Impact of the radial profile of atomic nuclei on observables in high-energy collisions

Abstract

In heavy-ion phenomenology, the nucleon density distribution in colliding nuclei is commonly described by a two-parameter Woods-Saxon (WS) distribution. However, this approach omits the detailed radial structure in the density distribution that arise from quantal filling patterns of neutrons and protons. These fine structures, as estimated by the Skyrme-Hartree-Fock density functional, cause small deviations in heavy-ion observables from the WS baseline, which cannot be captured by simply readjusting the WS parameters. These deviations are dependent on centrality and observable but often exhibit similar shapes for different nuclei. Such fine structures may introduce up to a 25% uncertainty in the measured differences in heavy-ion observables between the ⁹⁶Ru+⁹⁶Ru and ⁹⁶Zr+⁹⁶Zr mid-central collisions from the STAR Collaboration.

High-energy nuclear collisions at RHIC and LHC generate a hot and dense quark-gluon plasma (QGP), whose space-time expansion is governed by relativistic viscous hydrodynamic equations Heinz and Snellings (2013). These collisions, due to their extremely short timescales, effectively capture a snapshot of the nucleon spatial distribution in the nuclei, forming the initial condition of the QGP. This initial condition, through hydrodynamic expansion, imprints patterns on the particle momentum distributions observed in detectors Ollitrault (2023). Although this high-energy imaging concept is theoretically powerful, it was previously deemed impractical for determining nuclear structure. However, recent developments show that this can be achieved by comparing nuclei with similar masses but different internal structures Giacalone et al. (2021a). By constructing ratios of bulk observables between collisions of such nuclei, the effects of hydrodynamic response are canceled out, isolating the global shape and size features of the colliding nuclei Zhang et al. (2022). For instance, the STAR experiment successfully extracted the shape of ²³⁸U STAR Collaboration (2024), matching extractions based on low-energy spectroscopic measurements. Other studies of isobar or isobar-like systems of varying sizes also demonstrate the sensitivity of these ratios to the global characteristics of the nuclei involved Xu et al. (2018); Giacalone et al. (2021b); Xu et al. (2023); Aad et al. (2023); Liu et al. (2024); Giacalone et al. (2024).

In high-energy collisions, nuclear structure influences the geometric features of the QGP’s initial condition, such as ellipticity $\varepsilon_{2}$, triangularity $\varepsilon_{3}$, and the number of participating nucleons $N_{\mathrm{part}}$. These geometric feasures are linearly related to observables describing collective behavior, like elliptic flow $v_{2}$, triangular flow $v_{3}$, and charged particle multiplicity $N_{\mathrm{ch}}$ Boźek and Broniowski (2012); Niemi et al. (2013). Since isobar nuclei share the same mass number, deviations from unity in their ratios must arise from differences in nuclear structure, making them power tools for imaging nuclear structure.

In current high-energy phenomenology, the global structure of nuclei is often parameterized by the deformed Woods-Saxon (WS) function,

$$\rho(r,\theta,\phi))\propto[1+\exp(r-R_{0}(1+f(\theta,\phi))/a)]^{-1},$$ (1)

where $R_{0}$ and $a$ represent the half-density radius and skin thickness, respectively. The function $f(\theta,\phi)$ accounts for nuclear deformations by modulating the nuclear surface. For example, for axial-symmetric quadrupole and octupole deformation, $f(\theta,\phi)=\beta_{2}Y_{2,0}(\theta,\phi)+\beta_{3}Y_{3,0}(\theta,\phi)$. The ratios of observables between two isobar systems $\mathcal{O}=v_{n}$ and $N_{\mathrm{ch}}$ or $\varepsilon_{n}$ and $N_{\mathrm{part}}$, follow a scaling relation as a function of centrality Jia and Zhang (2023):

$$R_{\mathcal{O}}\equiv\frac{\mathcal{O}_{\mathrm{\mathrm{X}}}}{\mathcal{O}_{\mathrm{\mathrm{Y}}}}\approx 1+c_{1}\Delta R_{0}+c_{2}\Delta a+c_{3}\Delta\beta_{2}^{2}+c_{4}\Delta\beta_{3}^{2}\;,$$ (2)

where $\Delta R_{0}=R_{0,\mathrm{\mathrm{X}}}-R_{0,\mathrm{\mathrm{Y}}}$, $\Delta a=a_{\mathrm{\mathrm{X}}}-a_{\mathrm{\mathrm{Y}}}$, and $\Delta\beta_{n}^{2}=\beta_{n,\mathrm{\mathrm{X}}}^{2}-\beta_{n,\mathrm{\mathrm{Y}}}^{2}$. Indeed, STAR’s measurements in ⁹⁶Ru+⁹⁶Ru and ⁹⁶Zr+⁹⁶Zr isobar collisions Abdallah et al. (2022) align well with this scaling relation in both hydrodynamic and transport models with WS parameterizations Jia and Zhang (2023); Nijs and van der Schee (2023).

However, there is a catch. The actual distribution of the protons and neutrons in the nucleus is more accurately described by Density Function Theories (DFT) based on effective nuclear forces Bender et al. (2003). WS parameters derived from fitting DFT distributions miss detailed features Xu et al. (2021), as DFT distributions often show wiggly structures due to the orbital filling pattern of protons and neutrons. In the presence of nuclear deformation, the skin thickness may vary with the polar angle $\theta$ Liu et al. (2023). These fine strucures’ properties depend on specific DFT models and are assumed to have minimal impact on heavy-ion observables, often being ignored in heavy-ion phenomenology.

This paper aims to quantify the impact of the differences between DFT and WS parameterizations on heavy-ion observables. We consider spherical nuclei for simplicity and three observables: $\varepsilon_{2}$, $\varepsilon_{3}$, and $N_{\mathrm{part}}$. In this context, the scaling relation simplifies to:

$$R_{\mathcal{O}}\approx 1+c_{1}\Delta R_{0}+c_{2}\Delta a\;.$$ (3)

We compare the results of several representative nuclei with those obtained from the known differences between ⁹⁶Ru and ⁹⁶Zr, $\Delta a=a_{\mathrm{Ru}}-a_{\mathrm{Zr}}=-0.06$ fm and $\Delta R_{0}=R_{0,\mathrm{Ru}}-R_{0,\mathrm{Zr}}=0.07$ fmJia and Zhang (2023). We found that the differences between DFT and WS have a smaller impact on heavy-ion observables than $\Delta a$ but can be comparable to $\Delta R_{0}$.

Setup. There are many DFTs with many different forces one could choose from Dutra et al. (2012, 2014). For this study, we use the standard Skyrme-Hartree-Fock (SHF) model Chen et al. (2010). The SHF model defines the parameters of the Skyrme interaction in terms of ten macroscopic quantities, which have empirical values corresponding to the MSL0 force as in Ref. Chen et al. (2010); Liu et al. (2022). We generate spatial distributions of protons and neutrons, summing them to obtain the radial distributions of nucleons $\rho(r)$. The parameters $R_{0}$ and $a$ are determined by matching the first and second radial moments of the WS distribution to those of the SHF Xu et al. (2021). The $n^{\mathrm{th}}$ moment is defined as,

$$\left\langle r^{n}\right\rangle=\frac{\int r^{2+n}\rho(r)dr}{\int r^{2}\rho(r)dr}\;.$$ (4)

We scanned a wide range of near spherical nuclei and compared their SHF distributions with the WS fit. From this comparison, we identified three nuclei with characteristically different shapes in $\rho(r)$: ⁹⁶Zr, ¹¹²Sn and ¹⁹⁷Au, as shown in Fig. 1. The ⁹⁶Zr has a wiggle with a dip around $r=0$, while ¹⁹⁷Au exhibits a wiggle with a peak around $r=0$, both showing significant deviations from the WS distributions. In contrast, ¹¹²Sn has a relatively uniform distribution with smaller difference from the WS fit. The corresponding WS parameters are listed in Table 1.

	96Zr	112Sn	197Au
$R_{0}$ (fm)	5.059	5.538	6.610
$a$ (fm)	0.5774	0.5576	0.5552

Table 1: Woods-Saxon parameters obtained by matching the first and second moments of nuclear distributions to the SHF.

Figure 1: (a) SHF nucleon density distributions for ⁹⁶Zr, ¹¹²Sn, and ¹⁹⁷Au nuclei. (b) Ratio of SHF and WS nucleon density distribution via moment matching. The vertical lines indicate location $r=R_{0}-a$, where we combine the SHF to the left and WS to the right together to form $\rho_{\mathrm{mix}}$ (see text).

We used a standard Monte-Carlo Glauber model Miller et al. (2007) to simulate collisions of ⁹⁶Zr, ¹¹²Sn, and ¹⁹⁷Au, assuming a nucleon-nucleon cross-section of 42 mb. Nucleons are modeled with a hard core, maintaining a minimum separation of 0.4 fm. For each species, 100 million minimum bias events were generated, separately for SHF and WS distributions with their 1st and 2nd radial moments matched. To isolate the impact of the wiggle structure, we constructed another $\rho(r)$ distribution by combining SHF at low radii ($r<R_{0}-a$) and WS at higher radii ($r>R_{0}-a$) while maintaing the mass number, labeled as $\rho_{\mathrm{mix}}$. Additionally, simulations were repeated by adjusting $a$ by -0.06 fm and $R_{0}$ by +0.07 fm to model transitions from ⁹⁶Zr to ⁹⁶Ru, labeled as $\rho_{\mathrm{WSa^{-}}}$ and $\rho_{\mathrm{WSR^{+}}}$, respectively. Thus, five separate simulations were performed for each collision species.

For each event, we calculated $N_{\mathrm{part}}$, $\varepsilon_{2}$, and $\varepsilon_{3}$, defined from the transverse nucleon density $\rho_{\perp}(r_{\perp},\phi)=\int\rho dz$ as:

$$\varepsilon_{n}e^{in\Phi_{n}}\equiv-\frac{\int\rho_{\perp}(r,\phi)r^{n}e^{in\phi}rdrd\phi}{\int\rho_{\perp}(r,\phi)r^{n}rdrd\phi}\;.$$ (5)

We then constructed histograms of $\varepsilon_{n}\equiv\sqrt{\left\langle\varepsilon_{n}^{2}\right\rangle}$ and $p(N_{\mathrm{part}})$ as functions of $N_{\mathrm{part}}$. Note that these wiggly structures still remains in $\rho_{\perp}(r_{\perp})$, but with smaller mangnitudes in the inner region (see appendix). The ratios of observables from each distribution were divided by those from the default WS parameterization in Tab. 1, resulting in four ratios: $R_{\mathcal{O}}\{\mathrm{SHF}\}$, $R_{\mathcal{O}}\{\mathrm{mix}\}$, $R_{\mathcal{O}}\{\mathrm{WSa^{-}\}}$, and $R_{\mathcal{O}}\{\mathrm{WSR^{+}}\}$. These ratios are presented as $R_{\mathcal{O}}-1$ versus $N_{\mathrm{part}}$.

Figure 2: Glauber model results of $R_{\mathcal{O}}-1$ for $\mathcal{O}=\varepsilon_{2}$ (top row), $\varepsilon_{3}$ (middle row), and $p(N_{\mathrm{part}})$ (bottom row) for Zr+Zr (left column), Sn+Sn (middle column) and Au+Au (right column) collisions. Four ratios are shown: $R_{\mathcal{O}}\{\mathrm{SHF}\}-1$ (SHF vs. WS), $R_{\mathcal{O}}\{\mathrm{mix}\}-1$ (Mixed vs. WS), $R_{\mathcal{O}}\{\mathrm{WSa^{-}}\}-1$ (WS with $a$ decreased by 0.06 fm vs. without) and $R_{\mathcal{O}}\{\mathrm{WSR^{+}}\}-1$ (WS with $R_{0}$ increased by 0.07 fm vs. without), see text for explanation.

Results Figure 2 shows the relative differences, $R_{\mathcal{O}}\{\mathrm{SHF}\}-1,R_{\mathcal{O}}\{\mathrm{mix}\}-1$, $R_{\mathcal{O}}\{\mathrm{WSa^{-}}\}-1$, and $R_{\mathcal{O}}\{\mathrm{WSR^{+}}\}-1$, as functions of $N_{\mathrm{part}}$ for the three observables. In ⁹⁶A+⁹⁶A collisions, reducing $a$ by $0.06$ fm leads to a 2% enhancement in $\varepsilon_{2}$ in mid-central collisions, while increasing $R_{0}$ by $0.07$fm decreases $\varepsilon_{2}$ by about 0.4%. These behaviors align with the expectation that a more compact source increases $\varepsilon_{2}$ and vice versa Jia et al. (2023). The impact of both parameters is smaller in peripheral and central collisions. For larger ¹⁹⁷A+¹⁹⁷A collisions, the impact is more significant: the same variations in $a$ and $R_{0}$ lead to a 3.3% increase and a 0.9% decrease in $\varepsilon_{2}$, respectively, as larger systems have smaller $\varepsilon_{2}$ values, amplifying the influence of WS parameter variations. A similar stronger impact is observed for $\varepsilon_{3}$ (middle row of Fig. 2), with the sign of the relative change opposite to that for $\varepsilon_{2}$ and occurring at different $N_{\mathrm{part}}$ values. These findings are qualitatively consistent with previous studies Shou et al. (2015); Xu et al. (2018).

The impact on $p(N_{\mathrm{part}})$ can be understood as follows: reducing $a$ sharpens the overlap region’s edge, increasing the probability for mid-central and ultra-central events while reducing that in peripheral events, resulting in a broad bump in the ratio, similar to the STAR isobar data Abdallah et al. (2022). The bump’s magnitude is purely geometrical and nearly independent of the system size. In contrast, increasing $R_{0}$ slightly reduces $p(N_{\mathrm{part}})$ in mid-central collisions.

Figure 3: Relative difference $R_{\mathcal{O}}-1$ between SHF and WS with moment matching, where $\mathcal{O}$ is $\varepsilon_{2}$ (left), $\varepsilon_{3}$ (middle), and $p(N_{\mathrm{part}})$ (right) in Zr+Zr, Sn+Sn and Au+Au collisions. They are taken directly from Fig. 2.

Figure 4: Relative difference $R_{\mathcal{O}}-1$ between $\rho_{\mathrm{mix}}$ and $\rho_{\mathrm{WS}}$ (i.e differing only in the inner region), where $\mathcal{O}$ is $\varepsilon_{2}$ (left), $\varepsilon_{3}$ (middle), and $p(N_{\mathrm{part}})$ (right) in Zr+Zr, Sn+Sn and Au+Au collisions. They are taken directly from Fig. 2.

The difference between SHF and WS, represented by $R_{\mathcal{O}}\{\mathrm{SHF}\}-1$, depends on the observable and is shown by open black symbols in Fig. 2, which are more conveniently compared across systems in Fig. 3. Because $\left\langle r^{2}\right\rangle$ is matched between SHF and WS, the impact on $\varepsilon_{2}$, which has an $r^{2}$ weight in its definition (Eq.5), is minimal. However, $\varepsilon_{3}$ from SHF is about 2% smaller than from WS in mid-central collisions, due to the $r^{3}$ weight in $\varepsilon_{3}$ enhancing contributions from nucleons at the edge of the overlap region. As seen in Fig. 1b, the WS distribution has more contribution from the large $r$ region, where $\rho_{\mathrm{SHF}}<\rho_{\mathrm{WS}}$, significantly affecting mid-central collisions. The impact on $p(N_{\mathrm{part}})$ is small except in ⁹⁶Zr, where it reaches about 0.4–0.6% in mid-central collisions. The larger outer bump around $R_{0}$ in Fig. 1b, more significant in ⁹⁶Zr, enhances $p(N_{\mathrm{part}})$ in mid-central collisions. The overall shape of $R_{\mathcal{O}}\{\mathrm{SHF}\}-1$ shows similarities across different collision systems, particularly for $\varepsilon_{3}$. But even for $\varepsilon_{2}$, the shape in Zr+Zr is similar to Au+Au just with an opposite sign.

Given the prevalent wiggly structures in $\rho_{\mathrm{SHF}}/\rho_{\mathrm{WS}}$ (Fig. 1b), we investigate the impact of the inner region ($r<R_{0}-a$) and the surface region ($r>R_{0}-a$) of $\rho(r)$. For this, we examine $R_{\mathcal{O}}\{\mathrm{mix}\}-1$, which includes only the inner region’s impact. These results, represented by open red symbols in Fig. 2 and compared across systems in Fig. 4, show small magnitudes (¡0.5%) and are observable-dependent, but they exhibit the same $N_{\mathrm{part}}$ dependent shape across collision systems. The differences between $R_{\mathcal{O}}\{\mathrm{SHF}\}-1$ and $R_{\mathcal{O}}\{\mathrm{mix}\}-1$ (Fig. 2) represent the surface region’s contribution, which clearly dominates the $\varepsilon_{3}$ ratio differences between SHF and WS.

We also consider whether the differences in $\varepsilon_{2}$, $\varepsilon_{3}$, or $p(N_{\mathrm{part}})$ between SHF and WS can be absorbed by adjusting the WS parameters $R_{0}$ and $a$ alone:

$$R_{\mathcal{O}}\{\mathrm{SHF}\}\!-\!1 =a_{1}(R_{\mathcal{O}}\{\mathrm{WSa^{-}}\}\!-1\!)\!+\!a_{2}(R_{\mathcal{O}}\{\mathrm{WSR^{+}}\}\!-\!1)$$

$$R_{\mathcal{O}}\{\mathrm{mix}\}\!-\!1 =b_{1}(R_{\mathcal{O}}\{\mathrm{WSa^{-}}\}\!-1\!)\!+\!b_{2}(R_{\mathcal{O}}\{\mathrm{WSR^{+}}\}\!-\!1)\;,$$ (6)

where $a$ and $b$ are numerical coefficients common to all observables for a given collision species. We find that this adjustment can only be done for one observable at a time. For example, $\varepsilon_{2}$ in Fig. 4a can be adjusted by varying $a$, but this would induce a large reduction in $\varepsilon_{3}$ and an enhancement in $p(N_{\mathrm{part}})$, which are not observed. Therefore, the influence of these fine structures on observables is real and can only be accurately studied via actual DFT calculations.

Overall, the impact of DFT’s wiggly structures on observables is smaller than the expected variation in the skin parameter $a$ in Zr+Zr and Ru+Ru isobar collisions but could be comparable to the variation induced by the radius parameter $R_{0}$. Including wiggly structures in SHF induces changes that are about four times smaller for $\varepsilon_{2}$ (or 25%) and twice smaller for $p(N_{\mathrm{part}})$ (or 50%) compared to a $0.06$ fm variation in $a$. However, a comparable influence is observed for $\varepsilon_{3}$. Thus, current WS parameterizations of DFT are sufficient for studying the structure impact on $\varepsilon_{2}$ and $p(N_{\mathrm{part}})$ but not $\varepsilon_{3}$ 

Summary. We have demonstrated that the Woods-Saxon (WS) parameterization used to describe nucleon density distributions in heavy atomic nuclei fails to capture the detailed radial wiggle structure present in realistic Density Functional Theory (DFT) calculations. The differences between DFT (based on the Skyrme-Hartree-Fock density functional) and WS parameterizations result in characteristic variations in observables for the initial state of heavy-ion collisions, such as $\varepsilon_{2}$, $\varepsilon_{3}$, and $p(N_{\mathrm{part}})$. These differences, on a few percent level, are centrality- and observable-dependent, with similar shapes across different nuclei but distinct behaviors among the three observables. These variations cannot be simultaneously absorbed by merely re-adjusting the WS parameters.

The magnitude of the changes induced by the differences between SHF and WS is much smaller than the expected variations between isobar collisions, such as ⁹⁶Ru+⁹⁶Ru and ⁹⁶Zr+⁹⁶Zr, a golden dataset for studying the impact of nuclear structure. However, these differences do limit the precision with which nuclear structure can be constrained using heavy-ion collisions based on WS parameterization. Therefore, actual DFT calculations should be used directly to gauge the systematic uncertainties.

Our work, together with earlier ones Shou et al. (2015); Xu et al. (2023), is the first step in quantifying whether the WS parameterization is sufficient in heavy-ion phenomenology. Future directions include considering the influence of nuclear deformation in DFT versus deformed WS parameterization Ryssens et al. (2023), as well as studying other bulk observables such as mean transverse momentum fluctuations. More importantly, simulations based on state-of-the-art hydrodynamic models are required to quantify the impact of realistic DFT versus WS on the initial condition of QGP and the extraction of its transport properties. The uncertainties in nuclear structure input should be included as part of the Bayesian analysis for heavy-ion collisions along the line of Ref. Giacalone et al. (2023).

We thank G. Giacalone for useful comments. Z. Y and J.J are supported by DOE under Grants No. DE-SC0024602. JX is supported by the Strategic Priority Research Program of the Chinese Academy of Sciences under Grant No. XDB34030000, the NSFC under Grants No. 12375125, and the Fundamental Research Funds for the Central Universities.

Appendix

Figure 5a shows the nucleon density in the transverse (r,$\phi$) plane, $\rho_{\perp}(r_{\perp})=\int 2\pi r_{\perp}\rho dzdr_{\perp}$ for SHF. This distribution is directly used to calculate $\varepsilon_{n}$ and $N_{\mathrm{part}}$. Notably, in the interior region, the density of ⁹⁶Zr remains higher than that of ¹¹²Sn, even after integration along the z-axis. The corresponding ratios between the SHF and WS distributions in the transverse plane are shown in Fig. 5b. Compared to Fig. 1b, the wiggle structures at $r_{\perp}\lesssim R_{0}$ are reduced but not eliminated, while the structures at $r_{\perp}\gtrsim R_{0}$ remain more or less unchanged.

Figure 5: (a) SHF nucleon density distributions in the transverse plane for ⁹⁶Zr, ¹¹²Sn, and ¹⁹⁷Au nuclei. (b) Ratios of SHF and WS density distributions after moment matching and projection to the transverse plane. The vertical lines indicate the location $r_{\perp}=R_{0}-a$.