Presolar grains
Presolar grains (also called stardust) are minerals that predate the formation of our solar system and formed within outer gases of earlier stars. They can be seen as a palpable source of information on stellar processes, complementary to the observation of electromagnetic radiation collected by telescopes. Although very rare and very small (micrometer to submicrometer-sized), they can be recognizable by their unusual isotopic compositions, which are completely different from isotopic compositions of materials from our solar system. Presolar grains form within outflowing and cooling gases from evolved stars, such as red giants or supernovae. Their isotopic compositions are key to understanding nucleosynthesis and the galactic chemical evolution. Nuclear reactions in stars produce elements with a wide range of isotopic compositions, which are dependent on the properties of the star. Among the different mineral species that presolar grains make, silicon carbides (SiC) are the most studied grains, due to their relatively easy isolation from meteorites. In the last decades, the isotopic analysis of presolar SiC grains have allowed to define different types of grains having specific ranges of isotopic compositions, which are indicative of the different stellar environments where they form. While the origin of these grains are still uncertain, their classification into different types unable discussing their properties and evaluating their origin. In two studies published in 2021 and 2022 in the Astrophysical Journal Letters (Boujibar et al. 2021), we have used machine learning algorithms to evaluate the classification of presolar SiC grains.
The classification of presolar SiC grains was evaluated using cluster analysis, a machine learning algorithm that statistically determines groups of samples that have similar features. Figures on the right show the comparison between the traditional classification (e.g. Zinner et al. 2014, Treatise on Geochemistry) and the results of cluster analysis (Boujibar et al. 2021, ApJL). Using density distribution plots, we found that the presolar grain database (PGD) contains several analysis of grains which were likely contaminated (see peak). These grains were removed from the data before running cluster analysis. Liu et al. (2021, ApJL) subsequently showed that it likely represents surface contamination that can be overcome by presputtering to expose clean surfaces.
One main finding for this clustering is that most Y and Z grains are clustered jointly, suggesting that previous use of a divider at 12C/13C = 100 was arbitrary. Other findings include three clusters for mainstream (MS) grains, in which one cluster has a compact core of grains with a narrow range of C and Si isotopic compositions (see density distribution below cluster 9 in orange). This cluster highlights the possible enhanced production of stars of similar mass and metallicity in our solar neighborhood (Boujibar et al. 2021).
Results mentioned above were derived from isotopic ratios of C (C12/C13), N (14N/15N), Si (29Si/28Si and 30Si/28Si). These are most common data for presolar SiC grains. Considering the additional inferred Al isotopic ratios 26Al/27Al produces a smaller dataset and slightly different results. The most important result was the discovery of a cluster of MS grains having simultaneous 15N and 26Al enrichments (see pink cluster 8 on the left) .