QUARKONIUM SUPPRESSION USING 3+1D ANISOTROPIC HYDRODYNAMICS
A dissertation submitted to Kent State University in partial fulfillment of the requirements for the degree of Doctor of Philosophy
by
Brandon Krouppa
August, 2018
c Copyright
All rights reserved
Except for previously published materials Dissertation written by
Brandon Krouppa
BS, Jacksonville University, 2013
MA, Kent State University, 2015
PhD, Kent State University, 2018
Approved by
, Chair, Doctoral Dissertation Committee Dr. Michael Strickland
, Members, Doctoral Dissertation Committee Dr. Spyridon Margetis
Dr. Declan Keane
Dr. Khandker Quader
Dr. Edgar Kooijman
Dr. Diane Stroup
Accepted by
, Chair, Department of Physics Dr. James T. Gleeson
, Dean, College of Arts and Sciences Dr. James L. Blank Table of Contents
Table of Contents ...... iii
List of Figures ...... vi
List of Tables ...... xiii
List of Publications ...... xv
Acknowledgments ...... xvi
1 Introduction ...... 1
1.1 Units and notation ...... 1
1.2 The standard model ...... 3
1.3 Quantum Electrodynamics (QED) ...... 4
1.4 Quantum Chromodynamics (QCD) ...... 6
1.5 The running coupling in QED and QCD ...... 7
1.6 Hadrons ...... 8
1.6.1 Heavy quarkonia ...... 10
1.7 The QCD phase diagram ...... 12
1.8 Heavy-ion collisions ...... 13
1.8.1 Stages of heavy-ion collisions ...... 14
1.8.2 Centrality ...... 15
1.8.3 A hydrodynamical description of the quark-gluon plasma ...... 18
1.9 Quarkonium suppression ...... 20
1.9.1 Other suppression effects ...... 23
iii 2 The heavy quark potential ...... 24
2.1 Heavy quarkonia ...... 24
2.2 A perturbative inspired heavy-quark potential (Strickland-Bazow) ...... 25
2.3 A lattice QCD vetted heavy-quark potential (Rothkopf) ...... 26
2.4 Non-equilibrium corrections to the lattice-vetted heavy quark potential . . . 34
2.5 Solving for binding energies ...... 35
2.5.1 Strickland-Bazow potential discretizations ...... 37
2.5.2 Rothkopf potential discretizations ...... 38
2.6 Discussion on difference between potentials ...... 38
3 3+1d anisotropic hydrodynamics ...... 40
3.1 Anisotropic hydrodynamics ...... 40
3.1.1 The anisotropic equation of state ...... 46
3.2 aHydro initial conditions ...... 47
3.2.1 Determining the initial temperature-like scales ...... 49
4 Computing the suppression factor RAA ...... 52
4.1 Computing RAA ...... 52
4.1.1 Regenerative effects ...... 56
5 Results for quarkonium suppression ...... 69
5.1 Charmonium ...... 69
5.2 Bottomonium ...... 70
6 Discussion & outlook ...... 79
6.1 Charmonium ...... 79
6.2 Bottomonium ...... 80
6.3 Outlook ...... 81
iv 7 Appendix ...... 82
7.1 Finite initial anisotropy parameter ...... 82
7.2 Adjustment of the formation time of heavy quarkonia ...... 83
7.3 Model sensitivity of the final cutoff temperature for survival probability inte-
gration ...... 84
Bibliography ...... 85
v List of Figures
1.1 The Standard Model as of June 2018...... 4
1.2 The three fundamental vertices found in QCD. On the left we see the three-
gluon vertex, in the middle we see the four-gluon vertex, and on the right we
see the quark-gluon vertex. Figure taken from Reference [1]...... 6
2 1.3 A summary of the QCD running coupling constant αs(Q ) from [2]...... 9
1.4 Various types of hadron examples and their classification...... 9
1.5 A plot of the pNRQCD-based heavy quark potential for three temperatures.
The real part seen in this plot starts with the vacuum Cornell potential at
T = 0 GeV, and as the temperature increases, the quarkonia are subject to
melting...... 11
1.6 The phase diagram, as currently understood, with various high-energy exper-
iments outlined to probe various parts of the diagram. Figure taken from
[3]...... 12
1.7 Stages of heavy-ion collisions. Figure taken from [4]...... 14
1.8 The transverse profile of two colliding nuclei. The overlap is measured by the
impact parameter, b, the distance between the center of the two nuclei. When
calculated with the Glauber model, the overlapping region can be quantified
by the number of partons participating in the collision, Npart...... 16
1.9 The number of participants, Npart, as a function of the impact parameter, b,
for various collision systems studied in this work...... 17
vi 1.10 Charge-hardon multiplicity as a function of pseudorapidity, compared to AL-
ICE 2.76 TeV/nucleon collisions. The percentage ranges given in the figure
are various centrality classes, with 0 5% relating to the 5% most central − collisions, and so on. Figure taken from [5], with ALICE data provided from
[6, 7]...... 19
1.11 As a function of the dimuon invariant mass, we see three resonances appear in
proton-proton reference data, and then disappear in Pb-Pb collisions at 5.02
TeV. This suppression of states is thought to be due to the formation of the
QGP. Figure taken from [8]...... 21
1.12 An estimate of shadowing (a cold nuclear matter effect) on Υ(1S) suppression
at 2.76 TeV/nucleon LHC collisions of Pb-208 nuclei...... 23
2.1 The in-medium heavy quark potential [real part (left) and imaginary part
(right)] in full QCD with Nf = 2 + 1 light quark flavors based on ensembles
by the HotQCD collaboration (colored points, shifted for better readability).
By adjusting the Debye mass mD, the lattice QCD values of the real part are
adequately reproduced with the Gauss-law parameterization as shown with
the solid lines, at all temperature and length scales. Theoretical error bars (the
shaded region around the solid lines) are due to the uncertainty of the Debye
mass fit. The imaginary part shows good agreement at high temperatures
and small distances, while at temperatures close to the crossover transition,
deviations from the lattice data are seen. Note that the crossover temperature
on these lattices due to the relatively large pion mass of mπ 300MeV lies ≈
at TC = 172.5 MeV...... 31
vii 2.2 The Re[V] (left) and Im[V] (right) part of the in-medium heavy quark poten-
tial which has been vetted against lattice QCD. The different lines correspond
to different values of the Debye mass of the medium. The vacuum parameters
αs, σ, and c at mD = 0 are tuned to reproduce the PDG charmonium and bot-
tomonium spectra. String breaking is enforced at rsb = 1.25 fm. Ultimately,
only one parameter governs the temperature-dependence which modifies the
vacuum (Cornell) potential. Such thermal effects lead to Debye screening
of the real part of the potential and induce a finite imaginary part which
asymptotes at large distances...... 34
2.3 Comparison between Strickland-Bazow Υ(1S) and Rothkopf Υ(1S) for the
cases of an isotropic, ξ = 0, plasma (left), and a slightly oblate, ξ = 1 plasma.
In the Rothkopf model we observe weaker binding and a slightly stronger
screening effect from the imaginary part of the binding energy. Here Tc = 192
MeV which then sets the scale of the temperature-like scales of the calculation.
The discontinuities seen in Im[V] are due to the real part going negative. . . 39
3.1 Plotted with the dimensionless time variablew ¯ = τ = τT , we see varying τR 5η/s
L/ T initial conditions converge rather quickly to an attractor. Figure taken P P from Reference [9]...... 41
3.2 Shear tensor corrections calculated using second-order viscous hydrodynamics
Reference [10]...... 42
3.3 From the Glauber model, we calculate the transverse energy density profile
for a given impact parameter, b = 4.5...... 48
viii 3.4 Using a boost-invariant plateau with Gaussian tails, we can phenomenolog-
ically model the longitudinal profile for the initial energy density produced
in an ultra-relativistic heavy-ion collision for RHIC 200 GeV/nucleon Au-Au
collisions and LHC 2.76 TeV/nucleon and 5.02 TeV/nucleon Pb-Pb collisions. 50
4.1 Heavy-quarkonium production at the CMS and ATLAS experiments [11]. . . 53
4.2 cc¯ production across multiple collision energies and experiments [12]. . . . . 60
4.3 Using this plot, we can derive the factors going from 1.6 < y < 2.4 to y < | | | | 1.6 (x1.5), and 2.5 < y < 4.0 to y < 1.6 (x2). Linear interpolation was used | | | | to fill in the gaps from mid-rapidity to ultra-forward rapidity 2.5 < y < 4.0 | | [13]...... 61
4.4 J/ψ production in 5.02 TeV p-p collisions at CMS [14]...... 62
dσ 4.5 cc¯ dy production across two energies and a range of forward rapidities. With interpolation to mid-rapidity, this figure is used to derive the multiplicative
factors in going from 1.6 < y < 2.4 to y < 1.6 (x1.5), and 2.5 < y < 4.0 | | | | | | to y < 1.6 (x2) [15]...... 63 | |
4.6 χcJ production cross sections as a function of transverse momentum. Note
+ − that B = (χcJ J/ψγ) (J/ψ µ µ ) [16]...... 64 B → ·B → 4.7 Differential cross sections for the Υ(nS) states [17]...... 65
4.8 Differential cross sections for the Υ(nS) states [18]...... 66
4.9 Production cross sections for χb(mP ) states given as a ratio to Υ(1S) produc-
tion cross sections [19]...... 67
5.1 Raw suppression for the J/ψ, ψ(2S), and χc(1P ) charmonium states for LHC
2.76 TeV/nucleon Pb-Pb collisions. For this figure the states are modeled
with the Strickland-Bazow potential. The lack of suppression for the ψ(2S)
state is due to the large formation time...... 70
ix 5.2 Inclusive RAA calculation for the J/ψ state in 2.76 TeV/nucleon Pb-Pb colli-
sions at the LHC, as calculated by the Strickland-Bazow potential...... 71
5.3 Raw suppression for Υ(nS) and χb(mP ) states from LHC 2.76 TeV/nucleon
Pb-Pb collisions. With both potentials we see a sequential ordering of suppres-
sion and smooth transitions in RAA from peripheral to central collisions. The
order of sequential suppression is the same in both potentials, except for the
pileup of largely-suppressed states, which is more evident with the Rothkopf
potential. This pileup is due to a “halo” survival effect for bottomonia formed
near the edge of the QGP...... 72
5.4 Bottomonium suppression for RHIC 200 GeV/nucleon collisions as seen by
the STAR collaboration. The Strickland-Bazow potential features an error
band which corresponds to varying values of η/s while the Rothkopf potential
features a band corresponding to uncertainty in the Debye mass calculation
as a function of temperature. STAR data taken from [20]...... 73
5.5 Strickland-Bazow potential calculation (left), compared to the Rothkopf po-
tential calculation (right). Comparisons between our model calculations and
bottomonium suppression seen at LHC 2.76 TeV/nucleon Pb-Pb collisions as
viewed by the CMS collaboration. CMS data taken from [21]...... 74
5.6 Strickland-Bazow potential calculation (left), compared to the Rothkopf po-
tential calculation (right). In LHC 2.76 TeV/nucleon Pb-Pb collisions, we
make a comparison between model calculations and bottomonium suppres-
sion data gathered from the ALICE collaboration. ALICE data taken from
[22]...... 75
x 5.7 Strickland-Bazow potential calculation (left), compared to the Rothkopf po-
tential calculation (right). In this figure, we can state within reason, there
is no significant rapidity dependence for Υ(1S) suppression for LHC 2.76
TeV/nucleon collisions. With some tension from forward-rapidity ALICE
data, the more strongly suppressed Rothkopf Υ(1S) provides a reasonable
description of suppression. CMS and ALICE data taken from [21, 22]. . . . . 76
5.8 Strickland-Bazow potential calculation (left), compared to the Rothkopf po-
tential calculation (right). For transverse momentum comparisons, we see no
significant pT dependence in the data while the model calculations show a
decreasing suppression for larger pT due to larger formation times effectively
shielding formed bottomonia from thermal suppression. CMS data taken from
[21]...... 76
5.9 Strickland-Bazow potential calculation (left), compared to the Rothkopf po-
tential calculation (right). Predictions made with the Strickland-Bazow po-
tential agree quite well with CMS data at 5.02 TeV/nucleon Npart data for
all observed Υ(nS) states. Not seen is the Υ(3S) confidence intervals for a
completely suppressed Υ(3S). CMS data taken from [8]...... 77
5.10 Strickland-Bazow potential calculation (left), compared to the Rothkopf po-
tential calculation (right). For 5.02 TeV/nucleon collisions, we see the Strickland-
Bazow potential calculation successfully predicts ALICE data. The Rothkopf
potential calculation slightly overestimates suppression. ALICE data taken
from [23]...... 77
xi 5.11 The Strickland-Bazow potential calculation (left) predicted CMS and ALICE
rapidity data successfully. The Rothkopf potential calculation (right) slightly
overestimates Υ(1S) suppression, but explains Υ(2S) and Υ(3S) suppression
successfully. Note that confidence levels for a fully suppressed Υ(3S) state
are not shown here. CMS and ALICE data taken from [8, 23]...... 78
5.12 There is a slight increase in Υ(1S) experimental data as well as the Strickland-
Bazow potential model calculation (left). The Rothkopf potential model calcu-
lation describes Υ(1S) for low-pT , but overall seems to overestimate suppres-
sion for the Υ(1S). Both potential calculations describe Υ(2S) suppression
for this given collision system. CMS data taken from [8]...... 78
7.1 We make an adjustment to the initial geometry of the collision by deforming
the initial momentum-space distribution to an oblate spheroidal distribution. 82
7.2 As we adjust the formation time of the bottomonium states, the inclusive RAA
is affected by 10%. Adjusting this quantity measures the “shielding” effect ∼ the formation time has for heavy quarkonia...... 83
7.3 The survival probability is cut off at a given temperature, Tf . As we adjust
Tf , we see a small effect on final inclusive Υ(1S) RAA which becomes less
sensitive as the cutoff temperature decreases...... 84
xii List of Tables
1.1 The relationship between impact parameter, b, and centrality class, c, includ-
ing the average impact parameter for each centrality class in a √sNN = 2.76
TeV/nucleon Pb-Pb collision modeled using optical Glauber. Table taken
from [24]...... 17
3.1 Λ0 values (in GeV) for various collision energies, ξ0, and η/s...... 51
4.1 Feed down fractions which have been averaged over pT , for the J/ψ and ψ(2S)
states used in the determination of the final measured yields [11]. Note that
the feed-down fractions for the ψ(2S) are estimates due to lack of measured
feed down data...... 55
4.2 Feed down fractions which have been averaged over pT , for the Υ(nS) states
used in the determination of the final measured yields [11]. Note that the
feed-down fractions for the Υ(3S) are estimates due to lack of measured feed
down data...... 55
4.3 cc¯ production cross sections calculated from Figure 4.2. The mid-rapidity
value for 200 GeV collisions was taken directly from [25]. The mid-rapidity
value for 2.76 TeV collisions was taken directly from [26]. The mid-rapidity
value for 5.02 TeV collisions was derived via interpolating data given in Figure
4.2. The interpolation yields a 40% increase from 2.76 TeV to 5.02 TeV. . . . 60
J/ψ 4.4 Fixed via a ratio of cc¯ = 0.6%, we compute the differential cross sections of J/ψ states...... 62
σ(ψ(2S)) 4.5 Fixed via a ratio of σ(J/ψ) = 0.17, we compute the cross sections of ψ(2S) states...... 63
4.6 Cross sections for the χc(1P ) state computed via χc/ψ(2S) = 1.7%...... 65
xiii dσ 4.7 Differential cross sections dy from [27] for the Υ(1S) and Υ(2S) states. Υ(3S) states are computed from [17] for 2.76 TeV and [18] for 5.02 TeV collisions.
For RHIC collision energy, we take an extrapolated value for the differential
cross section...... 66
4.8 Calculated cross sections from [27] for the Υ(1S) and Υ(2S) states. Υ(3S)
states are computed from [17] for 2.76 TeV and [18] for 5.02 TeV collisions.
For ALICE rapidity cuts, the stated rapidity coverage for dimuons (from [28])
is 2.5 < y < 4.0...... 67
4.9 Calculated cross sections for the χb(mP ) states...... 68
xiv List of Publications
1. B. Krouppa, R. Ryblewski and M. Strickland, “Bottomonia suppression in 2.76 TeV
Pb-Pb collisions,” Phys. Rev. C 92, no. 6, 061901 (2015) doi:10.1103/PhysRevC.92.061901
[arXiv:1507.03951 [hep-ph]].
2. B. Krouppa and M. Strickland, “Predictions for bottomonia suppression in 5.023
TeV Pb-Pb collisions,” Universe 2, no. 3, 16 (2016) doi:10.3390/universe2030016
[arXiv:1605.03561 [hep-ph]].
3. B. Krouppa, R. Ryblewski and M. Strickland, “Bottomonium suppression in heavy-
ion collisions,” Nucl. Phys. A 967, 604 (2017) doi:10.1016/j.nuclphysa.2017.05.073
[arXiv:1704.02361 [nucl-th]].
4. B. Krouppa, A. Rothkopf and M. Strickland, “Bottomonium suppression using a lattice
QCD vetted potential,” Phys. Rev. D 97, no. 1, 016017 (2018) doi:10.1103/PhysRevD.97.016017
[arXiv:1710.02319 [hep-ph]].
xv Acknowledgments
I am eternally grateful for the love and support I have received from those who made this achievement possible. There are too many stories of encouragement, prayer, listening, and sharing of wisdom, to include in this section, but I sincerely hope each of you know the impact you had on my life, and the fact that I could not have done this without you.
I would like to thank my advisor, Dr. Michael Strickland, for his guidance, support, and supervision during my doctoral research. I also extend a thank you to the collaborators who have made this work possible, including each person who pushed our understanding of physical laws governing the universe a little further. This work was made possible by the
Department of Energy Office of Science.
I would also like to thank my parents for their love and support in every endeavor I take on, whether professionally or personally.
Deo ac Veritati (For God and for Truth),
Brandon Krouppa
xvi Chapter 1
Introduction
In this chapter I will give a brief introduction to high-energy physics. The standard model is discussed along with an overview of elementary particles and the quantum physics which governs such particles. I will then introduce the quark-gluon plasma (QGP), and the hydrodynamics which effectively models ultrarelativistic heavy-ion collisions. I will end with the description of heavy quarks and mesons which contain a heavy quark-antiquark pair, i.e. heavy quarkonia, and general phenomenology of such mesons in the QGP.
1.1 Units and notation
In the following work we utilize a natural system of units which simplifies notation sig- nificantly. The main feature of these units provide simplification as c = ~ = kb = 1. The occasional translation of units uses a factor of ~c and is worth listing as such conversions are frequently used this work
~c = 0.1973269788(12) GeV fm. (1.1)
Timescales in this work are typically denoted by fm/c. We make a note of the following unit in SI units to provide an ordinary perspective of timescales seen in heavy-ion collisions
1 fm/c 3.33 10−24 s. (1.2) ≈ ×
1 Other conversions to SI units can be done in the following manner
1 MeV = 1.602177 10−13 J, (1.3) ×
1 MeV/c2 = 1.78266 10−30 kg, (1.4) ×
1 MeV = 1.16045 1010 K. (1.5) ×
Providing further insight, one can make conversions to ordinary SI units
The region where we can define a critical temperature for the creation of a quark-gluon • 12 plasma is T0 150 MeV, which is 1.74 10 K. ∼ ∼ ×
The mass of the proton is 938 MeV = 1.67 10−27 kg. • ∼ ×
The proper time the quark-gluon plasma evolves before hadronization is 10 fm/c • ∼ 3 10−23 s. ∼ ×
It takes 8.263 104 J 5.157 1017 MeV to heat a cup of water from room • ∼ × ∼ × temperature ( 294 K) to its standard pressure boiling point at 373.15 K. ∼
In this work, we use the particle-physics Minkowski-space metric, i.e. “mostly negative,”
µν µ g = diag(+1, 1, 1, 1). As such, the four-velocity is normalized with uµu = 1. Milne − − − coordinates in Minkowski space are defined by xµ = (τ, x, y, ς), which is sometimes seen as
µ 2 2 x = (τ, x⊥, ς), with x⊥ = (x, y). Longitudinal proper time in this space is τ = √t z , − spacetime rapidity ς = arctanh(z/t), with the full metric being ds2 = dτ 2 dx2 dy2 τ 2dς2. − − − In addition to the spacetime rapidity ς, we will make use of the pseudorapidity, which is defined as θ η ln tan , (1.6) ≡ − 2
2 where θ is the angle between the particle three-momentum p and the beam-line axis.1 As a function of three-momentum, one can write the pseudorapidity as
1 p + pL pL η = ln | | = arctanh , (1.7) 2 p pL p | | − | |
where pL is the longitudinal momentum, i.e. the momentum component along the beam-line
axis. One can also simplify things by using the ultrarelativistic limit where m p, such that E p. Finally, we can define rapidity as ≈
1 E + p p y = ln L = arctanh L (1.8) 2 E pL E −
where E is the energy of the particle, E = pm2 + p2. Given such a definition, one can
complete the set of equations in relation to other quantities such as E = mT cosh(y), pT = pT ,
p 2 2 and pL = mT sinh(y), where mT = pT + m is the transverse mass.
1.2 The standard model
The four fundamental forces which govern the standard model are gravity, the strong
force, the electromagnetic force, and finally the weak force. Below I will give a brief descrip-
tion of each fundamental force.
The theory of gravity is largely described by the curvature of the spacetime “mesh” that
makes up the whole universe. While the macroscopic principles of the theory of gravitation
are successfully described by this, gravity remains unquantized and is not properly under-
stood in conjunction with the other three fundamental forces. The weak force is responsible
for interactions in nuclear matter, including the description of radiative decay of unstable
nuclei. The carriers of this force are the W and Z bosons. The electromagnetic force is
carried by the massless photon and pertains to other radiative effects largely known as the
1Note that in ultrarelativistic heavy-ion collisions, backscattering is completely negligible.
3 Figure 1.1: The Standard Model as of June 2018. electromagnetic spectrum. Quantization of this force has led to the theory of Quantum Elec- trodynamics (QED) and is regarded as one of the more successful quantizations, mainly due
1 to the nature of the running coupling being small at low momentum transfer, i.e. αQED . ≈ 137 The strong force is carried by the gluon and the length scale of this force probes quarks,
fundamental fermionic particles in the standard model. The quantization of the strong force
is described in the theory of Quantum Chromodynamics (QCD).
1.3 Quantum Electrodynamics (QED)
As explained in Section 1.2, Quantum Electrodynamics is the fundamental quantization
of the electromagnetic field and the leptons that couple to it, e.g. electrons. The Lagrangian
4 for QED is made from two parts, the Dirac part which is given by
¯ Dirac = ψ(i∂/ m)ψ, (1.9) L −
where ψ is the Dirac spinor, ψ¯ is the conjugate of the Dirac spinor and is defined as ψ¯ ψ†γ0, ≡ µ and we have employed the Feynman slash notation, where in general, A/ γ Aµ. The ≡ electromagnetic portion of the QED Lagrangian is
1 µν µ Maxwell = FµνF AµJ , (1.10) L −4 −
where Fµν = ∂µAν ∂νAµ is the field strength tensor for the electromagnetic field, Aµ is the − vector potential, and J µ is the 4-current source term. To get the full QED Lagrangian, we
sum Equations (1.9) and (1.10), with J µ = eψγ¯ µψ −
¯ 1 µν ¯ QED = ψ(i∂/ m)ψ FµνF + eψAψ./ (1.11) L − − 4
We can further simplify by introducing the covariant derivative
Dµ ∂µ ieAµ. (1.12) ≡ −
With the covariant derivative notation, the QED Lagrangian can be written as
¯ 1 µν QED = ψ(iD/ m)ψ FµνF . (1.13) L − − 4
5 1.4 Quantum Chromodynamics (QCD)
The QCD Lagrangian is
Nf X 1 = ψ¯ (iD/ m )ψ F a F µν, (1.14) QCD i i i 4 µν a L i=1 − −
where the subscript i denotes flavors of quarks and thus Nf is the number of quark flavors
a with masses mi. We can define the field strength Fµν as
a a a abc b c F = ∂µA ∂νA + gf A A , (1.15) µν µ − µ µ ν where A are the QCD fields and f abc are the structure constants of the SU(3) group.
Figure 1.2: The three fundamental vertices found in QCD. On the left we see the three-gluon vertex, in the middle we see the four-gluon vertex, and on the right we see the quark-gluon vertex. Figure taken from Reference [1].
Two key aspects of the theory of QCD are confinement and asymptotic freedom. Confine- ment is the principle which keeps quarks bound together. As quarks separate with distance, ever increasing amounts of energy are required to further separate the pair from one another.
The second property of QCD, asymptotic freedom, is the opposite of confinement. Asymp- totic freedom describes the fact that the QCD coupling constant becomes weak at high energies. Quantum chromodynamics also introduces additional degrees of freedom known as
6 color charge, as the prefix “chromo” suggests.
The quark model was developed in 1964, independently, by Gell-Mann and Zweig. Ver- ification of the model came rapidly with multiple experiments, most notably with the ob- servation of the Ω− particle at Brookhaven National Laboratory (BNL) [29], and with deep inelastic scattering at the Stanford Linear Accelerator Center (SLAC). With those two ex- periments, physicists have known that nuclear matter, i.e. protons and neutrons, are not elementary particles. The constituents of protons and neutrons, now known as up and down quarks, make up just two of the six flavors of quarks which have been experimentally ob- served. Heavier flavors of quarks were theorized along with the quark model in the 1960s, and since then have been observed in a wide range of experiments. The quark model was expanded with the inclusion of a force carrier, the gluon, completing the theory of QCD largely as it stands today. As the present-day Standard Model stands, we have experimen- tally observed six quarks, six leptons, four gauge bosons, and one scalar boson. For this work we will primarily focus two of the six quarks seen in Figure 1.1, namely the charm and bottom quarks.
1.5 The running coupling in QED and QCD
Derived from first-principle calculations [30, 31], we know the one-loop running coupling in QED to be
α(Q0) α(Q) = 2 , (1.16) α(Q0) Q 1 log 2 − 3π Q0 and the one-loop running coupling in QCD is
αs(Q0) αs(Q) = 2 , (1.17) αs(Q0) 2Nf Q 1 + 11 log 2 4π − 3 Q0
7 where Q is the momentum transfer scale which is used in the renormalization to compute the one-loop coupling, and Q0 is a reference scale which acts as an initial condition for the running coupling. This initial condition is gathered from experiment where scatterings can measure the coupling, or theoretical calculations which utilize a first-principles approach such as lattice QCD. The difference between the QED and QCD running coupling is found in the denominator of Equations (1.16) and (1.17). For 11 2Nf /3 > 0, i.e. Nf < 33/2 = 16.5, − the sign is opposite from that of QED. While QED flourished with theoretical methods such as perturbation theory due to the small coupling constant at low momentum transfer, i.e. αQED 1/137, QCD only sees this advantage for large temperatures with ordinary ≈ nuclear matter being described with other methods. The coupling for QCD, αs, has been studied extensively and a summary of measurements and theoretical efforts to calculate the running coupling have led to Figure 1.7. In Figure 1.7, we see the trend we expect from the phenomenon of confinement, notably when the momentum transfer Q becomes large, the coupling constant in QCD becomes small. We also see strong agreement between experiment and the one-loop running coupling theory curve.
1.6 Hadrons
Under normal circumstances, quarks are confined inside bound states called hadrons.
Hadrons are bound by the strong force which is described by QCD. There are various classes of hadrons which correspond to how many quarks make up the bound state, and can be seen in Figure 1.4. Mesons are made from quark-antiquark pairs, qq¯ and are largely the study of this work. Baryons are a combination of three quarks and make up the bulk of “ordinary” nuclear matter we see today, i.e. protons and neutrons. Other exotic hadrons have been observed in experiment such as the tetraquark (made of four quarks) and the pentaquark
(five quarks). The tetraquark is currently discussed to either by a molecular particle, i.e. a bound set of mesons, or a true bound state where gluons are exchanged between a quark
8 April 2016 2 τ decays (N3LO) αs(Q ) DIS jets (NLO) Heavy Quarkonia (NLO) – 0.3 e+e jets & shapes (res. NNLO) e.w. precision fits (N3LO) (–) pp –> jets (NLO) pp –> tt (NNLO) 0.2
0.1 QCD αs(Mz) = 0.1181 ± 0.0011 1 10 100 1000 Q [GeV]
2 Figure 1.3: A summary of the QCD running coupling constant αs(Q ) from [2]. and the other three quarks in the state.
quark
Antiquark
Figure 1.4: Various types of hadron examples and their classification.
Perturbative approaches to describe hadrons are, in large part, unsuccessful due to regime not being valid. However, we’ll see a few special properties which allow us to use perturbative approaches to extract physics from mesons formed with heavy quarks.
9 1.6.1 Heavy quarkonia
Heavy quarkonia are bound states of heavy quarks with their antimatter heavy anti-quark
partner, so they are generally characterized as mesons. Heavy quarks are the three most
massive quarks, q = c, b, t , the charm quark, bottom quark, and top quark, respectively. { } The charm and bottom quark have been seen with their antimatter partners to form charmo-
nia and bottomonia, however the top quark has not been observed to form the hypothetical
particle toponia, due to the ultra-short lifetime of the top quark. Therefore modern physics
understanding suggests toponia may never be formed.
Heavy quarkonia were first detected in 1974 with an excess of dimuons corresponding
to a dimuon invariant mass of 3.1 GeV, now known as the J/ψ meson which is a bound ∼ state between a charm quark and an anti-charm quark. Lately, heavy quarkonia have been
produced and studied using many collision systems at RHIC and the LHC.
Heavy quarkonia are rather small states with state radii less than 1 fm. Moreover,
their relative velocity is small and can be described with an inherent separation of scales,
2 mq mqv prel mqv Ebind. This hierarchy of scales along with small bound state ∼ | | ∼ radii allow us to exploit the heavy quarkonia with potential-based non-relativistic QCD
(pNRQCD), especially since pNRQCD approaches are well-defined for r < 1/mD
It is now known that the heavy-quark pair potential is complex-valued, with the imaginary part of the potential stemming from the state’s proclivity to dissociate when the surrounding temperature dampens gluonic interactions, effectively causing a quark to no longer “see” its antimatter partner. However, we can get a basic understanding of thermal effects on heavy quarkonia by examining the real part of the potential.
In Figure 1.5, we see the real part of a potential extracted from potential-based non- relativistic QCD (pNRQCD) heavy-quark potential. At T = 0 GeV, we obtain the Cornell
10 ���
���
) ��� ��� ]( � [ ���
-��� � = ����� ��� � = ����� ��� � = ����� ��� -��� ��� ��� ��� ��� ��� ��� ��� � (��)
Figure 1.5: A plot of the pNRQCD-based heavy quark potential for three temperatures. The real part seen in this plot starts with the vacuum Cornell potential at T = 0 GeV, and as the temperature increases, the quarkonia are subject to melting.
potential which is generally given by
a V (r) = + br, (1.18) −r
where a and b are parameters related to the strong coupling constant and QCD string
tension, respectively. Its general features are a linear part for large r which corresponds to
QCD confinement, followed by the 1/r part, known as the Coulombic part, which is induced
by the one-gluon exchange between the quark-antiquark pair.
Theoretical and phenomenological models to predict heavy-quark production utilize a
spectrum of techniques from effective theories to pNRQCD approaches which lean more on
first principle approaches. Recent studies from lattice QCD suggest pNRQCD models give
accurate descriptions of heavy quarkonium.
11 In more recent studies, a lattice-vetted potential has been used to compute the effects of the QGP on heavy quarkonia. This was an effort to more properly treat the heavy-quark po- tential in a hot QGP medium. Our work resulted in comparisons between theoretical models and experimental data seen in RHIC and LHC collisions at √sNN = 200 GeV/nucleon, 2.76
TeV/nucleon, and 5.02 TeV/nucleon [32].
1.7 The QCD phase diagram
Figure 1.6: The phase diagram, as currently understood, with various high-energy experi- ments outlined to probe various parts of the diagram. Figure taken from [3].
As mentioned above, normally quarks are confined inside hadrons, however, if one heats or compresses a system of hadrons, one finds that there can be a transition to a deconfined state. This can be summarized in the phase diagram of QCD, as seen in Figure 1.6. In the QCD phase diagram, two dependencies are typically seen: temperature and the baryon
12 chemical potential, which is related to the net baryon density. The main discussion of the
QCD phase diagram lies within the transition between hadronic matter and the QGP. For larger net baryon density and relatively low temperatures, there’s believed to be a first order phase transition. However, as the net baryon density is lowered toward zero, there’s a smooth crossover between hadronic matter and the QGP. Between the first order phase transition and the crossover, there’s a possibility for a critical point (as common among other phase diagrams in physics), so the search for such a point is currently underway at the Relativistic Heavy-Ion Collider (RHIC) with their Beam Energy Scan program [33].
Lattice studies performed suggest this region might be marked with a critical temperature of Tc = 150 170 MeV [34, 35]. The focus of this work is to discuss a theory model − which describes an observable seen at two colliders which probe ultra-low net baryon density and high temperatures with heavy-ion collisions. This region of the QCD phase diagram is relevant to understanding the early universe in accordance with the big bang theory where, initially, the universe was hot and neutral with nearly equal numbers of baryons and anti- baryons. As the early universe cooled, matter started dominating over antimatter and is now in a region where T 0 and µB 1 GeV. ∼ ∼
1.8 Heavy-ion collisions
Heavy-ion collisions are used to study various physical principles and theories such as
QED, QCD, the early universe, and other fundamental particle interactions which are used to provide a more comprehensive picture of the physics governing the universe. Observables such as elliptic flow, charged-hadron multiplicity, heavy quarkonium suppression, etc. are used to describe the hot QGP generated in the short timescales, and each probe a different aspect of the QGP. Below I will give an overview of heavy-ion collisions and how we model them using a realistic hydrodynamics framework.
13 1.8.1 Stages of heavy-ion collisions
Freezeout time τ > 10 fm/c
Hot Hadron Gas 6 < < 10 fm/c time
Equilibrium QGP beam direction 2 < < 6 fm/c
Non-equilibrium QGP 0.3 < < 2 fm/c
Semi-hard particle production 0 < < 0.3 fm/c beam direction
Figure 1.7: Stages of heavy-ion collisions. Figure taken from [4].
Various stages of heavy-ion collisions can be defined by the dominant features during the
timescale.
Before impact. At this time, the heavy ions are traveling close to the speed of light in
opposite directions. In the lab frame the two nuclei are highly Lorentz contracted with
γ = 1/p1 v2/c2. Collisions at RHIC and at the Large Hadron Collider (LHC) at CERN − see γ-values which are approximated by γ √sNN /2, which lead to γ 100 for √sNN = 200 ' '
GeV Au-Au collisions. For larger collision energies at the LHC, we see for √sNN = 5020
GeV γ 2500. While the nuclei are Lorentz contracted in the lab frame, it is important ' to consider the time it takes for the nuclei to pass through one another. As an example,
for Pb-208 in LHC 5.02 TeV/nucleon collisions, we can use R = (1.12A1/3 0.86A−1/3) − to compute the radius which is R 6.49 fm, which will be the “longest” case for central ' collisions. The duration of the collision is R/γ = 13/2500 0.05 fm/c, and increases for ∼ ' lower collision energies.
14 Semi-hard particle production. In this stage, 0 < τ < 0.3 fm/c, semi-hard particles are
produced, including many of the heavy-quark pairs studied in this work. At this stage it is
required to accurately describe the non-equilibrium modes of the QGP.
Pre-equilibrium. Immediately following the nuclear impact, the system is characterized
by a strong momentum-space anisotropy given by a negative ratio L/ T , where L is the P P P
local rest frame pressure in the longitudinal (beam) direction, and T is the local rest frame P pressure in the transverse direction. This timescale is also marked by strong color fields and studied theoretically using color-glass condensate and glasma models [36, 37, 38, 39, 40, 41].
Equilibrium. This stage is described as being close to equilibrium and locally thermalized.
In this timescale, pressure gradients are small enough to provide a consistent study of the
QGP using anisotropic hydrodynamics, a hydrodynamical model which incorporates local pressure anisotropies in its 3+1d evolution.
Freeze-out. Simply put, this stage is marked by the transition from QGP to free stream- ing nuclear matter. Collective flow of particles and inelastic collisions are ceased, and the timescale between collisions becomes larger than the expansion timescale, τcoll τexp. Par- ≥ ticles during this time become free streaming and subsequently travel to particle detectors roughly unimpeded.
1.8.2 Centrality
The determination of impact parameter is not measured directly, but rather modeled.
In this work we will use the optical Glauber model, which is employed to map the impact parameter, b, to various centrality classes. The optical Glauber model describes nuclei with smooth nucleon densities in which individual partons are not located at specific spatial co- ordinates. There are models which track individual partons such as Monte Carlo Glauber, however, averaging over (108+) events smooths out the randomness associated with lumpy O
15 partons in nuclear distributions. In experiments, 0% centrality corresponds to the most- central (overlapping) collisions, i.e. b = 0 (see Figure 1.9, while 100% centrality maps to peripheral collisions where the nuclei pass by each other. In the later case, peripheral colli- sions occur when the impact parameter reaches a threshold which depends on the colliding system. Table 1.1 contains the relationship between the impact parameter, b, and centrality class, c, for a given collision system with √sNN = 2.76 TeV/nucleon Pb-Pb collisions using optical Glauber. One can also make a connection to the number of participants in a given collision, Npart, via the Glauber model.
Figure 1.8: The transverse profile of two colliding nuclei. The overlap is measured by the impact parameter, b, the distance between the center of the two nuclei. When calculated with the Glauber model, the overlapping region can be quantified by the number of partons participating in the collision, Npart.
16 ��� ��� ��� ��-�� ���� ��� ��-�� ���� ��� ��-�� ���
���� ��� �
���
� � � �� �� �� �(��)
Figure 1.9: The number of participants, Npart, as a function of the impact parameter, b, for various collision systems studied in this work.
cmin cmax bmin bmax b cmin cmax bmin bmax b h i h i 0. 0.05 0 3.473 2.315 0.5 0.6 10.983 12.031 11.515 0.05 0.1 3.473 4.912 4.234 0.6 0.7 12.031 12.995 12.519 0.1 0.2 4.912 6.946 5.987 0.7 0.8 12.995 13.893 13.449 0.2 0.3 6.946 8.507 7.753 0.8 0.9 13.893 14.795 14.334 0.3 0.4 8.507 9.823 9.181 0.9 1. 14.795 20 15.608 0.4 0.5 9.823 10.983 10.414
Table 1.1: The relationship between impact parameter, b, and centrality class, c, including the average impact parameter for each centrality class in a √sNN = 2.76 TeV/nucleon Pb-Pb collision modeled using optical Glauber. Table taken from [24].
To calculate Npart, we start with the Woods-Saxon nuclear profile function.
n n (r) = 0 , (1.19) A 1 + e(r−R)/d
−3 where n0 = 0.17 fm is the central nucleon density and is determined via the normalization
R 3 1/3 −1/3 limA→∞ d r nA(r) = A, R = (1.12A 0.86A ) fm is the nuclear radius, and d = 0.54 −
17 fm is the skin depth of the nucleus [42]. Using Eq. (1.19), we construct the thickness function
Z ∞ p 2 2 2 TA(x, y) = dz nA x + y + z . (1.20) −∞
We then utilize the impact parameter, b, to offset the two nuclei.
nAB(x, y, b) = TA(x + b/2, y)TB(x b/2, y). (1.21) −
We can then integrate Equation (1.21) over the transverse plane to obtain Npart as a function of b. Z ∞ Npart(b) = dx dy nAB(x, y, b). (1.22) −∞
We now have a framework for computing centrality classes and the alternatively-seen Npart and will use such calculations extensively to translate b to Npart, as well as average over
Npart to obtain centrality classes and centrality-averaged heavy quarkonium suppression. In
Figure 1.9, we see the relation between the number of participants, Npart, and the impact parameter, b, for three collision systems explored by this work.
1.8.3 A hydrodynamical description of the quark-gluon plasma
Dissipative hydrodynamical codes, which describe the full 3+1d evolution of the temperature- like scale and the degree of momentum-space anisotropy, can be used to simulate the back- ground evolution of the hot media generated in ultrarelativistic heavy-ion collisions from just before the formation times of heavy quarkonia until the plasma becomes free streaming and the QGP “freezes out” into cold matter. The momentum-space anisotropy, ξ(x), character- izes the breaking of symmetry between the longitudinal (beam-line) direction and the trans- verse plane. Momentum-space anisotropies in the local rest frame of the QGP directly affect
p 2 2 the partonic one-particle distribution function, faniso(x, p) = fiso( pT + [1 + ξ(x)]pz/Λ(x)),
18 where Λ(x) is the local temperature-like scale of the QGP. The resulting local rest frame distribution is spheroidal in momentum space. This particular anisotropy parameter was found in studies to be the dominant dissipative effect in hydrodynamic simulations [10] and results in important modifications of the in-medium heavy-quark potential.
Analysis of heavy quarkonium suppression and regeneration provide insight into cer- tain hydrodynamic parameters like the initial temperature and the initial momentum-space anisotropy of the plasma, as well as the transport coefficient η/s which is the shear viscosity to entropy density ratio. Initial studies have shown η/s to have a lower bound of 1/4π and
η/s is generally believed to be between 1/(4π) and 3/(4π). Recent phenomenological work to describe the experimentally observed hadron spectra with quasiparticle aHydro have pinned down this value to be approximately 2/(4π) [43], which is consistent with theory calculations which compute η/s using second-order hydrodynamics using NLO equations [44].
300 0-5% 40-50% (a) (b) 250 1500 5-10% 200
η 10-20% 50-60% d / 1000 150 20-30% dN 100 60-70% 500 30-40% 50 70-80% aHydro ALICE 0 0 80-90% - 6 - 4 - 2 0 2 4 6 - 6 - 4 - 2 0 2 4 6 η η
Figure 1.10: Charge-hardon multiplicity as a function of pseudorapidity, compared to ALICE 2.76 TeV/nucleon collisions. The percentage ranges given in the figure are various centrality classes, with 0 5% relating to the 5% most central collisions, and so on. Figure taken from [5], with ALICE− data provided from [6, 7].
As we will see in our model, bottomonium suppression within this framework is consistent with other work which suggests 4πη/s = 2. As seen in Figure 1.10, aHydro has been
19 shown to reproduce charged-hadron multiplicity as a function pseudorapidity for ALICE 2.76
TeV/nucleon collisions at the LHC. Dissipative hydrodynamical codes provide critical insight
into the interaction of a heavy quarkonium state with a hot medium which is anisotropic in
momentum space.
1.9 Quarkonium suppression
The idea of heavy quarkonia deconfining in a hot media can trace its origin to Matsui
and Satz [45]. In their pioneering work, Matsui and Satz suggested that the hot medium
generated in URHICs resulted in color screening for closed cc¯ states. Ultimately, their work
would go on to suggest that suppression of the J/ψ charmonium state in URHICs would
indicate an unambiguous signature for the formation of a new state of matter known as
the quark-gluon plasma. This suppression is due to the formation of the QGP, a state of
matter where hadronic particles “melt,” resulting in a sharp increase in the relative number
of degrees of freedom. High-energy heavy ion collisions would form the heavy-quark pairs,
but the hot QGP around heavy quarkonium states would dampen the exchange gluon and
cause the cc¯ pair to dissociate. This effect is known as thermal suppression. Due to the short
formation times and relatively long decay times in the dimuon channel, heavy quarkonium
production is considered to be a reliable probe of the QGP temperature. The QGP is
characterized by “melting” of hadrons into their constituent parts known as quarks and
gluons. While ordinary hadronic matter contains light quarks (q = u, d ), the energies found { } in URHICs are high enough to generate significant amounts of heavy quarks (q = s, c, b ) { } formed with their antimatter quark pairs (¯q = s,¯ c,¯ ¯b ). Recent URHIC experiments at { } Brookhaven National Laboratory’s Relativistic Heavy Ion Collider (RHIC) and CERN’s
Large Hadron Collider (LHC) study heavy ions and the resulting hot and deconfined matter that is generated in the process.
In Figure 1.11, we see the number of dimuon events detected by the CMS collaboration as a
20 -1 ×103 PbPb 368 µb (5.02 TeV)
9 µµ p < 30 GeV T µµ CMS 8 |y | < 2.4 µ p > 4 GeV T 7 µ |η | < 2.4 6 Centrality 0-100% PbPb Data Total fit 5 Background 4 RAA scaled 3 Events / (0.1 GeV) 2 1 0 8 9 10 11 12 13 14
mµ+µ- (GeV)
Figure 1.11: As a function of the dimuon invariant mass, we see three resonances appear in proton-proton reference data, and then disappear in Pb-Pb collisions at 5.02 TeV. This suppression of states is thought to be due to the formation of the QGP. Figure taken from [8]. function of the dimuon invariant mass. In scaled proton-proton reference data (dashed line), we see three peaks appear which correspond to the Υ(1S), Υ(2S), and Υ(3S) bottomonium states. This is compared to Pb-Pb collisions at the same energy (solid line with data points) where we see peaks appear for the Υ(1S) and Υ(2S), but they are strongly suppressed. The
Υ(3S) has not be observed in significant amounts in Pb-Pb 5.02 TeV/nucleon collisions as the collection of events around the Υ(3S) mass is well within the background for the CMS experiment. The suppression of states is known as heavy quarkonium suppression and is believed to be a strong signature of the formation of a QGP.
21 Based on fits to other observables such as pion spectra, hydrodynamic simulations of
√sNN = 2.76 TeV/nucleon Pb-Pb collisions show the initial temperatures of the generated
QGP are T0 500 600 MeV, which are consistent with the work presented in this model. ∼ − With recent calculations of bottomonium state widths, it is suggested that the ground state
Υ(1S) may survive temperatures through 600 MeV. The indication that some Υ(1S) states deconfine in the hot QGP show the presence of an additional physical effect leading toward further decay of heavy quarkonia in the QGP. Furthermore, the benefits of working with heavy quarkonia are derived from their bound states being dominated by short range physics, as well as their relatively large mass which suppresses statistical regeneration of states based on large numbers of open heavy qq¯ pairs (for Au-Au 200 GeV/nucleon collisions at RHIC).
The presence of regeneration can be clearly seen in experimental data for the J/ψ state
when moving from 200 GeV/nucleon Au-Au collisions at RHIC, to 2.76 TeV/nucleon Pb-
Pb collisions at the LHC. With the recent upgrades at the LHC to expand nucleus-nucleus
collisions to 5.02 TeV/nucleon, bottomonia show hints of regeneration in some experiments,
while others show more suppression of states. Even with stronger thermal suppression from
the hotter QGP that has been formed, small, but significant, amounts of regeneration may
be possible for bottomonia at 5.02 TeV/nucleon Pb-Pb collisions.
The model was then extended to make predictions for 5.02 TeV/nucleon Pb-Pb collisions
at LHC [46, 47]. This was done by using extrapolations which determined the increase in
the initial temperature of the generated QGP and bottomonium distribution throughout the
QGP compared to lower energy collisions. The accuracy of such predictions compared to
subsequent experimental measurements were critical in understanding the lack of a regen-
eration effect for bottomonia, as the increase in energy did not result in less suppression in
most experimental data [47].
22 EPS09 NLO shadowing, Pb-Pb 2.76 TeV, R. Vogt, Priv. Comm. 1.4
1.2
1.0 AA
R 0.8 L s 1 H
U 0.6
0.4
0.2
0.0 -4 -2 0 2 4 y
Figure 1.12: An estimate of shadowing (a cold nuclear matter effect) on Υ(1S) suppression at 2.76 TeV/nucleon LHC collisions of Pb-208 nuclei.
1.9.1 Other suppression effects
Heavy quarkonium production can be modified by other effects such as cold nuclear matter effects and co-movers. For the purposes of this work, we will be neglecting such effects, as they are less important for bottomonium suppression.
One of these effects, cold nuclear matter shadowing, has been calculated using EPS09 nuclear parton distribution functions with NLO shadowing and is seen in Figure 1.12. It’s noted that the effect for the rapidity window using this model predicts a maximal 10% ∼ effect for raw Υ(1S) RAA. Therefore, to good approximation, we can ignore cold nuclear matter effects for the purposes of this work for the bottomonium states. Small collision systems such as p-Pb at the LHC suggest cold-nuclear matter effects such as shadowing play an important role for charmonium [48]. However, since this work focuses on nuclear collisions and bottomonium, we can neglect such effects due to limited impact on final suppression.
23 Chapter 2
The heavy quark potential
2.1 Heavy quarkonia
Heavy quarkonia are a hard probe of the quark-gluon plasma due to their formation early in the plasma’s proper time evolution, and exit into the hadronization stage of the QGP.
Heavy quarkonia then decay into dilepton pairs and then are detected by particle detectors surrounding the collision region of the particle collider. The mass scales of the relatively large charm and bottom quarks ensure that charmonium and bottomonium evolve in the plasma without losing vital information regarding properties of the QGP such as the temperature profile during the whole 3+1d evolution.
Heavy quarkonia have long been discussed as having a complex potential which is the result of gluonic dampening from the hot quark-gluon plasma. A “proper” heavy quark potential should feature the two main properties of QCD, namely confinement and asymp- totic freedom. If such a potential is not derived via first principles, e.g. lattice QCD, free parameters should be fit in an attempt to phenomenologically describe certain features in experiment.
The presence of a hydrodynamical attractor (see Fig. 3.1) in momentum-space is critical for not only the QGP evolution description using hydrodynamics, but also the modeling of heavy quarkonia as they are formed in the non-equilibrium phase of the plasma. As we will see in each respective section, the presence of this anisotropy has been computed in the
Strickland-Bazow potential analytically to all orders in the momentum-space anisotropy, ξ, for the real part and to first order in the imaginary part of the potential. Recent calculations have extended this result to include an analytic calculation of the imaginary part to all orders
24 in ξ [49].
2.2 A perturbative inspired heavy-quark potential (Strickland-Bazow)
In previous work seen in References [50, 51], and factoring in the significant presence
of momentum-space anisotropies in the QGP, the free-energy-based heavy quark potential
models, as originally described by Karsch, Matsui, and Satz, do not reproduce quarkonium
suppression data from experiments at both RHIC or the LHC. For this reason the pNRQCD
model used herein, also known as the perturbative-inspired (Strickland-Bazow) heavy quark
potential, utilizes the internal energy of the quarkonium pair U = F + TS to provide the
model potential.
The masses of heavy quarks allow charmonium and bottomonium pairs to be treated
using pNRQCD methods which allow for relativistic corrections. The model presented herein
represent the short- and medium- range gluonic screening of the heavy-quark potential in a
plasma which is anisotropic in momentum-space [51, 52, 53, 54, 55]. The long-range effects of
the potential are modeled by the original Karsh–Mehr-Satz (KMS) form for the free energy.
The real part internal-energy-based model, in it’s final form, is given by,
a −µr 2σ −µr −µr 0.8σ Re[V] = (1 + µr)e + [1 e ] σre 2 , (2.1) − r µ − − − mqr
where µ = (ξ, θ)mD [51, 53, 54], with θ quantifying the angle between the beam-line G direction and the line connecting the quark–antiquark pair, mD is the isotropic leading-
2 order Debye mass, a = 0.385, σ = 0.223 GeV [56], for q = c we have mc = 1.29 GeV, the mass of the charm quark, and for q = b, we have mb = 4.7 GeV, the mass of the bottom quark. The final term factors in the correction for the finite quark mass. Though this term is not important for bottomonium states, it is included for historical continuity with previous works. Recent progress with the imaginary part include an expansion which is accurate to
25 all orders in ξ. However, for the model presented in this work we utilize a small-ξ expansion of the heavy-quark potential [55, 57, 58].
n o Im[V] = αsCFT φ(ˆr) ξ[ψ1(ˆr, θ) + ψ2(ˆr, θ)] , (2.2) − −
wherer ˆ = rmD, φ(ˆr) is defined as
Z ∞ z sin (zrˆ) φ(ˆr) = 2 dz 2 2 1 , (2.3) 0 (z + 1) − rˆ
and ψ1 and ψ2 are defined as follows
Z ∞ z 3 2 sin(z rˆ) 2 ψ1(ˆr, θ) = dz 2 2 1 sin θ + (1 3 cos θ)G(ˆr, z) , (2.4) 0 (z + 1) − 2 z rˆ −
Z ∞ 4 3 z 2 2 sin(z rˆ) 2 ψ2(ˆr, θ) = dz 2 3 1 3 cos θ + (1 3 cos θ)G(ˆr, z) . − 0 (z + 1) − 3 − z rˆ − (2.5)
2.3 A lattice QCD vetted heavy-quark potential (Rothkopf)
In this section I will describe a lattice QCD vetted, non-relativistic potential-based de- scription of heavy quarkonium in a QGP which can trace its roots from a systematic treat- ment of heavy quarkonium in QCD from the effective field theories non-relativistic QCD
(NRQCD) and potential NRQCD [59, 60]. The inherent separation of scales between the rest mass of the heavy quark, mc,b, the medium temperature, and the characteristic scale of
QCD, ΛQCD, are exploited to adequately simplify the physical picture of heavy quarkonium.
One does not have to consider a full quantum field theoretical boundary value problem for
Dirac fields, but rather use an initial-value problem for two-component Pauli spinors which follows from NRQCD. Following, this non-relativistic theory is matched to terms of coupled
26 color singlet ψS(r, t) and the color octet ψO(r, t) wave functions which results in pNRQCD.
In the latter pNRQCD treatment, the interaction among heavy quarks, and the interaction with the hot medium, is obtained in both potential and non-potential contributions. The specific scale hierarchy of this problem means the potential contributions dominate, therefore the evolution of heavy quarkonium reduces to a Schr¨odingerequation.
In realistic high-temperature environments, such as heavy-ion collisions at RHIC and the
LHC, perturbation theory cannot be used to fully determine the matching coefficients to obtain the potential. Withal, vital insight has been gained by pNRQCD with the hard-loop approximation [57, 60]. Moreover, the in-medium potential must be complex at large tem- peratures, as discussed with an isotropic medium in [57] and a medium which is anisotropic in momentum space in [53, 58, 55, 49]. Therefore, purely real potential models such as the color singlet free energies, or the internal energies, are not valid descriptions of heavy quarkonium physics in a medium.
The heavy quark potential in this section is related to the rectangular Wilson loop which is a real-time QCD quantity. The relation is matched with a correlation function in the effective theory pNRQCD and the underlying microscopic QCD theory which follow identical physical content at the appropriate scale. In the case of a lattice QCD vetted potential, the unequal time correlation function of a heavy quarkonium singlet state can be identified with the following Wilson loop in the static limit
Z D E m→∞ D h µ a aiE ψS(r, t)ψS(r, 0) W(r, t) = Tr exp ig dx AµT . pNRQCD QCD ≡ −
Since the Wilson loop obeys a simple equation of motion [57]
i∂tW(r, t) = Φ(r, t)W(r, t), (2.6)
27 with a space- and time-dependent complex function Φ(r, t), the potential picture is valid, considering Φ asymptotes toward a constant value at late times, i.e. time independent. In general this value is complex, and the analogous potential is formally defined as
i∂tW (t, r) VQCD(r) = lim . (2.7) t→∞ W (t, r)
In its current form, Equation (2.7) is not amenable to a non-perturbative lattice QCD eval- uation which is evolved in Euclidean time which is an unphysical quantity. To counteract this, one must take a detour via spectral decomposition of the Wilson loop which relates the
Euclidean and Minkowski domains [61, 62],
Z Z W (τ, r) = dωe−ωτ ρ (ω, r) dωe−iωtρ (ω, r) = W (t, r). (2.8) ↔
Inserting Equation (2.8) into Equation (2.7) specifies that both the real and imaginary part of the potential are related to the position and width of the lowest lying peak within the
Wilson loop spectrum. If the potential picture is applicable, the Wilson loop spectrum contains a well-defined lowest-lying peak skewed Lorentzian form, from which the potential values can be extracted via a χ2-fit [63].
With the recent success of a Bayesian approach [64], the extraction of spectral functions from Euclidean lattice data has only been possible as of late due to an inverse problem.
Reconstruction robustness was improved significantly compared to attempts made in the past with the Maximum Entropy Method [62]. One considers Wilson line correlators fixed in the Coulomb gauge, instead of the Wilson loop on the lattice. This practice produces results which are free from a class of divergences impeding the numerical determination of the Wilson loop. With this method, potential values have been extracted in quenched QCD based on the Wilson action [65], as well as for QCD with Nf = 2 + 1 light quark flavors based on
28 recent work by the HotQCD collaboration. Both cases confirmed the appropriateness of the potential picture at all temperatures considered due to the well-defined peak of Lorentzian shape from the Wilson spectral functions.
Two more steps are needed before utilizing discrete potential values obtained from lattice
QCD, which have been explicitly laid out in [66, 67]. First, we parameterize values of the potential with an analytic formula which allows for the evaluation of Re[V] and Im[V] at intermediate separation lengths not resolved by lattice methods. Second, we correct the parameters in the analytic parameterization for artifacts remaining from finite volume and
finite lattice spacing, as the continuum extrapolated lattice QCD determination of the in- medium heavy quark potential has yet to be achieved.
The parameterization to an analytic expression we used is one based on a generalized
Gauss law for the vacuum heavy quark potential. Lattice QCD studies have shown, for phenomenologically relevant distances, that the quarkonium vacuum (T = 0) potential is reproduced by the Cornell ansatz which consists of a Coulombic region dominating the small-r physics, and a linearly rising term which is consistent with confinement in QCD. The effects of a running coupling constant in QCD are mimicked by allowing the linear terms to contribute down to the smallest distances between the quark pair. Following the Gauss law example, we consider the divergence of the auxiliary (color) electric field E = qra−1rˆ emerging from either the Coulombic, a = 1, q = αs, [αs] = 1 part, or the string-like part − with a = 1, q = σ, [σ] = GeV2.
E = 4π q δ(r). (2.9) ∇ ra+1
The parameters which enter this expression characterize the non-perturbative vacuum physics of the heavy quarkonium bound state are the strong coupling αs, the string tension σ, and a constant shift c. It is noted that the CF factor is absorbed into the definition of the strong
29 2 g CF coupling, i.e. αs = 4π . One can perform a Fourier transform on Gauss’ law and modify the right-hand side by dividing it with an in-medium permittivity, . Such a treatment, which is a well-known prescription in classical electrodynamics, can be used to introduce the effects of a thermal medium. This permittivity of a QCD medium which has been introduced is computed in hard-thermal loop perturbation theory,
2 2 −1 p p mD ε (p, mD) = 2 2 iπT 2 2 2 . (2.10) p + mD − (p + mD)
Non-perturbative physics of the bound state are encoded in the Cornell potential which is driven by a weakly-coupled gas of quarks and gluons. The combination of Equations (2.10) and (2.9) leads to integro-differential equations for the in-medium modified Coulombic and string-part of the vacuum potential as discussed in [68]. With a complex in-medium permit- tivity, the in-medium potential is also complex. Unlike the purely perturbative computations, the in-medium potential considered here receives a contribution to its real and imaginary part from the string-like portion of the Cornell potential. The expressions for the Coulombic part are
e−mDr Vc(r) = αs mD + + iT φ(mDr) , (2.11) − r with
Z ∞ z sin(xz) φ(x) = 2 dz 2 2 1 , (2.12) 0 (z + 1) − xz which coincide with the results of Reference [57].
30 ��� shifted lQCD Nf =2+1 asqtad data & T=0.86TC T=0.95TC T=1.06TC � Gauss-law fit T=1.19TC T=1.34TC T=1.41TC
� ��� T=1.66TC
] �
��� ��� ][
� ][ ��� ] ���� ���� � [ � ��� �� �� [ � �
T≈0(β=6.9) T≈0(β=7.48) T=0.86TC ���
� T=0.95TC T=1.06TC T=1.19TC
T=1.34TC T=1.41TC T=1.66TC � ��� ��� ��� ��� ��� ��� ��� ��� ��� � � � � � �� � [��] �[��]
Figure 2.1: The in-medium heavy quark potential [real part (left) and imaginary part (right)] in full QCD with Nf = 2 + 1 light quark flavors based on ensembles by the HotQCD collaboration (colored points, shifted for better readability). By adjusting the Debye mass mD, the lattice QCD values of the real part are adequately reproduced with the Gauss- law parameterization as shown with the solid lines, at all temperature and length scales. Theoretical error bars (the shaded region around the solid lines) are due to the uncertainty of the Debye mass fit. The imaginary part shows good agreement at high temperatures and small distances, while at temperatures close to the crossover transition, deviations from the lattice data are seen. Note that the crossover temperature on these lattices due to the relatively large pion mass of mπ 300MeV lies at TC = 172.5 MeV. ≈
31 The additional and string-like contribution, on the other hand, reads
1 1 Γ[ 4 ] σ Γ[ 4 ] σ ReVs(r) = 3 D 1 √2µr + , (2.13) − 2 3 −2 4 √π µ 2Γ[ 4 ] µ
4 2 for the real part, where the parameter µ = mDσ/αs characterizes the strength of the in- medium modification. For the imaginary part we obtain
2 σmDT ImVs(r) = i ψ(µr) = iαsT ψ(µr), (2.14) − µ4 −
where ψ corresponds to the following Wronskian
Z x 2 ψ(x) = D−1/2(√2x) dy ReD−1/2(i√2y)y φ(ymD/µ) 0 Z ∞ 2 +ReD−1/2(i√2x) dy D−1/2(√2y)y φ(ymD/µ) x Z ∞ 2 D−1/2(0) dy D−1/2(√2y)y φ(ymD/µ). − 0
A key characteristic of the above expressions is once the vacuum parameters of the Cornell
potential are fixed, the remaining temperature dependent parameter is the Debye mass mD.
The parameterization to analytic form was done in a straightforward fashion, however, it relies on some assumptions and needs to be vetted with lattice QCD data before proceeding.
It has been shown that lattice values of the resulting potential, with the Gauss law parame- terization, are significantly reproduced in both quenched [65] and full QCD simulations with
Nf = 2 + 1 light flavors [66]. The agreement can be seen in Fig.2.1. After fixing the three parameters αs, σ, and c with low-temperature ensembles, the Re[V] is fitted by tuning the
Debye mass mD. After mD is fixed, the given parameterization predicts the imaginary part of the potential Im[V]. The prediction results in agreement with quenched QCD simulations at high temperatures and becomes less accurate for temperatures around the phase transition
32 at zero net baryon chemical potential. To date, there has not been a vigorous determination of the imaginary part of the potential using full QCD. However, the values extracted in the procedure above show good agreement with the Gauss-law parameterization at high and in- termediate temperatures [66]. The Debye mass related to the full QCD in-medium potential show coherent deviations from perturbative predictions in the temperature regime relevant to this study, TC < T < 3TC .
Due to lattice artifacts mentioned above, the potential as it stands may not be directly applied to phenomenological computations, despite the already successful parameterization which reproduces the lattice QCD in-medium potential. The parameters αs, σ and c could be determined from first principles if one were to perform a continuum extrapolation of the
T = 0 potential in the thermodynamic limit. Instead, here we narrow these parameters phenomenologically such that the vacuum bottomonium spectrum is reproduced below the
D-meson and B-meson threshold for charmonium and bottomonium, respectively. The mass of the correct quark masses is the renormalon subtracted mass which then reflect the repro- duction of the spectra. They take on the value mRS0 = 4.882 0.041GeV for the bottom b ± quark and, for the charm quark, we compute the value to be mRS0 = 1.472 0.008GeV. c ± With this parameterization. With the vacuum potential in full QCD only robustly repro- duced up to distances of r 1 fm, we enforce, by hand, asymptotics which are flat due to ≈ string breaking at rSB = 1.25 fm. With all the above, the superlative set of parameters are given by
c = 0.1767 0.0210 GeV, αs = 0.5043 0.0298, √σ = 0.415 0.015 GeV. (2.15) − ± ± ±
Finally, the Debye mass is computed in a manner which is self-consistent with the dy- namical evolution of the QGP medium and use its value to apply the in-medium modification of the vacuum potential with the parameters given above. Figure 2.2 shows the real part of
33 Real-part of the heavy-quark potential from the Gauss-Law Imaginary-part of the heavy-quark potential from the Gauss-Law ��� ����
����
] ��� ���� ��� ][ ��� ] ][
��� ���� ��� ��� � [ ��
mD=0 mD=400MeV - �� [ � ���� mD=0 mD=400MeV α_s=0.504(29) -��� mD=100MeV mD=500MeV mD=100MeV mD=500MeV σ =0.415(15)GeV mD=200MeV mD=600MeV ���� mD=200MeV mD=600MeV c=-0.177(21)GeV mD=300MeV mD=300MeV ���� -��� � � � � � � � � � � � [��] �[��]
Figure 2.2: The Re[V] (left) and Im[V] (right) part of the in-medium heavy quark potential which has been vetted against lattice QCD. The different lines correspond to different values of the Debye mass of the medium. The vacuum parameters αs, σ, and c at mD = 0 are tuned to reproduce the PDG charmonium and bottomonium spectra. String breaking is enforced at rsb = 1.25 fm. Ultimately, only one parameter governs the temperature-dependence which modifies the vacuum (Cornell) potential. Such thermal effects lead to Debye screening of the real part of the potential and induce a finite imaginary part which asymptotes at large distances.
the potential (left) and the imaginary part of the potential (right) which has been used for
various values of the Debye mass. To determined which values of the Debye mass are signif-
icant in the evolution of heavy quarkonium, it is noted that lattice QCD studies show, in a
thermal QCD medium close to the crossover transition temperature, the ratio is mD/T 1 ≈
and grows to mD/T 2 as temperature approaches T = 2TC . ≈
2.4 Non-equilibrium corrections to the lattice-vetted heavy quark potential
The Debye mass in an isotropic medium is defined in the following manner
2 Z ∞ 2 g 2 dfiso mD = 2 dp p , (2.16) −2π 0 dp
where p is defined in the following manner which follows the breaking of symmetry along
the longitudinal axis and transverse plane, p p2 = p2 + p2. Lattice QCD studies to date ≡ ⊥ z
34 are restricted to isotropic systems where an explicit temperature can describe the medium.
In the case of the anisotropy used in this study, we instead have a temperature-like scale.
The effect of an anisotropic parameter, ξ, on the Strickland-Bazow potential derived from
pNRQCD has been analytically calculated which obtained the following result
µ(θ)−4 2b(a 1) + (1 + ξ)1/8 c(θ)(1 + ξ)d = 1 + ξ a − b 1 + 2 , (2.17) mD − (3 + ξ) (1 + eξ )
with a = 16/π2, b = 1/2, d = 3/2, e = 1/3, and the angular-dependence is factored into a
function which is defined as
3π2 cos(2θ) + (9 + 4√3 4√6)π2 + 64(√6 3) c(θ) = − − . (2.18) 4√3(√2 1)π2 16(√6 3)) − − −
The implementation is handled by tabulating the real and imaginary parts of the potential
as a function of mD and r, and then simply replacing mD with µ in the isotropic potential to extend the potential for plasma anisotropies.
2.5 Solving for binding energies
The algorithm from References [69, 70] is used to solve the resulting 3d Schr¨odinger equation on a regular lattice by transforming to imaginary time and using the finite difference time domain (FDTD) method. Using this method, we compute the real and imaginary parts of the binding energy over a range of temperature-like scales, Λ, from 144 MeV to 1037 MeV.
For each Λ, we compute the real and imaginary binding energies for a range of anisotropies,
ξ, from 0.3 to 200, with an irregular spacing which accounts for the fact that the majority − of the time the system probes small values of ξ.
Determination of wave functions of the heavy quarkonia are performed computationally
35 by solving the eigenvalue problem found in the time-independent Schr¨odingerequation
ˆ Hφν(x) = Eνφν(x), (2.19) where Hˆ is the Hamiltonian and is given in the form used to solve two-body problems
2 ˆ H = ∇ + V (x) + m1 + m2, (2.20) − 2µ
where m1 and m2 are the masses of the quarks, µ = m1m2/(m1 + m2) is the reduced mass of the quarkonium system which reduces to mq/2 since m1 = m2 = mq, and the potential
V (x) is the complex-valued potential mentioned above, V = Re[V] + iIm[V]. The index ν found on eigenfunctions φν and energies Eν represent a combination of all quantum numbers, however, we will utilize the standard labeling for heavy quarkonia, i.e. nS and mP for s- and p-wave heavy quarkonia with n and m being integers. To obtain the eingenfunctions, we solve the 3-dimensional Schr¨odingerequation
∂ i ψ(x, t) = Hψˆ (x, t), (2.21) ∂t
which can be expanded in terms of the eigenfunctions φν,
X −iEν t ψ(x, t) = cνφν(x)e . (2.22) ν
To extract the lowest-lying states, we perform a Wick rotation, τ it. Using this rotation ≡ to imaginary time, Equations (2.21) and (2.22) are written as
∂ ψ(x, τ) = Hψˆ (x, τ), (2.23) ∂τ −
36 X −Eν τ ψ(x, t) = cνφν(x)e . (2.24) ν
This process of using a Wick rotation effectively projects out the lowest-lying states since as
imaginary time is evolved, the excited-state wave functions become exponentially dampened.
The real and imaginary binding energies are extracted using [51]
φν V∞(θ) φν Eν,bind Eν m1 m2 h | | i , (2.25) ≡ − − − − φν φν h | i
where
V∞(θ) lim Re[V(θ, r)], (2.26) ≡ |r|→∞
and for the case of the Rothkopf potential we use the last point in r for V which is a ∞ good approximation as the potential is tabulated for distances up to 8 fm. Negative values
of Im[Ebind] only occur for large values of ξ in the Strickland-Bazow potential, which is a
consequence of the small-ξ expansion. Large values of ξ correspond to a nearly free streaming
quark-gluon plasma, so it is expected that the widths of quarkonium states return to vacuum
values, which are on the order of keV, which effectively allows us to set Im[Ebind] = 0 for ∼ this specific case.
2.5.1 Strickland-Bazow potential discretizations
For charmonia, we use a N 3 = 2563 lattice with lattice spacing a = 0.15 GeV−1 0.03 ≈ fm and a = 0.20 GeV−1 0.04 fm for a total box length of L = Na 7.68 fm and ≈ ≈ L = Na 10.08 fm, for s-wave and p-wave states, respectively. For bottomonia, we use a ≈ N 3 = 2563 lattice with lattice spacing a = 0.1 GeV−1 0.02 fm and a = 0.15 GeV−1 0.03 ≈ ≈ fm for a total box length of L = Na 5.04 fm and L = Na 7.56 fm, for Υ(nS) and ≈ ≈
χb(mP ) states, respectively. The potential is put onto this lattice and evolved with the
3-dimensional Schr¨odinger equation in imaginary time. The grid is filled with wave function
37 which is randomly initialized and then evolved in imaginary time until the ground state
converges within a given tolerance for changing binding energies in time. The imaginary-
time step size for the solving algorithm is taken to be ∆τ = a2/8 which has been shown to
keep the temporal evolution stable while retaining a reliable physical picture of the state.
Using snapshots of the wave function we can project out the first- and second-excited states
which are low-lying in the wave function due to the imaginary time evolution. We also
employ symmetry constraints which allow the code to solve for the s-wave quarkonia with
symmetric initial conditions and likewise we use antisymmetric initial conditions for the p-
wave quarkonia. This property is exploited using the fact symmetry cannot be broken by
the Hamiltonian evolution [69].
2.5.2 Rothkopf potential discretizations
To solve the Schr¨odingerequation with this potential, we employ the same 3-dimensional
code in which we solve the Schr¨odingerequation in imaginary time. We use a N 3 = 2563
lattice with grid spacing a = 0.15 GeV−1 for Υ(nS) states and a = 0.175 GeV−1 for the
χb(mP ) states. Since non-equilibrium corrections to this potential are handled on the po- tential level, the process for extracting the binding energies is the same for both potentials studied in this work.
2.6 Discussion on difference between potentials
The Strickland-Bazow potential is formulated using the internal energy of the heavy quarkonium pair which provides a much stronger binding when the potential is used to calculate suppression of states in heavy ion collisions [51]. While the Rothkopf potential is more weakly bound compared to the internal energy formulation, the Rothkopf potential remains bound stronger than the free energy description of heavy quarkonia.
From the results of Fig. 2.3, we see more-weakly bound Υ(1S) in the Rothkopf binding
38 ��� ��� ◆ ◆
● Strickland-Bazow, Real Part ● Strickland-Bazow, Real Part ��� ��� ■ Strickland-Bazow, Imaginary Part Υ(��) ■ Strickland-Bazow, Imaginary Part Υ(��) ◆ Rothkopf, Real Part ξ=� ◆ Rothkopf, Real Part ξ=� ▲ Rothkopf, Imaginary Part ▲ Rothkopf, Imaginary Part ● ��� ▲ ▲ ��� ▲ ▲ ◆ ▲ ▲ ◆ ▲ ▲ ● ▲ ▲ ▲ ▲ ▲ ● ▲ ▲ ▲ ▲ ● ▲ ▲ ▲ ▲ ▲ ��� ● ▲ ��� ◆ ▲ ▲ ▲ ▲ ● ▲ ▲ ◆ ▲ ● ◆ ● ◆● ● ◆ ● ● ��� ◆ ��� ◆ ● ● ◆ ▲ ● ● ◆ ������� ������ [ ��� ] ������� ������ [ ��� ] ● ◆ ● ◆ ● ◆ ● ■ ■ ◆ ● ■ ■ ■ ● ▲ ◆ ● ▲ ◆ ● ▲ ■ ■ ��� ● ▲ ▲ ■ ■ ��� ◆ ● ▲ ◆ ● ▲ ■ ■ ■ ◆ ● ▲ ▲ ■ ◆ ● ▲ ■ ■ ■ ◆ ●▲ ■ ■ ■ ◆ ▲ ●▲ ● ■ ■ ■ ◆ ▲ ▲ ● ● ■ ■ ■ ▲ ◆ ● ■ ■ ■ ▲ ◆ ● ■ ■ ■ ■ ▲ ◆ ■ ■ ●■ ▲ ▲ ◆ ●■ ●■ ■ ▲ ▲ ■ ■ ■ ◆ ● ● ▲ ◆■ ■ ■ ■ ● ● ▲ ■▲ ■ ■ ■ ◆ ● ● ▲ ▲ ■ ■ ■ ■ ■ ◆ ● ● ● ■ ▲■ ▲■ ■ ◆ ● ● ● ▲ ■▲ ■▲ ■ ■ ■ ◆ ● ● ● ■ ▲■ ▲■ ▲ ◆ ● ● ● ■ ▲■ ▲■ ▲■ ■▲ ■ ◆ ● ● ● ● ���▲ ◆ ◆ ◆ ◆ ◆ ◆ ◆ ◆ ◆ ◆● ◆● ◆● ◆● ◆● ◆● ◆● ◆● ◆● ���▲ ◆ ◆ ◆ ◆ ◆ ◆ ◆ ◆ ◆ ◆ ◆ ◆ ◆ ◆ ◆● ◆●
� � � � � � � � � � � � Λ/�� Λ/��
Figure 2.3: Comparison between Strickland-Bazow Υ(1S) and Rothkopf Υ(1S) for the cases of an isotropic, ξ = 0, plasma (left), and a slightly oblate, ξ = 1 plasma. In the Rothkopf model we observe weaker binding and a slightly stronger screening effect from the imaginary part of the binding energy. Here Tc = 192 MeV which then sets the scale of the temperature- like scales of the calculation. The discontinuities seen in Im[V] are due to the real part going negative.
energies for both of isotropic and anisotropic plasma. We also see the Rothkopf potential
results in a stronger imaginary part which corresponds to stronger screening effects in the
hot QGP. Therefore, we expect to see more suppression for the Rothkopf heavy quarkonia
compared to Strickland-Bazow heavy quarkonia. The jump for each Im[V] is due to con-
vergence issues when Re[V] < 0. Such regions of discontinuity are not factored into the
RAA calculation as we will come to see later in this paper. Finally, the sensitivity for the
momentum-space anisotropy parameter, ξ, is different for the Strickland-Bazow Υ(1S) and
the Rothkopf Υ(1S). This is due to the non-equilibrium correction being substituted for the
Debye mass mD for the Rothkopf potential.
39 Chapter 3
3+1d anisotropic hydrodynamics
3.1 Anisotropic hydrodynamics
Throughout this study, a hydrodynamic background is used to simulate ultrarelativistic heavy-ion collisions. This hydrodynamic background consists of massless conformal quarks and gluons which obey an ideal equation of state which reduces to εiso = 3Piso in the isotropic state. Due to rapid expansion along the longitudinal beam-line axis of the colliding nuclei, compared to the relatively slow initial transverse expansion, one expects high-energy heavy- ion collisions to result in a plasma which is anisotropic in momentum space. In fact, recent studies have shown, with this degree of freedom established in the equations which govern the background evolution, regardless of the initial condition of L/ T , i.e. the local rest P P frame (LRF) pressure in the longitudinal (beam-line) direction and the LRF pressure in the transverse plane direction, respectively, the QGP evolution converges to an anisotropic
“attractor” which drives the approach toward isotropic equilibrium (see Figure 3.1) at late times [9].
The presence of such an attractor is critical as heavy quarkonia are formed rather early in the QGP evolution (τ < 1 fm/c). Across a wide range of initial L/ T , we see the presence P P of a momentum-space anisotropy is important for the hydrodynamical background. Further- more, an observable like quarkonium suppression can probe the early time dynamics of the
QGP, for example, the initial energy density and the initial momentum-space anisotropy of the QGP.
40 1.0 aHydro attractor(Tinti Matching) NS 0.8 Numerical solution
0.6 T / L
0.4
0.2
0.0 0.5 1 2 5 w
Figure 3.1: Plotted with the dimensionless time variablew ¯ = τ = τT , we see varying τR 5η/s L/ T initial conditions converge rather quickly to an attractor. Figure taken from Reference [9].P P
The following one-particle distribution function is used in the typical, anisotropic, “Romatschke-
Strickland” form 2 p 2 2 faniso(p , ξ, Λ) = fiso p + ξ(p ˆn) , Λ , (3.1) ·
where Λ is transverse momentum scale, which characterizes the temperature-like scale of
the plasma, and ξ is the momentum-space anisotropy parameter. For the case of breaking symmetry between the longitudinal axis and the transverse plane, ξ is given by the following
relation where ˆn = z. 1 p2 ξ = h ⊥i 1 (3.2) 2 p2 − h zi
where pz p ˆz and p⊥ p pz. ≡ · ≡ −
41 A spheroidal treatment of the momentum-space anisotropy of the plasma is an approxi- mation, as there are two other diagonal components of the anisotropic tensor as well as three off-diagonal components. The effect of each component of this tensor has been studied in the context of viscous hydrodynamics and is shown in detail in Figure 3.2.
Σ 0.1 Au+Au, b=7 fm SM-EOS Q 〉
/(e+p) 0 ∆
mn 0.004 τx 〉 π 〈π τy π ττ
/(e+p) π 2 ηη 0
mn xy τ π π -0.1 〈π -0.004 ∆ mn mn initialized by π =2ησ mn 0 5 10 initialized by π =0 τ−τ0(fm/c) 0 5 10 τ−τ0(fm/c)
Figure 3.2: Shear tensor corrections calculated using second-order viscous hydrodynamics Reference [10].
Figure 3.2 shows the evolution of the shear tensor components from a realistic second- order viscous hydrodynamics simulation as a function of proper time. Relative to the other curves, two curves stand out much more than the rest, which are Σ and τ 2πηη.1 Here Σ is the sum of the spacelike components xx and yy, i.e. Σ πxx + πyy, and the third spacelike ≡ component, zz, being found in the τ 2πηη curve, i.e. τ 2πηη = πzz. The correction to the trans- verse pressure is simply Σ/2 and πzz provides the correction for the longitudinal pressure.
This alone provides the motivation for breaking the symmetry between the longitudinal axis
1Here η is the spatial rapidity.
42 and the transverse plane. The transverse plane also contains an anisotropy which is the
curve ∆ πxx πyy. The quantity ∆ is smaller than the two curves mentioned prior up to ≡ − around 7 fm/c after the initial proper time, which also roughly corresponds to the time at which freezeout will occur in the aHydro simulation of minimum-bias collisions. Therefore, for the purposes of this study, we can ignore shear tensor corrections beyond an anisotropy which characterizes the breaking of symmetry between the longitudinal beam-line axis and the transverse plane.
This spheroidal description of momentum-space anisotropy creates three cases for the plasma: ξ < 0 describing a prolate plasma, ξ = 0 describing an isotropic plasma, and ξ > 0 describing a plasma that is oblate in momentum-space. The isotropic distribution function
−E/T is given by Boltzmann statistics, i.e. fiso = e .
The equations of motion for the anisotropic system are derived by starting from kinetic theory and the assumption that the distribution function of the plasma is known. This is done by taking moments of the Boltzmann kinetic equation in the relaxation-time approximation
(RTA) µ µ p uµ p ∂µf = , (3.3) τeq
where τeq is the local relaxation time of the plasma which can depend on proper time, and
uµ is the local four-velocity of the plasma. The zeroth moment of the Boltzmann kinetic
equation produces the particle production equation
µ µ µ Neq N ∂µN = uµ − , (3.4) τeq
which shows a system where particles are not conserved. The equilibrium particle flux is
defined as follows
µ µ Neq = nequ , (3.5)
43 and likewise,
N µ = nuµ (3.6)
The first moment of the Boltzmann kinetic equation yields a set of partial differential equa-
tions which ensures energy-momentum conservation
µν ∂µT = 0, (3.7)
where at leading-order, the energy-momentum tensor has the following form for a plasma
which has a spheroidal anisotropy
µν µ ν µν µ ν T = (ε + P⊥)u u P⊥g (P⊥ Pk)z z . (3.8) − − −
When used together, Equations (3.4) and (3.7) result in a set of five partial differential
equations
ν Dµε = (ε P⊥)θu + (P⊥ Pk)uνDzz , (3.9) − − −
ν DzPk = (P⊥ Pk)θz + (ε + P⊥)zνDuu , (3.10) − " # u⊥ p⊥ ⊥P⊥ ν Duu⊥ = · ∇2 + DuP⊥ + (P⊥ Pk)uνDzz , (3.11) −ε + P⊥ u⊥ − ux 1 Du = 2 (ux∂y uy∂x)P⊥, (3.12) uy uy(ε + P⊥) −
Duξ 3DuΛ 1 h 3/4 p i = θu + 1 (ξ) 1 + ξ . (3.13) 2(1 + ξ) − Λ τeq − R
In these equations, note that the subscript denotes two-dimensional vectors in the trans- ⊥ verse plane, e.g. ⊥ (∂x, ∂y). Also note that the convective derivative is given by ∇ ≡ µ µ µ Dµ u ∂µ, the longitudinal derivative is Dz z ∂µ, and the expansion scalars are θu ∂µu ≡ ≡ ≡ µ and θz ∂µz . Without any symmetry constraints, we have the following parameterization ≡
44 of the four-velocity uµ and four-vector zµ in the lab frame,
µ u = (u0 cosh ϑ, u⊥, u0 sinh ϑ) (3.14)
zµ = (sinh ϑ, 0, cosh ϑ) (3.15) where the local longitudinal rapidity of the fluid is given by ϑ. A normalization condition is
µ applied to the four-velocity u uµ = 1 resulting in
q 2 u0 = 1 + u⊥, with (3.16)
q 2 2 u⊥ u + u . (3.17) ≡ x y
With the parameterizations given in Equations (3.14) and (3.15), quantities appearing in
Equations (3.9)-(3.13), ˆ Du = u⊥ ⊥ + u0L1, (3.18) · ∇
ˆ ˆ θu = ⊥ u⊥ + L1u0 + u0L2ς, (3.19) ∇ ·
ˆ Dz = L2, (3.20)
ˆ θz = L1ϑ, (3.21)
ν ˆ uνDzz = u0L2ϑ, (3.22)
ν ˆ zνDuu = u0 u⊥ ⊥ + u0L1 ϑ, (3.23) − · ∇ where the two linear differential operators are given by
ˆ ∂ς L1 = cosh(ς ϑ)∂τ sinh(ς ϑ) (3.24) − − − τ
45 ˆ ∂ς L2 = sinh(ς ϑ)∂τ cosh(ς ϑ) . (3.25) − − − τ
Lastly, the following relation is used between the relaxation time τeq and transport coefficient, shear viscosity to entropy density ratioη ¯ η/s, ≡
5¯η τ = . (3.26) eq 2T
3.1.1 The anisotropic equation of state
In order to properly close the dynamical equations above, an equation of state is needed to relate the energy density and pressure of the QGP. For this work, we use an ideal equation of state which relates the two as = 3P . The standard kinetic theory definitions are used E for the particle four-current, N µ, and the energy-momentum tensor, T µν,
Z d3p N µ pµf, (3.27) ≡ (2π)3p0
Z d3p T µν pµpνf. (3.28) ≡ (2π)3p0
With these definitions in Equations (3.27) and (3.28) and the tensor decompotions given in
Equations (3.5) and (3.8), the thermodynamics quantities of the system are given by
n n(Λ, ξ) = iso(Λ) , (3.29) √1 + ξ
ε(Λ, ξ) = (ξ)εiso(Λ), (3.30) R
P⊥(Λ, ξ) = ⊥(ξ)Piso(Λ), (3.31) R
Pk(Λ, ξ) = (ξ)Piso(Λ), (3.32) k
46 where niso, εiso, and Piso are the isotropic particle density, energy density, and pressure, respectively, and the anisotropic special functions required are
1 1 tan−1 √ξ (ξ) + , (3.33) R ≡ 2 1 + ξ √ξ
3 1 + (ξ2 1) (ξ) ⊥ − R , (3.34) R ≡ 2ξ ξ + 1 3 (ξ + 1) (ξ) 1 k R − . (3.35) R ≡ ξ ξ + 1
The factoring of the anisotropic parts from the thermodynamic quantities of the system
is a consequence of the conformal fluid assumed in this work. Extensions to a non-ideal
equation of state within the framework of quasiparticle anisotropic hydrodynamics have
been developed and shown to reproduce particle-spectra, elliptic flow coefficients, etc. from
ultra-relativistic heavy-ion collisions [43].
3.2 aHydro initial conditions
We consider two symmetric lead nuclei (A = 197 for RHIC Au-Au collisions and A = 208
for Pb-Pb LHC collisions). Each nuclei is modeled using a Woods-Saxon distribution for the
transverse nuclear profiles. For one nuclei, we have
n n (r) = 0 , (3.36) A 1 + e(r−R)/d
−3 where n0 = 0.17 fm is the central nucleon density and is determined via the normalization
R 3 1/3 −1/3 limA→∞ d r nA(r) = A, R = (1.12A 0.86A ) fm is the nuclear radius, and d = 0.54 − fm is the skin depth of the nucleus [42]. Using Eq. (3.36), we construct the thickness function
Z ∞ p 2 2 2 TA(x, y) = dz nA x + y + z . (3.37) −∞
47 We can then build the overlap density function for the two colliding nuclei with centers
separated by an impact parameter b chosen to be along thex ˆ direction, b = bxˆ,
nAB = TA(x + b/2, y)TB(x b/2, y). (3.38) −
The overlap density function is used as the probability weight for bottomonium production
in the transverse plane and is used to calculated the number of participants for a given
impact parameter b.
��
�
� y ( fm )
-�
-�� -�� -� � � �� x(fm)
Figure 3.3: From the Glauber model, we calculate the transverse energy density profile for a given impact parameter, b = 4.5.
For aHydro initial conditions we use a smooth linear combination, κbinary = 0.145 for
RHIC Au-Au collisions and κbinary = 0.15 for LHC Pb-Pb collisions, of Glauber wounded-
nucleon and binary collision scaling to set the initial energy density profile of the transverse
48 inel plane in the QGP [51]. The inelastic cross section is taken to be σNN = 42 mb for RHIC
inel 200 GeV Au-Au collisions, σNN = 62 mb for LHC Run One 2.76 TeV Pb-Pb collisions, and
inel finally σNN = 67 mb for LHC Run Two 5.02 TeV Pb-Pb collisions. In the spatial rapidity direction, a boost-invariant plateau is used at central rapidities along with Gaussian tails at large rapidities, which is consistent with limited fragmentation [71]. This is explicitly given by the following function,
( ς ∆ς)2 f(ς) exp − − 2 Θ( ς ∆ς) , (3.39) ≡ − 2σς | | −
with ∆ς = 1.0 and σς = 1.3 used for RHIC 200 GeV collisions, and ∆ς = 2.5 and σς = 1.4 used for LHC Run One 2.76 TeV collisions, both of which are tuned to reproduce the experimental pseudorapidity distributions of charged particles. For LHC Run Two 5.02
TeV collisions, we made predictions based on [72]. Using [72], one finds that going from 2.76
TeV collisions to 5.02 TeV collisions, there is a 12% increase in the plateau halfwidth which gives ∆ς5.02 TeV = 2.8, and no observable difference in the Gaussian halfwidth which results
5.02 TeV in σς = 1.4. Differences between longitudinal distributions among experiments can be seen in Figure 3.4.
The combination of a Glauber profile in the transverse plane along with Equation (3.39) for the longitudinal profile, we now have a full description of the initial 3+1d energy density profile for the QGP modeled using aHydro.
3.2.1 Determining the initial temperature-like scales
The initial temperatures for √sNN = 2.76 TeV collisions were fixed against soft hadron production using aHydro [73, 43]. Since these initial temperatures were fixed, we needed a self-consistent way to determine whether scans for the temperature-like scale Λ0 would result in reliable numbers. To do this, we run the aHydro data files through a code which finds the
49 ���
���
���
��� ��� ������������
���� ��� ��� ��� ��� ���� ��� ��� ���� ���
��� -� � � ������� ��������
Figure 3.4: Using a boost-invariant plateau with Gaussian tails, we can phenomenologically model the longitudinal profile for the initial energy density produced in an ultra-relativistic heavy-ion collision for RHIC 200 GeV/nucleon Au-Au collisions and LHC 2.76 TeV/nucleon and 5.02 TeV/nucleon Pb-Pb collisions.
freezeout hypersurface at a given temperature, and then integrate the number of conformal
partons on the surface given by
Z µ N = dΣµu ndens(ξ, Λ) (3.40)
n (Λ) √iso where ndens is the anisotropic number density and is given by ndens = 1+ξ . This num- ber would become the target for adjusting η/s and the initial momentum-space anisotropy
parameter, ξ0, i.e. we hold the final gluon number fixed as we vary the initial conditions.
However, during this work the LHC expanded their heavy-ion runs to 5.02 TeV/nucleon,
and little information was available about the initial temperatures in LHC Run Two Colli-
sions, so we made predictions based on a few simple assumptions about the QGP.
50 √sNN 0.2 TeV 2.76 TeV 5.02 TeV ξ 0 0 10 50 0 10 50 0 10 50 4πη/s 1 0.442 0.613 0.744 0.552 0.765 0.925 0.641 0.888 1.076 2 0.440 0.608 0.739 0.546 0.752 0.909 0.632 0.869 1.053 3 0.439 0.609 0.742 0.544 0.748 0.906 0.629 0.863 1.046
Table 3.1: Λ0 values (in GeV) for various collision energies, ξ0, and η/s.
We assume that the QGP created forms a rough ellipsoidal shape with end cap area A
and total initial volume V0 = τ0A. The energy of a particular collision is given by the beam
energy √sNN. However, only a certain fraction of beam energy is turned into energy in the
QGP system. This is given by E = γ√sNN, where γ is the fraction associated with QGP
formation from a heavy-ion collision. We assume that γ is constant as a function of center
of mass beam energy √sNN from 2.76 TeV/nucleon to 5.02 TeV/nucleon. We start with the
fact,
0 ENN QGP = √sNN, (3.41) E V0 ∝
0 4 and as we understand T √sNN, we can then draw conclusions that lead to the EQGP ∝ 0 ∝ following equation in which
1/8 T0 = sNN. (3.42)
We then took the initial temperature from 2.76 TeV Pb-Pb collisions for 4πη/s = 1 and