-field free region p+p data in 2008/9 Au+Au data in 2009/10 Improve S/B by rejecting combinatorial background

Effect of CF 4 Scintillation on HBD Performance Material of HBD 2-3% conv. probability 3-4x larger conversion background Most material behind radiator Reduced  -rejection in RICH Effect centrality dependent Scintillation ~ 10pe/pad in central AuAu Subtract statistically; multiple algorithms; 2 presented at QM11 Fluctuation in scintillation results in centrality dependent rejection In central collisions estimate 90% rejection at 65% efficiency Axel Drees 22 Rely on veto by HBD Scintillation light limits HBD veto HBD module in AuAu before subtraction: HBD module after subtraction: 10% central, 62 GeV: efficiency 60% Rejection 90% e-e- e-e-

Status at Quark Matter 2012 Background subtraction with rel. acceptance corrected like-sign pairs Systematic uncertainty to large to obtain result in central collisions Observe over subtraction in central collisions Ongoing analysis improvements Component-by-component background subtraction (run-4 AuAu) Improved pion rejection Increased statistics Axel Drees 23 Peripheral (60-92%) Au+Au Semi central (20-40%) Au+Au

Analysis Improvements Since QM2012 Improved data analysis of RICH information Issue: parallel track point to same ring in RHIC New algorithm resolves ring MC simulation: Purity 70%  90% at 80% pair efficiency Include TOF information for hadron rejection EMCal 450 ps; TOF ~120 ps Small improvement of S/B Increased statistics by 1.25 Axel Drees 24 PbSc timing central AuAu New run of data production completed

Quantitative Understanding of Background Simulation of background from   production  0  e - e +  e - e + Full PHENIX Geant simulation Unlike sign pairs Like sign cross pairs Work in progress Absolute normalization Quantitative understanding of rejection Simulation of jet background Once completed  look at unlike sign data Axel Drees 25 X Blind analysis conversions Dalitz data dito with HBD rejection Central (10-20%) AuAu Data and MC normalized to each other Late conversions rejection factor 6 late conversion early conversion no HBD rejection early conversion rejection factor 4

Summary and Outlook PHENIX measured ee-pairs at √s=200 GeV Well understood baseline in pp and dAu collisions Within 20-30% consistent with hadron decays & heavy flavor production Constrains heavy flavor production Puzzles in AuAu collisons larger excess beyond contribution from hadronic phase with medium modified  -meson properties soft momentum distribution Thermal photon puzzle (see talk by Gabor David) Large thermal yield with T > 220 MeV (10-20% of decay photons) Large elliptic flow (v2) HBD analysis moving towards completion Expect results this year Axel Drees 26

Backup slides Axel Drees 27

Axel Drees 28 Centrality Dependence of Enhancement CERES 95/96 Warning this is for illustration only!!! 0-30% cross section N part  N ch CERES p T > 200 MeV/c PHENIX In+In

Axel Drees 29 Acceptance Function Polarized unpolarized: difference smaller than 10%

Axel Drees 30 Acceptance for Virtual Photons Data presented as e+ and e- in acceptance, this is not the same as virtual photon in acceptance! Physical distribution requires that virtual photon is in acceptance! detector ** ee ee Virtual photon and electron and positron in the acceptance B-field ** ee ee Virtual photon in acceptance electron and/or positron NOT in the acceptance detector B-field Case A Case B Acceptance depends on pair dynamics!

Axel Drees Combinatorial Background: Like Sign Pairs --- Foreground: same evt N Background: mixed evt B++ l Shape from mixed events l Excellent agreements for like sign pairs l also with centrality and p T l Normalization of mixed pairs l Small correlated background at low masses from double conversion or Dalitz+conversion normalize B ++ and B  to N ++ and N  for m > 0.7 GeV Normalize mixed  pairs to l Subtract correlated BG l Systematic uncertainties l statistics of N ++ and N -- : 0.12 % l different pair cuts in like and unlike sign: 0.2 % Normalization of mixed events: systematic uncertainty = 0.25% Au-Au 31

Axel Drees Au-Au Raw Unlike-Sign Mass Spectrum Mixed unlike sign pairs normalized to: Unlike sign pairs data  signal /signal =  BG /BG * BG/signal as large as 200!! 0.25% Systematic errors from background subtraction:  up to 50% near 500 MeV arXiv: Run with added Photon converter 2.5 x background Excellent agreement within errors! 32

Axel Drees 33 Centrality Dependence of Background Subtraction For all centrality bins mixed event background and like sign data agree within quoted systematic errors!! Evaluation in 0.2 to 1 GeV range Compare like sign data and mixed background Similar results for background evaluation as function p T

Axel Drees Background Description of Function of p T Good agreement 34