Lifetimes of interstellar dust from cosmic ray exposure ages of presolar silicon carbide

We processed our noble gas data from 27 SiC grains and reprocessed data from published results from 22 SiC grains (20) to calculate an internally consistent set of presolar cosmic ray exposure ages for nearly 50 grains with the improved cosmogenic nuclide production rates and nuclear recoil corrections (SI Appendix, Table S1). The cosmogenic Ne component can be clearly resolved from the two other main components, nucleosynthetic Ne (Ne−G) and adsorbed atmospheric Ne based on distinct isotopic Ne compositions (SI Appendix, Fig. S8A). Ne−G is implanted into circumstellar grains from the hot post-AGB star wind emanating from the exposed He shell, and its concentrations decrease with increasing grain size (SI Appendix, Fig. S8B and refs. 19 and 20). Using C, N, and Si isotopes, all but three grains have been classified as mainstream SiC, originating in the outflows of low- to intermediate-mass (post) AGB stars (14, 35) (SI Appendix, Fig. S1 and Table S2). The three other grains are of AB type, based on their low ¹²C/¹³C ratios (SI Appendix). All newly analyzed grains are mainstream SiC.

We determined ³He and ²¹Ne exposures ages (T₃ and T₂₁) of 30 and 24 grains, respectively, and obtained upper age limits for 12 (³He) and 16 (²¹Ne) grains (SI Appendix, Table S3). For 18 grains, we have obtained both T₃ and T₂₁. Nominal recoil-corrected T₃ for 16 out of these 18 grains are higher than recoil-corrected T₂₁, whereas uncorrected ages show an opposite trend (Fig. 2). Helium is more easily lost through heating and through recoil than Ne, so both effects would result in lower nominal T₃ than T₂₁ before a recoil correction. Hence, a recoil correction will be larger for ³He than for ²¹Ne. SI Appendix, Fig. S2 shows that, for grains of <10 µm, nominal cosmogenic ³He recoil losses are >94% for the smallest grains analyzed here, whereas corresponding losses for ²¹Ne are >40%. Hence, any uncertainties in recoil corrections will result in a larger uncertainty of T₃. Heating of grains to high temperatures (≥900 K) would result in near-complete He loss (36). Helium loss works in the opposite direction of the trend we see in the data. This implies that, while some He loss cannot be excluded, no significant loss occurred; otherwise, much more He than Ne would have been lost, and even overcorrected T₃ would be smaller than T₂₁. The T₃ are less reliable than T₂₁, mainly because of larger uncertainties in the ³He recoil correction. The 16 recoil-corrected T₃ exceeding recoil-corrected T₂₁, consequently, indicate an overestimation of the recoil loss for ³He. The reason for this may be that these grains were actually irradiated in the ISM as parts of larger grains or as grain aggregates, or the grains were coated with large mantles of ices and organics while in the ISM. We estimate the original sizes of the irradiated objects in the ISM by varying the grain size and modeling the resulting recoil correction until the recoil-corrected T₃ and T₂₁ match (SI Appendix, Fig. S3). The estimated object diameters during irradiation are factors of ∼3× up to ∼30× higher than those of the analyzed grains. This results in ages of 44 to 85% of the original recoil-corrected ages (Fig. 3). In principle, it would be possible to test this result with cosmogenic Xe that has a much smaller recoil loss. Unfortunately, the amounts of cosmogenic Xe produced are below current detection limits for single-grain analyses, due to the low amounts of suitable target elements for Xe production in SiC (23). Bulk analyses of SiC give mixed signals and are not useful in this regard, as these do not resolve cosmogenic gas contributions from grains with different lifetimes. Seventy-five percent of the 16 analyzed grains that were part of much larger objects have euhedral shapes, which indicates they are not fragments of larger grains and were more likely parts of aggregates. The remainder look like they are shattered fragments of larger SiC grains (SI Appendix, Table S3 and Dataset S1), but, given the large object sizes estimated during ISM irradiation, larger than any known presolar SiC grain, they were likely also part of aggregates. Aggregates of minerals, suspected by some to be presolar, in an organic material matrix were recently observed in interplanetary dust particles (37). Bernatowicz et al. (38) observed organic coatings on ∼60% of pristine presolar SiC that they physically separated from their host meteorite without the use of chemical reagents. However, no aggregates or clustering of larger presolar grains have yet been observed during the in situ ion imaging searches of polished sections of meteorites (e.g., ref. 16). The lack of such clustering of larger grains could be due to preferred breakup of larger clusters of several dozen to hundred micrometers during accretion onto planetesimals in the early Solar System, while smaller clusters composed of smaller grains which have lower inertia, such as the ones observed by Ishii et al. (37), stayed intact. We propose that grains in the size range we analyzed formed in the outflows of (post) AGB parent stars (39) and coagulated there with organic matter to form larger aggregates. While large SiC dust grains are rare in the ISM, they are consistent with observations of circumstellar dust around AGB and post-AGB stars (40). Far-infrared excess associated with such dust may indicate the presence of up to millimeter-sized grains (41). Up to 5-mm-large dust grains were proposed to explain radio observations of dust around the Egg Nebula, a post-AGB star (42). Jura et al. also propose that the high-density winds from post-AGB stars are the sources of the large presolar SiC grains, such as the size fraction studied here.

We also obtained Li isotope data for 19 SiC grains. Many of these grains have a ⁷Li/⁶Li ratio below the chondritic (“solar”) value of 12.06 ± 0.03 (43), indicating the contribution of a cosmogenic Li component [the end-member cosmogenic ⁷Li/⁶Li ratio is ∼1.2 (29)]. However, the nominal Li ages determined from different spots on the same grains are highly variable. Li ages also correlate with the total, noncosmogenic Li concentration (SI Appendix, Fig. S4 and Table S4). These observations could be due to a combination of contamination with terrestrial or Solar System Li, matrix effects (25), or additional, unidentified Li components that would have contributed to the measured Li concentration. Because of low concentrations of cosmogenic Li and high abundance of normal Li, a reliable determination of cosmogenic Li is very difficult. Currently, Li does not allow us to obtain reliable ages, as discussed in more detail in SI Appendix.

Evidently, the Ne ages are more reliable than the Li and He ages, and we will base the following discussion mainly on ²¹Ne ages. They range from 4 ± 2 Ma (±1σ) to 3,200 ± 2,300 Ma (Figs. 2 and 3), and upper limits range from 3 to 3,300 Ma. We obtained ²¹Ne ages for two out of three AB grains; the calculated ages, 65 ± 9 Ma and 260 ± 59 Ma, fit into the age range of the mainstream grains. No age was determined for the third AB grain, due to an insufficient gas amount; the 2-µm-sized grain was the smallest one analyzed in this study.

Overall, the ²¹Ne age distribution trend (Fig. 3) is similar to what was previously reported for a smaller sample set (20), with most exposure ages below 300 Ma (60%) and 50% below 200 Ma. This is consistent with most theoretical lifetime estimates for much smaller, <1 μm interstellar dust of 100 to 300 Ma (4–9), but in contrast to the longer lifetimes expected for large grains (10–12). Assuming constant dust production rates from AGB stars and constant dust destruction rates, we would expect to encounter younger grains more frequently than older grains simply because older ones have a higher probability of encountering a destructive process. However, our age distribution does not fit any of the assumed steady-state models for different average lifetimes (SI Appendix, Fig. S5A). Having many large grains in a relatively narrow age range seems to require an explanation other than simply a lifetime effect, which would apply to small grains. We propose that this age distribution can be explained by these large grains being late-stage products of AGB stars with initial masses of ∼2 M_☉ that formed together. While less massive stars were more abundant, their evolutionary lifetimes were too long to reach the dust-producing AGB phase before the formation of the Solar System, and, hence, their dust has not been incorporated into meteorite parent bodies. The rarer, more massive AGB stars (with initial masses of >3 M_☉) are not likely to be a source of large SiC grains, as their higher radiation pressures would have ejected circumstellar grains before they grew to the large grain sizes observed here (44). It was previously suggested that the grains’ parent stars originated in a presolar starburst that could have been triggered by a galactic merger (20, 24), which Clayton (45) first proposed to explain the Si isotopic compositions of presolar mainstream SiC. Most observational and theoretical work on the history of the star formation rate (SFR) of our galaxy does not see evidence of a large starburst event in presolar times as hypothesized previously (20, 24), nor a flat SFR, but most studies conclude that the SFR only mildly fluctuated (46–50). These studies find a moderately enhanced SFR around 7 to 9 Ga ago. Several of the observational studies show that this broad peak consists of two peaks, with the more recent one close to 7 Ga ago (SI Appendix, Fig. S5 and ref. 46). Recent modeling work by Noguchi (49) based on observations of the chemical compositions of stars in the solar neighborhood reveals a moderately enhanced SFR that peaked around 7 Ga ago. In this model, this enhancement was caused by streams of cold matter that accreted onto the galactic disk from the halo (49). Based on stellar main-sequence lifetime calculations, we estimate that stars with ∼1.6_☉ to ∼1.9 M_☉, that formed together during this enhanced SFR episode ∼7 Ga ago, reached their dust-producing AGB phase between ∼4.9 and ∼4.6 Ga ago (SI Appendix, Fig. S5). These dust grains would then have been exposed to interstellar GCR for <300 Ma before being shielded in the forming Solar System. What we are seeing in the SiC age peak are the first arrivers of dust formed in the late stages of stars originating in the presolar enhanced SFR peak (SI Appendix, Fig. S5). The rest of the peak must be more recent than the start of the Solar System and was not sampled in the presolar grain population. Although speculative, this scenario is consistent with our data and, barring another explanation, may be a plausible reason for the observed presolar SiC age distribution for large grains with presolar ages of <300 Ma. While we see older grains, we do not see older peaks (other than from individual grains) in our age distribution. We explain this by two effects. First, grain destruction reduced the number of surviving old grains, and, second, the signal to noise ratios farther back in time are currently too low to show peaks within our dataset. Older interstellar ²¹Ne exposure ages obtained for at least 7 grains are >300 Ma and, for a few grains (3 out of 24, excluding those with upper limits; 5 out of 40 including upper limits), are consistent with what is expected for large grains (10, 12). In particular, if these grains were >100-μm aggregates in the ISM, long lifetimes are expected. Erosion by sputtering is slower than the time the grain is exposed to shock-heated gas, but large grains can erode significantly when they get slowed down in the cooled postshock gas and experience rare collisions with other large grains (10). Gradual erosion by collisions with smaller grains would leave cratered surfaces (10), something that has not been observed with SiC grains to date (38). Possible evidence of a microimpact crater was so far only found in a large presolar aluminum oxide grain (51). Some of the old grains could have been shielded from destructive processes in clumps. Such protective density inhomogeneities have been observed astronomically in shocked regions of the ISM (e.g., ref. 52).

The oldest grains based on both ³He and ²¹Ne ages are the smallest, and an inverse trend between age and grain size is apparent (Fig. 4 and SI Appendix, Fig. S6), consistent with the preliminary trend observed by Ott and Begemann (23) in Xe from bulk SiC analyses but in contrast to the prediction by Hirashita et al. (12). The trend persists in the recoil-corrected data and in the size-corrected subset but gets less prominent in the latter. Smaller grains are more abundant than larger grains in the ISM (3), resulting in a higher number of smaller grains that are old compared to larger ones. We can exclude a sampling bias, as we have not disproportionally analyzed small grains; on the contrary, only 12 of the 49 grains are <4 µm.

Gyngard et al. (53) proposed that grains with presolar ages older than the sun’s galactic year [∼230 Ma (54)] might have had the time to radially migrate from the inner parts of the galaxy toward the galactocentric distance of the forming Solar System. Because of the compositional gradient within our galaxy, we would expect these grains to reflect the metallicity of their parent stars. However, we do not observe a correlation between age and Si isotopic composition, which is a proxy for metallicity of stellar sources (55). Either our dataset is too small to reveal such a trend, the grains did not migrate as suggested, or there is no galactic gradient for Si isotopic composition, in contrast to O isotopic composition (56) and [Fe/H] (57). Recent astronomical observations (58) did not find a galactocentric δ²⁹Si trend within ∼200‰, a range that was less than expected from galactocentric variations in other isotope ratios but similar to the one measured in presolar SiC mainstream grains.

We should highlight that, at the end of their interstellar journey, the presolar grains could have been exposed to enhanced particle radiation from the young sun. Based on cosmogenic He and Ne concentrations in hibonite, an aluminum−calcium oxide, from the Murchison meteorite, the solar cosmic ray (SCR) flux these grains might have been exposed to was orders of magnitudes higher than today, consistent with what is expected during the T Tauri phase of the sun (30–32, 34). Hibonites were among the first condensates in the protoplanetary disk (59) and were transported to the disk surface far enough from the sun to evade significant heating, where they were irradiated by an enhanced SCR flux (34). If the presolar SiC grains had a similar exposure history to solar energetic particles in the protoplanetary disk as the hibonites, they would also have acquired a similar concentration of SCR-produced noble gases. The difference in irradiation time on the disk surface between presolar SiC and hibonites is not known. Given that the high-temperature condensate hibonite was present very early in the disk (59), the short disk lifetime of a few megayears (60, 61), and the exposure required to explain the cosmogenic hibonite data (34), we consider that the time difference between the hibonite and SiC exposure duration was probably small. Our results show that the majority of the cosmogenic ²¹Ne was acquired during presolar GCR exposure (SI Appendix, Fig. S7 and Calculation of He and Ne Exposure Ages). Specifically, at least 80% of the cosmogenic ²¹Ne for grains with ²¹Ne ages greater than 100 Ma was acquired by presolar GCR exposure. For these grains, the amount that might have been acquired during early Solar System formation is smaller than the uncertainty of the presolar exposure ages and, hence, not detectable. These findings only apply if the presolar grains were exposed to the early active sun at all. At most, the five grains with the lowest ages might have acquired all their cosmogenic ²¹Ne in the early Solar System (SI Appendix, Fig. S7). Early Solar System exposure does not significantly affect our interpretation of presolar ages, except, possibly, for these five grains.

However, our observation has implications for the origin of hibonites that formed in the solar nebula: The cosmogenic nuclide concentrations in the hibonites are typically much lower than that observed in presolar SiC grains, indicating that the irradiated hibonites are indeed early Solar System products and not of presolar origin.

We note that a presolar exposure age of a SiC grain is a nominal age and that the actual residence time in the ISM might have been shorter if the grains were exposed to a high energetic particle flux from other nearby stars in addition to background GCR exposure. We estimate that the chances of such a close encounter for the average interstellar SiC are low and that such exposure could have also led to destruction of the grain. Modeling of this probability is difficult due to many unknowns and beyond the scope of this work.

All data discussed in the paper is available in the SI Appendix.