
2
In the hybrid description, the transition from the fluid
description back to the transport description occurs on
an iso-energy-density hypersurface = 30, where 0≈
145 MeV/fm3. The hypersurface is then used to sample
hadrons according to the Cooper-Frye equation [21,22]
which then continue to interact within the cascade part
of the UrQMD model, until reactions cease and kinetic
freeze-out is reached. For the hydro part we use an equa-
tion of state that contains a smooth crossover between
a hadron resonance gas and a deconfined quark-gluon-
plasma [23].
A. Thermal model
The thermal model of particle production in heavy-
ion collisions assumes that their primordial abundances
are fixed at the stage of chemical freeze-out and corre-
spond to the hadron resonance gas model in chemical
equilibrium [24,25]. The only changes to the final abun-
dances come from decay feed-down. The model param-
eters – the temperature T, baryochemical potential µB,
and the freeze-out volume V– are extracted at each col-
lision energy by fitting the experimental data. The ther-
mal model is used to describe light (anti-)(hyper-)nuclei
production by incorporating these objects as explicit de-
grees of freedom in the partition function [26,27]. Under
the assumption that chemical freeze-out of light nuclei
happens simultaneously with other hadrons1, the model
provides predictions for light nuclei abundances in cen-
tral collisions of heavy ions without introducing further
parameters. In many cases, the model shows good agree-
ment with the experiment [30,31]. Augmented with the
canonical treatment of baryon number conservation, the
model can also describe features of light nuclei produc-
tion in small systems at the LHC [32].
In the present work, we confront the predictions of
the UrQMD coalescence approach both with the ther-
mal model and experimental data. For making predic-
tions of the midrapidity yields dN/dy at various colli-
sion energies, one has to specify the thermal model pa-
rameters T,µB, and Vas a function of √sNN. To this
end, we utilize the chemical freeze-out curve of Ref. [33]
which parameterizes the collision energy dependence of
the temperature and baryochemical potential, T(√sNN)
and µB(√sNN). In principle, this parametrization is
sufficient to study the collision energy dependence of
any yield ratio since the remaining volume parameter
V(√sNN) cancels out in any such ratio. Nevertheless, it
can also be helpful to study thermal model predictions for
absolute yields, for which one has to additionally specify
the V(√sNN) dependence. We fix V(√sNN) for 0-5% cen-
tral Au-Au/Pb-Pb collisions in the following way. First,
we use the world data [34–39] on the collision energy de-
pendence of charged pion multiplicity to parameterize its
collision energy dependence from 2.4 GeV to 5.02 TeV.
We take a fit function from [40] where it was used to pa-
rameterize the energy dependence of charged multiplicity.
The fit to the pion data yields
dNπ+
dy +dNπ−
dy =a sb
NN ln(sNN)−c. (1)
Here sNN is in the units of GeV2, and the parameter val-
ues are a= 49.84903, b= 0.04110131, and c= 61.48846.
Then, at each √sNN we fix V(√sNN) to a value such
that the thermal model reproduces dNπ+
dy +dNπ
−
dy from
Eq. (1). We also check that total baryon number dNB/dy
calculated at a given energy does not exceed the number
of participants, Npart = 360. If it does, the volume is
rescaled down such that dNB/dy =Npart. This rescaling
is only necessary at very low energies, √sNN .2.8 GeV.
The effect of exact local conservation of strangeness
becomes important for strange particles, such as hyper-
nuclei, at low collision energies where the amount of
the produced strangeness is small. Here we incorporate
this effect through the strangeness-canonical ensemble,
which enforces the exact conservation of net strangeness
in a correlation volume Vc. We take a correlation radius
Rc= 2.4 fm (Vc=4π
3R3
c), as inferred from the recent
measurements of the φ/K−ratio at √sNN = 3 GeV [41].
We also use the thermal model to study the system-
size dependence of light (anti-)(hyper-)nuclei ratios at
LHC energies. Canonical suppression effects drive this
dependence. Here we use the canonical statistical model
of Ref. [32], with a constant temperature T= 155 MeV
across all multiplicities, and the canonical correlation vol-
ume of Vc= 1.6dV/dy suggested by recent measurements
of antiproton-antideuteron correlations [42].
All our thermal model calculations are performed using
the open-source Thermal-FIST package [43]. These cal-
culations optionally include feed-down from the decays
of excited nuclei, as described in [44].
B. Coalescence approach
The coalescence approach to (hyper-)nuclei production
assumes that these nuclei are produced after the kinetic
freeze-out (last scattering or decay) of their constituents
[45–61]. If the full phase space information on the nucle-
ons and hyperons at this time is known, the probability
of a pair or triplet of baryons forming a bound nucleus
can be estimated from the coalescence formula [62]
1This assumption can be relaxed to allow light nuclei production
at later stages, if one takes partial chemical equilibrium into account. In such a scenario one obtains similar results as in the
standard thermal model [28,29].