Detection of Cosmic Fullerenes in the Almahata Sitta Meteorite: Are They an Interstellar Heritage?

For this study, 13 clasts from the AhS meteorite were used to track chemical diversity among them. Seven of them (AhS #04, #22, #24, #27, #28, #38, and #48) are ureilites with a high concentration of aggregates of carbonaceous material (up to 500 μm in diameter) composed of graphite and minor diamonds (Jenniskens et al. 2009; Zolensky et al. 2010). The remaining six samples are enstatite and ordinary chondrites, which are classified as follows: AhS #41 (EL6), #58 (H4-5), #1001 (Metal+Sulfide), #1002 (EL4-5), #1054 (LL3-4), and #2012 (EH4-5). In addition, a bulk sample from the interior of the Allende CC was used. For each analysis, fresh fragments of a few milligrams were crushed using a mortar and pestle. The powder was then attached to a 10 mm disk of stainless steel with conductive copper tape. The disk holding the powder was then mounted on the sample holder to be inserted inside the instrument via an automated vacuum interlock system. After introduction, the sample was positioned and moved in two axis using a motorized XY-manipulator with a minimum step of 100 μm. It was demonstrated that the copper tape did not produce a background signal that could interfere with the analysis.

AROMA, the Astrochemistry Research of Organics with Molecular Analyzer, is a unique experimental setup developed to study, with microscale resolution, the content in carbonaceous molecules of cosmic dust analogs and meteoritic samples (Sabbah et al. 2017). Mass spectrometry data and chemical analysis tools for all studied samples are available to the public in the AROMA database at http://aroma.irap.omp.eu.

AROMA consists of a microprobe laser desorption ionization source and a segmented linear quadrupole ion trap (LQIT) connected to an orthogonal time of flight (oTOF) mass spectrometer, as shown in Figure 1. Ions are produced at a low background pressure (10⁻⁶ mbar) by performing a one- or two-step LDI. The laser-generated ions are collimated by a set of lenses at high DC voltage and are thermalized in a radio-frequency (RF) octupole, which maximizes ion transmission and reduces the fragmentation that occurs in a typical LDI source. The ions are stored in the LQIT and processed if desired. A set of RF-DC optics is used to transfer the ions to high vacuum, and finally, they are monitored on an oTOF mass analyzer equipped with a two-stage reflectron and a fast microchannel plate detector. The total mass spectrum of an experiment is the superposition of multiple scans recorded over a given m/z range. The amplitude and RF frequency applied to the LQIT electrodes are optimized to maximize the ion signal in this mass range. Each scan is the result of 50 laser shots. The interaction of the desorption laser with the sample is controlled by a mechanical shutter. The laser hits the sample once every 2 s. The sample is then moved so that the next laser shot hits a fresh spot.

Zoom In Zoom Out Reset image size

In order to perform LDI, AROMA uses a pulsed (5 ns) infrared (IR) laser (Nd:YAG at 1064 nm) focused on the sample with a spot size of 300 μm to cause rapid and localized heating, promoting thermal desorption rather than decomposition. The typical IR laser desorption fluence used in this work is F_des = 300 mJ cm⁻² (60 MW cm⁻² in irradiance). This low fluence is one to two orders of magnitude lower than the explosive vaporization threshold that leads to plasma formation during ablation processes (Hoffman et al. 2014). The interaction of the laser with the sample is able to efficiently produce ions from the metallic phases, which are present in minerals, salts, and the matrix of natural samples (Koumenis et al. 1995; Tulej et al. 2011; Jayasekharan & Sahoo 2013; Shi & Deng 2016). In this step, C clusters can also be generated from the decomposition of a pure carbonaceous phase (Šedo et al. 2006). In the L2MS analysis, a pulsed (5 ns) ultraviolet (UV) laser (fourth harmonic of a Nd:YAG at 266 nm) perpendicularly intercepts the expanding plume. This leads to the selective ionization of species that can undergo (1+1) resonance-enhanced multiphoton ionization (REMPI), which is the case for fullerenes and aromatic species. The laser ionization fluence used in this work is F_ion = 20 mJ cm⁻² (4 MW cm⁻² in irradiance).

For each detected m/z peak with a signal-to-noise ratio greater than 10, a chemical formula is assigned using the mMass software, an open-source mass spectrometry tool (Strohalm et al. 2010). The chemical formula assignment is performed with a typical accuracy of 0.01 between measured and calculated m/z values. Peaks corresponding to various metallic elements (most intense peaks: Fe, K, Na, Ca, Al, Cr, etc.) seen as atomic ions or small clusters (e.g., Fe²) have been identified thanks to their mass defect and the high mass resolution provided by AROMA. This family is referred to hereafter as "metals" and reflects the different silicate phases found in AhS. As is well known in LDI mass spectrometry, the signal from elements, in particular Na and K, is ubiquitous in natural and standard laboratory samples. For the molecular family analysis, we calculate in addition the DBE (Marshall & Rodgers 2008) for each molecular formula, which is defined as

$$\begin{eqnarray}&&\mathrm{DBE}={\rm{C}}\#-{\rm{H}}\#/2+{\rm{N}}\#/2+1,\end{eqnarray} \tag{ 1 }$$

with C#, H#, and N#, the number of C, H, and N atoms in molecules (in our analysis, we only include C and H because the mass resolution of the AROMA setup is not sufficient to disentangle the mass of N from that of CH₂). The DBE is representative of the unsaturation level of the molecules and thus corresponds to a direct measure of their aromaticity. It is equal to the number of rings plus double bonds involving carbon atoms (as each ring or double bond results in the loss of two hydrogen atoms). In the recorded mass spectra, only a few species containing oxygen, not exceeding C₇, could be firmly identified. The DBE calculation allows us to sort the detected pure carbon and hydrocarbon ions into different molecular families (Martínez et al. 2020; Sabbah et al. 2021). Pure carbon species are sorted into C clusters (C# < 30) and fullerenes (C# ≥ 30). The transition between C clusters and fullerenes is somewhat arbitrary. von Helden et al. (1993) have shown that fullerenes appear in the C# range of 30–40 but this does not mean that all species in this range are actually fullerenes. Hydrocarbons are classified into three categories, HC clusters, PAHs, and aliphatics species. This classification is done using empirical factors and DBE limits established for hydrocarbon mixtures in complex natural organic matter (Koch & Dittmar 2006; Hsu et al. 2011; Lobodin et al. 2012). Hydrocarbons with 0.5 ≤ DBE/C# ≤ 0.9 are considered to be PAHs. HC clusters and aliphatic species are located at DBE/C# > 0.9 and <0.5, respectively.

Figure 2 gathers the data that were recorded in the range m/z = [500, 1300] for the seven ureilites. We could not detect any ion signal in this range for the other clasts. Figure 2 shows that species from C₄₂ to about C₁₀₀ (C₈₀ for #22 and #24) are observed with a mass spacing of 24 m/z (C₂), which is characteristic of the fullerene series. Figure 3 shows the mass spectra corresponding to the molecular content associated with a highly ordered pyrolitic graphite (HPOG) sample, which is revealed by one-step LDI. Different values of the desorption laser fluence were used. For F_des < 300 mJ cm⁻², almost no signal is observed (except two peaks at the noise level corresponding to C₁₀ and C₁₂). At F_des = 300 mJ cm⁻², a distribution of C clusters is observed, extending to C₂₈. Increasing the fluence to F_des = 1 J cm⁻² leads to higher peak intensities and a distribution extending to C₃₅. These observations indicate that ablation of carbonaceous material and chemistry within the plume can produce C clusters but not fullerenes. We expect similar results for other carbonaceous materials (e.g., diamond; Šedo et al. 2006).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

We also studied the evolution as a function of laser conditions of the size distribution of fullerenes in the AhS #04 sample (see Figure 4). For this sample, a two-step LDI scheme with wavelengths and fluences typical of those used to analyze PAHs and fullerenes in complex hydrocarbon mixtures without significant fragmentation (Lykke et al. 1993; Homann 1998; Faccinetto et al. 2011) was used. The fullerene ion signals were summed considering three size ranges covering medium (42 ≤ C# ≤ 58), large (60 ≤ C# ≤ 70), and very large (C# > 70) sizes. Figure 4 shows that the fullerene size distributions remain comparable within a factor of typically 2, independent of the laser conditions. These calibration measurements support the fact that fullerenes are intrinsic to the samples. By examining the relative fullerene ion signal between the seven AhS ureilite samples (Figure 4), we conclude that the fullerene size distribution is comparable in all samples, with the exception of an obvious lack of very large fullerenes in AhS #22 and #24, in which the detected species only go up to C# = 80. We cannot conclude whether this difference reflects an intrinsic difference in fullerene sources. Rather, it indicates the difficulty of extracting very large fullerenes from bulk samples and the possible role of sample heterogeneity.

Zoom In Zoom Out Reset image size

Figure 5 shows that hundreds of peaks are detected in the low-m/z mass range [10–500] of the 13 AhS samples. These peaks include a large diversity of carbonaceous species. We sorted these molecules into families by applying the DBE method (see Section 2.3). Examples of DBE/C# plots are provided in Figure 6. The families observed include aliphatics, HC clusters, PAHs, C clusters, and fullerenes as described in Sabbah et al. (2021). The aliphatic family is not considered because we could detect only a few species falling into this category: five species in AhS #04 and three in AhS #1001, all at C# ≤ 7. The sum of peak intensities for the different families is presented in Figure 7 and compared to the sum of peak intensities of the metals. Note that Figure 7 shows that the intensity of metals in non-ureilites is higher compared to ureilites. Conversely, the ion signal of carbonaceous species is much weaker in non-ureilites, with the exception of PAHs, which show a similar sum of peak intensities in both clast types and HC clusters in AhS #1001. The observation of PAHs in the porous chondritic components of the asteroid was previously explained as being due to the mobilization of organic matter during impacts (Sabbah et al. 2010). The PAHs detected are relatively small with a maximum of C# = 24 and should be much easier to evaporate than the larger fullerenes.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Similar to the fullerene analysis (Figure 4), we summed the C clusters' peak intensities for the two size categories, small (8 ≤ C# ≤ 17) and medium (18 ≤ C# ≤ 29). As shown in Figure 8, the ion signal of these species is large and generally exceeds the ion signal of fullerenes by a factor of 10. The contribution to the ion intensity of the family with 30 ≤ C# ≤ 40 (which contains both large C clusters and small fullerenes) is small. Adding this contribution to the C cluster family or the fullerene family, as was done in Figure 7, should therefore not affect the results.

Zoom In Zoom Out Reset image size

The observed variations in the sum of peak intensities with cluster sizes (Figure 8) may indicate different origins for small- and medium-sized clusters. We have seen that, under our experimental conditions, C clusters can be formed by the interaction of the IR laser with a solid carbonaceous material (e.g., HPOG) whereas fullerenes cannot (see Section 3.1). This does not exclude a contribution from C clusters trapped in molecular form in the samples, as suggested in our earlier work on laboratory analogs of stardust (Martínez et al. 2020). The different possible origins of the carbon cluster ion signal could explain why this signal is not closely correlated with the fullerene ion signal. If fullerenes are associated with only one specific carbonaceous phase among several, the fullerene ion signal would depend on the concentration of that phase and its accessibility by the desorption laser to achieve the extraction of its molecular content. Therefore, the apparent absence of fullerenes in non-ureilites could be only a sensitivity problem.

Previously reported detections of fullerenes in the Allende (Becker et al. 1994) and Murchison meteorites (Becker et al. 2000) have been questioned both because of the analytical techniques used (Hammond & Zare 2008) and the nondetection of these species by other groups (Buseck 2002). We did not detect fullerenes or C clusters in our previous analysis of a bulk sample of Murchison with the AROMA setup (Sabbah et al. 2017). To complete this result, we analyzed a powder from the interior of the Allende meteorite. The molecular composition of the carbonaceous species (see the mass spectrum in Figure 9) is dominated by PAHs and a few m/z peaks attributed to HC clusters (four peaks) and aliphatics (three peaks). There is no fullerene signature in the mass spectrum. The PAH distribution peaks around m/z = 202.08, which corresponds to pyrene and its isomers. It is found to be similar to that previously detected in several samples from the interior of Allende (Zenobi et al. 1989). Figure 10 shows the DBE/C# values in the C# range of 5 to 35 for the Murchison, Allende, AhS #1002, and AhS #04 samples. The molecular families of these samples are clearly different, suggesting possible differences in the initial chemical reservoirs or processes undergone by their parent body. Each plot of the DBE/C# pattern is indeed unique. PAHs dominate in three of the four selected samples, but they show some differences. Murchison contains a substantial fraction of methylated PAHs, while Allende contains more pyrene and isomers. In AhS #04, the ion signal is comparable for pyrene and indene, the smallest PAH (see Figure 10). Using the DBE/C# values, we sorted the carbonaceous molecules into families (for m/z greater than 100), and the results obtained for the four samples in Figure 10 are presented in Figure 11. The total PAH ion signal in Murchison is the highest among the four samples, due to the greater variety of PAHs detected (Figure 10). The fullerene ion signal in AhS #04 is lower than that of PAHs in Allende and AhS #1002 by a factor of 2–3 and by a factor of 5–6 than that of PAHs in Murchison. It is accompanied by a high signal of C clusters, at the same level as that of PAHs in Allende and AhS #1002.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Quantifying species abundance as a function of ion signal remains a challenge in LDI techniques (Elsila et al. 2004). The ion signal depends on several parameters, including the desorption and ionization efficiencies of the associated species, the nature of the analyte, the properties of its bounding with the surface, and the nature of the substrate itself. Signal dispersion can therefore be attributed to a combined effect of sample heterogeneity and the fact that mainly species from the sample surface are desorbed. This is different from destructive techniques such as ICP-MS using laser ablation in which all material from the laser crater is analyzed (Xiao et al. 2021). Relating the ion signal to abundances (on the surface) therefore requires calibration tests using several internal standards (Elsila et al. 2004) and samples of known concentrations (Koumenis et al. 1995).

The inferred mass concentration for PAHs in the Murchison meteorite ranges from 15 to 28 ppm, with an average of 22 ppm (Sephton 2002). Because the Murchison and AhS samples were analyzed with AROMA using the same experimental conditions and similar mass of each meteoritic sample, the sum of peak intensities for PAHs from the two experiments can be directly compared (see Figure 11). Only two samples show a value slightly higher than the Murchison value, while the others are lower by a factor of about 2 (six of them) and up to 20 for two of them. Considering the number of factors that can affect the absolute intensities (Elsila et al. 2004), we can conclude that the PAH mass concentrations are lower but comparable in AhS compared to Murchison.

In a previous study on soot, we demonstrated the ability to identify fullerene species in nascent soot particles produced in a slightly sooting ethylene/air premixed flame (Sabbah et al. 2021). In addition, we were able to follow the evolution of carbonaceous molecular families, including fullerenes, by analyzing soot collected at different heights above the burner. In this study, we have shown that AROMA can quantitatively trace the thermal processing of large PAHs (C# ≥ 50) leading to the formation of fullerenes in this hydrocarbon flame and that the detected molecules mainly originate from the surface of soot particles. This suggests that the efficiency of AROMA in detecting both molecular species (i.e., large PAHs and fullerenes) is similar (at least when associated with soot particles; Sabbah et al. 2021). Assuming this is the case for the AhS samples, the mass concentrations of fullerenes can be deduced by considering that the summed ion signal for fullerenes and PAHs is proportional to their respective concentrations. Therefore, the mass concentration of fullerenes is most probably comparable to that of PAHs in AhS, i.e., a few ppm (after correction of the ratio between the average mass of PAHs (m/z ∼ 212) and that of fullerenes (m/z ∼ 740)). However, this value should be taken with caution because PAHs and fullerenes are distributed in different phases within the AhS meteorite, PAHs being quite widespread while fullerenes are supposed to be associated with a specific solid carbon phase.

In the following, we discuss the pros and cons of possible scenarios that could explain the presence of fullerenes in the AhS meteorite. One possible scenario for the presence of fullerenes in AhS, which is suggested by the unique history of the ureilite parent asteroid (Goodrich et al. 2015), is the introduction of a primitive (CC-like) impactor that was responsible for the catastrophic disruption of the UPB. The timing of the UPB disruption led to the suggestion (Yin et al. 2018) that the impactor was an outer SS body, associated with a large-scale migration of outer SS bodies into the inner SS driven by the growth and/or migration of the giant planets during the gaseous disk phase (Walsh et al. 2012). Some impactor material may have been added to the ureilitic daughter bodies formed at this time and then redistributed to other clast types by subsequent impact gardening of the regolith. However, we have only observed fullerenes in ureilites and have not been able to detect fullerenes in two well-known CC, Murchison and Allende.

An alternative scenario is the possibility that the fullerenes were produced in the ureilites themselves sometime during their history. The UPB was a carbon-rich asteroid that experienced igneous processing involving temperatures up to ∼1250 °C–1300 °C (Collinet & Grove 2020). During this processing, graphite was formed from the primitive carbon-rich materials that accreted onto the body. However, there is no known mechanism by which the fullerenes could have formed during igneous processing. Subsequently, the UPB experienced a number of shock events, in particular, the large impact that resulted in a major disruption of the parent body (Goodrich et al. 2015); such events offer a potential mechanism for the formation of fullerenes. It has been proposed that the diamonds present in most ureilites are formed by the transformation of graphite during this shock (e.g., Lipschutz 1964; Nabiei et al. 2018, and references therein). A recent laboratory study (Popov et al. 2020) reports a zone of instability of diamond, for a range of temperatures, at pressures of 55–115 GPa. This instability leads to the transformation of diamonds into fullerene-like onions, which consist of multiple shells whose number decreases with temperature to a minimum value of 2–3 at a temperature of 2400 K or 2127 °C (Popov et al. 2020). This suggests the possibility that fullerenes in ureilites formed almost immediately after the formation of diamonds in an impact environment. However, it is unlikely that the extreme conditions necessary for the creation of these fullerene-like onions were achieved in ureilites. Nestola et al. (2020) have shown that diamonds in ureilites could have formed by impact shock events at much lower pressure and temperature. Furthermore, onion species formed by this mechanism differ from molecular fullerenes as shown by Raman spectroscopy (Popov et al. 2020). The idea that fullerenes could have formed by hypervelocity impacts in space has been investigated by the analysis of impacts collected on the Long Duration Exposure Facility (LDEF). Fullerenes were detected in a single LDEF impact crater, which motivated experimental simulations (Radicati di Brozolo et al. 1994). Although these experiments were limited to impacts at a velocity of 6 km s⁻¹, the authors concluded that the fullerenes observed in the LDEF crater are unlikely to have formed during a hypervelocity impact. Because this crater also contained chondritic elements, these authors suggested that the fullerenes were instead intrinsic to the impacting micrometeoroid.

Another scenario would be that these fullerenes formed much earlier and were inherited from the interstellar medium at the time of SS formation. The abundance of carbon in the form of C₆₀ in the diffuse interstellar medium is estimated to be on the order of a few 10⁻⁴ to a few 10⁻³ (Berné et al. 2017). This value is substantially lower than the abundance value of 10⁻¹ for interstellar PAHs, which are the carriers of the aromatic infrared bands. Interstellar PAHs are expected to be large, typically containing more than 50 C due to processing by UV photons (e.g., Montillaud et al. 2013). Such large PAHs are not detected in our study. We therefore expect the concentration of interstellar PAHs and/or fullerenes to be very low in AhS (as in Murchison and Allende). Another possibility is that these species are included in a phase from which they cannot be extracted by the AROMA desorption laser. The small PAHs (less than 30 carbon atoms in size) detected in AhS (as well as in Murchison and Allende) are more likely products of active cold chemistry in the dense molecular cloud from which the SS formed. This suggestion of Woods & Willacy (2007) was supported by the recent detection of small PAHs in the cold prestellar core TMC-1, in particular indene C₉H₈ (Burkhardt et al. 2021; Cernicharo et al. 2021a), as well as the availability of chemical pathways to form these small aromatic species by neutral neutral reactions at low temperatures (Cernicharo et al. 2021a, 2021b).

Finally, a local interstellar heritage has been proposed by several authors to account for anomalies in short-lived radionuclides (SLRs) such as ²⁶Al (e.g., Podosek 2005; Huss et al. 2009). More precisely, the idea arose that the SS prenatal cloud was seeded by small dust grains produced at the end of the life of nearby very massive stars (Arnould et al. 2006; Gaidos et al. 2009; Dauphas et al. 2010; Tatischeff et al. 2010; Gounelle & Meynet 2012). It is interesting to note that the production of fullerenes at the end of the life of very massive stars, at the presupernova stage of WR stars or at the supernova stage, has been suggested by modelers (Cherchneff et al. 2000; Clayton et al. 2001; Clayton & Meyer 2018) as discussed in Section 1. The model by Clayton & Meyer indicates that C-rich regions are strongly enriched in ¹²C, although processes leading to mixing with ¹³C may occur. This would help to understand the fact that, although some of the presolar grains of SN origin are strongly enriched in ¹²C, a significant fraction of them have an isotopic ratio close to the solar value (Croat et al. 2003; Lodders & Amari 2005). We derived a value of 0.68 ± 0.04 for the ratio of ¹³C¹²C₅₉/¹²C₆₀ for the seven AhS samples, to be compared with the value of 0.62 ± 0.04 derived from molecular analysis of terrestrial samples. The two ratios are therefore similar. A systematic study would be needed to refine these values and extend them to the entire fullerene population. However, we emphasize that the inferred ¹²C/¹³C ratio may be underestimated due to a possible contribution from fullerane species (Becker & Bunch 1997). At this stage, a scenario in which the fullerenes present in AhS originate from massive stars can therefore not be demonstrated on the basis of these isotopic ratios or other arguments. But neither can it be contradicted. This scenario has obvious advantages from an astrochemical perspective. It could rationalize why C₆₀ is observed in some astronomical environments and not in others. Its detection is mainly in evolved stars but in a limited number of them (a few percent of planetary nebulae; Otsuka et al. 2014) including H-rich but not H-poor circumstellar environments (García-Hernández et al. 2011). C₆₀ has also been detected in the environment of a number of massive young stellar objects (Sellgren et al. 2010; Roberts et al. 2012). This diversity is difficult to rationalize and may indicate star-forming regions associated with the shell of a WR bubble (Dwarkadas et al. 2017).