Hadron Freeze-out and QGP Hadronization

Johann Rafelski and Jean Letessier

Department of Physics, University of Arizona, Tucson, AZ 85721

and
Laboratoire de Physique Théorique et Hautes Energies
Université Paris 7, 2 place Jussieu, F–75251 Cedex 05.
arXiv:hep-ph/9902365v1 16 Feb 1999

February 16, 1999

Abstract
Abundances and m⊥ -spectra of strange and other hadronic particles emanating from central 158–
200 A GeV reactions between nuclei are found to originate from a thermally equilibrated, deconfined
source in chemical non-equilibrium. Physical freeze-out parameters and physical conditions (pressure,
specific energy, entropy, and strangeness) are evaluated. Five properties of the source we obtain are as
expected for direct hadron emission (hadronization) from a deconfined quark-gluon plasma phase.

1 Introduction
Quark-gluon plasma (QGP) is, by the meaning of these words, a thermally equilibrated state consisting
of mobile, color charged quarks and gluons. Thermal equilibrium is established faster than the particle
abundance (chemical) equilibrium and thus in general quark and gluon abundances in QGP can differ from
their equilibrium Stefan-Boltzmann limit. This in turn impacts the hadronic particle production yields, and
as we shall see, the chemical non-equilibrium is a key ingredient in the successful data analysis of experiments
performed at the CERN-SPS in the past decade. We address here results of 200 A GeV Sulphur (S) beam
interactions with laboratory stationary targets, such as Gold (Au), Tungsten (W) or Lead (Pb) nuclei and,
Pb–Pb collisions at 158 A GeV. In these interactions the nominal center of momentum (CM) available
energy is 8.6–9.2 GeV per participating nucleon.
Considerable refinement of data analysis has occurred since last comprehensive review of the field [1].
Our present work includes in particular the following:

1. We considered aside of strange also light quark (q = u, d) chemical non-equilibrium abundance [2]
and introduce along with the statistical strangeness non-equilibrium parameter γs , its light quark
analogue γq .

2. Coulomb distortion of the strange quark phase space has been understood [3].

Incorporating these developments, we accurately describe abundances of strange as well as non-strange

hadrons, both at central rapidity and in 4π-acceptance. We are thus combining in the present analysis
the strangeness diagnostic tools of dense matter with the notion of entropy enhancement in deconfined
phase [4].
As particle emerge from the QGP hadronization, not only their abundance but also their spectral shape
is of interest. Our analysis considers the impact of explosive radial flow on the spectra of particles at
high m⊥ . This contributes significant information about the fireball dynamics and the possibly deconfined
nature of the hadron source.

1
2 J. Rafelski and J. Letessier

2 The Coulomb effect in the quark fireball

The diverse statistical chemical parameters that we need to consider will in a self-explanatory way be
now introduced, considering the concept of strangeness balance: since strangeness can only be made and
destroyed by hadronic interactions in pairs, the net abundance in the hadronic fireball must vanish. We
consider a hot gas of free quarks, and evaluate the difference between strange and anti-strange quark
numbers (net strangeness). The Coulomb potential originating in the initial proton abundance distorts
slightly the Fermi particle distributions: strange quarks (negative charge) are attracted, while anti-strange
quarks repelled. Allowing for this slowly changing potential akin to the relativistic Thomas-Fermi phase
space occupancy model at finite temperature, the energy of a quark depends on both the momentum and
the Coulomb potential V : q
1
Ep = m2 + p2 − V . (1)
3
It is helpful to see here in first instance the potential V as a square well, Eq. (1) within the volume of
interest. Counting the states in the fireball we obtain:
" #
d3 rd3 p 1 1
Z
hNs − Ns̄ i = gs √ − √ . (2)
Rf (2π)3 1 + γs−1 λs−1 e( m2 +p2 − 31 V )/T −1
1 + γs λs e( m2 +p2 + 31 V )/T

In Eq. (2) the subscript Rf on the spatial integral reminds us that only the classically allowed region within
the fireball is covered in the integration over the level density, gs (= 6) is the quantum degeneracy of strange
quarks. The magnitude of the charge of strange quarks (Qs /|e| = 1/3) is shown explicitly, the Coulomb
potential refers to a negative integer probe charge.
The fugacity λs of strange quarks enters particles and antiparticles with opposite power, while the
occupancy parameter γs enters both term with same power. For γs < 1 one obtains a rather precise result
for the range of parameters of interest to us (see below) considering the Boltzmann approximation:
h V V
i
d3 r λs e 3T − λ−1 d3 pd3 x −√p2 +m2 /T
R −
s e 3T
Z
Rf
hNs − Ns̄ i ≃ gs γs e . (3)
d3 r (2π)3
R
Rf

The meaning of the different factors is now evident. γs controls overall abundance of strange quark pairs,
multiplying the usual Boltzmann thermal factor while λs controls the difference between the number of
strange and non-strange quarks. Since strangeness is produced as s, s̄-pair, the value of λs fulfills the
constraint Z
V V
h i
d3 r λs e 3T − λ−1
s e− 3T
= 0, (4)
Rf

which is satisfied exactly both in the Boltzmann limit Eq. (3), and for the exact quantum distribution
Eq. (2), when:
V
3
R
Rf d re
T
1/3
λ̃s ≡ λs λQ = 1 , λQ ≡ R 3
. (5)
Rf d r

λQ is not a fugacity that can be adjusted to satisfy a chemical condition, its value is defined by the applicable
Coulomb potential V . More generally, in order to account for the Coulomb effect, the quark fugacities within
a deconfined region should be renamed as follows in order to absorb the Coulomb potential effect:
1/3 1/3
λs → λ̃s ≡ λs λQ , λd → λ̃d ≡ λd λQ ,
q
−2/3 −1/6
λu → λ̃u ≡ λu λQ , λq → λ̃q ≡ λ̃u λ̃d = λq λQ . (6)

Since Qd = Qs = −1/3, the first line is quite evident after the above strangeness discussion, the second
follows with Qu = +2/3. Note that for a negatively charged strange quark the tilded fugacity Eq. (5)
contains a factor with positive power 1/3 but the potential that enters the quantity λQ is negative, and
thus λ̃s < λs . Because Coulomb-effect acts in opposite way on u and d quarks, its net impact on λq is smaller
Hadron Freeze-out and QGP Hadronization 3

than on λs , and it also acts in the opposite way with λ̃q > λq . To see the relevance of the tilde fugacities for
light quarks, note that in order to obtain baryon density in QGP one needs to use the tilde-quark fugacity
to account for the Coulomb potential influence on the phase space.
It is somewhat unexpected that for the Pb–Pb fireball the Coulomb effect is at all relevant. Recall that
for a uniform charge distribution within a radius Rf of charge Zf :
 "   #
 3 Zf e2 1 r 2
 − 1− , for r < Rf ;


V = 2 Rf 3 Rf (7)
Z e2
 − f ,

for r > Rf .


r
Choosing Rf = 8 fm, T = 140 MeV, ms = 200 MeV (value of γs is practically irrelevant) we find as solution
of Eq. (2) for hNs − Ns̄ i = 0 for Zf = 150 the value λs = 1.10 (precisely: 1.0983; λs = 1.10 corresponds
to Rf = 7.87 fm). We will see that both this values within the experimental precision arise from study of
particle abundances. For the S–W/Pb reactions this Coulomb effect is practically negligible.
In the past we (and others) have disregarded in the description of the hadronic final state abundances
the electrical charges and interactions of the produced hadrons. This is a correct way to analyze the
chemical properties since, as already mentioned, the quantity λQ is not a new fugacity: conservation of
flavor already assures charge conservation in chemical hadronic reactions, and use of λi , i = u, d, s exhausts
all available chemical balance conditions for the abundances of hadronic particles. As shown here, the
Coulomb deformation of the phase space in the QGP fireball makes it necessary to rethink the implications
that the final state particle measured fugacities have on the fireball properties.

3 Freeze-out of hadrons
The production of hadrons from a QGP fireball occurs mainly by way of quark coalescence and gluon
fragmentation, and there can be some quark fragmentation as well. We will explicitly consider the recom-
bination of quarks, but implicitly the gluon fragmentation is accounted for by our allowance for chemical
nonequilibrium. The relative number of primary particles freezing out from a source is obtained noting
that the fugacity and phase space occupancy of a composite hadronic particle can be expressed by its
constituents and that the probability to find all j-components contained within the i-th emitted particle is:

γj λj e−Ej /T ,
Y Y Y
Ni ∝ λi = λj , γi = γj . (8)
j∈i j∈i j∈i

Experimental data with full phase space coverage, or central rapidity region |y − yCM | < 0.5, for m⊥ >
1.5 GeV are considered; recall that the energy of a hadron ‘i’ is expressed by the spectral parameters m⊥
and y as follows, q q
X
Ei = Ej , Ei = m2i + p2 = m2i + p2⊥ cosh(y − yCM ) ,
where yCM is the center of momentum rapidity of the fireball formed by the colliding nuclei.
The yield of particles is controlled by the freeze-out temperature Tf . This freeze-out temperature is
different from the m⊥ -spectral temperature T⊥ , which also comprises the effect of collective matter flows
originating in the explosive disintegration driven by the internal pressure of compressed hadronic matter. In
order to model the flow and freeze-out of the fireball surface, one in general needs several new implicit and/or
explicit parameters. We therefore will make an effort to choose experimental variables (compatible particle
ratios) which are largely flow independent. This approach also diminishes the influence of heavy resonance
population — we include in Eq.(8) hadronic states up to M = 2 GeV, and also include quantum statistical
corrections, allowing for first Bose and Fermi distribution corrections in the phase space content. It is hard
to check if indeed we succeeded in eliminating the uncertainty about high mass hadron populations. As
we shall see comparing descriptions which exclude flow with those that include it, our approach is indeed
largely flow-insensitive.
4 J. Rafelski and J. Letessier

We consider here a simple radial flow model, with freeze-out in CM frame at constant laboratory time,
implying that causally disconnected domains of the dense matter fireball are synchronized at the instant
of the collision. Within this approach [5], the spectra and thus also multiplicities of particles emitted are
obtained replacing the Boltzmann exponential factor in Eq.(8),

1 γv Ej
Z
−Ej /T (1+~
vc ·~
pj /Ej )
e → dΩv γv (1 + ~vc · p~j /Ej )e− T , (9)
2π
p
where as usual γv = 1/ 1 − ~vc2 . Eq. (9) can be intuitively obtained by a Lorentz transformation between an
observer on the surface of the fireball, and one at rest in the general CM (laboratory) frame. One common
feature of all flow scenarios is that, at sufficiently high m⊥ , the spectral temperature (inverse slope) T⊥ can
be derived from the freeze-out temperature Tf with the help of the Doppler formula:

T⊥ = Tf γv (1 + vc ) . (10)

In actual numerical work, we proceed as follows to account for the Doppler effect: for a given pair of
values Tf and vc , the resulting m⊥ particle spectrum is obtained and analyzed using the spectral shape and
procedure employed for the particular collision system by the experimental groups, yielding the theoretical
inverse slope ‘temperature’ T⊥j .
Once the parameters Tf , λq , λs , γq , γs and vc have been determined studying available particle yields,
and m⊥ slopes, the entire phase space of particles produced is fully characterized within our elaborate
statistical model. Our model is in fact just an elaboration of the original Fermi model [6], in fact all
we do is to allow hadronic particles to be produced in the manner dictated by the phase space size of
valance quarks. With the full understanding of the phase space of all hadrons, we can evaluate the physical
properties of the system at freeze-out, such as, e.g., energy and entropy per baryon, strangeness content.

4 Results of data analysis

As noted our analysis requires that we form particle abundance ratios between what we call compatible
hadrons. We considered for S–Au/W/Pb reactions 18 data points listed in table 1 (of which three comprise
results with Ω’s). For Pb–Pb we address here 15 presently available particle yield ratios listed in table 2
(of which four comprise the Ω’s). We believe to have included in our discussion most if not all particle
multiplicity results available presently.
The theoretical particle yield results shown columns in tables 1 and 2 are obtained looking for a set of
physical parameters which will minimize the difference between theory and experiment. The resulting total
error of the ratios R is shown at the bottom of tables 1 and 2:
j !2
Rth j
− Rexp
χ2T
X
= j . (11)
j ∆Rexp

It is a non-trivial matter to determine the confidence level that goes with the different data analysis
approaches since some of the results considered are partially redundant, and a few data points can be
obtained from others by algebraic relations arising in terms of their theoretical definitions; there are two
types of such relations:
Ω+Ω Ω 1 + Ω/Ω Λ Λ Ξ Ξ
= · , = · · . (12)
Ξ + Ξ̄ Ξ 1 + Ξ/Ξ Λ Ξ Ξ Λ
However, due to smallness of the total error found for some of the approaches it is clear without detailed
analysis that only these are statistically significant.
In addition to the abundance data, we also explored the transverse mass m⊥ -spectra when the collective
flow velocity was allowed in the description, and the bottom line of tables 1 and 2 in columns Dv , Fv
includes in these cases the error found in the inverse slope parameter of the spectra. The procedure we
used is as follows: since within the error the high m⊥ strange (anti)baryon inverse slopes are within error,
Hadron Freeze-out and QGP Hadronization 5

Table 1: Particle ratios studied in our analysis for S–W/Pb/Au reactions: experimental results with
references and kinematic cuts are given, followed by columns showing results for the different strategies of
analysis B–F. Asterisk ∗ means a predicted result (corresponding data is not fitted). Subscript s implies
forced strangeness conservation, subscript v implies inclusion of collective flow. The experimental results
considered are from:
1 S. Abatzis et al., WA85 Collaboration, Heavy Ion Physics 4, 79 (1996).
2 S. Abatzis et al., WA85 Collaboration, Phys. Lett. B 347, 158 (1995).
3 S. Abatzis et al., WA85 Collaboration, Phys. Lett. B 376, 251 (1996).
4 I.G. Bearden et al., NA44 Collaboration, Phys. Rev. C 57, 837 (1998).
5 D. Röhrich for the NA35 Collaboration, Heavy Ion Physics 4, 71 (1996).
6 S–Ag value adopted here: T. Alber et al., NA35 Collaboration, Eur. Phys. J. C 2, 643 (1998).
7 A. Iyono et al., EMU05 Collaboration, Nucl. Phys. A 544, 455c (1992) and Y. Takahashi et al., EMU05

Collaboration, private communication.

Ratios Ref. Cuts [GeV] Exp.Data B C D Ds F Dv Fv

Ξ/Λ 1 1.2 < p⊥ < 3 0.097 ± 0.006 0.16 0.11 0.099 0.11 0.10 0.11 0.11
Ξ/Λ̄ 1 1.2 < p⊥ < 3 0.23 ± 0.02 0.38 0.23 0.22 0.18 0.22 0.23 0.22
Λ̄/Λ 1 1.2 < p⊥ < 3 0.196 ± 0.011 0.20 0.20 0.203 0.20 0.20 0.20 0.20
Ξ/Ξ 1 1.2 < p⊥ < 3 0.47 ± 0.06 0.48 0.44 0.45 0.33 0.44 0.44 0.43
Ω/Ω 2 p⊥ > 1.6 0.57 ± 0.41 1.18∗ 0.96∗ 1.01∗ 0.55∗ 0.98 1.09∗ 1.05∗
Ω+Ω
Ξ+Ξ̄
2 p⊥ > 1.6 0.80 ± 0.40 0.27∗ 0.17∗ 0.16∗ 0.16∗ 0.16 0.13∗ 0.13∗
+
K /K − 1 p⊥ > 0.9 1.67 ± 0.15 2.06 1.78 1.82 1.43 1.80 1.77 1.75
Ks0 /Λ 3 p⊥ > 1 1.43 ± 0.10 1.56 1.64 1.41 1.25 1.41 1.38 1.39
Ks0 /Λ̄ 3 p⊥ > 1 6.45 ± 0.61 7.79 8.02 6.96 6.18 6.96 6.81 6.86
Ks0 /Λ 1 m⊥ > 1.9 0.22 ± 0.02 0.26 0.28 0.24 0.24 0.24 0.24 0.24
Ks0 /Λ̄ 1 m⊥ > 1.9 0.87 ± 0.09 1.30 1.38 1.15 1.20 1.16 1.18 1.17
Ξ/Λ 1 m⊥ > 1.9 0.17 ± 0.01 0.27 0.18 0.17 0.18 0.17 0.16 0.17
Ξ/Λ̄ 1 m⊥ > 1.9 0.38 ± 0.04 0.64 0.38 0.38 0.30 0.37 0.35 0.35
Ω+Ω
Ξ+Ξ̄
1 m⊥ > 2.3 1.7 ± 0.9 0.98∗ 0.59∗ 0.58∗ 0.52∗ 0.58 0.72∗ 0.75∗
p/p̄ 4 Mid-rapidity 11 ± 2 11.2 10.1 10.6 7.96 10.5 10.6 10.4
Λ̄/p̄ 5 4π 1.2 ± 0.3 2.50 1.47 1.44 1.15 1.43 1.58 1.66
h−
p−p̄ 6 4π 4.3 ± 0.3 4.4 4.2 4.1 3.6 4.1 4.2 4.2
h+ −h−
h+ +h− 7 4π 0.124 ± 0.014 0.11 0.10 0.103 0.09 0.10 0.12 0.12
χ2T 264 30 6.5 38 12 6.2 11

overlapping we decided to consider just one ‘mean’ experimental value T⊥ = 235±10 for S–induced reactions
and T⊥ = 265 ± 15 for Pb–induced reactions. Thus we add one experimental value and one parameter,
without changing the number of degrees of freedom. Once we find values of Tf and vc , we evaluate the
j
slopes of the theoretical spectra. The resulting theoretical Tth values are in remarkable agreement with
j
experimental T⊥ , well beyond what we expected, as is shown in table 3. An exception is the fully strange
Ω + Ω spectrum. We note in passing that when vc was introduced we found little additional correlation
between now 6 theoretical parameters. The collective flow velocity is a truly new degree of freedom and it
helps to attain a more consistent description of the experimental data available.
Although it is clear that one should be using a full-fledged model such as Dv , we address also cases B and
C. The reason for this arises from our desire to demonstrate the empirical need for chemical non-equilibrium:
in the approach B, complete chemical equilibrium γi = 1 is assumed. As we see in tables 1 and 2 this
approach has rather large error. Despite this the results in column B in tables 1 and 2 are often compared
favorably with experiment, indeed this result can be presented quite convincingly on a logarithmic scale. Yet
as we see the disagreement between theory and experiment is quite forbidding. With this remark we wish to
demonstrate the need for comprehensive and precisely measured hadron abundance data sample, including
abundances of multi-strange antibaryons, which were already 20 years ago identified as the hadronic signals
6 J. Rafelski and J. Letessier

Table 2: Particle ratios studied in our analysis for Pb–Pb reactions: experimental results with references
and kinematic cuts are given, followed by columns showing results for the different strategies of analysis
B–F. Asterisk ∗ means a predicted result (corresponding data is not fitted or not available). Subscript s
implies forced strangeness conservation, subscript v implies inclusion of collective flow. The experimental
results considered are from:
1 I. Králik, for the WA97 Collaboration, Nucl. Phys. A 638,115, (1998).
2 G.J. Odyniec, for the NA49 Collaboration, J. Phys. G 23, 1827 (1997).
3 P.G. Jones, for the NA49 Collaboration, Nucl. Phys. A 610, 188c (1996).
4 F. Pühlhofer, for the NA49 Collaboration, Nucl. Phys. A 638, 431,(1998).
5 C. Bormann, for the NA49 Collaboration, J. Phys. G 23, 1817 (1997).
6 G.J. Odyniec, Nucl. Phys. A 638, 135, (1998).
7 D. Röhrig, for the NA49 Collaboration, “Recent results from NA49 experiment on Pb–Pb collisions at

158 A GeV”, see Fig. 4, in proc. of EPS-HEP Conference, Jerusalem, Aug. 19-26, 1997.
8 A.K. Holme, for the WA97 Collaboration, J. Phys. G 23, 1851 (1997).

Ratios Ref. Cuts [GeV] Exp.Data B C D Ds F Dv Fv

Ξ/Λ 1 p⊥ > 0.7 0.099 ± 0.008 0.138 0.093 0.095 0.098 0.107 0.102 0.110
Ξ/Λ̄ 1 p⊥ > 0.7 0.203 ± 0.024 0.322 0.198 0.206 0.215 0.216 0.210 0.195
Λ̄/Λ 1 p⊥ > 0.7 0.124 ± 0.013 0.100 0.121 0.120 0.119 0.121 0.123 0.128
Ξ/Ξ 1 p⊥ > 0.7 0.255 ± 0.025 0.232 0.258 0.260 0.263 0.246 0.252 0.225
(Ξ+Ξ̄)
(Λ+Λ̄)
2 p⊥ > 1. 0.13 ± 0.03 0.169 0.114 0.118 0.122 0.120 0.123 0.121
Ks0 /φ 3,4 11.9 ± 1.5 6.3 10.4 9.89 9.69 16.1 12.9 15.1
+ −
K /K 5 1.80 ± 0.10 1.96 1.75 1.76 1.73 1.62 1.87 1.56
p/p̄ 6 18.1 ±4. 22.0 17.1 17.3 17.9 16.7 17.4 15.3
Λ̄/p̄ 7 3. ± 1. 3.02 2.91 2.68 3.45 0.65 2.02 1.29
Ks0 /B 3 0.183 ± 0.027 0.305 0.224 0.194 0.167 0.242 0.201 0.281
h− /B 3 1.83 ± 0.2 1.47 1.59 1.80 1.86 1.27 1.83 1.55
Ω/Ξ 1 p⊥ > 0.7 0.192 ± 0.024 0.119∗ 0.080∗ 0.078∗ 0.080∗ 0.192 0.077∗ 0.190
Ω/Ξ 8 p⊥ > 0.7 0.27 ± 0.06 0.28∗ 0.17∗ 0.17∗ 0.18∗ 0.40 0.18∗ 0.40
Ω/Ω 1 p⊥ > 0.7 0.38 ± 0.10 0.55∗ 0.56∗ 0.57∗ 0.59∗ 0.51 0.60∗ 0.47
(Ω+Ω)
(Ξ+Ξ̄)
8 p⊥ > 0.7 0.20 ± 0.03 0.15∗ 0.10∗ 0.10∗ 0.10∗ 0.23 0.10∗ 0.23
χ2T 88 24 1.6 2.7 19 1.5 18

Table 3: Particle spectra inverse slopes: theoretical values Tth are obtained imitating the experimental
procedure from the Tf , vc -parameters. Top portion: S–W experimental T⊥ from experiment WA85 for
kaons, lambdas and cascades; bottom portion: experimental Pb–Pb T⊥ from experiment NA49 for kaons
and from experiment WA97 for baryons. The experimental results are from:
D. Evans, for the WA85-collaboration, APH N.S., Heavy Ion Physics 4, 79 (1996).
E. Andersen et al., WA97-collaboration, Phys. Lett. B 433, 209, (1998).
S. Margetis, for the NA49-collaboration, J. Physics G, Nucl. and Part. Phys. 25, 189 (1999).
0
TK TΛ TΛ TΞ TΞ T Ω+Ω
T⊥ [MeV] 219 ± 5 233 ± 3 232 ± 7 244 ± 12 238 ± 16 —
Tth [MeV] 215 236 236 246 246 260
T⊥ [MeV] 223 ± 13 291 ± 18 280 ± 20 289 ± 12 269 ± 22 237 ± 24
Tth [MeV] 241 280 280 298 298 335

of QGP phenomena [7]. The strange antibaryon enhancement reported by the experiment WA97 fully
confirms the role played by these particles [8].
In the approach C, we introduce strangeness chemical non-equilibrium [9], i.e., we also vary γs , keeping
Hadron Freeze-out and QGP Hadronization 7

Table 4: Statistical parameters which best describe the experimental S–Au/W/Pb results shown in table 1.
Asterisk (∗ ) means a fixed (input) value or result of a constraint. In approaches B to D, particle abundance
ratios comprising Ω are not considered. In case Ds strangeness conservation in the particle yields was
enforced. In case F the three data-points with Ω are considered. Lower index v implies that radial collective
flow velocity has been considered.
S–W Tf [MeV] λq λs γs /γq γq vc χ2T
B 144 ± 2 1.53 ± 0.02 0.97 ± 0.02 1∗ 1∗ 0∗ 264
C 147 ± 2 1.49 ± 0.02 1.01 ± 0.02 0.62 ± 0.02 1∗ 0∗ 30
D 143 ± 3 1.50 ± 0.02 1.00 ± 0.02 0.60 ± 0.02 1.22 ± 0.06 0∗ 6.5
Ds 153 ± 3 1.42 ± 0.02 1.10∗ ± 0.02 0.56 ± 0.02 1.26 ± 0.06 0∗ 38
F 144 ± 3 1.49 ± 0.02 1.00 ± 0.02 0.60 ± 0.02 1.22 ± 0.06 0∗ 12
Dv 144 ± 2 1.51 ± 0.02 1.00 ± 0.02 0.69 ± 0.03 1.41 ± 0.08 0.49 ± 0.02 6.2
Fv 145± 2 1.50 ± 0.02 0.99 ± 0.02 0.69 ± 0.03 1.43 ± 0.08 0.50 ± 0.02 11

Table 5: Statistical parameters which best describe the experimental Pb–Pb results shown in table 2 .
Asterisk (∗ ) means a fixed (input) value, or result of a constraint. In approaches B to D, particle abundance
ratios comprising Ω are not considered. In case Ds strangeness conservation in the particle yields was
enforced. In case F the four data-points with Ω are considered. Lower index v implies that radial flow
velocity has been considered.
Pb–Pb Tf [M eV ] λq λs γs /γq γq vc χ2T
B 142 ± 3 1.70 ± 0.03 1.10 ± 0.02 1∗ 1∗ 0∗ 88
C 144 ± 4 1.62 ± 0.03 1.10 ± 0.02 0.63 ± 0.04 1∗ 0∗ 24
D 134 ± 3 1.62 ± 0.03 1.10 ± 0.02 0.69 ± 0.08 1.84 ± 0.30 0∗ 1.6
Ds 133 ± 3 1.63 ± 0.03 1.09∗ ± 0.02 0.72 ± 0.12 2.75 ± 0.35 0∗ 2.7
F 334 ± 18 1.61 ± 0.03 1.12 ± 0.02 0.50 ± 0.01 0.18 ± 0.02 0∗ 19
Dv 144 ± 2 1.60 ± 0.02 1.10 ± 0.02 0.86 ± 0.03 1.72 ± 0.08 0.58 ± 0.02 1.5
Fv 328 ± 17 1.59 ± 0.03 1.13 ± 0.02 0.51 ± 0.03 0.34 ± 0.13 0.38 ± 0.31 18

γq = 1. A nearly valid experimental data description is now possible, and indeed, when the error bars were
smaller this was a satisfactory approach adapted in many studies [1]. However, only the possibility of light
quark non-equilibrium in fit D produces a statistically significant data description.
It is interesting to note that a significant degradation of χ2T occurs, especially in the Pb–Pb data, when
we require in column F that the particle ratios comprising Ω and Ω-particles are also described. We thus
conclude that a large fraction of these particles must be made in processes that are not considered in the
present model.
Another notable study case is shown in column Ds , with strangeness conservation enforced. Remarkably,
the S–W data, table 1, do not like this constraint. A possible explanation is that for S-induced reactions, the
particle abundances are obtained at relatively high p⊥ . Thus only a small fraction of all strange particles is
actually observed, and therefore the overall strangeness is hard to balance. Similar conclusion results also
when radial flow is explicitly allowed for, with a significant unbalanced strangeness fractions remaining,
as we shall discuss below. On the other hand, this constraint is relatively easily satisfied for the Pb–Pb
collision results, table 2, where a much greater proportion of all strange particles is actually experimentally
detected.
The statistical parameters associated with the particle abundances described above are shown in the
table 4 for S-W/Pb and in table 5 for Pb–Pb reactions. The errors of the statistical parameters shown are
those provided by the program MINUIT96.03 from CERN program library. When the theory describes the
8 J. Rafelski and J. Letessier

Table 6: Tf and physical properties (specific energy, entropy, anti-strangeness, net strangeness, pressure
and volume) of the full hadron phase space characterized by the statistical parameters given in table 4 for
the reactions S–Au/W/Pb. Asterisk ∗ means fixed input.

S–W Tf [MeV] Ef /B Sf /B s̄f /B (s̄f − sf )/B Pf [GeV/fm3 ]

B 144 ± 2 8.9 ± 0.5 50± 3 1.66 ± 0.06 0.44 ± 0.02 0.056 ± 0.005
C 147 ± 2 9.3 ± 0.5 49± 3 1.05 ± 0.05 0.23 ± 0.02 0.059 ± 0.005
D 143 ± 3 9.1 ± 0.5 48± 3 0.91 ± 0.04 0.20 ± 0.02 0.082 ± 0.006
Ds 153 ± 3 8.9 ± 0.5 45 ± 3 0.76 ± 0.04 0∗ 0.133 ± 0.008
F 144 ± 2 9.1 ± 0.5 48 ± 3 0.91 ± 0.05 0.20 ± 0.02 0.082 ± 0.006
Dv 144 ± 2 8.2 ± 0.5 44 ± 3 0.72 ± 0.04 0.18 ± 0.02 0.124 ± 0.007
Fv 145 ± 2 8.2 ± 0.5 44 ± 3 0.73 ± 0.05 0.17 ± 0.02 0.123 ± 0.007

data well, this is a one standard deviation error in theoretical parameters arising from the experimental
measurement error. In tables 4 and 5 in cases in which γq 6= 1 we present the ratio γs /γq , which corresponds
approximately to the parameter γs in the data studies in which γq = 1 has been assumed. It is notable that
whenever we allow phase space occupancy to vary from the equilibrium, a significant deviation is found. In
the S–W case, table 4, there is a 25% excess in the light quark occupancy, while strange quarks are 25%
below equilibrium. In Pb–Pb case, table 5, the ratio of the nonequilibrium parameters γs /γq ≃ 0.7 also
varies little (excluding the failing cases F with Ω, Ω data), though the individual values γs , γq can change
significantly, even between the high confidence cases.
We note that in the Pb–Pb reaction, table 5, γs > 1. This is an important finding, since an explanation
of this effect involves formation prior to freeze-out in the matter at high density of near chemical equilibrium,
γs (t <√tf ) ≃ 1. The ongoing rapid expansion (note that the collective velocity at freeze-out is found to
be 1/ 3) preserves this high strangeness yield, and thus we find the result γs > 1. In other words the
strangeness production freeze-out temperature Ts > Tf . Thus the strangeness equilibration time is proven
implicitly to be of magnitude expected in earlier studies of the QGP processes [10]. It is hard, if not really
impossible, to arrive at this result without the QGP hypothesis. Moreover, inspecting figure 38 in [10] we
see that the yield of strangeness we expect from the kinetic theory in QGP is at the level of 0.75 per baryon,
the level we indeed will determine below.
Another notable results is λ̃s ≃ λs ≃ 1.0 in the S–Au/W/Pb case, see table 4, and λs ≃ 1.1 in the Pb–Pb
case, see table 5, implying here λ̃s = 1, see section 2. We see clearly for both S- and Pb-induced reactions
a value of λs , characteristic for a source of freely movable strange quarks with balancing strangeness.

5 Physical properties of the fireball at chemical freeze-out

Given the precise statistical information about the properties of the hadron phase space, we can determine
the physical properties of the hadronic particles, see tables 6 and 7. We show for the same study cases
B–F, along with their temperature, the specific energy and entropy content per baryon, and specific anti-
strangeness content, along with specific strangeness asymmetry, and finally pressure at freeze-out. We note
that it is improper in general to refer to these properties as those of a ‘hadronic gas’ formed in nuclear
collisions, as the particles considered may be emitted in sequence from a deconfined source, and thus there
may never be a evolution stage corresponding to a hadron gas phase. However, the properties presented
are those carried away by the emitted particles, and thus characterize the properties of their source.
The energy per baryon seen in the emitted hadrons is, within error, equal to the available specific energy
of the collision (8.6 GeV for Pb–Pb and about 8.8–8.9 GeV for S–Au/W/Pb). This implies that the fraction
of energy left in the central fireball must be the same as the fraction of baryon number. We further note
that hadrons emitted at freeze-out carry away a specific s̄ content which is determined to be 0.72 ± 0.04
in S–Au/W/Pb case and 0.78 ± 0.04 for the Pb–Pb collisions (cases Dv ). Here we see the most significant
Hadron Freeze-out and QGP Hadronization 9

Table 7: Tf and physical properties (specific energy, entropy, anti-strangeness, net strangeness, pressure
and volume) of the full hadron phase space characterized by the statistical parameters given in table 5 for
the reactions Pb–Pb. Asterisk ∗ means fixed input.
Pb–Pb Tf [MeV] Ef /B Sf /B s̄f /B (s̄f − sf )/B Pf [GeV/fm3 ]
B 142 ± 3 7.1 ± 0.5 41 ± 3 1.02 ± 0.05 0.21 ± 0.02 0.053 ± 0.005
C 144 ± 4 7.7 ± 0.5 42 ± 3 0.70 ± 0.05 0.14 ± 0.02 0.053 ± 0.005
D 134 ± 3 8.3 ± 0.5 47 ± 3 0.61 ± 0.04 0.08 ± 0.01 0.185 ± 0.012
Ds 133 ± 3 8.7 ± 0.5 48 ± 3 0.51 ± 0.04 0∗ 0.687 ± 0.030
F 334 ± 18 9.8 ± 0.5 24 ± 2 0.78 ± 0.05 0.06 ± 0.01 1.64 ± 0.06
Dv 144 ± 2 7.0 ± 0.5 38 ± 3 0.78 ± 0.04 0.01 ± 0.01 0.247 ± 0.007
Fv 328 ± 17 11.2 ± 1.5 28 ± 3 0.90 ± 0.05 0.09 ± 0.02 1.40 ± 0.06

impact of flow, as without it the specific strangeness content seemed to diminish as we moved to the larger
collision system. We have already alluded repeatedly to the fact that for S–Au/W/Pb case the balance of
strangeness is not seen in the particles observed experimentally. The asymmetry is 18%, with the excess
emission of s̄ containing hadrons at high p⊥ > 1 GeV. In the Pb–Pb data this effect disappears, perhaps
since the p⊥ lower cut-off is smaller. One could also imagine that longitudinal flow which is stronger in
S–Au/W/Pb is responsible for this effect.
The small reduction of the specific entropy in Pb–Pb compared to the lighter S–Au/W/Pb is driven
by the greater baryon stopping in the larger system, also seen in the smaller energy per baryon content.
Both systems freeze out at E/S = 0.185 GeV (energy per unit of entropy). Aside of Tf , this is a second
universality feature at hadronization of both systems. The overall high specific entropy content agrees well
with the entropy content evaluation made earlier [4] for the S–W case. This is so because the strange
particle data are indeed fully consistent within the chemical-nonequilibrium description with the 4π total
particle multiplicity results.

6 Current status and conclusions

We have presented detailed analysis of hadron abundances observed in central S–Au/W/Pb 200 A GeV
and Pb–Pb 158 A GeV interactions within thermal equilibrium and chemical non-equilibrium phase space
model of strange and non-strange hadronic particles. In the analysis of the freeze-out the structure of the
particle source was irrelevant. However, the results that we found for the statistical parameters point rather
clearly towards a deconfined QGP source, with quark abundances near but not at chemical equilibrium:
statistical parameters obtained characterize a strange particle source which, both for S–Au/W/Pb and for
Pb–Pb case, when allowing for Coulomb deformation of the strange and anti-strange quarks, is exactly
symmetric between s and s̄ quark carriers, as is natural for a deconfined state.
There are similarities in freeze-out properties seen comparing results presented in tables 6 and 7 which
suggest universality in the extensive physical properties of the two freeze-out systems [3]. Despite consider-
ably varying statistical parameters we obtain similar physical conditions such as Tf and E/S corresponding
possibly to the common physical properties of QGP at its breakup into hadrons. The precision of the
data description we have reached, see tables 1–3 strongly suggests that despite its simplicity the model
we developed to analyze experimental data provides a reliable image of the hadron production processes.
When flow is allowed for the freeze-out temperature is identical in both physical systems considered, even
though, e.g., the baryochemical potential µB = 3T ln λq and other physical parameters found are slightly
different. It is worth to restate the values we obtained, case Dv (6 parameter description, no Ω, Ω):

Tf = 144 ± 2 MeV, µB = 178 ± 5MeV , for S–Pb/W ,

Tf = 144 ± 2 MeV, µB = 203 ± 5MeV , for Pb–Pb ,

10 J. Rafelski and J. Letessier

Even though there is still considerable uncertainty about other freeze-out flow effects, such as longi-
tudinal flow (memory of the collision axis), the level of consistency and quality of agreement between a
wide range of experimental data and our chemical non-equilibrium, thermal equilibrium statistical model
suggests that, for the observables considered, these effects do not matter.
The key results we found analyzing experimental data are:

1. λ̃s = 1 for S and Pb collisions ;

2. γsPb > 1, γq > 1 ;

√
3. vcPb = 1/ 3

4. S/B ≃ 40 ;

5. s̄/B ≃ 0.75 ;

are in remarkable agreement with the properties of a deconfined QGP source hadronizing without chemical
reequilibration. The only natural interpretation of our findings is thus that hadronic particles seen at 158–
200 GeV A nuclear collisions at CERN-SPS are emerging directly from a deconfined QGP state and do not
undergo a chemical re-equilibration after they have been produced.

Acknowledgments:
This work was supported in part by a grant from the U.S. Department of Energy, DE-FG03-95ER40937 .
LPTHE, Univ. Paris 6 et 7 is: Unité mixte de Recherche du CNRS, UMR7589.

References
[1] J. Sollfrank, J. Phys. G 23, 1903 (1997); [nucl-th/9707020], and references therein.
[2] J. Letessier and J. Rafelski, Phys. Rev. C 59, 947 1999; [hep-ph/9806386] .
[3] J. Letessier and J. Rafelski, Acta Phys. Pol.; B30, 153 (1999), [hep-ph/9807346] , and J. Phys. Part. Nuc. G25,
295, (1999), [hep-ph/9810332] .
[4] J. Rafelski, J. Letessier and A. Tounsi, Dallas–ICHEP (1992) p. 983 (QCD161:H51:1992); [hep-ph/9711350];
J. Letessier, A. Tounsi, U. Heinz, J. Sollfrank and J. Rafelski Phys. Rev. Lett. 70, 3530 (1993); and Phys. Rev.
D 51, 3408 (1995).
[5] E. Schnedermann, J. Sollfrank and U. Heinz, pp175–206 in Particle Production in Highly Excited Matter, NATO-
ASI Series B303, H.H. Gutbrod and J. Rafelski, Eds., (Plenum, New York, 1993)
[6] E. Fermi, Progr. theor. Phys. 5 570 (1950).
[7] J. Rafelski, pp 282–324, GSI Report 81-6, Darmstadt, May 1981; Proceedings of the Workshop on Future
Relativistic Heavy Ion Experiments, held at GSI, Darmstadt, Germany, October 7–10, 1980, R. Bock and R.
Stock, Eds., (see in particular section 6, pp 316–320); see also:
J. Rafelski, pp 619–632 in New Flavor and Hadron Spectroscopy, Ed. J. Tran Thanh Van (Editions Frontiers
1981), Proceedings of XVIth Rencontre de Moriond — Second Session, Les Arcs, March 21–27, 1981;
J. Rafelski, Nucl. Physics A 374, 489c (1982) — Proceedings of ICHEPNC held 6–10 July 1981 in Versailles,
France.
[8] E. Andersen et al., WA97-collaboration, Phys. Lett. B 433, 209, (1998);
Strangeness enhancement at mid-rapidity in Pb-Pb collisions at 158 A GeV/c, E. Andersen et al., WA97-
collaboration, Phys. Lett. B in press, CERN-EP preprint, January 5, 1999.
[9] J. Rafelski, Phys. Lett. B 262, 333 (1991); Nucl. Phys. A 544, 279c (1992).
[10] J. Rafelski, J. Letessier and A. Tounsi, Acta Phys. Pol. B 27, 1035 (1996), and references therein.
[11] J. Letessier, A. Tounsi and J. Rafelski, Phys. Rev. C 50, 406 (1994);
J. Rafelski, J. Letessier and A. Tounsi, Acta Phys. Pol. A 85, 699 (1994).