Charmed Meson measurements using a Silicon Tracker in Au+Au collisions at √sNN = 200 GeV in STAR experiment at RHIC Jaiby Joseph Ajish 11/9/2011

Outline Introduction Physics at RHIC Why collide nuclei at high energies? RHIC and STAR Physics at RHIC Important observations Heavy quark sector Charm measurement using Silicon Tracker Secondary vertexing Proof of principle with Ks0 Results New results using TOF detector Future

Why collide nuclei at high energies? Phase Diagram Study the Strong Interaction at high temperatures/densities Understand how matter behaved at the dawn of the Universe (~10-6 s) Create and study the properties of the Quark-Gluon Plasma (QGP), a new phase of nuclear matter. (Net Baryon) Density Tc is about 2 Trillion degrees Celsius. QGP is supposed to be a new ‘state’ of matter….like vapor is a new state of water at high temperatures Temperature achieved at RHIC is about 7.2trillion degrees at RHIC Quantum Chromo Dynamics (QCD) is the theory of strong interactions. A phase transition is predicted at high temperatures and/or densities. Collisions of heavy ions at relativistic speeds creates extreme temperatures/densities: Nuclear Matter Quark-Gluon Plasma (deconfined partonic matter) Lattice QCD predicts the phase transition at: Tc ~ 150 -170 MeV and ρ ~ 1GeV/fm3

Relativistic Heavy Ion Collider (RHIC) BRAHMS PHOBOS PHENIX STAR AGS TANDEMS 1 km exploring nuclear matter at extreme conditions over the last decade 2000-2010

RHIC Collisions Initial Conditions STAR Detector Initial high Q2 view of the event Initial high Q2 interactions Hadronization Freeze-Out Partonic matter QGP Experimental approach to induce the QCD phase transition: collide nuclei like Au+Au How to vary the T ? the Volume ? vary energy, Nr of participant Nucleons of Colliding Nuclei Collision systems used at RHIC are: Au+Au, Cu+Cu, d+Au and p+p at energies (7.7 GeV to 500 GeV for p+p) Chemical freezout : When inelastic collisions stop, ie average kinetic energy below pion mass. Relative abundances fixed Hadronic freezout: When elastic collisions stop, ie mean free path larger than system size (ie system is dilute). Momentum spectra freezes HEAVY flavor is created during the initial/high Q^2 interactions since a lot of energy is needed to create heavy quark masses. Processes dominated by perturbative QCD.

STAR Detector (in 2007) The tracking system consisted of : TPC : provides momentum, particle identification Silicon detectors : 1 layer of silicon strip detectors (SSD) and 3 layers of silicon drift detectors (SVT). higher spatial resolution : pointing resolution of 250µm in transverse direction (at 1GeV) was achieved.

STAR Detector (in 2010) MRPC ToF barrel BBC PMD FPD FMS EMC barrel EMC End Cap DAQ1000 FGT Completed Ongoing MTD HFT TPC FHC HLT Time of Flight –Full Barrel (Excellent Particle ID) Previous generation Silicon Detectors are removed and Heavy Flavor Tracker is being built for exclusive charm measurement

Physics @ RHIC New with Heavy Ions Hard Parton Scattering Jets and mini-jets (from hard-scattering of partons)  30 - 50 % of particle production high pt leading particles Extends into perturbative regime Calculations reliable hadrons q Leading particle leading particle schematic view of jet production In p-p collisions hard parton scattering will lead to two jets emerging back-to-back with about equal energy

Physics @ RHIC New with Heavy Ions hadrons q leading particle leading particle schematic view of jet production Scattered partons that propagate through hot and dense nuclear matter will radiate (lose) energy in colored medium interaction of parton with partonic matter will lead to: suppression of angular correlations suppression of high pT particles aka “parton energy loss” or “jet quenching” vacuum QGP

Physics @ RHIC Important observations (light quarks [u,d,s]) Suppression of angular correlations In central Au+Au collisions the light hadrons in away-side jets are suppressed. Not the case in p+p and d+Au partons lose energy via gluon radiation Energy loss depends on properties of medium (gluon densities, size) and properties of “probe” (color charge, mass) Medium created at RHIC has very high opacity

Physics @ RHIC Important observations (light quarks [u,d,s]) suppression of high pT particles In addition, a measurement of energy loss of high pT partons using RAB shows significant suppression Not the case in d+Au. NOT a cold nuclear matter effect. partons lose energy via gluon radiation Nuclear modification factor RAB → energy loss in partonic mater RAB = (A-B pT spectra)/(p-p pT spectra * “volume”) Medium created at RHIC has very high opacity

Physics @ RHIC – Heavy Quark Sector Heavy flavor is mostly produced at the earlier stages of the collision via gluon fusion : not affected by chiral symmetry restoration (i.e. mass is the same in-medium and vacuum) production cross-section found to binary scale ideal (calibrated) probe the medium created in heavy ion collision Look at the energy loss (RAA) [and elliptic flow (v2)] of heavy quarks. Theoretical models predicted gluon radiative energy loss for heavy quarks to be smaller than of light quarks, which is not experimentally observed. 1) Non-photonic electrons (NPE) Method - decayed from charm and beauty hadrons 2) At pT ≥ 6 GeV/c, RAA(NPE) ~ RAA(h±) !!! 3) Surprising Results: contradicts pQCD predictions challenges our understanding of the energy loss mechanism Needs Direct measurement of D and B mesons Raa (charmed) > Raa (light) equals charmed hadrons were expected to lose less energy than light hadrons

Measurement via Semi-leptonic (indirect) channels Indirect measurement through Semi-leptonic decay channels: D0  e+ + X (BR : 6.9 %) D+/- e+/- + X (BR : 17.2%) ✔ Large pT range. Use of specific triggers Relative contribution of electrons from B and D mesons are unknown. Measurement using azimuthal correlation of D mesons with e- Azimuthal correlations of e-h and e-D0 can be utilized to disentangle the charm and bottom contributions ✔ Triggers on high pT electrons 13

Measurement via hadronic (direct) channels Direct measurement using a combinatorial method (combining K and π tracks) Measurement of hadronic decay modes via invariant mass analysis. D0 (D0bar)K-+(K+-) BR : 3.8 % D+/-K BR : 9.2% ✔ C and B contributions separated. Limited to low momentum range. No triggers, no decay vertex reconstruction Challenging due to small decay length Cu+Cu 200 GeV TPC Only (Low pT) (transition) :In this new effort, we try to use the si detectors information to fully reconstruct the charm decay and improve the signal to noise

Measurement using Silicon Vertex Detector and decay vertex fit ✔ SVT/SSD not designed (thickness, geometry) for charm measurement ✔ Full reconstruction/fit of the decay vertex by combining K and π tracks – Some particle ID capabilities obtained from TPC dE/dx dE/dx bands of Kaon, Pion Challenges: Very short lived particles (average decay-length 70 μm) coupled to marginal SVT resolution (the pointing resolution is about 250 μm/GeV) Poor PID: Lack of TOF+SVT data sets, dE/dx has limited resolving power. Poor PID

D0 decay length (Simulation) For <pT> ~ 1GeV/c, βγ ~ 0.54 average decay length ~ 65μm (in the transverse plane) un-boost in the collider!

Distance of Closest Approach resolution run 7 Au+Au@200GeV (MinBias trigger). DCA resolution as a function of inverse momentum. Reflect the resolution and Multiple Coulomb Scattering. Distance of closest approach of the extrapolated tracks event vertex. This defines the event vertex  primary vertex of the D0 particle (since D0 have very small life time they decay close to the event vertex) STAR preliminary Including the silicon detectors in the tracking improves the pointing resolution. with 4 silicon hits, the pointing resolution to the interaction point ~ 250 μm at P = 1GeV/c.

D0 Decay Topology We introduced the Full reconstruction/fit of the decay vertex We used the full track error matrix (inside the beam pipe) for best error estimates Cut Optimization is based on Monte Carlo studies Better resolution in secondary vertex position is achieved with the fit method compared to usual helix swimming methods

Secondary vertex fit (MC Data - pure D0 Events) Mean of the difference reconstructed -MC Rms of the difference reconstructed -MC Correlation between the reconstructed decay and MC Old Silicon Detectors have marginal capabilities but served as a playground for gaining Alignment, Fitting etc experience. There is no systematic shift in reconstructed quantities. The standard deviation of the distribution is flat at ~ 250 m , which is of the order of the resolution of (SSD+SVT). 19

Reconstructed Quantities (example) (MC Data - pure D0 Events) pT Resolution Invariant Mass Reconstructed Decay Vertex Resolution Resolution : Inv Mass sigma ~ 13 MeV (0.7%, after a gauss fit) Decay Vertex Coordinates ~ 220 μm (transverse) ~ 200 μm (z-direction) The reconstructed parameters behave as expected with the current detector resolution.

Proof of principle with K0s Test with K0s decay reconstruction (about 200x the decay distance of D0s): c = 2.68 cm Signed decay length : an excess can be observed on the positive side of the decay length distribution, indicating the presence of long-lived decays. use the decay length significance L/L to improve the signal. more appropriate because of the momentum dependence of the decay length. Before cut After cut Signal+background background  After using a cut SL > 10, a clear peak at the K0S mass is observed. 21

D0+D0bar Signal (in 2007 Data) 24 Million Au+Au @ 200 GeV/c events are used for this analysis. 3rd degree polynomial fit is used for background estimation. Fit yielded an apparent signal significance of 10-σ (combined D0+D0bar signal) Signal remained stable as cuts are varied. Pol3 + gaus Pol3 Gaussian Mean = 1864.19 ± 10 MeV gaus fit

D0 and D0bar separately D0bar/D0 Ratio ~ 1.18 ± 0.24 ( compatible with unity indicating a vanishing μB) Attempt to extract physics with polynomial fit method revealed some problems: A robust background estimation method needed to see if the peak observed was an artifact A Multi Variate analysis is in progress to measure signal using same sign background estimation method. * \mu_b about 30 MeV at RHIC and 250 at CERN-SPS and 500 MeV at AGS * \mu_b is the energy you need to add a new particle in a thermal system…Vanishes if equal quark/anti quark pairs. The fact that about a third of the SVT/SSD system was dead during Run-7, combined with the marginal resolution of the previous generation silicon detector and combinatorial background is challenging to our efforts.

Ongoing Analysis with Multivariate Analysis (TMVA) TMVA is a tool for simultaneous optimization of many correlated cuts. Training samples for signal (pure D0) and background (`HIJING Au+Au’) are provided. It then produces a classifier output with weight files for signal and background. After training, testing can be done with Data sample (MC Embedding/Real) MC D0 Embedding 2007 Au+Au Data (1-2% of available data) Preliminary results looks promising but… work in progress.

Recent charm measurements with Time Of Flight (TOF) Detector STAR Time Of Flight (TOF) detector provides better particle ID (measures particle velocity β[β=v/c]) dE/dx + TOF offers excellent K, π separation up to p ~ 1.5 - 2 GeV/c New results use ~ 250 Million Au+Au Events from year 2010 and ~ 210 Million p+p events from 2009

Recent charm measurements with Time Of Flight (TOF) Detector Corrected pT Spectra RAA RAA shows charm suppression at ~ 4 GeV/c Cross-section is found to binary scale, indicating its production via initial hard scattering at RHIC Charm Cross Section

Future Heavy Flavor Tracker (HFT) STAR is undergoing a detector upgrade for the unambiguous measurement of charm – The Heavy Flavor Tracker (HFT) Low mass detector designed to identify mid-rapidity Charm and Beauty mesons and baryons through direct reconstruction, with unprecedented pointing resolution. CMOS sensors will provide single track resolution ~ 20-30 µm at <pT> ~ 1GeV/c.

Key Measurements of the HFT (1) Energy loss of direct D0 - Rcp (2) Elliptic flow, v2 (3) Charmed Baryon to Meson Enhancement The methods we have developed here are directly applicable in HFT

Summary For the 1st time in STAR, a secondary vertex reconstruction with full fit is developed using silicon vertex detectors. Results obtained looks interesting, work in progress to wrap up the analysis with the most advanced tools in High Energy Physics – Multi Variate Analysis Pioneering work that is directly applicable to the upcoming upgrade to STAR – Heavy Flavor Tracker (HFT)

Thank you

Back-Up

Physics @ RHIC Important observations (Light flavors) Partonic Collectivity Substantial elliptic flow (v2) signal observed for a variety of particle species. Rapid Thermalization v2 scaled by the number of valance quarks shows an apparent scaling Development of anisotropy in the partonic stage of collision Behaves like an ideal fluid

Charm Cross-Section Comparison at 200 GeV NLO Ref: R. Vogt, arXiv:0709.2531v1 [hep-ph] STAR and PHENIX do not agree about total charm production x-section Need precise, exclusive measurements

Secondary Vertex Resolution Plots (x,y,z) Fit Method (central region) σXY ~ 55μm σZ ~ 25μm X Y Z Helix Swimming Method – using global parameters (central region) σXY ~ 150μm σZ ~ 135μm Helix Swimming Method – using DCAGeometry (central region) σXY ~ 140μm σZ ~ 125μm Simulation results shows that a factor of two was gained in secondary vertex Resolution

Strategy of Reconstruction Cuts are applied in the analysis code to reduce background and to increase the candidate pool Select Event – Apply Event Level Cuts Select Trigger Cuts on Z-Vertex Position and its error Loop over Tracks – Apply Track Level Cuts Number of Silicon Hits Transverse DCA (DCAXY) Track Momentum etc. Pair Association - D0 Candidate Level Cuts rapidity, Cosine of Kaon decay angle etc. Decay Vertex Fit – Decay fit Level Cuts probability of fit, decay length error of decay length etc. Particle Identification – Apply PID Cuts |nσK|, |nσπ| Output Saved for offline Analysis

Attempt to extract physics Uncorrected pT spectra: A normalized pT Spectra corrected for acceptance and efficiency would be used to: - extract total charm cross-section, freeze out parameters etc. - calculate energy loss RAA The results with a polynomial background estimate seem to be inconsistent. A robust background estimation method needed to see if the peak observed was an artifact – a “same sign” background subtraction method was performed

2007 Production MinBias Cuts in 1st Production Cuts in 2nd Production cut changed new cut Cuts in 1st Production Cuts in 2nd Production EVENT level triggerId : 200001, 200003, 200013 Primary vertex position along the beam axis : |zvertex| < 10 cm Resolution of the primary vertex position along the beam axis: |zvertex|< 200µm TRACKS level Number of hits in the vertex detectors : SiliconHits>2 (tracks with sufficient DCA resolution) Transverse Momentum of tracks: pT >.5GeV/c Momentum of tracks: p >.5GeV/c Number of fitted: TPC hits > 20 Pseudo-rapidity :||<1 (SSD acceptance) dEdxTrackLength>40 cm DCA to Primary vertex (transverse), DCAxy< .1 cm EVENT level triggerId : 200001, 200003, 200013 Primary vertex position along the beam axis : |zvertex| < 10 cm Resolution of the primary vertex position along the beam axis: |zvertex|< 200µm TRACKS level Number of hits in the vertex detectors: SiliconHits>1 Transverse Momentum of tracks: pT >.5GeV/c Momentum of tracks p >.8GeV/c Ratio TPC hits Fitted/Possible > 0.51 Pseudo-rapidity :||<1.2 dEdxTrackLength>40 cm DCA to Primary vertex (transverse), DCAxy< .2 cm Radius of first hit on track : < 9 cm if number of silicon hits =2 < 13 cm else

Cuts from Previous production Continued.. Cuts from Previous production Cuts in New Production D0 candidate |y(D0)|<1 |cos(*)|<0.8 DECAY FIT level Probability of fit >0.01 && |sLength|<.1cm Particle ID : ndEdx :|nK|<2.5, |nπ|<2.5 DECAY FIT level Probability of fit >0.1 && |sLength|<.1cm Particle ID : ndEdx :|nK|<2, |nπ|<2 In both productions we made a pico file for further analysis. Cuts Used for making a pico file Previous Production New Production ndEdx :|nK|<2, |nπ|<2 |cos(*)|<0.6 DCA daughters < 300 µm |D0Eta|<1.85 |Cos(θ*)<0.6

Uncorrected pT Spectra

Physics @ RHIC Important observations (Light flavors) Partonic Energy Loss Partonic Collectivity Substantial elliptic flow (v2) signal observed for a variety of particle species. Rapid Thermalization v2 scaled by the number of valance quarks shows an apparent scaling Development of anisotropy in the partonic stage of collision In central Au+Au collisions the light hadrons in away-side jets are suppressed. Different for p+p and d+Au In addition, a measurement of energy loss of high pT partons using RAB shows significant suppression partons lose energy via gluon radiation Medium created at RHIC has very high opacity Behaves like an ideal fluid

‘Same sign’ background subtraction

Heavy Quark Energy Loss Puzzle – NPE Method Still the main method at RHIC STAR: Phys. Rew. Lett, 98, 192301(2007) and nucl-ex/0607012v3 1) Non-photonic electrons (NPE) decayed from - charm and beauty hadrons 2) At pT ≥ 6 GeV/c, RAA(NPE) ~ RAA(h±) !!! Contradicts naïve pQCD predictions Surprising results - - challenge our understanding of the energy loss mechanism - force us to RE-think about the elastic-collisions energy loss - Requires direct measurements of c- and b-hadrons.

Measurement via Semi leptonic (indirect) channels Indirect measurement through Semi-leptonic decay channels: D0  e+ + X (BR : 6.9 %) D+/- e+/- + X (BR : 17.2%) ✔ Large pT range. Relative contribution of electrons from B and D mesons are unknown. Use of specific triggers Measurement using azimuthal correlation of D mesons with e- Azimuthal correlation of open charm mesons with non-photonic Electron can be utilized to disentangle the charm and bottom contributions[3] ✔ Triggers on high pT electrons Any information from direct reconstruction of D and B-mesons would help

My contributions Charm Analysis Service Work QA, Problem fixing, Resolution studies Detailed studies for ‘online’/‘offline’ Cut optimization Data Productions (Micro/Pico-DSTs) First observation charmed meson signal in real data (from 2007 Au+Au dataset) Signal extraction, optimization, fitting, pT binning Embedding QA, Study of Systematics and Physics Analysis Service Work acceptance of D-mesons with a prototype design for the HFT upgrade

An explanation for the non-consistent physics results Same cuts are used to produce this picture that were used in the polynomial fit case Red = Signal Blue = [(++) + (--)] 0.5 for D0 only or D0bar only. This was a big surprise!!! The fact that about a third of the SVT/SSD system was dead during Run-7, combined with the marginal resolution of the previous generation silicon detector and combinatorial background limits our efforts. A final effort to measure the signal using a multivariate analysis is in progress.

Charm cross section vs √sNN

Charm and beauty from e-D0 azimuthal correlations PYTHIA BEAUTY sign(e) ≠ sign(K) Like Sign Δϕ ~ π Like Sign Δϕ ~ 0 sign(e) = sign(K) Unlike Sign Δϕ ~ π A Mischke, Phys. Lett. B671, 361 (2009) 23/06/2010 Kent State University, 9 November 2010, S. Kabana