betaFeedback

WorksPhysics and Astronomy

Temperature-resolved sensitivities of 56Ni^{56}{\rm Ni} production to helium-burning reactions in pair-instability supernovae

Abstract

We propose a temperature-resolved Monte Carlo (MC) approach to identify the temperature regimes in which low-energy helium-burning reaction rates most strongly affect nucleosynthesis in very massive stars that undergo pair-instability supernovae (PISNe). By performing MC simulations of PISNe, we quantify how temperature-dependent variations in key helium-burning reaction rates, i.e., the triple-αα and 12C(α,γ)16O^{12}{\rm C}(α,γ)^{16}{\rm O} rates, influence 56Ni^{56}{\rm Ni} synthesis. Thousands of stellar evolution calculations using MESA\texttt{MESA} reveal that both the 12C(α,γ)16O^{12}{\rm C}(α,γ)^{16}{\rm O} and triple-αα reactions exhibit their strongest sensitivity at T≃2.5×108 KT \simeq 2.5 \times 10^{8}\,{\rm K}, but with opposite correlation signs. We show that this temperature corresponds to the regime in which the ratio of the sampled rate multipliers is most clearly imprinted on the pre-carbon-burning C/O composition. This demonstrates that PISN nucleosynthesis can probe helium-burning reaction rates in specific low-temperature regimes.

Introduction

Very massive stars (VMSs), with progenitor masses of ∼200M⊙\sim200M_{\odot}, are theoretically predicted to end their evolution as pair-instability supernovae (PISNe) (e.g., [2, 13, 17]). A PISN is triggered by the creation of electron–positron pairs in the massive C/O core, which reduces the radiation-pressure support. This induces rapid core contraction and explosive oxygen burning, ultimately leading to the complete disruption of the star. Such a thermonuclear explosion produces a large amount of radioactive iron-group nuclei, predominantly 56Ni^{56}\mathrm{Ni}, which serves as the main energy source of the supernova (SN) light curve. As PISNe are expected to be significantly brighter than ordinary SNe, their observational identification is of particular importance. Recently, a possible PISN candidate has been observed (Schulze et al. 2024), highlighting the need for further theoretical investigation of the observational properties of PISNe.

Although theoretical models are essential for interpreting PISNe, simulations of PISN progenitor evolution still suffer from large uncertainties in several physical aspects. In addition to the basic modeling of stellar evolution, nuclear reaction rates, in particular, remain significantly uncertain, even for the main nuclear-burning reactions, such as 12C(α,γ)16O^{12}\mathrm{C}(\alpha,\gamma)^{16}\mathrm{O} and triple-α\alpha (see, e.g., Tur et al. 2009, 2010; Dotter & Paxton [9]; Suda et al. 2011; Kikuchi et al. [24]; Tanikawa et al. 2023; Tognini et al. 2023; Xin et al. 2025). Despite extensive efforts over several decades in nuclear physics, determining these rates remains highly challenging both experimentally and theoretically (e.g., deBoer et al. 2017; Fynbo et al. 2005; Kibíde et al. 2020).

The amount of 56Ni^{56}\mathrm{Ni} synthesized in a PISN is one of the key observables, because it largely determines the brightness of the SN light curve. However, theoretical predictions of the 56Ni^{56}\mathrm{Ni} yield are affected by various physical uncertainties in stellar evolution and explosion models, which complicates a definitive comparison between theory and observations. In particular, previous studies ([22] (Takahashi 2018)) have shown that the amount of 56Ni^{56}\mathrm{Ni} synthesized in PISNe strongly depends on helium-burning reaction rates. This dependence implies that uncertainties in helium-burning reaction rates directly propagate into uncertainties in the predicted 56Ni^{56}\mathrm{Ni} yield, and hence into the expected luminosity and observability of PISNe.

For a realistic investigation of such reaction-rate uncertainties, the rates should not be treated as uniformly uncertain over the entire temperature range. Throughout stellar evolution and explosion, different temperature ranges may play different roles, and the sensitivity of the final nucleosynthesis products to an individual reaction rate can vary accordingly. Such temperature dependence is also well known in nuclear physics, where reaction cross sections depend strongly on energy and where the dominant physical processes governing the rates may differ between energy ranges. Thus, identifying the temperature ranges in which reaction-rate variations most strongly affect 56Ni^{56}\mathrm{Ni} production is essential both for interpreting future PISN observations and for guiding nuclear cross-section measurements. In previous studies, the 12C(α,γ)16O^{12}\mathrm{C}(\alpha,\gamma)^{16}\mathrm{O} and triple-α\alpha reaction rates were artificially scaled uniformly up and down over the full temperature range, allowing the authors to estimate upper and lower bounds on 56Ni^{56}\mathrm{Ni} production under such extreme assumptions. However, actual reaction rates are unlikely to deviate so strongly over the entire temperature range. Moreover, such uniform scaling does not identify which temperature ranges are responsible for the resulting change in the final 56Ni^{56}\mathrm{Ni} yield.

During helium burning, the triple-α\alpha reaction produces 12C^{12}\mathrm{C}, while the 12C(α,γ)16O^{12}\mathrm{C}(\alpha,\gamma)^{16}\mathrm{O} reaction converts 12C^{12}\mathrm{C} into 16O^{16}\mathrm{O}. The two reactions therefore act as the production and destruction channels of carbon during the formation of the CO core. This makes it necessary to consider both reactions simultaneously when investigating how helium-burning reaction-rate uncertainties affect PISN nucleosynthesis. In this work, we develop a temperature-resolved Monte Carlo (MC) approach to identify the temperature ranges in which helium-burning reaction-rate variations most strongly affect 56Ni^{56}\mathrm{Ni} production in PISNe. Using thousands of stellar evolution models with independently sampled temperature-dependent multiplier factors for the triple-α\alpha and 12C(α,γ)16O^{12}\mathrm{C}(\alpha,\gamma)^{16}\mathrm{O} reactions, we quantify the statistical relationship between the rate variation at each temperature and the final 56Ni^{56}\mathrm{Ni} yield. We introduce the methodology in section 2, present the results in section 3, discuss their implications in section 4, and summarize our conclusions in section 5.

Method

Stellar evolution calculations with MESA

We perform stellar evolution simulations of VMSs using the stellar evolution code MESA, version r15140 [35, 36, 21] (Paxton et al. 2015; Paxton et al. 2018; Paxton et al. 2019). The equation of state (EOS) is constructed from several sources: OPAL (Rogers & Nayfonov 2002), SCVH (Saumon et al. 1995), FreeEOS [19], HELM (Timmes & Swesty 2000), and PC (Potekhin & Chabrier 2010). Radiative opacities are primarily taken from OPAL [18] (Iglesias & Rogers 1996), supplemented by low-temperature data from Ferguson et al. [12] and high-temperature, Compton-scattering-dominated opacities from Buchler and Yueh [4]. Electron-conduction opacities are taken from Cassisi et al. [5].

Nuclear reaction rates are based on JINA REACLIB [7], except for the triple-α\alpha and 12C(α,γ)16O^{12}\mathrm{C}(\alpha,\gamma)^{16}\mathrm{O} reactions, and are supplemented by several weak reaction rate tables [14, 33, 27]. Screening effects are treated according to the prescription of Chugunov, DeWitt, and Yakovlev [6], and thermal neutrino loss rates are adopted from Itoh et al. [20].

We adopt a hydrogen-envelope-stripped VMS model with an initial mass of 100 M⊙100\,M_{\odot}, following the scenario proposed by Marchant et al. [28], assuming a metallicity of Z=10−5Z=10^{-5} for both the stellar components and their environment. We employ a nuclear reaction network based on approx21_plus_co56.net, which follows the major α\alpha-chain isotopes and includes radioactive 56Ni^{56}\mathrm{Ni} and 56Co^{56}\mathrm{Co}. We note that the additional calculations for SN 2018ibb in section 4.2 use different helium-core masses and metallicities, as described there.

Each model is initialized following the standard MESA procedure, where the initial stellar model is constructed as a chemically homogeneous gaseous sphere [35]. A model is regarded as a successful PISN when all stellar material becomes unbound from the gravitational potential, which means that every mass shell has a velocity exceeding the escape velocity.

Temperature-resolved MC sampling of reaction rates

Our goal is to identify the characteristic temperature ranges of the triple-α\alpha and 12C(α,γ)16O^{12}\mathrm{C}(\alpha,\gamma)^{16}\mathrm{O} reactions that show the strongest sensitivity of the final 56Ni^{56}\mathrm{Ni} yield in the evolution of VMSs leading to PISNe. To this end, we perform thousands of stellar evolution simulations using MC-modified reaction rate tables. By correlat-

Image not extracted

Figure 1. Upper: Absolute reaction rates shown as piRp^{R}_{i} in equation (1). ⟨ααα⟩\langle\alpha\alpha\alpha\rangle denotes the reaction rate of 3α3\alpha, and ⟨12Cα⟩\langle{}^{12}\mathrm{C}\alpha\rangle denotes that of 12C(α,γ)16O^{12}\mathrm{C}(\alpha,\gamma)^{16}\mathrm{O}. Lower: Reaction rates normalized to the standard STARLIB values [1, 26] (Sallaska et al. 2013) shown as fiRf^{R}_{i} in equation (1). The green lines show the standard STARLIB values, while the other lines show examples of randomly generated reaction rate realizations.

AlT text: Four panels show the triple-alpha reaction rate on the left and the carbon-alpha reaction rate on the right as functions of logarithmic temperature. In the upper panels, the standard reaction rates are shown by green lines, while three randomly generated rate realizations are shown by magenta, cyan, and orange lines with cross markers for examples 1, 2, and 3, respectively. The lower panels show the corresponding ratios to the standard rates, with the same color coding, varying between approximately 0.5 and 2. ing the final 56Ni^{56}\mathrm{Ni} yield with the random multiplier assigned at each temperature, this method identifies the temperature windows in which uncertainties in the helium-burning reaction rates are most strongly propagated to PISN nucleosynthesis. These temperature windows indicate where improved nuclear physics inputs, or future observational constraints on PISN nucleosynthesis, may help identify the rate inputs most relevant to PISN predictions. This differs from studies that perform MC analyses by simultaneously varying multiple reaction rates involved in nucleosynthesis (e.g., Nishimura et al. [32]), where the rate-uncertainty factors are assumed to vary uniformly over the entire temperature range.

To evaluate the impact of reaction-rate uncertainties on nucleosynthesis, we generated MC-sampled tables based on STARLIB (Sallaska et al. 2013). We adopt the rate by Angulo et al. [1] for the triple-α\alpha reaction and the rate by Kunz et al. [26] for the 12C(α,γ)16O^{12}\mathrm{C}(\alpha,\gamma)^{16}\mathrm{O} reaction. For each stellar model ii, each reaction channel R\mathcal{R} (e.g., triple-α\alpha or 12C(α,γ)16O^{12}\mathrm{C}(\alpha,\gamma)^{16}\mathrm{O} reaction), and each temperature point TT, a multiplier fiR(T)f_i^\mathcal{R}(T) is sampled from a uniform distribution over [0.5,2.0][0.5, 2.0], thereby introducing distinct temperature-dependent variations in the reaction rates. In this work, we apply this procedure to the triple-α\alpha and 12C(α,γ)16O^{12}\mathrm{C}(\alpha,\gamma)^{16}\mathrm{O} reactions, which are both critical during the same evolutionary stage. The random multipliers for these two reactions are generated simultaneously in each model. This approach allows us to investigate whether, and to what extent, competition between the triple-α\alpha and 12C(α,γ)16O^{12}\mathrm{C}(\alpha,\gamma)^{16}\mathrm{O} reactions affects 56Ni^{56}\mathrm{Ni} synthesis.

The multiplicative factors were drawn from a uniform distribution over [0.5,2.0][0.5,2.0], corresponding to a variation within a factor of two of the recommended rate. This range was adopted as a simple exploratory envelope, comparable to the scale of the rate uncertainties relevant to the helium-burning reactions considered here.1 We emphasize that the resulting rate tables are not intended to represent statistically sampled realizations of the STARLIB rate uncertainties. In STARLIB, rate uncertainties are given as temperature-dependent log-normal factors, and realistic nuclear uncertainties are generally correlated across temperature. Our goal is instead to perform a temperature-resolved sensitivity analysis: by perturbing the rate at each temperature point independently, we identify the temperature ranges in which variations of the helium-burning rates are most strongly associated with changes in the final 56Ni^{56}\mathrm{Ni} yield.

Figure 1 shows illustrative examples of the randomly generated 12C(α,γ)16O^{12}\mathrm{C}(\alpha,\gamma)^{16}\mathrm{O} and triple-α\alpha reaction rate tables. The green line indicates the standard rate, while the colored lines represent different MC realizations. This demonstrates the extent of the random variations introduced in our analysis.

The perturbed reaction rates for model ii are then calculated as

piR(T)=pstdR(T)×fiR(T).(1)p_i^\mathcal{R}(T)=p_{\mathrm{std}}^\mathcal{R}(T)\times f_i^\mathcal{R}(T). \tag*{(1)}

Subsequent nucleosynthesis calculations are performed for each model to obtain the synthesized 56Ni^{56}\mathrm{Ni} mass at the end of the calculation, denoted by M56Ni,iM_{^{56}\mathrm{Ni},i}. For each reaction R\mathcal{R} and temperature TT, we investigate the relationship between M56Ni,iM_{^{56}\mathrm{Ni},i} and the corresponding sampled multiplier fiR(T)f_i^\mathcal{R}(T) and quantify their statistical correlation using the Pearson correlation coefficient:

rR(T)=∑i=1n(M56Ni,i−M‾56Ni)(fiR(T)−fR(T)‾)∑i=1n(M56Ni,i−M‾56Ni)2∑i=1n(fiR(T)−fR(T)‾)2.(2)r^\mathcal{R}(T)=\frac{\sum_{i=1}^{n}\left(M_{^{56}\mathrm{Ni},i}-\overline{M}_{^{56}\mathrm{Ni}}\right)\left(f_i^\mathcal{R}(T)-\overline{f^\mathcal{R}(T)}\right)}{\sqrt{\sum_{i=1}^{n}\left(M_{^{56}\mathrm{Ni},i}-\overline{M}_{^{56}\mathrm{Ni}}\right)^2}\sqrt{\sum_{i=1}^{n}\left(f_i^\mathcal{R}(T)-\overline{f^\mathcal{R}(T)}\right)^2}}. \tag*{(2)}

Here, nn denotes the number of stellar models included in the correlation calculation, namely models that successfully produced a final 56Ni^{56}\mathrm{Ni} yield. Models that failed to explode or whose calculations did not complete successfully account for less than about 10%10\% of the total attempted sample and are excluded (944 failed models in 10,000 trials in our work) from both nn and the sums in equation (2). The overbar indicates the average over the models included in the correlation calculation. This procedure provides a sensitivity analysis that identifies the temperature windows in which each reaction shows the strongest linear association with 56Ni^{56}\mathrm{Ni} synthesis. To satisfy detailed balance, the same factor is also applied to the reverse reaction rates.

Results

Temperature-dependent sensitivity of key reactions

Using the temperature-resolved MC sampling described in section 2, we performed 10,000 stellar evolution calculations with independently sampled rate multipliers. Figure 2 presents the temperature dependence of the Pearson correlation coefficient between the sampled multiplier and the final 56Ni^{56}\mathrm{Ni} yield for the triple-α\alpha and 12C(α,γ)16O^{12}\mathrm{C}(\alpha,\gamma)^{16}\mathrm{O} reactions. The vertical axis shows the correlation coefficient, while the horizontal axis corresponds to temperature. The blue and red curves show r3α(T)r^{3\alpha}(T) for the triple-α\alpha reaction and rCα(T)r^{\mathrm{C}\alpha}(T) for the 12C(α,γ)16O^{12}\mathrm{C}(\alpha,\gamma)^{16}\mathrm{O} reactions, respectively. Both reactions exhibit pronounced sensitivity around 2.5×1082.5 \times10^{8} K, yet with opposite signs: 12C(α,γ)16O{}^{12}\mathrm{C}(\alpha,\gamma){}^{16}\mathrm{O} is associated with a strong positive correlation (rCα(2.5×108 K)≃0.644r^{\mathrm{C}\alpha}(2.5 \times10^{8}\ \mathrm{K}) \simeq0.644), whereas the triple-α\alpha reaction shows a negative correlation (r3α(2.5×108 K)≃−0.548r^{3\alpha}(2.5 \times10^{8}\ \mathrm{K}) \simeq-0.548). This contrast suggests that the two reactions affect 56Ni{}^{56}\mathrm{Ni} synthesis in opposite directions within the same temperature range. Replacing M56Ni,iM_{{}^{56}\mathrm{Ni},i} in equation (2) with the final explosion energy, we find that the strongest correlations again occur at T≃2.5×108T \simeq2.5 \times10^{8} K, with Pearson coefficients of −0.604-0.604 for the triple-α\alpha reaction and 0.6410.641 for the 12C(α,γ)16O{}^{12}\mathrm{C}(\alpha,\gamma){}^{16}\mathrm{O} reaction.

The sign difference arises from their distinct nuclear roles. The triple-α\alpha reaction facilitates the production of 12C{}^{12}\mathrm{C} by consuming α\alpha-particles during helium burning, while 12C(α,γ)16O{}^{12}\mathrm{C}(\alpha,\gamma){}^{16}\mathrm{O} converts 12C{}^{12}\mathrm{C} into 16O{}^{16}\mathrm{O}, thereby reducing the available carbon. As a result, the two reactions act in competition, and their balance influences the pre-explosive composition of the CO core.

The 3α3\alpha reaction contributes to the growth of the CO core. Since heavier CO cores are generally associated with stronger explosions and increased 56Ni{}^{56}\mathrm{Ni} synthesis (Takahashi et al. 2016), this reaction might be expected to enhance 56Ni{}^{56}\mathrm{Ni} production. However, the negative correlation observed here suggests that, within this temperature range, the suppressive effect of increased carbon abundance on oxygen burning dominates over the indirect effect of core mass growth. This balance highlights the utility of correlation analysis in disentangling the relative importance of such competing nuclear effects.

These results indicate that the 56Ni{}^{56}\mathrm{Ni} yield is sensitive not only to the explosive oxygen-burning phase itself, but also to the progenitor composition established during helium burning. Variations in the helium-burning reaction rates around T≃2.5×108T \simeq2.5 \times10^{8} K can therefore affect the final 56Ni{}^{56}\mathrm{Ni} yield indirectly, by modifying the pre-explosive C/O composition. In the next subsection, we examine this compositional imprint directly.

Imprint of helium-burning rates on the pre-explosive C/O composition

The temperature sensitivity identified in figure 2 suggests that helium-burning reaction rates leave their strongest imprint on the progenitor composition around T≃2.5×108T \simeq2.5 \times10^{8} K. This imprint can be understood through the carbon–oxygen composition established before carbon burning. The triple-α\alpha reaction first produces 12C{}^{12}\mathrm{C} and initiates the formation of the CO core, while the subsequent 12C(α,γ)16O{}^{12}\mathrm{C}(\alpha,\gamma){}^{16}\mathrm{O} reaction consumes 12C{}^{12}\mathrm{C} and produces 16O{}^{16}\mathrm{O}. Therefore, the balance between these two reaction rates controls how much carbon remains after helium burning and how much material is converted into oxygen and heavier α\alpha-chain nuclei.

Figure 3 shows two-dimensional histograms of the relation between the rate-multiplier ratio and the effective number ratio of produced carbon to produced oxygen. For each model ii, the rate-multiplier ratio at temperature TT is defined as

Fi(T)=fi3α(T)fiCα(T).(3)\mathcal{F}_{i}(T)=\frac{f_{i}^{3\alpha}(T)}{f_{i}^{\mathrm{C}\alpha}(T)}. \tag*{(3)}

Here, fi3α(T)f_{i}^{3\alpha}(T) and fiCα(T)f_{i}^{\mathrm{C}\alpha}(T) are the sampled multipliers applied to the standard triple-α\alpha and 12C(α,γ)16O{}^{12}\mathrm{C}(\alpha,\gamma){}^{16}\mathrm{O} rates, respectively. The effective number ratio of produced carbon to produced oxygen is

Image not extracted

Figure 3. Two-dimensional histograms showing the relation between the rate-multiplier ratio and the effective C/O number ratio of produced nuclei at 5.0×1075.0 \times10^{7}, 1.0×1081.0 \times10^{8}, 2.5×1082.5 \times10^{8}, and 3.0×1093.0 \times10^{9} K. The horizontal axis indicates the rate-multiplier ratio, Fi(T)\mathcal{F}_{i}(T), defined in equation (3). The vertical axis indicates the effective produced C/O number ratio, evaluated from the total stellar species masses at Tc=108.7T_{\mathrm{c}}=10^{8.7} K, before the onset of carbon burning. All panels are plotted with the same binning, axis ranges, and color scale. The color indicates the number of models per bin, shown on a logarithmic scale. A distinct structured distribution is seen only at 2.5×1082.5 \times10^{8} K, whereas the distributions at the other temperatures are comparatively diffuse.

Alt text: Four two-dimensional histograms show the rate-multiplier ratio on the horizontal axis and the effective produced C/O number ratio on the vertical axis at 5.0×1075.0 \times10^{7}, 1.0×1081.0 \times10^{8}, 2.5×1082.5 \times10^{8}, and 3.0×1093.0 \times10^{9} K. The color scale gives the number of models per bin on a logarithmic scale. The 2.5×1082.5 \times10^{8} K panel shows a narrow positively sloped distribution, whereas the distributions in the other three panels are broader and less structured.

evaluated from the total stellar species masses when the central temperature reaches Tc=108.7T_{\mathrm{c}}=10^{8.7} K. This temperature is chosen because central helium burning has almost ended, whereas carbon burning has not yet started. Thus, apart from the negligible contribution of the initial metallicity, nuclei heavier than carbon can be regarded as products of successive α\alpha captures. We define the effective produced C/O number ratio as

M(12C)+34M(16O)+35M(20Ne)+36M(24Mg)34M(16O)+45M(20Ne)+46M(24Mg),(4)\frac{M({}^{12}\mathrm{C})+\frac{3}{4}M({}^{16}\mathrm{O})+\frac{3}{5}M({}^{20}\mathrm{Ne})+\frac{3}{6}M({}^{24}\mathrm{Mg})}{\frac{3}{4}M({}^{16}\mathrm{O})+\frac{4}{5}M({}^{20}\mathrm{Ne})+\frac{4}{6}M({}^{24}\mathrm{Mg})}, \tag*{(4)}

where M(X)M(X) denotes the total stellar mass of isotope XX at Tc=108.7T_{\mathrm{c}}=10^{8.7} K. The terms proportional to M(16O)M({}^{16}\mathrm{O}), M(20Ne)M({}^{20}\mathrm{Ne}), and M(24Mg)M({}^{24}\mathrm{Mg}) in the numerator represent the amount of carbon that must have been produced before subsequent α\alpha captures. Similarly, the terms proportional to M(20Ne)M({}^{20}\mathrm{Ne}) and M(24Mg)M({}^{24}\mathrm{Mg}) in the denominator represent the corresponding oxygen contribution. Since (M12C/12)/(M16O/16)=(4/3)(M12C/M16O)(M_{{}^{12}\mathrm{C}}/12)/(M_{{}^{16}\mathrm{O}}/16)=(4/3)(M_{{}^{12}\mathrm{C}}/M_{{}^{16}\mathrm{O}}), the final factor of 3/43/4 converts the mass ratio into a number ratio. We note that contributions from heavier α\alpha-chain nuclei are negligible at this stage. The total stellar mass of the next α\alpha-chain isotope, 28Si{}^{28}\mathrm{Si}, remains below 10−2 M⊙10^{-2}\,M_{\odot} in all models at Tc=108.7T_{\mathrm{c}}=10^{8.7} K, indicating that the α\alpha-chain has hardly proceeded beyond 24Mg{}^{24}\mathrm{Mg}.

Among the temperatures shown in figure 3, only T≃2.5×T \simeq2.5 \times 10810^{8} K exhibits a clearly structured distribution. This indicates that Fi(T)\mathcal{F}_i(T) is most directly imprinted on the C/O composition before carbon burning at this temperature. In contrast, the distributions at the other temperatures are much more diffuse, suggesting that rate variations at those temperatures do not control the pre-carbon-burning C/O composition in the same systematic way. This result provides a physical interpretation of the correlations shown in Figure 2.

Figure 2. Temperature dependence of the Pearson correlation coefficient between the temperature-dependent rate multipliers and the synthesized 56Ni^{56}\mathrm{Ni} mass. The solid curves represent r3α(T)r^{3\alpha}(T) (blue) and rCα(T)r^{\mathrm{C}\alpha}(T) (red). Positive values indicate a direct correlation between rate enhancement and 56Ni^{56}\mathrm{Ni} production, while negative values indicate an inverse correlation. Alt text: Correlation coefficient on the vertical axis as a function of logarithmic temperature in kelvin on the horizontal axis. The carbon-alpha curve (red line) has a sharp positive peak near 2.5×108 K2.5\times10^8\ \mathrm{K}, while the triple-alpha curve (blue line) has a negative minimum near the same temperature; both remain close to zero over most other temperatures.

The opposite signs of the correlations in Figure 2 can therefore be understood as a consequence of how the two reactions regulate the residual carbon abundance. Previous studies have shown that the amount of carbon remaining after helium burning is important for determining the strength of the subsequent PISN explosion and the synthesized 56Ni{}^{56}\mathrm{Ni} mass (Takahashi 2018; Kawashimo et al. 2024). A larger 12C(α,γ)16O{}^{12}\mathrm{C}(\alpha,\gamma){}^{16}\mathrm{O} rate at T≃2.5×108T \simeq2.5 \times10^{8} K shifts the helium-burning products toward oxygen-rich compositions, leaving less carbon available for pre-explosive carbon burning. The progenitor then retains a more compact structure until oxygen ignition, resulting in more violent oxygen burning and a larger 56Ni{}^{56}\mathrm{Ni} yield. In contrast, a larger triple-α\alpha rate shifts the composition toward carbon-rich models, where enhanced carbon burning injects energy before oxygen ignition, affects the stellar structure, and suppresses subsequent 56Ni{}^{56}\mathrm{Ni} production [22]. Our results therefore support the interpretation that the temperature at which the rate-multiplier ratio most clearly determines the pre-carbon-burning C/O composition is also the temperature at which the final 56Ni{}^{56}\mathrm{Ni} yield becomes most sensitive to the helium-burning reaction rates.

Discussion

Convergence

To validate the reliability of the computed results, we examined how the correlation values evolve with the number of stellar models. This analysis also provides insight into the computational demand associated with our method.

Image not extracted

Figure 4. Stepwise evolution of the correlation coefficients as the cumulative number of attempted stellar models is increased. At each step, the correlation coefficient is computed using only the successful models among the attempted models up to that point. Each line corresponds to a different temperature, as indicated by the color scale. The upper panel shows the 3α3\alpha series, and the lower panel shows the 12C(α,γ)16O{}^{12}\mathrm{C}(\alpha,\gamma){}^{16}\mathrm{O} series. Alt text: Two stacked panels show the correlation coefficient on the vertical axis as a function of the cumulative model number on the logarithmic horizontal axis, for the triple-alpha reaction in the upper panel and the carbon-alpha reaction in the lower panel. Curve color represents temperature according to the logarithmic color scale at right, ranging from 0.1×1080.1 \times10^{8} to 100×108100 \times10^{8} K. The curves fluctuate widely at small model numbers. At large model numbers, most curves lie close to zero, while the green curves corresponding to temperatures around 2.5×1082.5 \times10^{8} K remain clearly offset from zero, negatively for the triple-alpha reaction and positively for the carbon-alpha reaction.

Figure 4 presents the evolution of the correlation coefficient as a function of the cumulative number of attempted stellar models, shown on the horizontal axis. At each point, the correlation coefficient is evaluated using only the successful models among the attempted models up to that point, as in equation (2). The vertical axis shows the resulting correlation coefficient. Each curve is color-coded by the corresponding temperature, with blue and red representing lower and higher temperatures, respectively. Note that the model count includes cases that did not result in successful explosions. Since such failed-explosion models account for less than about 10% of the total sample, excluding them would change the effective model count only by a factor of order unity and would not qualitatively affect the convergence behavior.

Candidate temperature windows associated with the dominant correlation signals already emerge after O(102)\mathcal{O}(10^{2}) attempted models. This indicates that the method can efficiently identify promising temperature ranges where reaction-rate variations may strongly affect 56Ni{}^{56}\mathrm{Ni} synthesis, although larger samples are required for robust estimates of the correlation amplitudes.

SN 2018ibb

SN 2018ibb is one of the most promising PISN candidates, with long-term observational monitoring extending beyond 1000 days (Schulze et al. 2024). Several theoretical studies have also supported its interpretation as a PISN (e.g., Kozyreva et al. 2024; but see also Nagele et al. 2024). As an illustrative application of the temperature sensitivity identified above, we discuss the implications for SN 2018ibb under the PISN interpretation.

For this application, we perform additional stellar-evolution calculations separately from the temperature-resolved MC models described in section 2.2. For each reaction, the tabulated rate at T=2.5×108T = 2.5 \times10^{8} K is independently varied from 0.1 to 10 times its standard value, while the tabulated rates at all other grid points are kept at their standard values. The neighbouring grid points at 2.02.0 and 3.0×1083.0 \times10^{8} K thus retain their standard values, and MESA evaluates the rate between tabulated points using piecewise monotonic cubic interpolation [35, 36]. This wide parameter range is used only to visualize the model response over an extended parameter space and not as a realistic uncertainty range.

Figure 5 shows the synthesized 56Ni{}^{56}\mathrm{Ni} mass for 130 M⊙130\,M_{\odot} helium-core models in the plane of the 12C(α,γ)16O{}^{12}\mathrm{C}(\alpha,\gamma){}^{16}\mathrm{O} and triple-α\alpha reaction rates at T=2.5×108T = 2.5 \times10^{8} K. The red outlined grid cells in the right panel indicate model combinations whose synthesized 56Ni{}^{56}\mathrm{Ni} masses fall within the observationally inferred range for SN 2018ibb, ∼25\sim25–44 M⊙44\,M_{\odot}. SN 2018ibb is thought to have originated from the explosion of a progenitor with a helium-core mass of ∼120\sim120–130M⊙130 M_\odot (Schulze et al. 2024). The left panel assumes the metallicity at which PISNe are typically expected to occur, Z∼10−3Z \sim10^{-3} [16] (Simonato et al. 2025). Even when both reaction rates are varied over the wide range of 0.1–10 times their standard values, the synthesized 56^{56}Ni mass does not reach this observationally inferred range in the Z∼10−3Z \sim10^{-3} model.2 Note that the weak color variation in the left panel partly reflects the use of the same linear color scale in both panels, since the synthesized 56^{56}Ni masses there are generally below 10M⊙10 M_\odot and variations in this low-yield regime are more clearly seen on a logarithmic scale (see Kawashimo et al. [22]); the weak visual contrast alone therefore does not imply a shift of the characteristic temperature.

Image not extracted

Figure 5. Synthesized 56^{56}Ni mass as a function of the 12^{12}C(α,γ\alpha,\gamma)16^{16}O reaction rate (horizontal axis) and the triple-α\alpha reaction rate (vertical axis) at T=2.5×108T = 2.5 \times10^{8} K. The left and right panels show 130M⊙130 M_\odot helium-core models with Z=10−3Z = 10^{-3} and Z=1.42×10−4(=10−2Z⊙)Z = 1.42 \times10^{-4} (= 10^{-2} Z_\odot), respectively. The color map shows the synthesized 56^{56}Ni mass for each combination of reaction rates. The red boxes indicate the grid cells in the parameter space that are consistent with the 56^{56}Ni mass inferred for SN 2018ibb, 2525–44M⊙44 M_\odot. For the white regions in the upper-left and lower-right parts of the panels, see footnote 2. Alt text: Two heat maps show synthesized 56^{56}Ni mass as a function of the carbon-alpha reaction rate on the horizontal axis and the triple-alpha reaction rate on the vertical axis at 2.5×1082.5 \times10^{8} K, for metallicities Z=10−3Z = 10^{-3} on the left and Z=1.42×10−4Z = 1.42 \times10^{-4} on the right. Color indicates the synthesized 56^{56}Ni mass. Red outlines, which appear only in the right panel, mark grid cells corresponding to 2525–44M⊙44 M_\odot of 56^{56}Ni, while white regions indicate models without successful PISN explosions.

The right panel of figure 5 shows the corresponding result for a lower-metallicity model with Z=1.42×10−4(=10−2Z⊙)Z = 1.42 \times10^{-4} (= 10^{-2} Z_\odot). In this case, a region of parameter space appears that is consistent with the 56^{56}Ni mass inferred for SN 2018ibb. Since the outcome of a PISN depends sensitively on the progenitor mass and is therefore strongly affected by metallicity-dependent mass loss, this comparison suggests that SN 2018ibb may have originated from an unusually metal-poor environment, relative to the metallicities expected for the bulk of the PISN population. Although this discussion is only illustrative, it shows that the temperature-resolved reaction-rate sensitivity identified in this work can provide a useful framework for interpreting future PISN observations.

Conclusion

We conducted large-scale MC stellar evolution simulations to identify the temperature ranges in which helium-burning reaction rates most strongly affect 56^{56}Ni production in PISNe. The 12^{12}C(α,γ\alpha,\gamma)16^{16}O and triple-α\alpha reactions show peak sensitivity around T≃2.5×108T \simeq2.5 \times10^{8} K, with opposite correlation signs reflecting their competitive roles in setting the pre-explosive C/O composition. We showed that the rate-multiplier ratio at this temperature is most clearly imprinted on the effective C/O composition before carbon burning, explaining why localized variations in low-temperature helium-burning rates can significantly alter the final 56^{56}Ni yield. In nuclear-energy terms, the standard Gamow-window expressions (e.g., deBoer et al. 2017) give E0≃370E_0 \simeq370 keV with Δ≃210\Delta\simeq210 keV for 12^{12}C(α,γ\alpha,\gamma)16^{16}O at this temperature, while the triple-α\alpha reaction proceeds through the 8^{8}Be ground-state and Hoyle-state resonances at approximately 92 and 287 keV above the respective thresholds [3], placing the identified sensitivity in the few-hundred-keV regime relevant to stellar helium burning.

Our convergence analysis further shows that candidate temperature windows for the dominant sensitivities can already emerge after only O(102)\mathcal{O}(10^{2}) stellar-evolution calculations, although larger samples are required to determine the correlation amplitudes robustly. This rapid emergence of the sensitivity signal is an important advantage of the present temperature-resolved MC approach and makes it practical to identify the key temperature windows of uncertain nuclear reaction rates with a relatively small number of models.

As future wide-field surveys increase the number of PISN candidates, statistical comparisons with theoretical models may provide an observational route to probing low-energy helium-burning reaction rates. Such constraints would be particularly valuable given the remaining nuclear-physics uncertainties. The triple-α\alpha reaction has been extensively studied both experimentally and theoretically, with recent precision measurements of the Hoyle-state radiative decay properties further refining the reaction rate (e.g., Fynbo et al. [15]; Kibédi et al. [23]; Tsumura et al. 2021; Sakanashi et al. 2025). Despite this progress, the rate is still less tightly constrained at the lower temperatures relevant to progenitor evolution. The situation is more severe for the 12C(α,γ)16O{}^{12}\mathrm{C}(\alpha,\gamma){}^{16}\mathrm{O} reaction, whose rate remains difficult to determine directly at the relevant low energies (e.g., deBoer et al. 2017; Shen et al. 2023; Nan et al. [31]; Potterveld et al. 2024; Mukhamedzhanov et al. [29]), leaving substantial uncertainties. The present temperature-resolved MC framework can also be extended to more complex nucleosynthesis systems involving many coupled reactions, providing a systematic way to connect temperature-dependent nuclear physics uncertainties with observable nucleosynthesis signatures.

The robustness of the exact sensitivity peak against variations in progenitor properties and stellar-evolution physics remains to be systematically explored. Nevertheless, our results suggest that the sensitivity around T≃2.5×108 KT \simeq2.5 \times10^{8}\,\mathrm{K} persists within standard evolutionary scenarios: the 130 M⊙130\,M_{\odot} helium-core models in figure 5 remain clearly sensitive to rate variations at this temperature, and previous studies find ordinary variations in metallicity, internal mixing, and stellar winds to have smaller effects on pair-instability outcomes than uncertainties in key nuclear reaction rates (Farmer et al. [11]; Sakstein & Croon 2026). Qualitatively different evolutionary histories may lead to shifts in the exact peak; for example, the triple-α\alpha reaction can already operate during hydrogen burning in Population III massive stars (Ohkubo et al. [34]). Additional nuclear reactions and nonlinear couplings among multiple rates may likewise modify the pre-explosive composition. Extending the present temperature-resolved approach to a broader range of progenitor models and reaction networks will help establish how widely the sensitivity around T≃0.25 GKT \simeq0.25\,\mathrm{GK} applies across the PISN progenitor population. Within the models considered here, our results identify this temperature range as particularly sensitive to reaction-rate variations, highlighting the value of temperature-resolved analyses for understanding how nuclear reaction-rate uncertainties affect PISN outcomes.

Funding

This work is supported by JSPS KAKENHI grant Nos. JP24H00008, JP24H02245, JP24K00668, JP25H01273, JP25K01035, JP26K00723, and JP26KJ0832. HK was supported by the RIKEN Junior Research Associate Program.

Acknowledgments

HK thanks Masaaki Kimura for fruitful discussions.

Data availability

The data underlying this article will be shared upon reasonable request to the corresponding author.

References

  1. [1]Angulo, C., et al. 1999, Nucl. Phys. A, 656, 3
  2. [2]Barkat, Z., Rakavy, G., & Sack, N. 1967, Phys. Rev. Lett., 18, 379
  3. [3]Bertulani, C. A., & Kajino, T. 2016, Prog. Part. Nucl. Phys., 89, 56
  4. [4]Buchler, J. R., & Yueh, W. R. 1976, ApJ, 210, 440
  5. [5]Cassisi, S., Potekhin, A. Y., Pietrinferni, A., Catelan, M., & Salaris, M. 2007, ApJ, 661, 1094
  6. [6]Chugunov, A. I., DeWitt, H. E., & Yakovlev, D. G. 2007, Phys. Rev. D, 76, 025028
  7. [7]Cyburt, R. H., et al. 2010, ApJS, 189, 240
  8. [9]Dotter, A., & Paxton, B. 2009, A&A, 507, 1617
  9. [11]Farmer, R., Renzo, M., de Mink, S. E., Marchant, P., & Justham, S. 2019, ApJ, 887, 53
  10. [12]Ferguson, J. W., Alexander, D. R., Allard, F., Barman, T., Bodnarik, J. G., Hauschildt, P. H., Heffner-Wong, A., & Tamanai, A. 2005, ApJ, 623, 585
  11. [13]Fryer, C. L., Woosley, S. E., & Heger, A. 2001, ApJ, 550, 372
  12. [14]Fuller, G. M., Fowler, W. A., & Newman, M. J. 1985, ApJ, 293, 1
  13. [15]Fynbo, H. O. U., et al. 2005, Nature, 433, 136
  14. [16]Gabrielli, F., et al. 2024, MNRAS, 534, 151
  15. [17]Heger, A., Fryer, C. L., Woosley, S. E., Langer, N., & Hartmann, D. H. 2003, ApJ, 591, 288
  16. [18]Iglesias, C. A., & Rogers, F. J. 1993, ApJ, 412, 752
  17. [19]Irwin, A. W. 2012, Astrophysics Source Code Library, ascl:1211.002
  18. [20]Itoh, N., Hayashi, H., Nishikawa, A., & Kohyama, Y. 1996, ApJS, 102, 411
  19. [21]Jermyn, A. S., et al. 2023, ApJS, 265, 15
  20. [22]Kawashimo, H., Sawada, R., Suwa, Y., Moriya, T. J., Tanikawa, A., & Tominaga, N. 2024, MNRAS, 531, 2786
  21. [23]Kibédi, T., et al. 2020, Phys. Rev. Lett., 125, 182701
  22. [24]Kikuchi, Y., Hashimoto, M.-a., Ono, M., & Fukuda, R. 2015, PTEP, 2015, 063E01
  23. [26]Kunz, R., Fey, M., Jaeger, M., Mayer, A., Hammer, J. W., Staudt, G., Harissopulos, S., & Paradellis, T. 2002, ApJ, 567, 643
  24. [27]Langanke, K., & Martínez-Pinedo, G. 2000, Nucl. Phys. A, 673, 481
  25. [28]Marchant, P., Renzo, M., Farmer, R., Pappas, K. M. W., Taam, R. E., de Mink, S. E., & Kalogera, V. 2019, ApJ, 882, 36
  26. [29]Mukhamedzhanov, A. M., deBoer, R. J., Igaziev, B. F., Blokhintsev, L. D., Kadyrov, A. S., & Savin, D. A. 2024, Phys. Rev. C, 110, 055803
  27. [31]Nan, W., et al. 2024, Phys. Rev. C, 109, 045808
  28. [32]Nishimura, N., Fröhlich, C., & Rauscher, T. 2026, MNRAS, 546, stag152
  29. [33]Oda, T., Hino, M., Muto, K., Takahara, M., & Sato, K. 1994, At. Data Nucl. Data Tables, 56, 231
  30. [34]Ohkubo, T., Nomoto, K., Umeda, H., Yoshida, N., & Tsuruta, S. 2009, ApJ, 706, 1184
  31. [35]Paxton, B., Bildsten, L., Dotter, A., Herwig, F., Lesaffre, P., & Timmes, F. 2011, ApJS, 192, 3
  32. [36]Paxton, B., et al. 2013, ApJS, 208, 4

Paper details

vlnc:W_run59cjq
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents