Progenitor and explosion properties of SN 2023ixf estimated based on a light-curve model grid of Type II supernovae

Type II supernovae (SNe) are explosions of massive stars that retain hydrogen-rich envelopes until the time of their explosion. Type II SNe are the most frequently observed type of core-collapse SNe (e.g., [Perley et al. (2020)]), and they provide vital information to understand the standard explosion mechanism of core-collapse SNe (e.g., [Burrows & Vartanyan (2021)]). However, many mysteries related to Type II SNe remain unsolved. For example, photometric and spectroscopic observations of Type II SNe shortly after their explosion indicate the existence of confined circumstellar matter (CSM) around red supergiants (RSGs) that are about to explode (e.g., [Yaron et al. (2017), Förster et al. (2018)]). Concretely, optical light curves of Type II SNe rise much faster than those predicted without considering the existence of a CSM (González-Gaitán et al., 2015), and the fast rise can be explained by the existence of a dense, confined CSM around the progenitors of Type II SNe (e.g., Moriya et al. (2011); Nagy & Vinkó (2016); Morozova et al. (2017)). In addition, SN 2020tlf showed a bright precursor lasting for 130 days, during which the dense, confined CSM could have likely formed (Jacobson-Galán et al., 2022). However, the physical mechanisms forming the dense, confined CSM have not been revealed.

SN 2023ixf is one of the closest Type II SNe observed in the last decade, and their detailed observations provided us with valuable information on Type II SNe and their progenitors. SN 2023ixf was discovered immediately after the explosion (Itagaki, 2023), and its intensive follow-up observations were conducted in broad wavelengths (e.g., Berger et al. (2023); Grefenstette et al. (2023); Sgro et al. (2023); Hosseinzadeh et al. (2023); Koenig (2023); Teja et al. (2023); Jacobson-Galán et al. (2023); Hiramatsu et al. (2023); Zhang et al. (2023); Vasylyev et al. (2023); Smith et al. (2023); Bostroem et al. (2023); Yamanaka et al. (2023); Zimmerman et al. (2024); Chandra et al. (2024); Li et al. (2024); Singh et al. (2024)). These observations revealed the existence of a dense, confined CSM in SN 2023ixf. SN 2023ixf also demonstrated that the future multi-messenger observations of nearby SNe will also be interesting (e.g., Sarmah (2024); Guetta et al. (2023); Kheirandish & Murase (2023); Murase (2024); Ravensburg et al. (2024)). The progenitor of SN 2023ixf has been identified, but its nature is actively discussed (e.g., Kilpatrick et al. (2023); Jencson et al. (2023); Pledger & Shara (2023); Flinner et al. (2023); Van Dyk et al. (2024); Niu et al. (2023); Qin et al. (2023); Dong et al. (2023); Liu et al. (2023); Neustadt et al. (2024); Xiang et al. (2024)).

While nearby Type II SNe like SN 2023ixf can only appear once in a decade or so, thousands of distant Type II SNe are predicted to be discovered in a year in the era of the Vera C. Rubin Observatory’s Legacy Survey of Space and Time (LSST, LSST Science Collaboration et al. (2009)). It will be impossible to perform intensive follow-up observations of such a large number of Type II SNe individually as conducted for SN 2023ixf. A systematic approach is required to infer the properties of a large number of Type II SNe, for which only limited photometric information is expected to be available. To estimate the physical properties of a large number of Type II SNe, Moriya et al. (2023) constructed a model grid of Type II SNe containing about 228,000 explosion models with different combinations of progenitors, explosion properties, and CSM properties. This model grid has been used to estimate the properties of a large number of Type II SNe discovered by ongoing extensive transient surveys such as Zwickey Transient Facility (ZTF, e.g., Subrayan et al. (2023); Silva-Farfán et al. (2024)).

In this paper, we present our estimates of the progenitor and explosion properties of SN 2023ixf based on the model grid of Moriya et al. (2023). We discuss the distribution of the parameters estimated by the model grid and compare our parameter estimates with those estimated by other studies. This study allows us to test the ability of the model grid to estimate the properties of Type II SNe by using the well-observed Type II SN 2023ixf. We estimate the properties of SN 2023ixf by using only the pre-computed model grid in Moriya et al. (2023) without additional calculations. Using the pre-computed model grid, we can quickly identify the rough nature of SN 2023ixf, based on which more detailed modeling of its light curves and spectra can be performed.

The rest of this paper is organized as follows. We first present our methods to compare the model grid and the observations of SN 2023ixf in Section 2. We present the results of the model comparison and our parameter estimation in Section 3. We discuss our results in Section 4 and conclude this paper in Section 5.

Figure 1 presents the cumulative rank distributions of the progenitor and explosion properties obtained using the g band light curves. The best fit model has the ZAMS mass of 10 $\mathrm{M}_{\odot}$, explosion energy of $2.0~{}\mathrm{B}$, ${}^{56}\mathrm{Ni}$ mass of 0.06 $\mathrm{M}_{\odot}$, mass-loss rate of $10^{-2.0}~{}\mathrm{M}_{\odot}~{}\mathrm{yr^{-1}}$, $\beta$ of 3.5, and CSM radius of $6\times 10^{14}~{}\mathrm{cm}$. The CSM mass of the model is 0.85 $\mathrm{M}_{\odot}$. The best-fit model is presented in Figure 2. Although we do not consider velocity information when we compare the model grid, we can find that the photospheric velocity evolution also matches well in the best-fit model.

When we look into the cumulative ranking distribution, the smaller ZAMS mass progenitors are preferred, and most of the best-fit models have the ZAMS mass of 10 $\mathrm{M}_{\odot}$. Figure 3 presents the best-fit models for each progenitor model. The higher mass models tend to have slow-declining light curves than those observed for SN 2023ixf. Although the best-fit model has the explosion energy of $2.0~{}\mathrm{B}$, the models with $2.5~{}\mathrm{B}$ and $3.0~{}\mathrm{B}$ also tend to have small $\chi^{2}$ values. Thus, the explosion energy is estimated at $2.0-3.0~{}\mathrm{B}$. The most preferred ${}^{56}\mathrm{Ni}$ mass is 0.06 $\mathrm{M}_{\odot}$. The mass-loss rate of the best matching models are $10^{-2.0}~{}\mathrm{M}_{\odot}~{}\mathrm{yr^{-1}}$, but the models with $10^{-3.0}~{}\mathrm{M}_{\odot}~{}\mathrm{yr^{-1}}$ and $10^{-2.5}~{}\mathrm{M}_{\odot}~{}\mathrm{yr^{-1}}$ have relatively low ranks. Thus, the mass-loss rate can be estimated to be in the range of $10^{-2.0}-10^{-3.0}~{}\mathrm{M}_{\odot}~{}\mathrm{yr^{-1}}$. Although $\beta=\mathbf{3.5}$ gives the smallest $\chi^{2}$, the models with the other $\beta$ also provide a relatively low ranking, and we conclude that it is difficult to constrain $\beta$. Since $\beta$ determines the CSM density structure at the immediate vicinity of the progenitor, the CSM mass is affected by $\beta$. If we fix the progenitor mass ($10~{}\mathrm{M}_{\odot}$), mass-loss rate ($10^{-2.0}~{}\mathrm{M}_{\odot}~{}\mathrm{yr^{-1}}$), and CSM radius ($6\times 10^{14}~{}\mathrm{cm}$), the CSM mass ranges from $0.21~{}\mathrm{M}_{\odot}$ ($\beta=0.5$) to $1.3~{}\mathrm{M}_{\odot}$ ($\beta=5.0$). Thus, the CSM mass with the order of $0.1~{}\mathrm{M}_{\odot}$ is preferred, but the exact CSM density structure is not well constrained. The models with the CSM radius of $6\times 10^{14}~{}\mathrm{cm}$ give the best-fit models, but the CSM radii of $8\times 10^{14}~{}\mathrm{cm}$ and $10^{15}~{}\mathrm{cm}$ provide the models with relatively low ranking.

So far, we have only looked into the comparison results obtained using the g band light curve. Figure 4 shows the cumulative ranking distribution obtained by using the g and r band light curves. In this case, the best fit light-curve model has the ZAMS mass of 10 $\mathrm{M}_{\odot}$, explosion energy of $2.5~{}\mathrm{B}$, ${}^{56}\mathrm{Ni}$ mass of 0.06 $\mathrm{M}_{\odot}$, mass-loss rate of $10^{-3.0}~{}\mathrm{M}_{\odot}~{}\mathrm{yr^{-1}}$, $\beta$ of 5.0, and CSM radius of $10^{15}~{}\mathrm{cm}$. The CSM mass in this model is 0.52 $\mathrm{M}_{\odot}$. The best-fit light-curve model and its photospheric velocity evolution are presented in Figure 5. We find that adding the r band information does not change the results of the parameter estimations much. The best-fit parameters for the explosion energy, mass-loss rate, and $\beta$ change slightly, but the overall ranking distributions remain unchanged.

Finally, we show the comparison results based on the pseudo-bolometric light curve. The cumulative ranking distribution is shown in Figure 6, and the best fit light-curve model is presented in Figure 7. The best fit model has the ZAMS mass of 10 $\mathrm{M}_{\odot}$, explosion energy of $2.0~{}\mathrm{B}$, ${}^{56}\mathrm{Ni}$ mass of 0.04 $\mathrm{M}_{\odot}$, mass-loss rate of $10^{-3.0}~{}\mathrm{M}_{\odot}~{}\mathrm{yr^{-1}}$, $\beta$ of 4.5, and CSM radius of $10^{15}~{}\mathrm{cm}$. This model has a CSM mass of 0.44 $\mathrm{M}_{\odot}$. A large difference in the parameter estimations from those obtained by the g and r band light curves is in the ${}^{56}\mathrm{Ni}$ mass. While most of the best-fit models based on the g and r band light curves have the ${}^{56}\mathrm{Ni}$ mass of 0.06 $\mathrm{M}_{\odot}$, the best-fit models based on the pseudo-bolometric light curve has the ${}^{56}\mathrm{Ni}$ mass of 0.04 $\mathrm{M}_{\odot}$. The ranking distribution of the other parameters is not significantly different from those obtained by the g and r band light curves.

Based on the $\chi^{2}$ ranking distribution presented in the previous section, SN 2023ixf can be estimated to have the progenitor ZAMS mass of 10 $\mathrm{M}_{\odot}$, the explosion energy of $2.0-3.0~{}\mathrm{B}$, the ${}^{56}\mathrm{Ni}$ mass of $0.04-0.06~{}\mathrm{M}_{\odot}$, the mass-loss rate of $10^{-3.0}-10^{-2.0}~{}\mathrm{M}_{\odot}~{}\mathrm{yr^{-1}}$, and the dense CSM radius of $(6-10)\times 10^{14}~{}\mathrm{cm}$. The wind structure $\beta$ is not well constrained. These parameter estimations are obtained using the pre-computed model grid in Moriya et al. (2023) without additional computations. Still, the estimated parameters are close to those obtained by detailed numerical investigations. For example, Bersten et al. (2024) and Martinez et al. (2024) estimated the ZAMS mass of 12 $\mathrm{M}_{\odot}$, the explosion energy of $1.2\times 10^{51}~{}\mathrm{erg}$, the ${}^{56}\mathrm{Ni}$ mass of 0.05 $\mathrm{M}_{\odot}$, the mass-loss rate of $3\times 10^{-3}~{}\mathrm{M}_{\odot}~{}\mathrm{yr^{-1}}$ with the wind velocity of $115~{}\mathrm{km~{}s^{-1}}$, the dense CSM radius of $8\times 10^{14}~{}\mathrm{cm}$, and $\beta=5$ through the detailed numerical modeling. The estimated parameters are close to those we estimated in this paper. Our approach gives us a better idea of the uncertainties in the parameter estimation thanks to the $\chi^{2}$ ranking provided by the large model grid.

The parameter estimation approach we took in this paper is based on the pre-existing model grid. Thus, this paper does not aim to perfectly fit the light curves. As we can find in the best-fit models (Figs. 2 and 5), our light-curve models tend to decline slower than the observed light-curves, especially in the r band. The light-curve decline rate is closely related to the progenitor structure, including the hydrogen-rich envelope mass (e.g., Dessart et al. (2013); Moriya et al. (2016)). The assumed progenitor models from Sukhbold et al. (2016) might not have been suitable for precisely reproducing the light-curve properties of SN 2023ixf. Still, the parameter estimation by the model grid can provide a good first guess for starting the detailed modeling of individual SNe.

Our models prefer a low-mass progenitor for SN 2023ixf. The progenitor mass of SN 2023ixf has been debated because the progenitor mass estimates from the pre-explosion images range from low mass (around $10~{}\mathrm{M}_{\odot}$, e.g., Kilpatrick et al. (2023); Pledger & Shara (2023); Van Dyk et al. (2024); Xiang et al. (2024)) to high mass (around $20~{}\mathrm{M}_{\odot}$, e.g., Jencson et al. (2023); Niu et al. (2023); Qin et al. (2023); Soraisam et al. (2023); Liu et al. (2023)). Our mass estimate aligns with the low-mass progenitor estimates consistent with the results from the other light-curve modeling efforts (e.g., Bersten et al. (2024)). Our progenitor mass estimates are based on the progenitor model grid of Sukhbold et al. (2016). Adopting different progenitor models with different stellar evolution parameters can lead to different progenitor mass estimates due to inherent degeneracies with the parameters especially in the progenitor mass and radius (e.g., Goldberg et al. (2019); Goldberg & Bildsten (2020)).

The properties of the dense, confined CSM estimated with our method are consistent with those obtained by other studies. The previous estimates based on the early-phase light-curve properties are around $10^{-3}-10^{-2}~{}\mathrm{M}_{\odot}~{}\mathrm{yr^{-1}}$ if we adopt the wind velocity of 10 $\mathrm{km~{}s^{-1}}$ as assumed in our model grid (e.g., Teja et al. (2023); Hiramatsu et al. (2023); Jacobson-Galán et al. (2023); Zhang et al. (2023); Zimmerman et al. (2024); Soker (2023)). The estimated mass-loss rate is also consistent with that estimated by comparing spectroscopic observations and models (e.g., Jacobson-Galán et al. (2023); Bostroem et al. (2023)). The CSM density scales as $\rho_{\mathrm{CSM}}\propto\dot{M}/v_{\mathrm{wind}}$ (Eq. 1). In our model, we assume the terminal wind velocity of 10 $\mathrm{km~{}s^{-1}}$. If we adopt the terminal wind velocity of 100 $\mathrm{km~{}s^{-1}}$ as observed in the high-resolution spectra of SN 2023ixf (Smith et al., 2023), the mass-loss estimate increases by a factor of 10. However, the observed velocity of 115 $\mathrm{km~{}s^{-1}}$might be affected by the radiative acceleration of the unshocked CSM (e.g., Tsuna et al. (2023)). The radius of the dense, confined CSM we estimated, $(6-10)\times 10^{14}~{}\mathrm{cm}$, is also consistent with other estimates (e.g., Teja et al. (2023); Jacobson-Galán et al. (2023); Hiramatsu et al. (2023); Zhang et al. (2023); Bostroem et al. (2023)), although the early-phase ultraviolet observations provide a smaller estimate of the CSM radius (Zimmerman et al., 2024).

The mass-loss rate estimates based on the pre-explosion progenitor properties as well as X-ray observations are around $10^{-4}~{}\mathrm{M}_{\odot}~{}\mathrm{yr^{-1}}$ (e.g., Grefenstette et al. (2023); Chandra et al. (2024); Jencson et al. (2023); Niu et al. (2023); Qin et al. (2023); Soraisam et al. (2023)). This discrepancy likely originates from the aspherical nature of the dense, confined CSM identified by spectroscopic and polarimetric observations of SN 2023ixf (e.g., Vasylyev et al. (2023); Smith et al. (2023); Singh et al. (2024)). Early light curves and spectra in the optical wavelengths are affected mainly by the dense part of the CSM.

We have presented our results comparing the observed light curve of SN 2023ixf and the pre-existing synthetic light-curve model grid from Moriya et al. (2023). By simply evaluating the $\chi^{2}$ value for every model and rank the models based on the $\chi^{2}$ value, we estimated that SN 2023ixf have the progenitor ZAMS mass of 10 $\mathrm{M}_{\odot}$ (adopting the progenitor models from Sukhbold et al. (2016)), the explosion energy of $2.0-3.0~{}\mathrm{B}$, the ⁵⁶Ni mass of $0.04-0.06~{}\mathrm{M}_{\odot}$, the mass-loss rate of $10^{-3.0}-10^{-2.0}~{}\mathrm{M}_{\odot}~{}\mathrm{yr^{-1}}$, and the dense, confined CSM radius of $(6-10)\times 10^{14}~{}\mathrm{cm}$. The wind structure parameter $\beta$ is not constrained well. The estimated parameters match those estimated by the previous detailed modeling of SN 2023ixf.

Our parameter estimation is based on the pre-existing model grid, and we did not perform any additional calculations to estimate the properties of SN 2023ixf. Thus, we can find some discrepancies between the best-matching models and observations, but our approach provides a reasonable parameter estimation. Our simple approach is sufficient to obtain a rough parameter estimation of Type II SNe, building upon which we can perform more detailed modeling. In particular, such a simple, quick parameter estimation will be essential to characterize thousands of Type II SNe discovered each year in the era of LSST. This study demonstrated that we can have a reasonable quick parameter estimation for them by using the pre-existing model grid and extracting information on the explosion and mass-loss mechanisms of RSGs.

The authors thank Jared Goldberg and the anonymous referee for constructive comments that improved this paper. TJM is supported by the Grants-in-Aid for Scientific Research of the Japan Society for the Promotion of Science (JP24K00682, JP24H01824, JP21H04997, JP24H00002, JP24H00027, JP24K00668). This research was supported by the Australian Research Council (ARC) through the ARC’s Discovery Projects funding scheme (project DP240101786). This work was supported by the JSPS Core-to-Core Program (grant number: JPJSCCA20210003).