1Jim Thomas - LBL

Energy Loss and Flow in Heavy Ion Collisions at RHICPreparing to open a new chapter at RHIC with Charm + Beauty

Jim ThomasLawrence Berkeley National Laboratory

Berkeley, CA

Argonne National Laboratory

March 5th, 2008

2Jim Thomas - LBL

Who is RHIC and What Does He Do?

RHIC

• Two independent rings

• 3.83 km in circumference

• Accelerates everything, from p to Au

s L p-p 500 1032

Au-Au 200 1027

(GeV and cm-2 s-1)

• Polarized protons

• Two Large and two small detectors were built

h

BRAHMS

PHOBOS

PHENIX

STARSTAR

And for a little while longer, it is the highest energy heavy ion collider in the world

Long Island

Long Island

3Jim Thomas - LBL

STAR is a Suite of Detectors

Barrel EM Calorimeter

FTPCs

Time Projection Chamber

Silicon TrackerSVT & SSD

Endcap Calorimeter

Magnet

Coils

TPC Endcap & MWPC

Central Trigger Barrel & TOF

Beam Beam Counters

4.2 meters

Not Shown: pVPDs, ZDCs, and FPDs

A TPC lies at the heart of STAR

PMD

4Jim Thomas - LBL

RHIC Physics is Relativistic Nuclear Physics

5Jim Thomas - LBL

Unlike Particle Physics, the initial state is important

• The nucleus is not a point like particle, it is macroscopic

• Only a few of the nucleons participate in the collision as determined by the impact parameter

• The initial state is Lorentz contracted

• There is multiple scattering in the initial state before the hard collisions take place

– Cronin effect

• Cross-sections become coherent. – The uncertainty principle allows wee

partons to interact with the front and back of the nucleus

– The interaction rate for wee partons saturates ( ρσ = 1 )

• The intial state is even time dilated– A color glass condensate

• proton • neutron • delta • pion string

6Jim Thomas - LBL

Au on Au Event at CM Energy ~ 130 GeV*A

Two-track separation 2.5 cm

Momentum Resolution < 2%

Space point resolution ~ 500 m

Rapidity coverage –1.8 < < 1.8

A Central Event

Typically 1000 to 2000 tracks per event into the TPC

7Jim Thomas - LBL

Particle ID using Topology & Combinatorics

Secondary vertex: Ks + p +

+ + K e++e-

Ks + + - K + + K -

p + - + + -

from K+ K- pairs

K+ K- pairs

m inv

m inv

same event dist.mixed event dist.

background subtracted

dn/dm

dn/dm

“kinks”

K +

8Jim Thomas - LBL

Strange Baryons and Mesons: Au+Au @ 200 GeV

, , and yields .vs. pTPhys. Rev. Lett. 98 (2007) 060301

9Jim Thomas - LBL

Transverse Radial Expansion: Isotropic Flow

The transverse radial expansion of the source (flow) adds kinetic energy to the particle distribution. So the classical expression for ETot

suggests a linear relationship

-

K -

p

Au+Au at 200 GeV

Typical STAR Data

2KFOObs massTT

Slopes decrease with mass. <pT> and the effective temperature increase with mass.

T ≈ 575 MeV

T ≈ 310 MeV

T ≈ 215 MeV

10Jim Thomas - LBL

Kinetic Freezeout from Transverse Radial Flow

<ßr> (RHIC) = 0.55 ± 0.1 cTKFO (RHIC) = 100 ± 10 MeV

Explosive Transverse Expansion at RHIC High Pressure

Tth

[GeV

]< r

> [

c]

STA

RPH

EN

IX

Thermal freeze-out determinations are done with the blast-wave model to find <pT>

STAR Preliminary

11Jim Thomas - LBL

Dependent Distributions – Flow

• The overlap region in peripheral collisions is not symmetric in coordinate space

• Almond shaped overlap region

– Larger pressure gradient in the x-z plane drives flow in that direction

– Easier for high pT particles to emerge in the direction of x-z plane

• Spatial anisotropy Momentum anisotropy

• Perform a Fourier decomposition of the momentum-space particle distribution in the plane

– For example, v2 is the 2nd harmonic Fourier coefficient of the distribution of particles with respect to the reaction plane

))2cos(2)cos(21(2

121

2

3

3

vvdydpp

Nd

pd

dNE

TT

directed elliptic isotropic

12Jim Thomas - LBL

Interpreting Flow – order by order

n=1: Directed Flow has a period of 2 (only one maximum)

– v1 measures whether the flow goes to the left or right – whether the momentum goes with or against a billiard ball like bounce off the collision zone

n=2: Elliptic flow has a period of (two maximums)

– v2 represents the elliptical shape of the momentum distribution

))4cos(2)2cos(2)cos(21(2

1421

2

3

3

vvvdydpp

Nd

pd

dNE

TT

directed elliptic isotropic higher order terms

13Jim Thomas - LBL

v2 vs. pT and Particle Mass

• The mass dependence is reproduced by hydrodynamic models

– Hydro assumes local thermal equilibrium

– At early times

– Followed by hydrodynamic expansion

D. Teaney et al., QM01 Proc.P. Huovinen et al., nucl-th/0104020

PRL 86, 402 (2001) & nucl-ex/0107003

Anisotropic transverse flow is large at RHIC

• v2 is large– 6% in peripheral

collisions (for pions average over all pT )

• Flow is developed very rapidly

– Data suggests very early times ~ fm/c

• Hydro calculations are in good agreement with the data

– Hydro assumes local thermal equilibrium

– Followed by hydrodynamic expansion

– The mass dependence is reproduced by the models

M. Oldenberg, nucl-ex/0412001.P. Huovinen et al., QM04

14Jim Thomas - LBL

Elliptic Flow: in an ultra-cold Fermi-Gas

Li-atoms released from an optical trap exhibit elliptic flow analogous to what is observed in ultra-relativistic heavy-ion collisions

– Elliptic flow is a general feature of strongly interacting systems!

A Simulation of Elliptic Flow

15Jim Thomas - LBL

v2 at high pT shows meson / baryon differences

Asym. pQCD Jet Quenching

Bulk PQCD Hydro

qn Coalescence

16Jim Thomas - LBL

The Recombination Model ( Fries et al. PRL 90 (2003) 202303 )

The flow pattern in v2(pT) for hadrons

is predicted to be simple if flow is developed at the quark level

pT → pT /n

v2 → v2 / n ,

n = (2, 3) for (meson, baryon)

17Jim Thomas - LBL

A Macroscopic many body system: It Flows

Spatial anisotropy Momentum anisotropy

– For example, v2 is the 2nd harmonic Fourier coefficient of the distribution of particles with respect to the reaction plane

))2cos(2)cos(21(2

121

2

3

3

vvdydpp

Nd

pd

dNE

TT

directed elliptic isotropic

18Jim Thomas - LBL

Elliptic flow scales with the number of quarks

Implication: (uds) quarks, not hadrons, are the relevant degrees of freedom at early times in the collision history

19Jim Thomas - LBL

v4 Exhibits Constituent Quark Scaling, too

S.L. Huang QM08

))4cos(2)2cos(2)cos(21(2

1421

2

3

3

vvvdydpp

Nd

pd

dNE

TT

directed elliptic isotropic higher order terms

20Jim Thomas - LBL

Constituent Quark Scaling?

• Hadrons are created by the recombination of quarks and this appears be the dominant mechanism for hadron formation at intermediate pT

• Baryons and Mesons are produced with equal abundance at intermediate pT

• The collective flow pattern of the hadrons appears to reflect the collective flow of the constituent quarks.

• up, down, and strange quarks do it … despite the difference in their masses

Partonic Collectivity

21Jim Thomas - LBL

Hints of Elliptic Flow with Charm

• D e +X

Single electron spectra from PHENIX show hints of elliptic flow

Is it charm or beauty?

• Very profound, if it is true, because the Charm quark is rare and heavy compared to u, d, or s quarks

• Indicator for rapid thermalization

• Their will be RHIC upgrades to cut out large photonic backgrounds:

e+e-

and reduce other large statistical and systematic uncertainties

A Look to the Future: better if we can do direct topological identification of Charm

Shingo Sakai, QM 2006 PRL 98, 172301 (2007)

22Jim Thomas - LBL

Lets look at some collision systems in detail …

Final stateInitial state

Au + Au

d + Au

p + p

23Jim Thomas - LBL

Partonic Energy Loss and Jet Quenching

Energy loss suppression of leading hadron yieldThe jet can’t get out!

ddpdT

ddpNdpR

TNN

AA

TAA

TAA /

/)(

2

2

Binary collision scaling p+p reference

d+Au

Au+Au

No quenching

Quenching!

24Jim Thomas - LBL

Heavy Flavor Energy Loss … RAA for Charm

• Heavy Flavor energy loss is an unsolved problem

– Gluon density ~ 1000 expected from light quark data

– Better agreement with the addition of inelastic E loss

– Good agreement only if they ignore Beauty …

• Beauty dominates single electron spectra above 5 GeV

• In the future, RHIC upgrades will separate the Charm and Beauty contributions

Theory from Wicks et al. nucl-th/0512076v2

Where is the contribution from Beauty?

25Jim Thomas - LBL

Jet Physics … it is easier to find one in e+e-

Jet event in eecollision STAR Au+Au collision

26Jim Thomas - LBL

Angular Distribution: Peripheral Au+Au data vs. pp+flow

C2(Au Au)C2(p p) A *(1 2v22 cos(2))

Ansatz: A high pT triggered Au+Au event is a superposition of a high pT triggered p+p event plus anisotropic transverse flow

v2 from reaction plane analysis

“A” is fit in non-jet region (0.75<||<2.24)

27Jim Thomas - LBL

Angular Distribution: Central Au+Au data vs. pp+flow

C2(Au Au)C2(p p) A *(1 2v22 cos(2))

28Jim Thomas - LBL

Lessons learned – Dark Matter … its opaque

• The backward going jet is missing in central Au-Au collisions when compared to p-p data + flow

• The backward going jet is not suppressed in d-Au collisions

• These data suggest opaque nuclear matter and surface emission of jets

Surface emission

Suppression of back-to-back correlations in central Au+Au collisions

29Jim Thomas - LBL

Where does the Eloss go?

PHENIX p+p Au+Au

Lost energy of away-side jet is redistributed to rather large angles!

Trigger jetAway-side jet

30Jim Thomas - LBL

Mach Cone: Theory vs Experiment

• Hint of a Mach Cone?

STAR preliminary

0-12% 200 GeV Au+Au

away

near

Medium

mach cone

Mediumaway

neardeflected jets

31Jim Thomas - LBL

The Ridge (after v2 subtraction)

d+Au, 40-100% Au+Au, 0-5%

d+Au Au+Au

3 < pT(trig) < 6 GeV2 < pT(assoc) < pT(trig)

Jet

Ridge

STAR Preliminary

32Jim Thomas - LBL

The Ridge May be due to Very Strong Fields

33Jim Thomas - LBL

Strong Fields Run Away Thermalisation

34Jim Thomas - LBL

AdS/CFT correspondence (from H. Liu)

N = 4 Super-Yang-Mills theory with SU(N)

Maldacena (1997) Gubser, Klebanov, Polyakov, Witten

A string theory in 5-dimensional anti-de Sitter spacetime

NN = 4 Super-Yang-Mills (SYM):

anti-de Sitter (AdS) spacetime: homogeneous spacetime with a negative cosmological constant.

maximally supersymmetric gauge theory

scale invariant

A special relative of QCD

The value turns out to be universal for all strongly coupled

QGPs with a gravity description. It is a universal lower bound.

1

4s

35Jim Thomas - LBL

PHENIX PRL 98, 172301 (2007)

• RAA of heavy-flavor electrons in 0%–10% central collisions compared with 0 data and model calculations

• V2 of heavy-flavor electrons in minimum bias collisions compared with 0 data and the same models.

• Conclusion is that heavy flavor flow corresponds to /s at the conjectured QM lower bound

0 0

36Jim Thomas - LBL

Old Chinese Proverb

Beware of theorists waiting for data– Confusion

37Jim Thomas - LBL

Nuclear Fluid Dynamics ... with friction

• The energy momentum tensor for a viscous fluid

• Conservation laws: and where

• The elements of the shear tensor, , describe the viscosity of the fluid and can be thought of as velocity dependent ‘friction’

• Simplest case: scaling hydrodynamics– assume local thermal equilibrium

– assume longitudinal boost-invariance

– cylindrically symmetric transverse expansion

– no pressure between rapidity slices

– conserved charge in each slice

• Initially expansion is along the Z axis, so viscosity resists it– Conservation of T means that energy and momentum appear in the

transverse plane … viscosity drives radial flow

• Viscosity is velocity dependent friction so it dampens v2 – Viscosity (/z ) must be near zero for large elliptic flow to be observed

pguupT )(

0 T 0

j uj ii

38Jim Thomas - LBL

3D Hydro … new insights

• Romatschke2 performed relativistic viscous hydrodynamics calculations including the new 2nd order terms beyond Landau’s prescription

• Data on the integrated elliptic flow coefficient v2 are consistent with a ratio of viscosity over entropy density up to /s 0.16

• But data on minimum bias v2 seem to favor a much smaller viscosity over entropy ratio, below the bound from the anti–de Sitter conformal field theory conjecture

PRL 99, 172301 (2007)

39Jim Thomas - LBL

Which theory is correct? Only the data can tell.

40Jim Thomas - LBL

Caption: The viscosity to entropy ratio versus a reduced temperature.

Lacey et al. PRL 98:092301(07) hep-lat/0406009; hep-ph/0604138 Csernai et al, PRL97, 152303(06)

The universal tendency of flow to be dissipated due to the fluid’s internal friction results from a quantity known as the shear viscosity. All fluids have non-zero viscosity. The larger the viscosity, the more rapidly small disturbances are damped away.

Quantum limit: /sAdS/CFT ~ 1/4

pQCD limit: ~ 1

At RHIC: ideal (/s = 0) hydrodynamic model calculations fit to data

Perfect Fluid at RHIC?!

H2O N2

He

hadronicpartonic

Viscosity and the Perfect Fluid

41Jim Thomas - LBL

STAR Upgrades to keep the discoveries rolling …

• Forward Meson Spectrometer– Gluon density distributions, saturation effects, and transverse spin

• DAQ Upgrade– order of magnitude increase in rate

– extra bandwidth opens the door to ‘small’ physics

• Full Barrel MRPC TOF– extended particle identification at intermediate pT

• Forward GEM Tracker

– end cap tracker for W sign determination

• Heavy Flavor Tracker

– high precision Heavy Flavor Tracker near the vertex

– opens the door to direct topological ID of Charm & Beauty

• Muon Telescope

• Forward Reaction Plane Detector

• A Crystal Calorimeter for low E photons - HBT

In t

he

qu

eue

En

gin

es r

un

nin

gW

arm

ing

Up

42Jim Thomas - LBL

Opening New Frontiers at RHIC

• (Hint … Segue to new hardware)

• Direct topological reconstruction of open charm (explain why)

• The prevailing hypothesis is that we have partonic collectivity at RHIC

• Is there collective behavior in the charm sector, too?– If so, it would be extraordinary because charm is so heavy

• Beauty is accessible but difficult

• What is the challenge … with the hardware

43Jim Thomas - LBL

The Properties of the Open Charm Hadrons

Particle Decay Channel c (m) Mass (GeV/c2)

D0 K + (3.8%) 123 1.8645

D+ K + + (9.5%) 312 1.8694

K+ K + (5.2%)+ + - (1.2%)

150 1.9683

p K + (5.0%) 59.9 2.2865

+

SD

+

C

44Jim Thomas - LBL

Direct Topological Identification of Open Charm

The STAR Inner Tracking Upgrades will identify the daughters in the decay and do a direct topological

reconstruction of the open charm hadrons.

No ambiguities between charm and beauty.

Goal: Distinguish secondary from primary vertices by putting a high precision detector near the IP to extend the TPC tracks to small radius

50-150 m

45Jim Thomas - LBL

The Heavy Flavor Tracker = PXL + IST + SSD

• A new detector

– 30 m silicon pixelsto yield 10 m space point resolution

• Direct Topological reconstruction of Charm

– Detect charm decays with small c, including D0 K

• New physics

– Charm collectivity and flow to test thermalization at RHIC

– Charm Energy Loss to test pQCD in a hot and dense medium at RHIC

• The SSD … is part of the plan for tracking TPC HFT

• The technical design is evolving but converging rapidly to final form.

PXL: 2 layers of Si at small radii

IST: 1 layer of Si at intermediate radius

SSD: an existing detector at 23 cm radius

46Jim Thomas - LBL

Active Pixel Sensors from IPHC Strasbourg

• 30 m pixels

• 640 x 640 pixels per chip

• 10 chips per ladder

• 100 – 200 sec integration time

• 164 M pixels, total

• 300 kRad hardness (~1 yr under certain conditions)

Properties:

• Signal created in low-doped epitaxial layer (typically ~10-15 μm)

• Sensor and signal processing integrated in the same silicon wafer

• Standard commercial CMOS technology

Inner layer

Outer layer

20 cm

47Jim Thomas - LBL

Getting a Boost from the TPC

• The TPC provides good but not excellent resolution at the vertex and at other intermediate radii

~ 1 mm

• The TPC provides an excellent angular constraint on the path of a predicted track segment

– This is very powerful.

– It gives a parallel beam with the addition of MCS from the IFC

• The best thing we can do is to put a pin-hole in front of the parallel beam track from the TPC

– This is the goal for the Si trackers: SSD, IST, and PXL

• The SSD and IST do not need extreme resolution. Instead, the goal is to maintain the parallel beam and not let it spread out

– MCS limited

– The PXL does the rest of the work

TPC

MCS Cone

VTX

The Gift of the TPC

OFC

IFC

48Jim Thomas - LBL

Optimized HFT Configuration

The HFT configuration described in the Addendum

49Jim Thomas - LBL

Calculating the Performance of the Detector

• Billoir invented a matrix method for evaluating the performance of a detector system including MCS and dE/dx

– NIM 225 (1984) 352.

• The ‘Information Matrices’ used by Billoir are the inverse of the more commonly used covariance matrices

– thus, ’s are propagated through the system

• ITTF tracking software uses a similar method (aka a Kalman Filter)– The ‘hand calculations’ go outside-in

– STAR Software goes outside-in and then inside-out, and averages the results, plus follows trees of candidate tracks. It is ‘smart’ software.

MCS D M MCS D M MCS

50Jim Thomas - LBL

Hand Calculations .vs. GEANT & ITTF

- - - - PXL stand alone configurationPaper Proposal configuration

GEANT & ITTF Updated configuration … no significant changes in pointing at VTX

TPC alone

Full System

51Jim Thomas - LBL

MinBias Pileup – The PXL Layers Integrate over Time

A full study of the integrated hit loading on the PIXEL detector includes the associated pileup due to minBias Au-Au collisions and the integration time of the detector.

2022

00

1 1( , , , )

2 ( )2

za

a

dN dN dMinBias z r ZDC e dz

dA d r d z z

PIXEL-1 Inner Layer

PIXEL-2 Outer Layer

Radius 2.5 cm 8.0 cm Central collision hit density 17.8 cm-2 1.7 cm-2

Integrated MinBias collisions (pileup) 23.5 cm-2 4.2 cm-2

UPC electrons 19.9 cm-2 0.1 cm-2

Totals 61.2 cm-2 6.0 cm-2

Pileup is the bigger challenge

Integrate over time and interaction diamond

200 sec

2022

02 20

2720 1 1( , , , )

2 2 ( )

za

a

dNMinBias z r e dz

dA r r z z

52Jim Thomas - LBL

Efficiency Calculations in a high hit density environment

The probability of associating the right hit with the right track on the first pass through the reconstruction code is:

P(good association) = 1 / (1+S)

where S = 2 x y

P(bad association) = (1 – Efficiency) = S / ( 1 + S )

and when S is small

P(bad association) 2 x y

x is the convolution of the detector resolution and the projected track error in the ‘x’ direction, and is the density of hits.

The largest errors dominates the sum

x = ( 2xp + 2

xd )

y = ( 2yp + 2

yd )

Asymmetric pointing resolutions are very inefficient … try to avoid it

An areaA density, depends on and pileup

53Jim Thomas - LBL

The performance of the TPC pointing at the PXL

• The performance of the TPC acting alone to point at the PXL detector depends on the integration time of the PXL chips

P(good association) = 1 / (1+S) where S = 2 x y

Integration Time (sec)

Sin

gle

Lay

er E

ffic

ien

cy

The purpose of intermediate tracking layers is to make 55% go up to ~100%

depends on

2.5 cm

8.0 cm

54Jim Thomas - LBL

The performance of the TPC + SSD + PXL

• The performance of the TPC + HFT acting together depends on the integration time of the PXL chip … but overall the performance is very good

P(good association) = 1 / (1+S) where S = 2 x y

Integration Time (sec)

Sin

gle

Lay

er E

ffic

ien

cy

Random errors only included in hand calculations and in GEANT/ITTF simulations

Note that systematic errors are not included in the hand calculations nor in the GEANT Simulations

55Jim Thomas - LBL

The challenge is to find tracks in a high density environment with high efficiency because a D0 needs single track 2

Raw HFT Tracking Efficiency: 0.98 x 0.98 x 0.93 x 0.94 = 0.84

Geometric acceptance and TPC track finding efficiencies 0.9 x 0.9 x 0.8 = 0.65 In this example Tot = 0.55

– Goal: graded resolution and high efficiency from the outside in

– TPC – SSD – IST – PXL

– TPC pointing resolution at the SSD ~ 1 mm = 0.98

– SSD pointing at the IST is ~ 400 m = 0.98

– IST pointing at PXL 2 is ~ 400 m = 0.93

– PXL 2 pointing at PXL1 is ~ 125 m = 0.94

– PXL1 pointing at the VTX is ~ 40 m

The performance of the TPC + SSD + IST + PXL

~ 50 cm

56Jim Thomas - LBL

D0 reconstruction

θ

MD0 = 1.8645 GeV/c2 c = 123 m

V0(D0)

dca K

0.5 < pT < 1.5 GeV/c

57Jim Thomas - LBL

D0 expected invariant mass distributions

pT

pT distributions for (S,B) at high p

T are from power-law guess and Hijing, respectively.

D0 Background slope at high pT could be uncertain due to limited statistics in MC.

For 100 M Au+Au central collisions at 1x RHIC II luminosity

38BS

S

22 BS

S

69BS

S

58Jim Thomas - LBL

• The predicted absolute efficiency of the HFT detector. – The red squares show the efficiency for finding the D0 meson with the full set

of Geant/ITTF techniques. The black circles show the efficiency AFTER cuts.

• The tracking efficiency is improved by 20-30% compared to the simulation in the proposal. Mostly due to improved hit selection in PXL.

D0 Reconstruction Efficiency

Geant/ITTF

59Jim Thomas - LBL

Improved understanding of signal / background

Updated: S: D0 yield dN/dy = 2 Loosen the # of PIXEL hits selection D0 background in more real estimation

Assume perfect PID at pT<1.5 GeV and no PID at p

T>1.5 GeV/c

100 M Au+Au central @ 200 GeV

| | < 1, D0+D0_

60Jim Thomas - LBL

Estimate v2 sensitivity – focus on the error bars

From central to minimum bias, assume: D0 scaled by N

bin Hijing background scaled by N

part

61Jim Thomas - LBL

Estimate Rcp sensitivity: focus on the error bars

62Jim Thomas - LBL

The next level of difficulty: Charm baryon - c

M = 2.286 GeV/c2 c = 60 m

pT

63Jim Thomas - LBL

Nuclear Matter at RHIC … a perfect liquid … an sQGP

• Its hot– Chemical freeze out at 175 MeV & Thermal freeze out at 100 MeV

• Its fast– Transverse expansion with an average velocity greater than 0.55 c

– Large amounts of anisotropic flow (v2) suggest hydrodynamic expansion and high pressure at early times in the collision history

• Its opaque and strongly interacting– Saturation of v2 at high pT

– RAA … suppression of high pT particle yields relative to p-p

– Jet Quenching … suppression of the away side jet

• It has partonic degrees of freedom– Constituent quark scaling of v2 and v4 for u, d, s and perhaps charm

• There are hints that it is thermally equilibrated– Excellent fits to particle ratio data with equilibrium thermal models

• And it has nearly zero viscosity and perhaps a Mach cone– Perhaps it is at or below the quantum bound from the AdS/CFT conjecture

64Jim Thomas - LBL

Backup Slides

65Jim Thomas - LBL

TPC Pointing at the PXL Detector

• The TPC pointing resolution on the outer surface of the PXL Detector is greater than 1 mm … but lets calculate what the TPC can do alone

– Assume the new radial location at 8.0 cm for PXL-2, with 9 m detector resolution in each pixel layer and a 200 sec detector

– Notice that the pointing resolution on PXL-1 is very good even though the TPC pointing resolution on PXL-2 is not so good

• The probability of a good hit association on the first pass– 55% on PXL2

– 95% on PXL1

Radius PointResOn(R-)

PointResOn(Z)

Hit Density(cm-2)

8.0 cm 1.4 mm 1.5 mm 6.0

2.5 cm 90 m 110 m 61.5

This is surprising: The hard work gets done at 8 cm!

The purpose of the intermediate tracking layers is to make 55% go up to ~100%

All values quoted for mid-rapidity Kaons at 750 MeV/c

66Jim Thomas - LBL

D0 Decay Kinematics

PT of the Kaon (GeV/c)

PT o

f th

e P

ion

(G

eV/c

)

• D0’s thrown by Pythia for p-p collisions

• D0 pT shown by different color dots (e.g. Blue = 1.3 GeV D0s)

67Jim Thomas - LBL

• pT distributions of electrons from semi-leptonic decay of heavy flavor mesons (left D-mesons, right B-mesons) as a function of parent pT. The inserted plots represent the projections to the corresponding heavy flavor distributions. The widths of the electron pT windows are indicated by dashed boxes.

68Jim Thomas - LBL

3<pt,trigger<4 GeV

pt,assoc.>2 GeV

Au+Au 0-10%

preliminary

69Jim Thomas - LBL

The Development of a Weibal Instability

70Jim Thomas - LBL

0

0 . 2

0 . 4

0 . 6

0 . 8

1

1 . 2

1 . 4

1 . 6

-6 -4 -2 0 2 4 6y

dy

dn

Nomenclature: Rapidity vs xf

• xf = pz / pmax

– A natural variable to describe physics at forward scattering angles

• Rapidity is different. It is a measure of velocity but it stretches the region around v = c to avoid the relativistic scrunch

– Rapidity is relativistically invariant and cross-sections are invariant

)/(tanh 1 Epyor z

Rapidity and pT are the natural kinematic variable for HI collisions( y is approximately the lab angle … where y = 0 at 90 degrees )

When the mass of the particle is unknown, then y

1tanh yy

z

z

pE

pEy ln

2

1

β

71Jim Thomas - LBL

V1: Pions go opposite to Neutrons

• hi

62 GeV Data

At low energy, the pions go in the opposite direction to the ‘classical’ bounce of the spectator baryons

200 GeV Data

At the top RHIC energy, the pions don’t flow(v1 at =0 )but at ALICE, v1 may have a backward wiggle.Reveals the EOS

72Jim Thomas - LBL

Interpreting Flow – order by order

n=1: Directed Flow has a period of 2 (only one maximum)

– v1 measures whether the flow goes to the left or right – whether the momentum goes with or against a billiard ball like bounce off the collision zone

n=2: Elliptic flow has a period of (two maximums)

– v2 represents the elliptical shape of the momentum distribution

))2cos(2)cos(21(2

121

2

3

3

vvdydpp

Nd

pd

dNE

TT

directed elliptic isotropic

73Jim Thomas - LBL

TPC Gas Volume & Electrostatic Field Cage

• Gas: P10 ( Ar-CH4 90%-10% ) @ 1 atm

• Voltage : - 28 kV at the central membrane 135 V/cm over 210 cm drift path

420 CM

Self supporting Inner Field Cage: Al on Kapton using Nomex honeycomb; 0.5% rad length

74Jim Thomas - LBL

Cold nuclear matter:0 ~ 0.16 GeV/fm3

30x0

nucl-ex/0311017

PRL 87 (01) 52301

R2

dydz 0

Boost invariant hydrodynamics: Bjorken Estimate of Initial Energy Density

d

dNp

Rdy

dE

Rch

TT

2

31122

~ 0.2 - 1 fm/ctime to thermalize the

system

~ 6.5 fm

Bjorken Estimate of Initial Energy Density

3/5.4 fmGeV3/7.0 fmGeVC

75Jim Thomas - LBL

This gap is a manifestation of the approximate SU(2)R x SU(2)L chiral symmetry of QCD with pions as the Nambu-Goldstone bosons

QCD is a Rich Theory with Many Features

Hadron “level” Diagram

Hadron 'level' diagram

0

500

1000

1500

0 10 20 30 40

Degeneracy

Mass (MeV)

Kfo

{W. Zajc

MFFDiL a

aˆ~

4

1We have a theory of the strong

interaction

Low(er) energy nuclear physics uses OPEP or descriptions in terms of a pion gas. These worked because QCD is a theory with a mass gap.

76Jim Thomas - LBL

Something Funny Happens at T > mc2

An exponentially increasing density of hadronic states suggests– A “limiting temperature” TH

– A phase transition(?) in hadronic matter

This was noticed before quarks were identified as the constituents of matter– ( Hagedorn, Nuovo Cimento Supp., 3 (147) 1965 )

Fit this form with TH = 163 MeVDensity of States .vs. Energy

HTmaemdm

dnm /~)(

dmem

dmem

TTm

a

Tm

H

)11

(

/

~

)(

Which requires T < TH

Thermal equilibrium suggests

77Jim Thomas - LBL

Charm Cross Sections at RHIC

1) Large systematic uncertainties in the measurements2) Theory under predict by a factor ~ 2 and STAR ~ 2 x PHENIX3) Directly reconstructed charm hadrons Upgrades

78Jim Thomas - LBL

Au on Au Event at CM Energy ~ 130 GeV*A

Real-time track reconstruction

Pictures from Level 3 online display. ( < 70 mSec )

Data taken June 25, 2000.

The first 12 events were captured on tape!

79Jim Thomas - LBL

Identified Mesons and Baryons: Au+Au @ 200 GeV

and p yields .vs. pTPhys. Rev. Lett. 97 (2006) 152301

80Jim Thomas - LBL

The Large Detectors – PHENIX and STAR

STAR PHENIX

81Jim Thomas - LBL

Partonic energy loss via leading hadrons

Energy loss softening of fragmentation suppression of leading hadron yield

ddpdT

ddpNdpR

TNN

AA

TAA

TAA /

/)(

2

2

Binary collision scaling p+p reference

82Jim Thomas - LBL

Au+Au and p+p: inclusive charged hadrons

p+p reference spectrum measured at RHIC

PRL 89, 202301

83Jim Thomas - LBL

PHENIX data on the suppression of 0s

Factor ~5 suppression for central Au+Au collisions

lower energy Pb+Pb

lower energy

84Jim Thomas - LBL

Pixel & IST – optimizations and progress

2.5 cm radius

8 cm radius

Inner layer

Outer layer

End view ALICE style carbon

support beams (green)

See talks by

HH Wieman B Surrow

• One IST layer at 14 cm• Good performance• Utilizes the existing SSD

• Fewer channels• Lower cost• Extra space for PXL layers

• Basic Parameters– Short strips ( < 1 cm )

– Wide strips ( ~ 500 m )

– Approx 150 m x 2000 m resolution

The proposed changes and optimizations have been verified with hand calculations and are scheduled to be put thru a full system test with GEANT/ITTF simulations.

85Jim Thomas - LBL

Central Collisions: Density of hits on the Detectors

dN dN d

dz d dz

2 2

1( , )

dr z

dz r z

where

22

1 700( ) 17.8

2 2

dN dNCentral cm

dA dz r r

Au+Au Luminosity (RHIC-II) 80 x 1026 cm-2s-1

dn/d (Central) 700 dn/d (MinBias) 170 MinBias cross section 10 barns MinBias collision rate (RHIC-II) 80 kHz Interaction diamond size, σ 15 cm Integration time for Pixel Chips 200 sec

Radius Simple Formulas

HIJING thru GEANT

PXL 1 2.5 cm 17.8 cm-2 19.0 cm-2

PXL 2 8.0 cm 1.7 cm-2 1.8 cm-2

IST 14.0 cm 0.57 cm-2 0.66 cm-2

SSD 23.0 cm 0.21 cm-2 0.23 cm-2

The density of hits is not large compared to the number of pixels on each layer. The challenge, instead, is for tracking to find the good hits in this dense

environment.

Slightly conservative numbers

100,000 pixels cm-2

86Jim Thomas - LBL

Monte Carlo Simulation Strategy & Updates

Central Au-Au events from Hijing

Geometry definition in GEANT

Detector response simulation

Digitization to raw hits

STAR ITTF reconstruction chain

User's analysis code

Real data from DAQ

Association between rec and MC

PIXEL hits pileup

D0 Measurements: dN/dy per NN collision ~ 0.004 (STAR) we take half of this as our estimate of the rate

# Hits selection in PIXEL: MC hits and Rec hits can be > 2 we include these tracks

D0 Background: K from D decays and from other decays -- important at high p

T . D0 -> K- + X (53%)

PID with TOF: Assume perfect K/ at pT< 1.5 GeV/c, no PID for K/ beyond

that. Background also includes PID contamination.

87Jim Thomas - LBL

Geometry definition in MC and event sample

Hit position in silicon layers from MC

Segment sizes and resolutions

Central (b = 0-3 fm) Au-Au Hijing + 10 D0 per event (flat pT, eta)

|Vertex_z| < 5 cm BR=100%

9x99x9

88Jim Thomas - LBL

D0 S,B evolution with different pileup levels

89Jim Thomas - LBL

Pixel Prototype – Geant .vs. Hand CalculationsGEANT

• Three arm PXL prototype configuration (early deployment / engineering test)• Good acceptance around the expected mean pT of D0 ’s (i.e. ~1 GeV)• Ideal to measure charm cross-section via direct topological reconstruction

Hand Calculations

pT of the D0 (MeV/c)pT of the D0 (GeV/c)A

cc

ep

tan

ce

(f

or

D0s

th

row

n in

to t

he

TP

C

acce

pta

nce

)

90Jim Thomas - LBL

The Simplest ‘Simulation’ – basic performance check

• Study the last two layers of the system with basic telescope equations with MCS

– PXL 1 and PXL 2 alone ( no beam pipe )

– Give them 9 m resolution

0

( / )13.6mcs

MeV c x

p X

2 22 2 2 22 11 2 2 1

2 2

2 1sin ( )

mcs rr r

r r

hh

• In the critical region for Kaons from D0 decay, 750 MeV to 1 GeV, the PXL single track pointing resolution is predicted to be 20-30 m … which is sufficient to pick out a D0 with c = 123 m

• The system (and especially the PXL detector) is operating at the MCS limit

• In principle, the full detector can be analyzed 2 layers at a time …

TPC alone

PXL alone

91Jim Thomas - LBL

HFT Performance & Simulation Studies

A marriage of Intuition, Hand Calculations, and Detailed Geant Simulations

Jim ThomasLawrence Berkeley National Laboratory

February 25-26, 2008

92Jim Thomas - LBL

Graded Resolution from the Outside In

• A PXL detector requires external tracking to be a success– The TPC and intermediate tracking provide graded resolution from the

outside-in• The intermediate layers form the elements of a ‘hit finder’

– The spatial resolution is provided by the PXL layers• The next step is to ensure that the hit finding can be done

efficiently at every layer in a high hit density environment

TPCvtx

PXL alone

TPCSSD

SSDIST

ISTPXL2

PXL2PXL1

PXL1VTX

93Jim Thomas - LBL

Ghosting increases as pileup increases

Pile-up level: 1x RHIC II luminosity

94Jim Thomas - LBL

Secondary Vertex Resolution

cm

c

Co

un

ts

Left figure, observed decay length (including realistic pT weighting) Right figure, D0 decay length scaled by a factor of 1/ No beamline constraint required … In central AuAu collisions, the D0 secondary vertices are clearly

separated from the primary vertex

cm

95Jim Thomas - LBL

The Phase Diagram for Nuclear Matter

The goal is to explore nuclear matter under extreme conditions – T > mc2 , > 10 * 0 and net 0

• The goal at RHIC is to understand the QCD in the context of the many body problem

• Another goal is to discover and characterize the Quark Gluon Plasma

• RHIC is a place where fundamental theory and experiment can meet after many years of being apart

96Jim Thomas - LBL

Chemical and Kinetic Freeze-out

• Chemical freeze-out (first)– End of inelastic interactions

– Number of each particle species is frozen

• Useful data– Particle ratios

• Kinetic freeze-out (later)– End of elastic interactions

– Particle momenta are frozen

• Useful data– Transverse momentum distributions

– and Effective temperatures

space

tim

e

inelasticinteractions

Chemicalfreeze-out

elasticinteractions

Kineticfreeze-out

blue beam yellow beam

97Jim Thomas - LBL

Chemical Freeze-out – from a thermal model

• The model assumes a thermally and chemically equilibrated fireball at hadro-chemical freeze-out which is described by a temperature T and (baryon) chemical potential : dn ~ e-(E-)/T d3p

• Works great, but there is not a word of QCD in the analysis. Done entirely in a color neutral Hadronic basis!

Thermal model fits

Compare to QCD on the (old) Lattice:

Tc = 154 ± 8 MeV (Nf=3)

Tc = 173 ± 8 MeV (Nf=2)(ref. Karsch, various)

MeV 629(RHIC)μ

MeV 7177(RHIC)T

B

ch

MeV 270(SPS)μ

MeV 170160(SPS)T

B

ch

input: measured particle ratios output: temperature T and baryo-chemical potential B

98Jim Thomas - LBL

• Final-state analysis suggests RHIC reaches the phase boundary

• Hadron spectra cannot probe higher temperatures

• Hadron resonance ideal gas (M. Kaneta and N. Xu,

nucl-ex/0104021 & QM02)

– TCH = 175 ± 10 MeV

– B = 40 ± 10 MeV

• <E>/N ~ 1 GeV(J. Cleymans and K. Redlich, PRL 81, p. 5284, 1998 )

Lattice results

Neutron STAR

Putting RHIC on the Phase Diagram

We know where we are on the phase diagram but eventually

we want to know what other features are on the diagram

99Jim Thomas - LBL

-meson Flow: Partonic Flow

-mesons are special: - they show strong collective flow and - they are formed by coalescence of thermalized s-quarks ‘They are made via coalescence of seemingly thermalized quarks in central Au+Au collisions, the observations imply hot and dense matter with partonic collectivity has been formed at RHIC’

Phys. Rev. Lett. 99 (2007) 112301 and Phys. Lett. B612 (2005) 81

100Jim Thomas - LBL

The Suppression occurs in Au-Au but not d-Au

d+Au

Au+Au

No quenching

Quenching!

101Jim Thomas - LBL

Hand calculations assume the acceptance is flat in pT and assume a single track at = 0.5

Single Track Efficiencies & Ghosting

Au + Au central collisions @ 200 GeV

TPC tracking efficiency ~80-85%

Ghosting =

# of tracks with 2 PIXEL hits & either of 2 PIXEL hits is a wrong hit

# of track with 2 PIXEL hits

Hand Calculations Ghost Rate = 1 – Raw Tracking

Efficiency

102Jim Thomas - LBL

• The HFT is thin, unique, innovative and robust

• The design have been tested extensively with hand calculations and a few key examples have been simulated with GEANT/ITTF software

• Simulations … completed tasks A full Monte Carlo simulation + reconstruction chain with HFT in STAR Comprehensive study on the pointing resolution and single track

efficiency for the STAR system with HFT with full MC simulations. Comprehensive study on the D0 reconstruction in Au+Au central

collisions, including realistic signal/background study. D0 reconstruction efficiency in Au+Au Quantify the pile-up effect on the single track efficiency (ghosting), D0

background and signal significance.

• To do Improved understanding of single track efficiency and ghosting at low pT

Optimization of D0 reconstruction at low pT – improving efficiency

Systematic study of other Charm hadrons, such as the c, and Bottom p+p 200/500 GeV simulations, pile-up effect and improved vertex finders

Summary: A rich physics program with the HFT