Stellar Origin of ¹⁵N-rich Presolar SiC Grains of Type AB: Supernovae with Explosive Hydrogen Burning

Primitive meteorites contain small amounts of dust with highly anomalous isotopic compositions compared to solar system materials (e.g., Zinner 2014; Nittler & Ciesla 2016). These presolar grains formed around stars that ended their lives prior to the solar system formation. So far, tens of thousands of presolar silicon carbide (SiC) grains have been characterized for their C, N, and Si isotopic compositions, based on which five main groups have been defined (Hoppe et al. 1994). The isotopic signatures of mainstream (MS) and X grains definitively link them to low-mass asymptotic giant branch (AGB) stars and supernovae (SNe), respectively. The isotopic compositions of the other types of grains are less diagnostic of their progenitor stars, and of these, the AB group grains have the most ambiguous stellar origins (Amari et al. 2001b, hereafter A01b; Nittler & Ciesla 2016).

The AB SiC grains⁴ (∼5%–6% of all SiC) are highly enriched in ¹³C (¹²C/¹³C < 10) with a wide range of ¹⁴N/¹⁵N ratios (∼50 − >10,000). So far, J-type carbon stars and born-again AGB stars have been proposed as sources of AB grains (Alexander 1993; A01b). Hedrosa et al. (2013) reported observational evidence for higher-than-solar ¹⁴N/¹⁵N ratios in six of seven J-type carbon stars, but the lower-than-solar value for the one exception was a lower limit. State-of-the-art born-again AGB stellar models by Herwig et al. (2011) predict ¹⁴N/¹⁵N ratios up to ∼50,000 in the He intershell region with ¹²C/¹³C < 10 (Jadhav et al. 2013). Thus, neither scenario seems able to explain the lower-than-solar ¹⁴N/¹⁵N ratios of ∼50% of AB grains, though they cannot be completely ruled out due to poor statistics in the observational measurements and uncertainties in stellar models. Novae are well-known stellar explosions in which high-temperature H burning can produce extremely low ¹⁴N/¹⁵N (0.2−∼1000; José et al. 2004; José & Hernanz 2007) and ¹²C/¹³C (<2), and have been proposed as the parent stars of the so-called putative nova grains (PNGs; <0.1% of all SiC) that have extreme ¹³C and ¹⁵N enrichments (Amari et al. 2001a). Despite the systematic difference between ¹⁵N-rich AB grains and PNGs, Liu et al. (2016a, hereafter L16a) pointed out that the isotopic signatures of ¹⁵N-rich AB grains could also reflect nucleosynthetic processes in novae but probably with lower masses relative to PNGs.

Intriguingly, an enhancement in two p-process isotopes, ⁹²Mo and ⁹⁴Mo, was found in one AB grain studied by Savina et al. (2003), raising the possibility of an SN origin, the most likely stellar site for p-process nucleosynthesis. SN models (Woosley & Heger 2007, hereafter WH07; Pignatari et al. 2013b), however, generally predict extremely high ¹²C/¹³C ratios (>10⁵) in the SN C-rich He/C zone that must have contributed a majority of precursor material to the SiC (e.g., Table 2 of Xu et al. 2015) because thermodynamic equilibrium models require C/O > 1 to form SiC. Note that the stable CO molecule could be dissociated in SNe by electrons arising from ⁵⁶Co decay, allowing carbonaceous grains to grow even in O > C conditions (e.g., Clayton 2013). However, quasi-equilibrium calculations (Ebel & Grossman 2001) show that although graphite could be stable in O-dominated SN zones, SiC is not stable under such O-rich conditions. Moreover, these zones also have extremely high ¹²C/¹³C ratios. More importantly, AB and PNG grains show similar Raman characteristics to mainstream SiC grains from low-mass AGB stars after correcting for the isotope-induced peak shifts and broadening (Liu et al. 2016b), suggesting similar condensation environments in their parent stars, i.e., equilibrium condensation with C/O > 1. Nittler & Hoppe (2005) reported a presolar SiC grain with the Si and Ti isotopic signature of X grains from SNe, but C and N isotope ratios similar to PNGs, raising the possibility of explosive H-burning occurring in the C-rich He/C zone during an SN explosion. Furthermore, L16a found a subgroup of ¹³C- and ¹⁵N-rich presolar SiC, C2 grains (<0.1% of presolar SiC), with neutron-capture isotopic signatures (^29,30Si, ⁵⁰Ti excesses), pointing to an SN origin. Pignatari et al. (2015, hereafter P15) investigated H ingestion into the He/C zone during the pre-SN stage of a massive star's evolution followed by the core-collapse SN explosion. They found that a H-ingestion event before the SN explosion, which is predicted by the stellar model calculations, can switch off the convective mixing process, leaving some H in the He/C zone during the explosion (see P15 and L16a for model details). The combination of the pre-SN H-ingestion event followed by the SN explosive nucleosynthesis can lead to low ¹²C/¹³C and ¹⁴N/¹⁵N, and high ²⁶Al/²⁷Al in the H-ingested He/C zone. Thus, SNe should also be considered as possible sources of ¹⁵N-rich AB grains and PNGs.

The SiC grains in this study were extracted from the Murchison CM2 chondrite using the CsF isolation method described by Nittler & Alexander (2003). Three presolar SiC grain mounts, labeled #1, #2, and #3, were prepared by dispersing grains separated in size by sedimentation (∼1 μm) on high purity Au foils from a water suspension. The grains were subsequently pressed into the Au foils with a flat sapphire disk. The ¹³C- and ¹⁵N-rich presolar SiC grains analyzed in our previous study were all found on mount #3 (L16a). The AB grains and PNGs in this study were nondestructively identified with a Raman microscope by their lower-shifted Raman TO peak (Liu et al. 2016b). Isotopic analyses of C, N, Si, Mg–Al, and Ca–Ti were performed with the NanoSIMS 50L ion microprobe at the Carnegie Institution. Analytical procedures are described in L16a. The isotopic data are reported in Table 1 and all the plots with 1σ uncertainties. We relaxed the definition of AB grains (¹²C/¹³C < 10) to ¹²C/¹³C < 16 because, for instance, grain M2-A3-G1225 has a ¹²C/¹³C ratio of 13 but a ²⁶Al/²⁷Al ratio of 0.02, an order of magnitude higher than those of MS grains. Among the potential sources of AB grains discussed above, at present, only novae and SNe appear likely to produce ¹⁵N-rich N, indicating that it may be useful to re-divide the group into: group 1 grains with ¹⁴N/¹⁵N < 440 (solar value) and group 2 grains with ¹⁴N/¹⁵N ≥ 440. The ¹⁴N-rich and ¹⁵N-rich AB grains are therefore referred to hereafter as AB1 and AB2 grains, respectively.

In addition to AB grains, the rare PNGs and C2 grains also have ¹²C/¹³C < ∼10, but have quite restricted ¹⁴N/¹⁵N ratios (∼5 to ∼50) in contrast to AB grains (∼20–10,000; Figure 1(a)). Also, there exist distinctive differences between AB1 and AB2 grains: (1) presolar SiC with ¹²C/¹³C lower than the equilibrium CNO cycle value of 3.5 are all ¹⁵N-rich (Table 1); (2) for AB1 grains, the lower the ¹⁴N/¹⁵N ratio, the higher the ²⁶Al/²⁷Al ratio, whereas AB2 grains show no correlation in Figure 1(a); (3) the linear fits to AB grain Si three-isotope data show that compared to AB2 grains, AB1 grains are more enriched in ³⁰Si relative to ²⁹Si (Figure 1(b)). Our N and Al isotopic data for AB1 grains are well correlated and show less scatter than the literature data from A01b, likely indicating less contamination in our study. This is expected because we removed adjacent materials around the grains of interest with a focused ion beam instrument and performed isotope analyses in imaging mode to further exclude contamination during data reduction. Moreover, the size of the O⁻ beam of the NanoSIMS 50L used here is a factor of several smaller than that of the ims-3f instrument used by A01b, which likely further reduced the amount of contamination sampled during our analyses.

Zoom In Zoom Out Reset image size

Three out of eight AB1 grains and one PNG from this study had negative δ^33,34S values outside of 1σ errors (Figure 2(a)), i.e., ³²S excesses, which could result from the decay of radiogenic ³²Si (t_1/2 = 153 years). Novae are predicted to have normal or positive δ³³S compositions (Parikh et al. 2014) and thus cannot explain the ³²S excesses observed in these grains. Pignatari et al. (2013b) first identified ³²S excesses in type C SiC grains as most likely indicating the initial presence of radiogenic ³²Si that was produced by neutron capture in the so-called neutron-burst zone (Meyer et al. 2000) within the He/C zone during an SN explosion. Later, Fujiya et al. (2013) found such radiogenic ³²Si signatures in AB grains and pointed out that the ³²Si excesses could also be produced by the intermediate neutron-capture process (i-process; Cowan & Rose 1977) in born-again AGB stars with typical neutron densities of ∼10¹⁵ cm⁻³. Although the AB grains of Fujiya et al. (2013) were not measured for N, we used the ²⁶Al/²⁷Al ratios of these grains to tentatively separate AB1 and AB2 grains based on a dividing value of ²⁶Al/²⁷Al of 0.003 (Figure 1(a)), shown in Figure 2(a). As a matter of fact, two of the three grains from Fujiya et al. (2013) with ³²S excesses are most likely AB1 grains. While the third grain was not measured for its Mg–Al isotopes, it is more enriched in ³⁰Si relative to ²⁹Si, increasing the likelihood of it also being an AB1 grain (Figure 1(b)). In contrast, none of the AB2 grains show any ³²S excesses, and the ³²S excesses therefore seem to be mainly associated with the ¹⁵N-enriched AB1 grains, possibly challenging the born-again AGB scenario suggested by Fujiya et al. (2013). Finally, five out of 12 AB1 grains from our study (Figure 2(b)) show clear neutron-capture isotopic signatures, i.e., large excesses in ⁵⁰Ti (δ⁵⁰Ti ≥ ∼200‰). In contrast, although large ³²S excesses are found in two out of four PNGs, none of the eight PNGs studied so far show any ⁵⁰Ti enhancement. All the AB grains and PNGs show quite close-to-solar δ^46,47Ti values, indicating close-to-solar initial metallicities for their parent stars (Alexander & Nittler 1999).

Zoom In Zoom Out Reset image size

Overviews of nova nucleosynthesis in the context of PNG isotopic signatures are given by José et al. (2004) and José & Hernanz (2007). To summarize, although PNGs show isotopic signatures similar to nova nucleosynthetic yields, several fundamental inconsistencies make it problematic to link AB1 grains and PNGs to classical novae. (1) In the context of nova nucleosynthesis, the Si isotope ratios of PNGs indicate that these grains are more likely to originate from ONe novae (1.15–1.35 M_⊙), resulting in large excesses in ³⁰Si relative to ²⁹Si. Figure 3(a), however, shows that the N and Al isotopic data of PNGs can be best explained by CO novae with 1.00–1.15 M_⊙ and probably even lower masses, while the ¹⁵N and ²⁶Al correlation of AB1 grains can be better explained by predictions for a 1.15 M_⊙ ONe nova, both of which are completely contrary to the Si isotope constraints. (2) One potential solution to the inconsistency in (1) is that if in the parent CO novae of AB1 grains, the companion low-mass stars had evolved to their AGB phases, the accreted H-rich envelope material onto the white dwarfs would be ¹⁴N- and ²⁶Al-enriched (¹⁴N/¹⁵N > 1000, ²⁶Al/²⁷Al = 0.001–0.01; Palmerini et al. 2011) compared to the solar-composition envelope adopted in the models. This would lead to a better match with the AB1 grain data by CO nova models, but requires finely tuned timing since the AGB phase only lasts on the order of ∼10⁶ years. (3) A strong piece of evidence against a nova origin is the neutron-capture isotopic signatures preserved in these grains. In detail, AB1 grains generally show ³²S and ⁵⁰Ti excesses, both of which point to neutron irradiation that does not occur in novae (José & Hernanz 2007). Although none of the PNGs analyzed so far show Ti isotopic anomalies indicating neutron capture, two out of four PNGs had large ³²S excesses. The disagreement between isotopic signatures of Si and S measured in PNGs and nova models is illustrated in Figure 3(b). Thus, it is unlikely that AB1 grains and at least some of the PNGs originated from novae. Finally, stellar models for lower-than 1.00 M_⊙ CO novae predict higher-than-solar ¹⁴N/¹⁵N and could potentially explain AB2 grains that do not show neutron-capture isotopic signatures (Haenecour et al. 2016).

Zoom In Zoom Out Reset image size

P15 qualitatively investigated the effect of H ingestion in the He/C zone for a 25 M_⊙ SN by varying the amount of ingested H from 1.2% (-H) to 0.0024% (-H500) in the He/C zone, and also by adopting two SN shock models: 25T with a peak temperature at the bottom of the He/C zone of 23 × 10⁸ K (reproducing the SN shock condition of a 15 M_⊙ stellar model), and 25d with a peak temperature of 7 × 10⁸ K (the original SN shock condition of a 25 M_⊙ model). By comparing with the P15 models, L16a showed that the C, N, and Al isotope ratios of C2 grains can be matched by the 25d models. In addition, the neutron-burst process can still take place in regions of the He/C zone where insufficient amounts of H (≤0.0024%) are ingested. The C2 grain data, therefore, can be explained in the context of SN nucleosynthesis by considering heterogeneous conditions and asymmetries within the He/C zone before the SN shock, possibly triggered by the H-ingestion event.

AB1 grains show well-correlated ¹⁵N and ²⁶Al excesses, pointing to two end-members: ¹⁵N- and ²⁶Al-rich explosive H-burning products and certain ¹⁴N-rich material with ²⁶Al/²⁷Al < 0.001. In spite of differences in the details of the model predictions, existing SN models (WH07; Pignatari et al. 2013a, 2013b, 2015) generally agree that only material from the outermost layers is ¹⁴N-rich but ²⁶Al-poor within an SN (Lin et al. 2010; Xu et al. 2015). Thus, the model predictions in Figure 4(a) are shown as mixing lines between explosive H-burning products from the He/C zone with weight-averaged isotopic compositions and material from the outermost zones (8.4–13.3 M_⊙) in 25T and 25d models. Figure 4(a) shows that most of the PNGs and two AB1 grains with close-to-terrestrial ¹⁴N/¹⁵N ratios (if not caused by terrestrial N contamination) are well matched by the mixed ejecta.

Zoom In Zoom Out Reset image size

On the other hand, both the WH07 and Pignatari models show that the ²⁶Al/²⁷Al ratios of the outermost layers are significantly increased in SN models ≥20 M_⊙ relative to those <20 M_⊙ (Figures 4(b), (c)), resulting from the loss of the H envelope in the former with the H-burning zone exposed on the surface. Since the mismatch of the models with the C2 and AB1 grain data in Figure 4(a) is mainly caused by the high ²⁶Al/²⁷Al ratios of materials from the outermost layers, the lower ²⁶Al/²⁷Al ratios of SNe < 20 M_⊙ provide a solution. Because the compositions of the explosive H-burning products depend mainly on the peak temperature reached during explosions instead of the stellar mass, we took the same explosive H-burning products in Figure 4(a) and mixed them with the H-envelope material in the 12 M_⊙ WH07 model, shown in Figure 4(b). In this case, the N–Al isotope correlation of the AB1 grains is well reproduced by the C-rich 25d and 25d-H5 ejecta, implying that AB1 grains could have come from low-mass core-collapse SNe with H ingestion. Alternatively, AB1 grains could also be explained by SNe ≥ 20 M_⊙ if they incorporated a significant amount of materials from the surrounding ¹⁴N-rich H-envelope material with much lowered ²⁶Al/²⁷Al ratio lost during the pre-supernova phase. There is clear observational evidence for interactions between the SN ejecta and the ejected pre-supernova material (e.g., Smith et al. 2015).

More importantly, the N, Al, Si, and Ti isotopic compositions of C2 and AB1 grains strongly indicate that they share a common origin in SNe. Figure 4(b) shows that C2 and AB1 grains fall quite close to the same mixing line, with C2 grains lying closer to the explosive H-burning products end-member. Moreover, L16a found a δ⁵⁰Ti excess of 800 ± 200‰ in one C2 grain with a δ³⁰Si value of ∼400‰. In comparison, all the AB1 grains have Si isotope ratios within 100‰ of terrestrial values and δ⁵⁰Ti excesses up to 200‰, which could be explained by diluting the C2 isotopic signature with roughly four times more material from the outermost H envelope that have normal Si and Ti isotopic compositions. In comparison, most of the PNGs can be better explained by more massive SNe compared to AB1 grains (Figure 4(a)). Moreover, Figure 4(b) implies that AB1 grains with closer-to-solar N isotope ratios probably can be matched by C-rich ejecta in SNe that achieved peak temperatures lower than 7 × 10⁸ K (the value adopted in the 25d models) at the bottom of their He/C zones during explosions, implying higher-mass progenitor stars, lower explosive energies, and/or SN asymmetries.

Differences exist between the WH07 and P15 model predictions (Figures 4(c), (d)), which result from differences in the final stellar structures and the adopted nuclear reaction rates. For instance, P15 used the more recent LUNA rates for ¹⁴N(p, γ)¹⁵O that are lowered by up to 40% below 1.5 × 10⁸K (Imbriani et al. 2005) compared to the previously recommended values (Angulo et al. 1999), explaining the higher ¹⁴N/¹⁵N in the P15 models. In addition, both N and Al ratios in Figures 4(c), (d) are also affected by the amount of surface material lost during the pre-supernova phase.

Interestingly, grain M1-A7-G897 had the largest ³²S and ⁵⁰Ti excesses (Figure 2), strongly supporting the concomitant production of both nuclei by neutron capture in the parent SNe. Our results, therefore, support the initial presence of ³²Si in AB1 grains. This also agrees with the fact that AB1 grains are more enriched in³⁰Si relative to ²⁹Si than AB2 grains as a result of stronger neutron-capture nucleosynthesis. As with previous studies (Hoppe et al. 2012; Fujiya et al. 2013; Xu et al. 2015), the derived ³²Si/²⁸Si ratios of the three AB1 grains with ³²S excesses from our study are on the order of 10⁻⁴–10⁻³, which are consistent with the 15 M_⊙ SN models of Pignatari et al. (2013b), but are generally lower than those of type C grains. Also, in the 12 M_⊙ WH07 SN model, enough ³²Si is made in the C-rich He/C zone to explain the grain data, along with their Si and Ti isotope ratios (Figure 5 of L16a). In contrast, PNGs show large ³²S excesses but no measurable ⁵⁰Ti excesses. If a large amount of Ti from the neutron-burst zone within the He/C zone were removed by condensing TiC, for example, before mixing into the envelope, the decoupled ³²S and ⁵⁰Ti correlation of PNGs could be explained. We, however, do not favor this explanation because Liu et al. (2017) recently found well-correlated Si and Ti isotope ratios in X grains from SNe, suggesting insignificant amounts of fractionation between Si and Ti prior to mixing of ejecta from different zones. The correlated ³²S and ⁵⁰Ti excesses in grain M1-A7-G897 also seem not to support this scenario. On the other hand, the decoupled ³²S and ⁵⁰Ti excesses of PNGs could be explained if ³²S excesses were incorporated into the grains as ³²S made by alpha capture (Figure 8 of L16a).

Multi-element isotopic data clearly show distinctive differences between ¹⁴N-rich and ¹⁵N-rich AB grains. We thus propose to divide AB grains into AB1 grains with ¹⁴N/¹⁵N < 440 and AB2 grains with ¹⁴N/¹⁵N ≥ 440. Detailed comparisons of the new isotope data on these grains with state-of-the-art nucleosynthetic calculations show fundamental problems in linking AB1 grains and PNGs to classical novae. In contrast, there is a satisfactory agreement between the AB1, PNG, and C2 grain data and models for SNe with H ingestion from the envelope into the He/C zone prior to an SN explosion. Interestingly, one AB1 grain with the largest S and Ti isotopic anomalies found in this study, for the first time, shows concomitant overproduction of ³²S and ⁵⁰Ti in the parent star, providing additional evidence to the initial presence of radiogenic ³²Si in presolar grains.

We thank the anonymous referee for a careful and constructive reading of the manuscript. This work was supported by NASA's Cosmochemistry program (grant NNX10AI63G to L.R.N.) and benefited from the National Science Foundation under grant No. PHY-1430152 (JINA Center for the Evolution of the Elements). M.P. acknowledges significant support to NuGrid from grants PHY 02-16783 and PHY 09-22648 (Joint Institute for Nuclear Astrophysics, JINA), NSF grant PHY-1430152 (JINA Center for the Evolution of the Elements) and EU MIRG-CT-2006-046520, and from SNF (Switzerland). M.P. also thanks services of the Canadian Advanced Network for Astronomy Research (CANFAR), which in turn is supported by CANARIE, Compute Canada, University of Victoria, the National Research Council of Canada, and the Canadian Space Agency, and ongoing resource allocations on the University of Hulls High Performance Computing Facility viper.