Destratification in the Progenitor Interior of the Mg-rich Supernova Remnant N49B

Because both the observed Mg and Ne in the N49B ejecta should have been synthesized in the O-rich layer of the progenitor star, we assume that the Mg/Ne ratio purely reflects the elemental composition in that layer. Of course, although the explosive nucleosynthesis could change the elemental composition, the explosive Ne burning region is too narrow to contribute to a significant change in the abundance of the entire O-rich layer (see Appendix A and Figure 7). Therefore, we discuss here the cause of the high Mg/Ne ratio based on a comparison with that in pre-supernova models.

We analyzed 1,499 one-dimensional pre-supernova models (with the standard mass loss) provided by Sukhbold et al. (2018) to search for destratification events that reproduce a high Mg/Ne ratio as observed in N49B. As pointed out by Yadav et al. (2020), the disturbed shell activity is expected to reduce the amount of Ne in the O-rich layer as the Ne mixes inward and is burned. Not only in 3D simulations, some shell mergers has also been observed in 1D pre-supernova simulations (e.g., Woosley et al., 1995; Woosley & Heger, 2007; Rauscher et al., 2002; Tur et al., 2007; Sukhbold & Woosley, 2014), where the compactness is tend to be small (see section 3.4 in Sukhbold & Woosley, 2014). The interior of the star expands as nuclear burning is enhanced in the merged burning shell, which makes the compactness smaller. Therefore, some relationship between the Mg/Ne ratio and compactness is expected.

In Figure 2, we clearly see that a model with a high Mg/Ne mass ratio $>$ 1 has a small compactness, $M_{4}$ and $\mu_{4}$ (see the detail discussions in section 3.2). Here we have calculated the Mg/Ne mass ratios in the O-rich layer of pre-supernova models, which has a high oxygen mass fraction $>$ 0.4. The Mg-rich models appear around regions where the compactness changes drastically, except for the models that have a low He-core mass below 4 $M_{\odot}$ ($M_{\rm ZAMS}\lesssim 14M_{\odot}$).

In the low-mass models with $M_{\rm Hecore}<4~{}M_{\odot}$, the Mg-rich cases are mainly the result of Ne-burning shell intrusion into the outer region (Figure 3(a)). As an example, Figure 4 shows the mass fraction profiles and convective histories of the 12.02 $M_{\odot}$ and 12.75 $M_{\odot}$ pre-supernova models, where we have chosen the two similar-mass models that most easily compare the differences in elemental distributions. In the 12.02 $M_{\odot}$ model, the Ne-shell burning is not convective after the third O-shell burning (about 0.1 yr before the collapse), leaving a thin O/Mg/Si layer. On the other hand, in the 12.75 $M_{\odot}$ model, the Ne-shell burning becomes convective and breaks into the outer region shortly after $\sim$10^-3 yr before the collapse. The mixing process reduces the Ne fraction in the O-rich layer, resulting in a high Mg/Ne ratio in the O-rich layer.

In the high-mass models with $M_{\rm Hecore}>4~{}M_{\odot}$, the Mg-rich models are considered as the result of shell mergers between the O- and C-burning shells (Figure 3(b)). In this case, the strong O-shell burning produces a violent merger of shells. As an example, Figure 5 shows the mass fraction profiles and convective histories of the 15.28 $M_{\odot}$ and 15.90 $M_{\odot}$ pre-supernova models, where we have chosen the two similar-mass models that most easily compare the differences in elemental distributions. In the 15.28 $M_{\odot}$ model, the fourth convective C-burning shell ignites about 0.1 years before the collapse, expanding into the region between 2 $M_{\odot}$ and 3 $M_{\odot}$ at the same time as the first O-shell ignites. This C-burning shell does not merge with the inner shell until the collapse, forming a flat O/Ne layer in the region between $\sim$2 $M_{\odot}$ and $\sim$3 $M_{\odot}$. In the 15.90 $M_{\odot}$ model, the last Ne- and O-burning shells expand and merge with the outer C-burning shell after $\sim$10^-3 yr before collapse, producing a large convective zone that expands between $\sim$1.6 $M_{\odot}$ and $\sim$3 $M_{\odot}$. Here, the mixing is more extensive than in the low-mass cases, and Ne is burned more efficiently, resulting in a higher abundance of Mg.

The most significant difference between the shell destruction processes for the low-mass and high-mass models is the degree of Si intrusion into the O-rich layer. In the former, the Ne-burning shell, where the Si fraction is low, breaks into the outer shell, producing a low Si/O ratio in the O-rich layer. On the other hand, in the latter, the O-burning shell, where the Si fraction is high, merges with the outer C-burning shell. As a result, the O/Ne layer disappears and a new O/Si layer is formed (see the mass region between 2 $M_{\odot}$ and 3 $M_{\odot}$ in the bottom of Figure 5). In some extreme cases, the entire Si-rich layer is incorporated into the O-rich layer, which resembles the multi-dimensional simulation of Yadav et al. (2020). Therefore, the Si/O mass ratio would also be important to characterize these shell-merger phenomena.

Indeed, the Mg/Ne vs. Si/Ne plot (Figure 6) shows this tendency very well. We found that N49B has a low Si abundance, which currently supports the Ne-shell intrusion into the outer shell, which can be seen in the low-mass progenitors, as the origin of the Mg-rich ejecta in the remnant. However, there is a possibility that the Si abundance is diluted by mixing with the ISM gas, as seen in its sub-solar abundance. Also, in the low-mass progenitor case, the effects of explosive nucleosynthesis show unfavorable results for too large Mg/Ne mass ratios greater than 1, as seen in N49B (see Appendix A and Figure 7). In addition, there would be uncertainties in pre-supernova models, thus we refrain from drawing strong conclusions on this topic. We note that the uncertainty in nuclear reaction rates such as ¹²C($\alpha$,$\gamma$)¹⁶O also changes the values in Figure 6.

As shown in Figure 2, the destratification processes that produce a high Mg/Ne ratio in the pre-supernova interior would also affect the explosion itself. Remarkably, almost all the Mg-rich models have the lowest $\mu_{4}$ values (see also Figure 2 top right). This is thought to be achieved by the expansion of the convection zone due to the intense shell-burning activity, which would lead to a low infall-mass rate into the stalled shock. Another interesting feature would be a relatively high $M_{4}$ value, which implies a high neutrino luminosity from its proto-neutron star, in the Mg-rich models with $M_{\rm Hecore}<4M_{\odot}$ (see blue points in Figure 2). Both the low infall-mass rate and the high neutrino luminosity in the low-mass Mg-rich progenitors imply the easiest explosions among all the models, which may produce a more spherical and low-energy explosion. In the case of N49B, a relatively high explosion energy of $\sim$2–4$\times 10^{51}$ erg was estimated (Hughes et al., 1998) and its shape has a smooth spherical form, which does not support all the features. While accurate estimates of explosion energy are always difficult in supernova remnant studies, future systematic discussions of this topic will be very interesting to understand the impact of intense shell activity on supernova explosions.

In many O-rich supernova remnants, their O/Ne layer appears to be intact without destratification. For example, famous O-rich supernova remnants such as 1E 0102.2–7219, G292.0+1.8, and Puppis A show high O/Ne ratios of $\sim$0.4 (e.g., Flanagan et al., 2004; Kamitsukasa et al., 2014; Katsuda et al., 2010), suggesting that destratification processes may not be a universal phenomenon in the evolution of massive stars (see red data points in Figure 6 for the normal O/Ne layer cases). Future systematic studies of the elemental composition of the O-rich ejecta in core-collapse supernova remnants will provide a fruitful discussion of the fraction of massive stars that undergo the destratification processes.

Table 2 shows the rate of destratificated Mg-rich events in the pre-supernova models in Sukhbold et al. (2018), which shows that 10% of the pre-supernova models have a Mg-rich (Ne-poor) oxygen layer. In particular, the models in the mass ranges 12–13 $M_{\odot}$ and 15–16 $M_{\odot}$, where convection and shell burning conditions change significantly, have high Mg-rich event rates of 19% and 56%, respectively. In contrast to this, there are no Mg-rich events in the 19–22 $M_{\odot}$ and 24–27 $M_{\odot}$ mass regions where the compactness increases steadily. Given the initial mass function, the progenitors of $\lesssim$ 16 $M_{\odot}$ could account for the majority of destratification events. However, the model uncertainties are thought to be very large.

As pointed out in Sukhbold et al. (2018), the C- and O-convective shells are separated from coupling by only a single thin zone in some models, which may actually appear as other shell merger events. Therefore, differences in the treatment of multidimensional convection and overshoot may affect the appearance of shell mergers. This means that observational calibration will be important to understand the true rate of destratification events. As a first step in the observational approach to this topic, we discuss here two Mg-rich/Ne-poor candidates, G284.3-1.8 and Cassiopeia A, that are likely to have undergone the destratification during their pre-supernova evolution.

G284.3-1.8 is known as a Galactic supernova remnant hosting a high-mass X-ray binary (HMXB) in the central region. Williams et al. (2015) have reported a high Mg/Ne abundance ratio of $\sim$3.5 in the western region, which is converted to a mass ratio of $\sim$1.3 using the solar abundance ratios in Anders & Grevesse (1989). As can be seen in Figure 6, the elemental composition of its ejecta is similar to that of N49B. The most significant difference from the N49B case is that only a portion of the western side has the Mg-rich composition. If the shell merger causes the Mg-rich ejecta, the Mg-rich composition is expected to see in the entire O-rich layer because the O-rich layer is well mixed in by convection. To produce the asymmetric distribution of Mg/Ne abundance in the remnant, some special shell destruction process may be needed, however we cannot conclude it from the current observed facts alone.

Cassiopeia A is one of the most well-studied core-collapse supernova remnants in our Galaxy. Vink et al. (1996) have reported significantly low Ne and Mg abundances in the ejecta of Cassiopeia A. The Ne/O mass ratio was estimated to be $\sim$0.01-0.02 (see also Hwang & Laming, 2012), suggesting the possibility of the Ne consumption by shell merger. On the other hand, the Mg/Ne mass ratio of $\sim$0.3 is not large, in contrast to the theoretical predictions for shell mergers with Ne depletion. Therefore, it cannot be explained by the destratification process alone, as investigated in this paper.

As discussed above, few candidates can currently be concluded to have undergone shell destruction like N49B; a comprehensive survey of O-rich remnants to show what fraction of such events exist would be beneficial to the discussion of the internal structure of progenitors of supernova remnants.