STELLAR ORIGINS OF EXTREMELY ¹³C- AND ¹⁵N-ENRICHED PRESOLAR SIC GRAINS: NOVAE OR SUPERNOVAE?

Carbon-13 is produced in stars by proton capture on pre-existing ¹²C followed by β-decay, ¹²C(p,γ)¹³N(β⁺ν)¹³C, which occurs during H burning in the CNO cycle at temperatures ranging from 1.5 × 10⁷ to 3.5 × 10⁸ K. Although the ¹³C production is enhanced with increasing temperature, ¹³C is also more efficiently consumed via ¹³N(p,γ)¹⁴O(β⁺ν)¹⁴N at sufficiently high T during the nova thermal nuclear runaway (TNR). As a consequence, the ¹²C/¹³C production ratio does not depend strongly on temperature. On the other hand, ¹⁵N can only be significantly produced at high temperatures (¹⁴N/¹⁵N < 0.1). Thus, hot H burning during the CNO cycle (>10⁸ K) is needed in order to produce significant amounts of ¹³C and ¹⁵N (José et al. 2004). Amari et al. (2001a) discussed difficulties in explaining both low ¹²C/¹³C and ¹⁴N/¹⁵N ratios of putative nova grains by nucleosynthesis in AGB stars, CCSNe, and Wolf–Rayet stars, which led to the conclusion that novae, ONe novae in particular, are the most likely stellar source of highly ¹³C- and ¹⁵N-enriched presolar SiC grains.

The only ¹³C- and ¹⁵N-enriched stellar site identified by both astrophysical observations (e.g., Sneden & Lambert 1975) and nucleosynthetic simulations (e.g., José et al. 2004; Denissenkov et al. 2014) is classical novae. José et al. (2004) pointed out that the main effect of classical nova nucleosynthesis on the Si isotopes is an increase of δ³⁰Si/²⁸Si with δ²⁹Si/²⁸Si below or approaching zero. Putative nova, C1, and C2 grains are compared to ONe nova model predictions of José et al. (2004) with updated nuclear inputs for Si isotope ratios in Figure 3. The nova model predictions are shown as mixing lines between weighted average isotopic compositions of pure ONe nova ejecta and the solar isotopic composition. This figure shows that all the putative nova grains lie between the 1.15 and 1.35 M_⊙ ONe nova model predictions, although the pure nova ejecta need to be diluted with a large amount of isotopically normal material to match the grain data. In contrast, type C2 grains show enrichments in both ²⁹Si and ³⁰Si relative to ²⁸Si that are incompatible with the nova nucleosynthetic calculations. For instance, even though individual shells of nova ejecta are more variable with respect to their bulk compositions, the lowest ratio of δ³⁰Si/²⁸Si to δ²⁹Si/²⁸Si reached in individual shells of the 1.35 M_⊙ ONe nova model (∼6) is still a factor of three to six higher than those of C2 grains (Figure 5 of Amari et al. 2001a), which cannot be explained by mixing with materials of solar isotopic composition (i.e., isotopic dilution cannot cause the ratio of δ³⁰Si/²⁸Si to δ²⁹Si/²⁸Si to vary). Thus, ONe nova models are incompatible with the Si isotopic signatures of C2 grains.

Zoom In Zoom Out Reset image size

Other types of stars that can produce low ¹²C/¹³C ratios include born-again AGB stars and J-type stars. Born-again AGB stars (Fujimoto 1977) are post-AGB stars that undergo a very late thermal pulse (VLTP, e.g., Sakurai's object). At this stage, the remnant star has only a thin residual H shell left that could be ingested into the convective He-rich shell during the VLTP, producing extremely low ¹²C/¹³C ratios. However, models of such stars predict large excesses in ¹⁴N (Figure 10 of Herwig et al. 2011) and are thus unlikely as sources of the ¹³C- and ¹⁵N-rich SiC grains. J-type giants are C-rich stars that are mainly identified by their low ¹²C/¹³C and the absence of s-element enrichments (Abia & Isern 2000); their origin is poorly understood. Hedrosa et al. (2013) reported N isotope ratios for six J-type giants, one of which, WX cyg, may have a ¹⁴N/¹⁵N ratio as low as six and could show the same ¹⁵N-excess seen in the putative nova grains. Thus, C-rich J-type stars might have contributed ¹³C- and ¹⁵N-enriched SiC dust grains to the Solar System. However, according to the ^29,30Si excesses in C2 grains (see also Section 4.1.2 for details), s-process element enrichments are expected in these grains, while such enrichments are not observed in J-type stars (Abia & Isern 2000). Thus, at present they are less likely to be the stellar progenitors of C2 grains.

Similar to type C2 grains, another rare group of presolar SiC grains, type C grains (hereafter type C1 grains), also have large excesses in ²⁹Si and ³⁰Si, relative to ²⁸Si. However, C1 grains are enriched in ¹²C and ¹⁵N, relative to solar materials, similar to the isotopic signatures of most X grains (∼1%–2% of presolar SiC grains, see Zinner 2014 for references). Figure 3 shows that the Si isotopic compositions of C1 and C2 grains fall on similar trends. For reference, the slopes of the linear fits for C1 and C2 grains in Figure 3 are 0.89 ± 0.05 and 1.12 ± 0.57, respectively, which are consistent with each other within 95% confidence bands. In contrast to the ²⁸Si-enriched X grains that came from CCSNe, the rarer type C1 and C2 grains (<0.1%) are characterized by large excesses in ²⁹Si and ³⁰Si. These are most likely explained by neutron capture triggered by activation of the ²²Ne(α,n)²⁵Mg neutron source during CCSN explosions (Meyer et al. 2000; Pignatari et al. 2013b). The associated neutron-capture nucleosynthesis is called a neutron burst process, i.e., n process (Blake & Schramm 1976; Thielemann et al. 1979; Meyer et al. 2000). Figure 4 shows the 12 M_⊙, Z_⊙ CCSN model of Woosley & Heger (2007, W&H07), which predicts a neutron burst zone in the outer C/O and inner He/C zones (shaded area in the left panel of Figure 4). Model predictions for the Si isotope ratios of these shells are compared with those of C1 and C2 grains (right panel of Figure 4); all the grain data are well matched by mixing neutron-burst products with H envelope material (with an assumed initial solar Si isotopic composition). The good agreement is consistent with the ²⁹Si and ³⁰Si excesses of both C1 and C2 grains being the result of neutron-burst nucleosynthesis in CCSNe.

Zoom In Zoom Out Reset image size

Pignatari et al. (2013b) proposed that the large negative δ³³S/³²S and δ³⁴S/³²S values found in some C1 grains can be explained by the presence of ³²S from the in situ decay of short-lived ³²Si (t_1/2 = 153 a) produced by the n process (predicted ³²Si/²⁸Si on the order of 10⁻³, Figure 5 of Pignatari et al. 2013b), which could be incorporated into SiC grains as ³²Si and then decay to ³²S after grain condensation. This 12 M_⊙ CCSN model, as discussed in Xu et al. (2015), predicts high ³²Si/²⁸Si ratios in the neutron burst zone (Figure 4), in agreement with the 15 M_⊙ model by Pignatari et al. (2013a). The Si and S isotopic compositions of C1 grains therefore can be explained by the 12 M_⊙ CCSN model predictions for the C-rich neutron burst zone, if mixing with H envelope material is considered.

In addition to the good agreement between type C2 grain data and neutron burst predictions for Si isotopes, additional neutron-burst isotopic signatures were observed in the grain GAB and also strongly support the CCSN origin (Figure 5). To qualitatively compare model predictions with the grain data, we mixed materials from individual neutron burst shells of different CCSNe with different amounts of H envelope material (with solar Si and Ti isotopic compositions to dilute the n-process products) to match the measured δ²⁹Si/²⁸Si value of GAB (230 ± 6‰) and calculated predictions for the other neutron-rich Si and Ti isotopes. Figure 5 clearly shows that all of the baseline CCSN models (e.g., 12 and 25 M_⊙ CCSN model by W&H07) predict elevated δ³⁰Si/²⁸Si, ³²Si/²⁸Si, and δ⁵⁰Ti/⁴⁸Ti values in the neutron burst zone, and the 25 M_⊙ CCSN model predictions agree best with the grain data. Note that the 25 M_⊙ CCSN model of W&H07 predicts that the neutron burst zone is located within the C/O zone, where the C/O ratio is less than unity. So the main prerequisite for SiC formation (C > O) therefore cannot be satisfied. Overall, the Si, S, and Ti isotopic compositions of grain GAB are well reproduced by pure signatures of the neutron burst zone without the need of further mixing from deeper zones. In addition, no measurable excess in δ⁴⁴Ca/⁴⁰Ca was found in grain GAB, indicating that its parental material was mainly from outer zones of the supernova without significant incorporation of ⁴⁴Ti from deeper layers, in good agreement with the constraints from other n-process isotopic signatures.

Zoom In Zoom Out Reset image size

CCSN models, however, predict extremely high ¹²C/¹³C ratios for the C/O and He/C zones (10⁷–10⁹ for ¹²C/¹³C, Table 2 of Xu et al. 2015 and references therein), because ¹²C is abundantly produced as a result of the triple-alpha reaction, while the lack of H in these zones precludes abundant production of ¹³C. Xu et al. (2015) therefore highlighted difficulties in explaining ¹³C and ¹⁵N excesses in SiC X grains and low density graphite grains that were discussed in previous studies (e.g., Travaglio et al. 1999; Yoshida 2007). Nittler & Hoppe (2005) pointed out that the CCSN isotopic signatures (excesses in ²⁸Si, ⁴⁴Ca, and ⁴⁹Ti) observed in the ¹³C- and ¹⁵N-enriched grain M11-334-2 demonstrates that high-temperature H burning must occur in the He-shell zones of at least some CCSNe and they proposed that some kind of extra mixing process(es) brought extra H into the He shell of a pre-supernova star, which could eventually enhance ¹³C and ¹⁵N production in the He shell via proton-capture reactions at high temperatures. For the first time, CCSN models of P15 show that an H ingestion event in the pre-supernova phase can affect the usual He-burning signature, and the impact of the following explosive H burning in the He shell during the supernova shock is considered in the models. In the next section, we compare our new grain measurements on C, N, and Al isotopic abundances with theoretical predictions by P15, and we analyze their main nucleosynthetic features.

In addition to extra mixing processes occurring along the boundary of two convective zones caused by e.g., rotation (e.g., Meynet et al. 2006) and gravity waves (Arnett & Meakin 2011), more tumultuous mixing events can occur in stars, such as shell mergers (e.g., Woosley & Weaver 1995; Rauscher et al. 2002), and convective-reactive events following the ingestion of less processed material into hotter and deeper stellar zones (e.g., Herwig et al. 2014). For instance, Woosley & Weaver (1995) found that in one finely zoned 10 M_⊙ model of solar metallicity, the He shell is fully convectively linked with the H envelope, which could possibly bring a large amount of H into the He/C shell in CCSNe. Similarly, Pignatari et al. (2013a) found that H is ingested into the He shell in a 25 M_⊙ supernova model with solar metallicity.

Recently, P15 have qualitatively investigated the effect of H ingestion in the He shell material (e.g., Herwig et al. 2011, 2014; Stancliffe et al. 2011; Woodward et al. 2015) and the presence of residual H during explosive He burning in the 25 M_⊙ supernova model of Pignatari et al. (2013a). In particular, during pre-supernova evolution, H was ingested in the He shell and shell convection was then turned off, allowing survival of the ingested H in the He shell. P15 adopted two sets of supernova shock models (T_peak = 0.7 × 10⁹ K at the bottom of the He-rich zone in model 25d, and T_peak = 2.3 × 10⁹ K in model 25T, with the latter reproducing the stellar condition of the 15 M_⊙ CCSN model 15d in Pignatari et al. 2013a). The amount of H remaining from the pre-supernova evolution is varied from 1.2%(−H, the value predicted by the stellar model 25d), 0.12%(−H10), to 0.0024% (−H500) in the He shell. The predictions of these models for the He/C zone are compared to the ¹³C- and ¹⁵N-enriched presolar SiC grains from this study in Figure 6. Comparing the predictions of the −H500 (the least amount of H ingested) models with those of the other two models clearly indicates that explosive H burning greatly enhances the production of ¹³C, ¹⁵N, and ²⁶Al in the He/C zone.

Zoom In Zoom Out Reset image size

Three of the four C2 grains have ¹²C/¹³C ratios lower than three, which can only be reached by the 25T-H and 25d-H models. Although Al contamination could lower the initial ²⁶Al/²⁷Al ratios estimated for the grains, the inferred amount of Al contamination found in most SiC grains is only up to 65%, i.e., a factor of two higher at most (Groopman et al. 2015), which generally agree with the amount of Al contamination found in the grain GAB, (Figure 2(b)). By mixing ¹³C- and ¹⁵N-enriched shells with H envelope material to match C isotope ratios of C2 grains, we found that the 25d-H predictions are more comparable (²⁶Al/²⁷Al ≈ 0.02) to the grain data (²⁶Al/²⁷Al ≈ 0.01–0.03) than the 25T-H model predictions (²⁶Al/²⁷Al ≈ 0.1–0.3). Finally, we point out that as indicated by their ²⁸Si and ⁴⁴Ti excesses, X grains are likely to have incorporated more material from deeper layers, e.g., from the C/Si zone (Pignatari et al. 2013a). So even if explosive H burning also occurred in the outer He shells of their parent CCSNe, the isotopic signatures of explosive H burning can be modified by incorporating materials with ¹²C-rich explosive He-burning signatures in X grains.

Figure 5 shows differences in the predicted Si and Ti isotopes of the CCSN models for the neutron burst zone. In the 25d-H to 25d-H500 models, no resolvable neutron burst zone can be found and these are therefore not shown. A neutron burst zone exists in the 25T-H500 model, but there is a limited amount of H ingestion and the effect of proton-capture nucleosynthesis is negligible (Figure 5). Differences are found between P15 and W&H07 model predictions in Figure 5. For instance, while the 25d-H500 model predictions for ³²Si/²⁸Si and δ⁵⁰Ti/⁴⁸Ti lie between the 12 and 25 M_⊙ CCSN model predictions of W&H07, the corresponding predictions for δ³⁰Si/²⁸Si are higher than both the W&H07 models. Together with the intrinsic differences between the two sets of models, the variations that we consider here could be caused by different nuclear cross sections adopted for the nucleosynthetic calculations of W&H07 and P15. For neutron capture on ³⁰Si, W&H07 adopted the experimental MACS (Maxwellian Averaged Cross Section) at 30 keV from Bao et al. (2000), 6.5 ± 0.6 mb. On the other hand, P15 adopted the more recent ³⁰Si MACS values reported by Guber et al. (2003), which is more than a factor of three times lower (1.82 ± 0.33 mb) than that used by W&H07 at 30 keV. Because of the lowered probability for neutron capture on ³⁰Si by adopting the Guber et al. (2003) MACS values, the P15 models predict higher ³⁰Si abundance relative to that of ²⁸Si. Thus, the difference in the δ³⁰Si/²⁸Si predictions is mainly caused by different ³⁰Si MACS values adopted in the two sets of models.

When the ingested amount of H is increased from 0.0024% to 1.2% (25T-H model), the neutron burst nucleosynthesis is suppressed because the ²²Ne neutron source abundance (²²Ne(α,n)²⁵Mg) for the n process is greatly reduced due to competing proton captures on ²²Ne via ²²Ne(p,γ)²³Na (P15). Consequently, ⁴⁹Ti is more abundantly made than ⁵⁰Ti, and ³²Si/²⁸Si is about 20 orders of magnitude lower, because of the lowered neutron density and the consequent negligible production of ³²Si (Figure 5). Nucleosynthetic calculations based on one-dimensional (1D) stellar models by P15 therefore cannot explain the n-process isotopic signatures of the C2 grains. Highly likely, the discrepancy between the grain data and theoretical predictions is caused by the fact that 1D stellar models cannot obviously predict the large heterogeneities of H ingestion into the He shell that have been found in multi-dimensional hydrodynamic simulations (e.g., pre Herwig et al. 2014). Thus, although 1D nucleosynthetic calculations can provide robust predictions for specific nuclide abundances resulting from the H ingestion event when the relevant reaction rates are experimentally well determined (e.g., reactions to produce ¹²C, ¹³C, ¹⁴N, ¹⁵N, ²⁶Al, ²⁷Al), different neutron- and proton-capture isotopic signatures may exist in different regions of the He shell, resulting from the multi-dimensional nature of H-ingestion events, i.e., varying amounts of H mixed into different regions of the He shell due to the heterogeneous H-ingestion process predicted by multi-dimensional models.

Overall, the P15 models suggest that H has to be effectively injected into the He shell (∼1.2%) in order to account for the low ¹²C/¹³C isotope ratios of C2 grains. Furthermore, the neutron burst isotopic signatures of C2 grains strongly support the heterogeneous distribution of ingested H and asymmetries predicted by multi-dimensional simulations for the H-ingestion event. That is to say, C2 grains probably incorporated materials from different regions of the He shell, so proton-capture isotopic signatures are the result of explosive H burning in regions with higher amounts of ingested H (∼1.2%), while n-process isotopic signatures are records of nucleosynthesis during pre-supernova evolution in regions with negligible amounts of ingested H (<0.0024%). Therefore, although 1D stellar models provide important insights into different nucleosynthetic processes in CCSNe, complete multi-dimensional hydrodynamic simulations of the H ingestion event are needed to illustrate how parental materials of C2 grains are mixed prior to their condensation and to quantitatively compare nucleosynthetic predictions with C2 grain data. Currently, detailed multi-dimensional simulations for H ingestion are only available for the He shell in low mass stars. Therefore, similar calculations are also needed for massive stars with solar-like metallicity. To conclude, comparison of the C2 grain data with nucleosynthetic calculations based on 1D stellar models is critical to capture physical process(es) that are not previously considered, while the multi-element isotope data of C2 grains also presents a challenge for state-of-the-art 1D simulations of the hydrogen ingestion event, which provides a unique opportunity to constrain both neutron- and proton-capture environments prior to and during the supernova explosions for multi-dimensional hydrodynamic simulations.

A comparison of putative nova grains and nova models was given by Amari et al. (2001a). The new putative nova grains in our study span the same range as previous grains for C, N, and Si isotopic compositions (Figure 1), so the reader is referred to Amari et al. (2001a) for discussion on these isotope ratios. In this section, we focus on comparison of the grain data with nova model predictions for ²⁶Al/²⁷Al and S isotope ratios.

Figure 7(a) shows that mixing curves for ONe and CO nova model predictions follow different trajectories in the plot of ²⁶Al/²⁷Al versus ¹²C/¹³C, mainly because ²⁶Al is highly overproduced in ONe novae, which occur at higher outburst temperatures. Overall, four putative nova grains with ²⁶Al/²⁷Al > 0.06 can be roughly matched by ONe nova model predictions. The mixing ratios required to match the grain data indicate that if they are indeed ONe nova grains, they must have come from ONe novae with masses lower than 1.35 M_⊙, because the extreme δ³⁰Si/²⁸Si values predicted by the 1.35 M_⊙ ONe nova model are not observed (Figure 3). In addition, putative nova grains with ²⁶Al/²⁷Al ∼ 0.01 overlap with CO nova model predictions, in contrast to the previous argument that all of the putative nova grains originated from ONe novae solely based on Si isotope ratios (Amari et al. 2001a). The Si isotope anomalies of these CO nova grains are within ±100‰ with small excesses in ³⁰Si relative to ²⁹Si, while the main effect of CO nova nucleosynthesis is to decrease δ²⁹Si/²⁸Si while leaving δ³⁰Si/²⁸Si unchanged. Variations in the Si isotopes of mainstream presolar SiC grains (±200‰) are believed to result from the effect of GCE on their parent stars (Alexander & Nittler 1999; Lugaro et al. 1999; Zinner et al. 2006). Thus, Si isotope ratios of the companion main-sequence star in a binary system that forms a nova outburst later should fall along a line with a slope of 1.38 defined by the mainstream grain data in the plot of δ²⁹Si/²⁸Si versus δ³⁰Si/²⁸Si (Zinner et al. 2007). Consequently, CO nucleosynthesis could lower the δ²⁹Si/²⁸Si value during nova outburst and therefore result in excess ³⁰Si relative to ²⁹Si as observed in putative CO nova grains. As a result, the possibility of grains from CO novae mitigates one of the difficulties in linking putative nova grains to novae (problem 3 discussed in Section 1), but thermodynamic calculations for the innermost shell of CO novae predict that SiC cannot be condensed in such an oxidized environment (José et al. 2004). Condensation calculations including kinetic effects and mixing among different shells of CO novae, however, are needed to fully investigate this issue.

Zoom In Zoom Out Reset image size

More importantly, none of the nova models can explain the ³⁴S anomalies found in two putative nova grains (G270_2 and Ag2_6). As shown in Figure 1 of José (2016), the proton-capture path cannot reach the S mass region in novae with WD masses lower than 1.35 M_⊙, so that S isotope abundances are barely affected by lower-mass nova nucleosynthesis. In 1.35 M_⊙ ONe novae, the proton-capture nucleosynthesis overproduces S isotopes, and ^32,33S isotopes are more abundantly made than ³⁴S, corresponding to negative δ³⁴S/³²S values. The 1.35 M_⊙ ONe nova model, however, predicts huge excesses in δ³⁰Si/²⁸Si, inconsistent with the grain data (Figure 7(b)). In addition, neutron-rich isotopes cannot be made in novae because nova nucleosynthesis is dominated by proton-capture and β⁺ decay (Figure 2 of José 2016). Negative δ³⁴S/³²S values of nova grains therefore cannot be explained by ³²S excess from ³²Si β⁻ decay as argued for the C1 and C2 grains from CCSNe. Thus, the isotopic compositions of grains G270_2 and Ag2_6 seem to be inconsistent with predictions of classical nova nucleosynthesis. However, it is noteworthy to point out that the production of S isotopic abundances is still affected by nuclear uncertainties, e.g., the treatment of the ³⁴Cl ground and isomeric nuclear states (e.g., Coc et al. 2000). Furthermore, these nova nucleosynthetic models are all based on 1D hydrodynamic model calculations that cannot predict asymmetries during nova outbursts and do not account for the effect of rotation. Thus, the mismatch with the multi-element isotopic data by 1D nova model predictions could also indicate incorporation of products from regions with varying nucleosynthetic environments within the nova ejecta during a single nova outburst (similar to the discussion for CCSNe in Section 4.1.4). Thus, multi-dimensional hydrodynamic calculations of nova outbursts (e.g., Casanova et al. 2011) are needed to confirm the mismatch of nova model predictions with the isotopic compositions of G270_2 and Ag2_6.

In Figure 6, predictions for explosive H burning events in CCSNe are overlapped with most of the putative nova grain data, if one considers variations in the peak temperature and peak density during explosions, the amount of ingested H, and the percentage of mixed H envelope material. This demonstrates the strong similarities between explosive H burning events in novae and CCSNe. Thus, many isotopic signatures can be diagnostic for both stellar sources. The explosive H ingestion scenario in CCSNe, however, at the moment has many advantages over classical novae in explaining the putative nova grain data: (1) Explosive H burning in CCSNe only affects a thin shell of the He/C zone, and mixing with a large amount of isotopically close-to normal material should be a natural consequence of grain condensation during explosions. In contrast, a mechanism for mixing such a large amount of normal material with pure nova ejecta has not been identified; (2) A C-rich environment is obtainable by mixing explosive H ingestion products with extended He-shell layers and more external H-rich material. The main prerequisite for SiC formation (C > O) can therefore be satisfied.

While S isotopic anomalies cannot be explained by models for low-mass novae (<1.35 M_⊙), the negative δ³⁴S/³²S values of grains G270_2 and Ag2_6 could be matched by local mixing between the He/C zone and zones with ²⁸Si, ³²S excesses in CCSNe (Figure 8). Traditionally, the Si/S zone of the classical W&H07 models was considered as the only zone with large excesses in ²⁸Si and ³²S because of alpha captures. Pignatari et al. (2013a) have recently shown that increased shock velocities and consequently higher peak temperatures can result in efficient α-capture at the bottom of the He/C zone, and a C/Si zone can be formed there. The 25T-H model predicts such a C/Si zone (blue circles in Figure 8), which lies below the ¹³C- and ¹⁵N-enriched shells. As shown in Figure 8, local mixing between the He/C and C/Si zones can explain both the ³⁴S depletion (or ³²S excess) and ³⁰Si excesses of G270_2 and Ag2_6. Thus, the state-of-the-art model calculations suggest that grains G270_2 and Ag2_6 might have originated from CCSNe. Negligible amounts of ³²Si are produced in the He/C zone (Figure 5) and the C/Si zone of the 25T-H model, so the negative δ³⁴S/³²S values cannot be explained by ³²S excesses from radioactive ³²Si decay in this scenario.

Zoom In Zoom Out Reset image size

Finally, the stellar origin of ¹⁵N-enriched AB grains is even more ambiguous than that of the putative nova grains. J-type stars, CCSNe with H-ingestion, and CO novae are all possible stellar sites, while the CO novae are less favored because of the oxidized stellar environment, which make it less likely for SiC to condense. It should also be noted that there are consistent differences between the ¹⁵N-enriched AB grains and putative nova grains (higher ¹⁴N/¹⁵N, lower ²⁶Al/²⁷Al, no or smaller excesses in ³⁰Si relative to ²⁹Si in AB grains, see Section 3.1 for details). This seems to indicate that ¹⁵N-enriched AB grains are carriers of weaker proton-capture processes compared to putative nova grains.