The presolar grain inventory of fine‐grained chondrule rims in the Mighei‐type (CM) chondrites

We investigated the inventory of presolar silicate, oxide, and silicon carbide (SiC) grains of fine‐grained chondrule rims in six Mighei‐type (CM) carbonaceous chondrites (Banten, Jbilet Winselwan, Maribo, Murchison, Murray and Yamato 791198), and the CM‐related carbonaceous chondrite Sutter's Mill. Sixteen O‐anomalous grains (nine silicates, six oxides) were detected, corresponding to a combined matrix‐normalized abundance of ~18 ppm, together with 21 presolar SiC grains (~42 ppm). Twelve of the O‐rich grains are enriched in 17O, and could originate from low‐mass asymptotic giant branch stars. One grain is enriched in 17O and significantly depleted in 18O, indicative of additional cool bottom processing or hot bottom burning in its stellar parent, and three grains are of likely core‐collapse supernova origin showing enhanced 18O/16O ratios relative to the solar system ratio. We find a presolar silicate/oxide ratio of 1.5, significantly lower than the ratios typically observed for chondritic meteorites. This may indicate a higher degree of aqueous alteration in the studied meteorites, or hint at a heterogeneous distribution of presolar silicates and oxides in the solar nebula. Nevertheless, the low O‐anomalous grain abundance is consistent with aqueous alteration occurring in the protosolar nebula and/or on the respective parent bodies. Six O‐rich presolar grains were studied by Auger Electron Spectroscopy, revealing two Fe‐rich silicates, one forsterite‐like Mg‐rich silicate, two Al‐oxides with spinel‐like compositions, and one Fe‐(Mg‐)oxide. Scanning electron and transmission electron microscopic investigation of a relatively large silicate grain (490 nm × 735 nm) revealed that it was crystalline åkermanite (Ca2Mg[Si2O7]) or a an åkermanite‐diopside (MgCaSi2O6) intergrowth.


INTRODUCTION
Primitive meteorites, micrometeorites, interplanetary dust particles (IDPs), and cometary material contain small amounts of refractory dust grains which predate the formation of the Sun and our planetary system and are characterized by highly anomalous isotopic compositions that cannot be explained by chemical or physical processes within the solar system (e.g., Zinner 2014). Instead, they possess nucleosynthetic signatures of their stellar parents. These presolar or "stardust" grains condensed in the ejecta of evolved stars or stellar explosions (novae and supernovae). Subsequently, they became part of the interstellar medium (ISM) where they were exposed to high-energetic irradiation, exposure to shockwaves from nearby supernova explosions, and grain-grain collisions. The presolar dust that is found today in primitive solar system materials mostly escaped alteration and homogenization processes in the ISM as well as during the formation of the protosolar nebula and protoplanetary disk (e.g., Zinner 2014). The properties of these grains hold valuable information on stellar nucleosynthesis and evolution, grain formation in circumstellar environments, and the types of stars contributing material to the nascent solar system. Moreover, determining the abundances and distributions of the different types of presolar dust allows for the study of parent body alteration processes, and also potential heterogeneities in the distribution of presolar circumstellar matter within the nascent solar system (Zinner 2014;Floss and Haenecour 2016).
Silicates are the most abundant presolar grain species large enough for the isotopic analysis of single grains (e.g., Floss and Haenecour 2016). In contrast to the more refractory types of presolar dust (silicon carbide, graphite, Si 3 N 4 , oxides), silicates cannot be separated chemically from their host meteorites (Nguyen and Zinner 2004). Instead, they have to be analyzed in situ or among physically separated matrix grains with high spatial resolution techniques like NanoSIMS (nanoscale secondary ion mass spectrometry) ion imaging (e.g., Hoppe et al. 2013). Matrix-normalized presolar silicate abundances can exceed 200 parts per million (ppm) in the most primitive unequilibrated meteorites (e.g., Nittler et al. 2018). Oxide stardust grains are generally less abundant, with concentrations of up to tens of ppm in meteorites (Vollmer et al. 2009;Leitner et al. 2012a;Haenecour et al. 2018), and abundances of presolar silicon carbide (SiC) are typically in the range of tens of ppm in primitive meteorites (Floss and Stadermann 2009b;Nguyen et al. 2010;Leitner et al. 2012aLeitner et al. , 2018Zhao et al. 2013Zhao et al. , 2014Davidson et al. 2014a).
Most of the Group 3 grains, with 17 O and 18 O depletions, might have formed around low-mass (<1.4 M • ) AGB stars with Z<Z • . However, the isotopic compositions of several Group 3 grains could also indicate an origin from core-collapse supernova (CCSN) explosions (Nittler et al. 2008;Nguyen and Messenger 2014;Hoppe et al. 2015). Grains of Group 4 have significantly enhanced 18 O/ 16 O ratios, with CCSNe as the most probable stellar sources (Nittler et al. 2008;Vollmer et al. 2008;Nguyen and Messenger 2014). A small number of grains (<1%) show extreme enrichments of 17 O (>5-6 9 10 À3 ) that cannot be accounted for by nucleosynthesis in single AGB stars (Nittler et al. 2008;Gyngard et al. 2011;Palmerini et al. 2011). Binary stars experiencing mass transfer are potential sources for these "extreme Group 1" grains (Nittler et al. 2008); some of them could have formed in the ejecta of nova outbursts Nittler et al. 2008Nittler et al. , 2012Gyngard et al. 2010Gyngard et al. , 2011Leitner et al. 2012b;Nguyen and Messenger 2014).
Dust condensation around AGB stars occurs almost exclusively during the thermally pulsing (TP-) AGB phase. At the start of the TP-AGB phase, the stellar envelope is oxygen-rich (i.e., C/O < 1), facilitating the formation of silicate and oxide stardust. After each thermal pulse, material from the He-intershell dominated by 12 C is mixed to the stellar surface during the so-called "third dredge-up." After a number of dredge-up episodes, the star evolves into a carbon star (C/O > 1), now producing carbonaceous and SiC dust (e.g., Gail et al. [2009] and references therein). Based on their C-, N-, and Si-isotopic compositions, presolar SiC can also be divided into distinct populations (Hoppe and Ott et al., 1997). A detailed overview over the presolar SiC classification and the astrophysical background is given in Hoppe et al. (2010) and Zinner (2014).
CM chondrites are among the more primitive solar system materials and constitute the largest group of carbonaceous chondrites (Brearley and Jones 1998;Lauretta et al. 2000;Weisberg et al. 2006;Chizmadia and Brearley 2008). They have experienced varying degrees of aqueous alteration (McSween and Richardson 1977;Zolensky et al. 1993;Brearley and Jones 1998), representing mixtures of primary phases (e.g., chondrules, Ca-Al-rich inclusions, Fe,Ni metal, anhydrous silicates, sulfides), and secondary phases like phyllosilicates and Fe-oxides, often intergrown on the micrometer to submicrometer scale (Lauretta et al. 2000;Zega and Buseck 2003;Chizmadia and Brearley 2008). This makes them valuable targets for the investigation of the effects of aqueous alteration on primitive components of the solar nebula. One characteristic feature of CM chondrites is the presence of fine-grained dust rims (FGRs) around chondrules ( Fig. 1) and other similarly large constituents that are texturally distinct from the interchondrule matrix (Fuchs et al. 1973;Grossman et al. 1988;Metzler et al. 1992;Zolensky et al. 1993;Brearley 2000, 2001;Liffman and Toscano 2000;Hua et al. 2002;Vogel et al. 2003;Zega and Buseck 2003;Cuzzi 2004;Trigo-Rodriguez et al. 2006). Various different models of rim formation have been suggested, including parent-body scenarios like chondrule erosion or accretion in the regolith (Sears et al. 1993;Tomeoka and Tanimura 2000;Trigo-Rodriguez et al. 2006), but a nebular origin is now widely accepted (Metzler et al. 1992;Ciesla et al. 2003;Chizmadia and Brearley 2008;Hanna and Ketcham 2018). One fundamental obstacle in the discussion of these formation scenarios is the fact that different structures are designated as "fine-grained rims" or "fine-grained chondrule rims." The FGRs investigated in this study (Table 1) share certain properties with the FGRs studied in CO (Davidson [2009]; Haenecour et al. [2018]and references therein) and CR chondrites (e.g., Leitner et al. 2016), such as the presence of primitive components showing little signs of alteration, in close correlation (often at the submicrometer scale) with constituents that have experienced aqueous alteration. In contrast, the FGRs described by Tomeoka and Tanimura (2000) and Takayama and Tomeoka (2012) in the Tagish Lake carbonaceous chondrite and several CV (Vigarano-type) chondrites consisted primarily of phyllosilicates and had experienced severe aqueous alteration.
The presence of presolar grains, especially presolar silicates, in the CR and CO FGRs, and their different abundances in the FGRs and the respective interchondrule matrices of the studied meteorites clearly advocates for a nebular origin of these structures Haenecour et al. 2018). Neither chondrule erosion nor accretion in the regolith could account for the presence of stardust grains or the observed abundance variations. In a nebular setting, free-floating chondrules would accrete rims from finegrained dust (which contains presolar grains), and then form planetesimals/meteorite parent bodies together with "rimless" chondrules, other larger constituents, and the remaining fine-grained dust fraction. Presolar grains from CM chondrites, especially from the well-studied Murchison meteorite, have been the subject of investigations for the past three decades. The vast majority of these investigations have been conducted on grain separates, only suitable for the most refractory types of dust (SiC, graphite, aluminum-rich oxides, nanodiamonds; Bernatowicz et al. 1987;Lewis et al. 1987;Zinner et al. 1987Zinner et al. , 2003Amari et al. 1990;Nittler et al. 1993Nittler et al. , 1995. Nagashima et al. (2005) reported four presolar O-rich grains, potentially silicates, detected in situ in Murchison with a modified ims-1270 ion probe. They calculated an abundance of 3 ppm, but due to the low spatial resolution (500-1000 nm) in comparison to the 100-150 nm typically used for NanoSIMS ion imaging, this abundance is not comparable to NanoSIMS-derived results, and likely underestimates the true abundance, because smaller grains are likely to remain undetected. More recently, Liebig and Liu (2014) detected one presolar Al-oxide and three presolar SiC grains by NanoSIMS ion imaging in Murchison, but no presolar silicates were reported.
Here, we report the results of a comprehensive in situ study of oxygen-rich and SiC stardust in the finegrained chondrule rims within the CM chondrites Banten, Jbilet Winselwan, Maribo, Murchison, Murray, and Yamato 791198, and the CM-related Sutter's Mill meteorite (Jenniskens et al. 2012;Zolensky et al. 2014). Preliminary results were reported as conference papers (Leitner et al. 2013(Leitner et al. , 2014. The presence of stardust grains in the FGRs would further support a nebular origin of the rims and be in line with the detection of presolar silicates and oxides in FGRs from CO (Ornans-type) chondrites (Davidson 2009;Haenecour et al. 2018), CR (Renazzo-type) chondrites, and the ungrouped minimally altered C2 Acfer 094 . Investigation of the presolar grain inventory of fine-grained chondrule rims in CM chondrites will allow for further study of the effects of secondary processing on the nebular materials which provided the building blocks of the larger solar system bodies. Moreover, the CMs bear some spectral similarities to several classes of Main Belt asteroids (B, C, F, and G groups;Vilas and Gaffey 1989;Gaffey et al. 1993), therefore representing potential reference material for the investigation of future samples collected from such asteroid types.

SAMPLES AND EXPERIMENTAL
We studied thin sections of the CM chondrites Banten, Jbilet Winselwan, Maribo, Murchison (two separate sections), Murray, and Yamato (Y-) 791198, together with a thin section from Sutter's Mill (SM2-3)  (Table 1). Petrological subtypes from three different classification schemes (Rubin et al. 2007;Alexander et al. 2013;Howard et al. 2015) are included in Table 1. Prior to the SIMS analyses, suitable target areas within fine-grained rims were identified and documented by scanning electron microscopy. The two main criteria for area selection were (1) absence/preferably low occurrence of larger (d≫1 lm) constituents in the areas, and (2) little/no indications of (extensive) aqueous alteration. We selected five FGRs (out of 193) from Banten, seven (out of 737) from Jbilet Winselwan, eight (out of 282) from Maribo, six (out of 1695) from Murchison, three (out of 112) from Murray, five (out of 889) from Y-791189, and seven rims out of 20 from Sutter's Mill (Table 1).

Scanning Electron Microscopy and SEM-EDS
Samples (Table 1) were characterized by backscatter electron (BSE) imaging and X-ray elemental mapping with an LEO 1530 field emission scanning electron microscope (FE-SEM) equipped with an Oxford X-max 80 silicon drift EDS detector at the Max Planck Institute for Chemistry (MPIC) in Mainz. For data acquisition, Oxford's Aztec software suite was utilized.
Quantitative element data were acquired without external standards. The accuracy of the system was determined in a previous study (Leitner et al. 2018) by measuring a suite of standard minerals with known compositions.

Nanoscale Secondary Ion Mass Spectrometry (NanoSIMS)
Suitable FGRs were selected for O-isotopic analysis to search for presolar O-rich grains with the NanoSIMS 50 ion probe at the MPIC. Carbon and silicon isotopes were measured for SiC-candidates in Murray, Maribo, Jbilet Winselwan, Banten, and Sutter's Mill identified by 28 Si-hotspots (i.e., small high-intensity areas accompanied by low 16 O-intensities) during the O isotope measurements. The success rate for this approach depends on how much synthetic SiC has been deposited in the respective meteorite sections during sample preparation and polishing, but lies typically between 80% and 95%. However, it should be noted that this approach might introduce a bias to the sizes of detected grains, since SiC does not always display significantly enhanced 28 Si-intensities in comparison to the surrounding matrix silicates. This could lead to a  Ruzicka et al. (2015). i Ruzicka et al. (2014). j Specimen SM2-3 was collected before rain (Jenniskens et al. 2012). k Thin section produced from material supplied by Uli Ott (MPIC). l n.d.: Not determined. m Rubin et al. (2007); based on a set of mineralogic and textural properties (e.g., matrix silicate abundance, chondrule mesostasis, abundance and composition of tochilinite-cronstedite intergrowths). n Alexander et al. (2013). Numbers before the slash: Determined from the bulk-N-isotopic compositions, largely comparable to the Rubin et al. (2007) classification; numbers after the slash: classification based on bulk water/OH H contents; compatible with the classification by Howard et al. (2015). o Howard et al. (2015). preferential detection of larger grains. The thin sections of Murchison and Y-791198 had to be recoated after initial NanoSIMS investigation, and the measurement fields containing the SiC-candidates could not be relocated.
Prior to O-isotopic analyses, fields of 14 9 14 lm² were sputtered with a high current primary beam (~20 pA) to remove the carbon coating on selected sample areas. All measurements were conducted in ion imaging mode, where a primary Cs + beam (100 nm diameter,~1 pA) was rastered over 3 9 3 lm 2 -to 10 9 10 lm 2 -sized sample areas. Ion counting rates of 12 C À , 16 O À , and 28 Si À were corrected for quasi-simultaneous arrivals with correction factors according to Slodzian et al. (2004) and Hillion et al. (2008). Corrections were applied individually for each grain or region of interest.
O-isotopic analyses were performed in "chained analysis" mode; that is, after defining a starting position, measurements are conducted automatically, with the sample stage moving in a predefined pattern. For data reduction and processing, in-house software developed at the MPIC ("Nano50Isotopes") was used.

Oxygen Isotopic Measurements
Secondary ion images of 16 O À , 17 O À , 18 O À , 28 Si À , and 27 Al 16 O À were acquired in multicollection mode to identify O-anomalous presolar signatures. Each measurement consisted of five image planes of 11 minutes integration time each over 10 9 10 lm² areas (256 9 256 pixels). Presolar O-rich grains were then identified by their anomalous O-isotopic compositions with respect to surrounding material. Simultaneous detection of 28 Si À and 27 Al 16 O À allowed for a first-order distinction between silicates and Al-rich oxides (Fig. 2). The often-adopted 4r-criterion has been found to be insufficient for the identification of presolar grains . Thus, we used the 5.3rcriterion from the aforementioned study for the identification of presolar isotope anomalies, together with the requirement that the isotopic anomaly is present in at least three subsequent image planes. In the grain identification process, the 50% contour lines around identified anomalous grains (minima and maxima) were applied, and significance levels were calculated from the confidence limits for Poisson statistics of small numbers (Gehrels 1986). Generally, confidence limits should not be calculated simply from the square root of the detected counts when statistics of small numbers are involved. This is because the Poisson distribution is asymmetric for small mean number of counts. We will demonstrate this by the example of grain HR-E@1-11-1 from Hoppe et al. (2015). This grain has d 17 O = 785 AE 173 &, which gives a significance of 4.5 r when statistics based on the square root of the counts is applied. The procedure of Gehrels (1986) takes the asymmetry of the Poisson distribution into account. The equations use two parameters (b, c) which depend on the confidence limit (table 3 of Gehrels 1986). Both parameters are given up to confidence limits corresponding to 3.3 r. As these two parameters vary only smoothly for confidence limits larger than 3 sigma, we expanded the Gehrels approach to confidence limits larger than 3.3 r, using the b and c values for 3.3 r. Using the respective equations gives a significance of 5.4 r for the 17 O anomaly of grain HR-E@1-11-1. We have verified this approach by numerically integrating the Poisson distribution (106 counts measured for 17 O, 59.4 counts calculated for 17 O for solar system composition), which gives a significance of 5.5 sigma. This demonstrates that the Gehrels approximation is quite robust, even for significances exceeding 3.3 r. For larger number of counts, as in the present study, the difference between the significances calculated from the square root of the counts and Gehrels/numerical integration gets smaller. We note that all grains from the present study, except Y_2_16, have significances of more than 5.3 r, even when using "square root" statistics. We chose to use the 5.3 sigma limit here instead of the usually used 4 sigma limit to be consistent with the work of Hoppe et al. (2015), even if the measurements presented here were done with different measurement conditions. Moreover, typical numbers of 17 O counts are not very different (median values of 110 in Hoppe et al. [2015] versus 180 in the work presented here), which results from the fact that image dimensions were a factor of 2 lower and integration times longer in Hoppe et al. (2015). If we would have used a 4 r criterion here for presolar grain identification, only one additional grain would have been identified, that is, the selection of the r-limit has no big impact on the discussion of our data.
The measured O-isotopic compositions of presolar silicates and oxides were normalized to O-rich material from the surrounding matrix, which was assumed to have solar system isotopic ratios ( Table 2; the spread of O-isotopic compositions found in the vast majority of matter of unequivocally solar system origin is smaller than the typical measurement uncertainty of the NanoSIMS measurements reported here, so normalizing our grain data on a single isotope ratio assumed for the matrix is justified).
Not all presolar grain sizes reported here could be determined from SEM images; we had to rely on the secondary ion images instead. In these instances, the sizes are reported as "a 9 b" (in nm), with "a" and "b" being the major and minor axis of an elliptical approximation of the grain. To determine average grain sizes, we recalculated the grain diameters from the respective grain areas by assuming a circular outline with the same area.

Carbon and Silicon Isotopic Measurements
Secondary ion signals of 12 C À , 13 C À , 28 Si À , 29 Si À , and 30 Si À were detected simultaneously in multicollection mode. Five to ten image planes of 4 min integration time each were acquired over 3 9 3 lm²to 4 9 4 lm²-sized areas (128 9 128 pixels). Carbonisotopic ratios were normalized to those measured on a synthetic SiC-standard with a known composition of d 13 C PDB = -29&, and the Si-isotopic ratios of each grain were normalized to the ratios of the whole ion image (assuming terrestrial isotopic composition for the surrounding matrix material), after confirming that no significant deviations between the isotopic composition of the standard and the respective sample areas existed.

Auger Electron Spectroscopy
Major element compositions of six presolar grains from Murchison, Murray, and Jbilet Winselwan were analyzed with the PHI 700 Auger Nanoprobe at Washington University in St. Louis, using established procedures for presolar silicate grains Floss 2018). In brief, after surface sputter cleaning with a widely defocused 2 kV, 1 lA Ar + ion beam, Auger electron energy spectra were obtained with a 10 kV  Table 2; Tables A1 and  A2. (Color figure can be viewed at wileyonlinelibrary.com.) primary electron beam (0.25 nA), which was rastered over the grains of interest. The lateral resolution was 20-25 nm, with an information depth of a few nm. Twenty spectral scans over a small area of a given grain are combined to obtain a single Auger spectrum, which is subjected to a 7point Savitsky-Golay smoothing and differentiation routine prior to peak identification and quantification. Sensitivity factors for O, Si, Fe, Mg, Ca, and Al were determined from olivine and pyroxene standards of various compositions ). Errors are based on the 1r-uncertainties of the sensitivity factors (O: 3.6%; Si: 11%; Fe: 11.2%; Mg: 9.4%; Ca: 10.8%; Al: 24.9%) and should be considered lower limits, since they do not include corrections for background noise, or take into account other factors potentially affecting the quality of a spectrum (e.g., sample charging, residual surface contamination). Detection limits are element dependent, but are typically in the range of a few atom percent. In addition, elemental distribution maps (3 9 3 lm² field of view, 256 9 256 pixels) of two of the grains (JW01_02B_02 and MRY_34_19) and the surrounding meteorite matrix were acquired at 10 keV and 10 nA, with 5-30 scans (depending on the element mapped).

Focused Ion Beam (FIB) Preparation and Transmission Electron Microscopy (TEM)
One electron-transparent lamella was prepared successfully from presolar silicate grain JW01_SA4_54 from Jbilet Winselwan with the focused ion beam (FIB) technique (Wirth 2004;Zega et al. 2007), using an FEI Quanta 3D FEG at the Institute of Geosciences, Friedrich Schiller University (IGW-FSU), Jena. Sample preparation followed standard protocols (e.g., Harries and Zolensky 2016). Subsequent field-emission TEM analyses were performed with an FEI Tecnai G 2 FEG at IGW-FSU, and with a Zeiss Libra 200FE equipped with an in-column Omega energy filter at the Institute for Mineralogy, University of M€ unster. Both TEMs were operated at 200 kV. We used conventional brightfield (BF) TEM imaging for diffraction contrast imaging, and the "high angle annular darkfield" (HAADF) detector in scanning TEM mode for Z contrast imaging. Energy-dispersive Xray spectra were recorded in conventional TEM mode using a defocused beam with integration times of several tens of seconds to obtain sufficient counting statistics using an Oxford X-Max n 80T SDD-EDS detector.

Presolar O-Anomalous Grains
A total area of 85,880 lm² in FGRs from six CM chondrites (Table 1) has been searched for Oanomalous presolar grains (silicates and oxides), together with 10,130 lm² in Sutter's Mill. We identified 15 O-anomalous grains (nine silicates, six oxides) in the CMs (Table 2; Figs. 3 and 4), with sizes ranging from Table 2. Presolar silicates and oxides identified in the CM chondrites analyzed in this study. "Gr." denotes the group of the respective grain, in accordance with the classification by Nittler et al. (1997Nittler et al. ( , 2008. "Ph." gives the phase of the grain (sil.: silicate, Al-/Fe-ox.: Al-/Fe-oxide). Grain sizes were determined from SEM images unless noted otherwise. All errors are 1r. 180 to 600 nm for the silicates and 190-490 nm for the oxides; no O-anomalous grains were found in the areas analyzed in Banten or in Sutter's Mill. Grain diameters were determined either from high-resolution SEM images, or, if not available, estimated from secondary ion images (see Table 2). Murchison and Y-791198 contained three and two silicates, respectively. Three presolar grains were found in each of Maribo and Jbilet Winselwan (two Al-oxides and one silicate, and two silicates and one Fe-oxide, respectively), and four in Murray (one silicate, three Al-rich oxides). Twelve grains belong to Group 1, with 17 O/ 16 O ratios ranging from 4.88 AE 0.21 9 10 À4 to 14.78 AE 1.28 9 10 À4 , and 18 O/ 16 O ratios between 1.15 AE 0.11 9 10 À3 and 2.30 AE 0.12 9 10 À3 . One Al-oxide from Murray falls into Group 2, with 17 O/ 16 O = 6.15 AE 0.55 9 10 À4 and 18 O/ 16 O = 0.89 AE 0.11 9 10 À3 . Finally, three grains are from Group 4, with enhanced 18 O/ 16 O ratios compared to the terrestrial value, ranging from 2.43 AE 0.08 9 10 À3 to 2.85 AE 0.13 9 10 À3 , and 17 O/ 16 O varying between 3.27 AE 0.30 9 10 À4 and 5.83 AE 0.58 9 10 À4 .

Presolar Grain Abundances
The identified presolar silicate and oxide grains represent an average O-anomalous grain abundance of  (Table 3).
The average SiC abundance is 42 AE 10 ppm, with individual abundances between 21 (+20/-11) ppm and 64 (+51/-31) ppm. As we pointed out earlier, no SiC abundances could be determined for Murchison and Y-791198. The confidence level of all errors is 1r (unless stated otherwise) and is based on Poisson statistics. Asymmetrical errors were determined by the method for small numbers described by Gehrels (1986).

Elemental Compositions of Presolar O-Anomalous Grains
Major element compositions of six presolar grains from Murchison, Murray, and Jbilet Winselwan were obtained by auger electron spectroscopy (AES). Quantitative data could be determined successfully for five of these grains (Table A1 in Appendix), while JW01_02B_02 could only be qualitatively identified as Fe-oxide, with, potentially, a small Mg-content (Fig. 6a). Grain MUR_1B_3 turned out to be compositionally heterogeneous; the left part of the grain in Fig. 3b is Si-rich, with (Mg+Fe)/Si = 0.5, and Mgfree (with no Mg above detection limits), while the right part has an Fe-rich olivine-like composition, with Mg# = 28 (with Mg# = Mg/[Mg+Fe] 9 100). Silicate grain JW01_S4A_28 is also Fe-rich (Mg# = 29), but Sipoor, with (Mg+Fe)/Si = 2.5 (Table A1 in Appendix). Silicate MRY_20_07 is of the "intermediate" type , showing an (Mg+Fe)/Si ratio of 1.5 AE 0.2, and Mg-rich (with no Fe above detection limits). We also measured two Al-oxide grains, MRY_34_04 and MRY_34_19 (Fig. 6b), and found them to have spinel-like compositions ( Table A1 in the Fig. 4. Oxygen three-isotope plot showing the isotopic ratios of presolar silicates (red circles) and oxides (blue diamonds) found in five of the six CM chondrites of this study (no grains were found in Banten). The four main groups according to Nittler et al. (1997Nittler et al. ( , 2008 are marked by the dashed black lines. Literature data of presolar silicates and oxides (gray symbols, "literature"; Nittler et al. 1993Nittler et al. , 1994Nittler et al. , 1997Nittler et al. , 1998Nittler et al. , 2008Huss et al. 1994;Hutcheon et al. 1994;Strebel et al. 1997;Choi et al. 1998Choi et al. , 1999Alexander 1999, 2003;Krestina et al. 2002;Messenger et al. 2003Messenger et al. , 2005Nguyen et al. 2003Nguyen et al. , 2007Nguyen et al. , 2010Zinner et al. 2003Zinner et al. , 2005Mostefaoui and Hoppe 2004;Nagashima et al. 2004;Nguyen and Zinner 2004;Floss et al. 2006Floss et al. , 2008Marhas et al. 2006;Stadermann et al. 2006;Bland et al. 2007;Vollmer et al. 2007Vollmer et al. , 2008Vollmer et al. , 2009Yada et al. 2008;Busemann et al. 2009;Floss andStadermann 2009a, 2012;Bose et al. 2010aBose et al. , 2010bGyngard et al. 2010;Keller and Messenger 2011). Errors are 1r. The solar system average ratios are indicated by the dashed lines. SiC grains discovered in this study. The compositional range observed for MS, AB, and Z grains is marked by the oval shapes. Solar system composition is denoted by the dashed lines. All displayed errors are 1r. The blue line denotes the SiC mainstream line (Zinner et al. 2006). (Color figure can be viewed at wileyonlinelibrary.com.) Appendix), with Al/Mg ratios of 3.1 AE 0.9 and 2.6 AE 0.7, respectively.

Transmission Electron Microscopy and SEM-EDS Analysis of JW01_S4A_54
The investigation of the relatively large (490 nm 9 735 nm) silicate grain JW01_S4A_54 by AES was hampered by sample charging effects, but, due to its size, quantitative SEM-EDS was successful (spot analysis, lateral resolution~400 nm at 5 kV; information depth~300 nm). To monitor contributions from surrounding matrix material, we also obtained spectral data from four locations (about~0.5-1 lm distance) surrounding the presolar grain. The results are summarized in Table A2 in the Appendix and show that no Ca is present in the neighborhood of JW01_S4A_54. The grain is an Mg-Ca-bearing silicate, with a diopsidelike composition (Table A2 in the Appendix). The grain was subsequently prepared for TEM-analysis. Unfortunately, after some preliminary investigations, the TEM lamella was contaminated during handling, so no further structural data could be obtained. However, selected area electron diffraction (SAED) patterns indicate it to be crystalline. The observed lattice plane d-spacing of 0.548 nm is consistent with Mg-rich melilite (d 110 = 0.553 nm in endmember akermanite; Swainson et al. 1992) and rules out diopside as the main phase. A weaker reflection at d = 0.289 nm may belong to diopside, but this assignment is ambiguous. TEM-EDS spectra obtained before contamination qualitatively verify the SEM-EDS result, but were taken with a strongly defocused beam such that contributions from neighboring Fe-Mg silicates hamper the assessment of stoichiometry (Fig. 7a, approximate location from where the spectrum was obtained denoted by the yellow arrow). This is illustrated by the occurrence of Al, S, and Fe in the TEM-EDS spectrum, as well as by the presence of Cu from the copper grid.

Isotopic Signatures of Presolar Silicate and Oxide Grains
Most of the presolar silicates and oxides studied here (~75%) belong to Group 1 (Fig. 4), originating most likely from low-mass (~1.2-2.2 M • ) RGB/AGB stars (Nittler 2009;Palmerini et al. 2011). Their O isotope signatures are well explained by the "first dredge-up" (FDU) process, occurring when the star turns into a red giant star (Lattanzio and Boothroyd 1997;Boothroyd and Sackmann 1999). The star's surface composition is modified by the admixing of the products of partial H-burning, and its 17 O/ 16 O ratio is greatly enhanced (El Eid 1994;Boothroyd and Sackmann 1999). The stellar 18 O/ 16 O ratio is only marginally affected by the FDU, thus mainly representing the initial composition of the stellar envelope. However, recent Mg-isotope investigations found that a fraction (~10-15%) of the Group 1 grains originates from stellar sites undergoing explosive hydrogen-burning, most likely CCSN with H-ingestion (Leitner and Hoppe 2019), or super-AGB stars (M = 8-10 M • ), potentially ending their lives as so-called electron-capture supernovae (Nittler 2019;Verdier-Paoletti et al. 2019b). Thus, without Mg-isotopic data, there remains some uncertainty for the stellar sources of the Group 1 grains reported here. We compared the oxygen isotopic data of the grains from this study with presolar oxide grain data and the masses and initial Table 3. Abundances of presolar silicate, oxide, and SiC grains (given in ppm) for the individual meteorites studied here. Errors are 1r, and were calculated according to Gehrels (1986). In the last column, the numbers of grains of each type found in the respective meteorites are given.
The three Group 4 grains represent~19% of the grain inventory. This number is higher than the 10% typically observed for >200 nm-sized presolar O-rich grains in primitive chondrites (Floss and Haenecour 2016), but the significance of this difference is unclear due to the very limited statistics. The grains display moderately enhanced 18 O/ 16 O ratios, while the 17 O/ 16 O scatter around the solar system value, being consistent with an origin in a 15 M • CCSN, with contributions from the Si/S (~0.05%), O/Ne (~0.15%), O/C (~1.9%), He/C (~1.1%), He/N (~4.8%), and H (~92.0%) zones (Rauscher et al. 2002;Vollmer et al. 2008;Nguyen and Messenger 2014).

Silicon Carbide Grains
Eighteen out of 21 SiC grains (~86%) found in this study show isotopic characteristics compatible with the observations from other meteoritic sample sets for MS SiC, supporting an origin in 1.5-3 M • AGB stars with $ solar metallicity (Lodders 2003;Zinner et al. 2006) ( Table A3 in Appendix, Fig. 5). Three grains (~14%) are of type AB. Their origins are still under debate; possible sources include J-type carbon stars as well as born-again AGB stars, and an origin in CCSN e with explosive H-burning has been suggested (Amari et al. 2001;Fujiya et al. 2013;Liu et al. 2017aLiu et al. , 2017bNguyen et al. 2018;Schmidt et al. 2018).

Elemental Composition of Silicate and Oxide Stardust
The elemental compositions of presolar grains, especially silicates, can be used to gain insights into aqueous or thermal alteration affecting not only the grains themselves but (potentially) also the host meteorite. The three presolar silicates studied by AES and grain JW01_S4A_54 investigated by SEM-EDS and TEM represent, of course, too small a number to draw general conclusions about potential trends in grain compositions. Nevertheless, due to the relatively small number of presolar silicates and oxides studied by TEM compared to the number of grains identified by secondary ion mass spectrometry (<100 versus >1,500), any TEM data are a valuable addition to the existing data set. It should be noted that, despite the obvious aqueously altered state of the CMs, we still observed the full range of Mg-contents typical of presolar silicates from other chondrites, from Mg# = 0 to 100. The (Mg+Fe)/Si ratios also span the full compositional range observed for presolar silicates, from "Si-rich" (left part of MUR#001_1B_3; Fig. 3b Grain JW01_S4A_54 is a Mg-Ca-bearing silicate. TEM investigation showed its structure to be compatible with akermanite (Mg-rich melilite, Ca 2 Mg [Si 2 O 7 ]; Fig. 7b). In contrast, the previous SEM-EDS analysis indicated a diopside-like composition (MgCaSi 2 O 6 ; Table A2 in the Appendix); that is, the Ca content is smaller than expected for near-endmember akermanite. This conundrum can be resolved with the following considerations: We have to take into account that the excitation volume of the SEM-EDS measurement, even though conducted with 5 kV acceleration voltage, reaches deeper into the sample than the presolar grain ( Fig. 7a and 7c). Let us now assume that the underlying matrix is virtually free of Ca (as is the fine-grained material surrounding the presolar grain, Table A2 in the Appendix). In this case, the measured grain composition would be diluted by Mg and Si from outside the grain, and the measured Caconcentration would be lower than the true Ca content of JW01_S4A_54. Therefore, the undiluted composition of the grain could indeed be compatible with akermanite. The diffraction pattern (Fig. 7b) appears to indicate the presence of diopside besides the more pronounced akermanite-pattern. In addition, the SEM image (Fig. 3a) shows a slight variation of contrast over the grain: the lower left portion of the grain appears to be somewhat brighter than the rest. This could indicate a compositional and/or structural difference; thus, JW01_S4A_54 could indeed be a composite grain, consisting of an intergrowth of akermanite and diopside. Melilite and spinel are the first Mg-bearing phases condensing from a cooling gas in the typical pressure range for circumstellar outflows (p~10 À10 bar to 10 À6 bar) at temperatures of 1452 K and 1397 K, respectively. They are followed by anorthite (CaAl 2 Si 2 O 8 , 1387 K), forsterite (Mg 2 SiO 4 , 1354 K), and diopside (CaMgSi 2 O 6 , 1347 K; Gail andSedlmayr 1998, 1999;Lodders 2003). If we consider now total pressures and dust/gas-ratios in the typical range for circumstellar envelopes (e.g., p tot~1 0 À6 to 10 À8 bar; Jeong et al. 2003), and dust/gas = 4 9 10 À3 to 9 9 10 À2 (by mass; Gobrecht et al. 2015), diopside condenses in fact at higher temperatures than forsterite. And if we further assume that the entire Al was already bound in other earlier formed minerals, condensation of diopside around an akermanite core becomes feasible.
The two spinel-like grains from Murray both show Al/Mg ratios slightly higher than stoichiometric (3.1 AE 0.9 and 2.6 AE 0.7, calculated from Table A1 in the Appendix). Given the comparably large errors, this observation is not significant, although there are some indications that the presolar spinels from Murray might indeed have elevated Al/Mg ratios (Zinner et al. 2005;Gyngard et al. 2010). High temperature processing has been suggested ), similar to observations made for spinels in Ca-Al-rich inclusions (Simon et al. 1994). However, for the Murray spinels studied by Gyngard et al. (2010), it could not be fully ruled out that the higher Al/Mg ratios resulted from artifacts of the SIMS measurements (caused by the differing ionization efficiencies of Mg and Al in combination with measurement times not long enough to allow for full equilibration of the ion signals, e.g., Lin et al. 2010).

Origins of Fe-Rich Presolar Dust Grains
Two of the silicate grains studied by AES show comparably high Fe-abundances (12-18 atom%, Table A1 in the Appendix and Fig. 8). This is compatible with the observation that many of the stardust silicates found to date have relatively high Fe contents (Fig. 8), and stands in contrast to equilibrium condensation models predicting silicates with higher Mg contents as the first silicate phases to condense at high temperatures (~1000 K) in stellar environments (Gail 2003;Min et al. 2007). The origin of this Fe is still a matter of intense research, as it may be either a primary condensation or a secondary alteration feature Vollmer et al. 2009;Bose et al. 2010aBose et al. , 2010bOng and Floss 2015). One feasible primary mechanism for the production of Fe-rich silicates would be nonequilibrium grain condensation in stellar outflows (see also the discussion in Bose et al. 2010a). Gas and dust are rapidly driven outward by radiation pressure, resulting in a steep temperature drop where equilibrium may not be maintained. This would allow the incorporation of considerable amounts of Fe into condensing silicate grains (Gail and Sedlmayr 1999;Gail 2003). This is further supported by laboratory experiments on nonequilibrium condensation of silicates (Rietmeijer et al. 1999;Nuth et al. 2000), as well as astronomical observations of Fe-rich silicates (e.g., Kemper et al. 2001). The experiments showed that Mgand Fe-rich condensates with either high Mg-or Fecontents form directly from the vapor, while grains with intermediate or "mixed" Mg-Fe-compositions are not among the primary condensates (Rietmeijer et al. 1999). Furthermore, Rietmeijer et al. (1999) proposed the accretion of dust aggregates from endmember silicates, which subsequently are transformed by energetic reactions into ferromagnesian grains. Especially the heterogeneous composition of MUR_1B_3 (Si-rich Fesilicate, together with a fayalite-like, i.e., Fe-rich olivinelike, composition) might be indicative of an origin by nonequilibrium processes (Rietmeijer et al. 1999;Ferrarotti and Gail 2001).
Possible secondary processes responsible for the introduction of Fe into presolar grains include Fe ion implantation or preferential sputtering of Mg and Si relative to Fe (e.g., Jones 2000), assuming sufficiently long residence times of the dust grains in the ISM. This might have indeed played a major role for JW01_S4A_28, with its Si-poor-and Fe-rich (Mg# = 29) composition. For MUR_1B_3, however, this effect could only have played a minor role (if any). We would expect a Fe-rich and Si-poor composition instead of the Si-rich grain portion, and a rather pyroxene-like composition instead of the Fe-rich olivine-like grain portion (Table A1 in the Appendix), due to the aforementioned preferential removal of Si and Mg compared to Fe by sputtering processes. Secondary parent body processes like aqueous alteration and thermal metamorphism can also be responsible for the introduction of additional Fe via fluid-mineral interactions (Le Guillou et al. 2015;Vollmer et al. 2016;Hopp and Vollmer 2018), or diffusion (Jones and Rubie [1991]; Vollmer et al. [2009] and references therein). Especially for the CM chondrites, there is a growing body of evidence that at least some specimens experienced some degree of thermal metamorphism, in addition to the abundant aqueous alteration (Nagashima et al. 2005;Tonui et al. 2014;Schrader and Davidson 2017;King et al. 2019). Thus, their presolar grain inventory might have also been affected by such processes, resulting in grain destruction and/or chemical modification.
Presolar Fe-oxides are very rare in our current inventory of presolar dust; only a few grains have been observed so far Bose et al. 2010a;Zega et al. 2015;Haenecour et al. 2018). Five of them (Bose et al. 2012;Zega et al. 2015;Haenecour et al. 2018) originate from low-mass AGB stars, while three grains Bose et al. 2010a) belong to Group 4. Although a supernova origin is now widely accepted for the Group 4 silicate and oxides, Floss et al. (2008) showed that in the case of Fe-oxides, parent stars with supersolar metallicities might well be a viable alternative. A proper elemental quantification of grain JW01_02B_02 by AES was hampered by the fact that, due to the position of the analysis spot, there is an unknown contribution from surrounding Mg-(and potentially Fe-) bearing material (Fig. 6a). Thus, we could not determine whether JW01_02B_02 had a magnetite-, w€ ustite-, or magnesiow€ ustite-like composition, although magnesiow€ ustite appears less likely, judging from the AES element maps. Preparation of the grain for compositional and structural analysis by TEM was not successful, and no further information could be obtained.
As discussed by Floss et al. (2008), condensation of magnesiow€ ustite (Mg,Fe)O is likely to occur under nonequilibrium conditions around AGB stars with low mass-loss rates (Ṁ ≤ 4 9 10 À6 M • yr À1 ; Ferrarotti and Gail 2003). Its condensation is prevented at higher rates due to the early condensation of silicates. Calculations by Ferrarotti and Gail (2003) show that with decreasing temperature, higher concentrations of Fe can be incorporated into condensing magnesiow€ ustite grains; however, formation of almost pure FeO requires an additional explanation. Additional diffusion of Fe into the grain might well explain its composition, although we cannot constrain whether this already happened in the circumstellar envelope, or later on the meteorite parent body. Zega et al. (2015) discussed the formation conditions for circumstellar Fe 3 O 4 , around AGB stars. They found that oxidation of previously formed Fe dust in the circumstellar envelope is feasible, if residence times of~9,000-500,000 years are taken into account. However, due to the higher ejection velocities and the resulting faster dispersion of gas and dust in supernova environments, we do not consider such resident times as very likely in the present case. As mentioned earlier, Floss et al. (2008) concluded a more likely origin of their Fe-oxide grain from a high metallicity AGB-star, although its O-isotopic signature would also comply with a CCSN origin. Moreover, there is some observational evidence for the presence of supernova Fe-oxides: In IR spectra from the supernova remnant Cassiopeia A, FeO has been identified as one of the carriers of the 21 lm feature (Rho et al. 2008(Rho et al. , 2009Arendt et al. 2014). Therefore, we can neither confirm nor refute an AGB star or CCSN origin for our grain.

Presolar Grain Abundances and Grain Sizes
The occurrence of presolar grains in the FGRs of the CM chondrites gives evidence for their relatively primitive nature and thus for a nebular origin of the host rims, similar to the FGRs observed in CR (e.g., Leitner et al. 2016) and CO chondrites (e.g., Haenecour et al. 2018). The observed abundances of O-rich presolar grains in the fine-grained chondrule rims in CM chondrites are comparable with both the abundances in the interchondrule matrix (ICM) of moderately altered CR2 chondrites Koch and Floss 2017) and fine-grained clasts of petrologic types 1-2 in three metal-rich chondrites from the CR clan, Isheyevo (CH/CB [high metal/Bencubbinlike] 3), Acfer 182 (CH3), and Lewis Cliff (LEW) 85332 (C3-ungr.; Leitner et al. 2018; Fig. 9). The concentrations are lower than those observed for the FGRs in a set of CR2 chondrites , as well as those reported for the least altered chondrites (Floss and Stadermann 2009a;Vollmer et al. 2009;Nguyen et al. 2010;Zhao et al. 2013;Nittler et al. 2018; Fig. 9), which can be attributed to progressive parent body processing. Several classification schemes have been introduced to assign petrologic subtypes to the CM chondrites (Rubin et al. 2007;Alexander et al. 2013;Howard et al. 2015; Table 1). With the exception of Maribo and Sutter's Mill, all CMs studied here span a very narrow range of bulk petrologic subtypes, no matter which classification scheme is used (Table 1). Sutter's Mill experienced more severe alteration (CM2.0-2.1; Jenniskens et al. 2012) and is not considered a "classical" CM, while Maribo is, together Fig. 9. Abundances of presolar O-rich dust grains in parts per million (ppm) for the CM chondrites and Sutter's Mill from this study. For comparison, literature data for CM, CO, and CR chondrites; clast material from the metal-rich chondrites of the CR clan; as well as several ungrouped carbonaceous chondrites are shown (Floss and Stadermann 2005, 2009aMarhas et al. 2006;Vollmer et al. 2009;Nguyen et al. 2010;Mostefaoui 2011;Zhao et al. 2011Zhao et al. , 2013Zhao et al. , 2014Leitner et al. 2012aLeitner et al. , 2016Leitner et al. , 2018Davidson et al. 2014bDavidson et al. , 2015Heck et al. 2014;Koch and Floss 2017;Haenecour et al. 2018;Nittler et al. 2018 with the CM Paris (Hewins et al. 2014), apparently one of the least altered CMs, with Maribo being >CM2.6 (Haack et al. 2012), and Paris classified as~CM2.9 (Hewins et al. 2014). The vast majority of CM chondrites are brecciated (Bischoff et al. 2006), complicating the assignation of representative "bulk" petrologic subtypes (Bischoff et al. 2017). However, within error limits, there is no significant difference between the presolar O-rich grain abundances of Maribo and the other CMs investigated here. Moreover, the abundances determined for two lithologies from Paris (Verdier-Paoletti et al. 2019a; Fig. 9) also lie within the average range of abundances determined for the CMs (Fig. 9). Thus, the presolar silicate and oxide inventories of even the least altered CMs are significantly affected by aqueous alteration (and, potentially, by thermal metamorphism), if we assume a largely homogeneous initial distribution of presolar silicates and oxides in the protosolar nebula. Alternatively, given the relatively primitive nature of Paris and Maribo, the precursor material of their matrices could have contained a smaller initial abundance of presolar O-rich grains than the finegrained materials accreted, for example, into CR or CO chondrites. This could have been the result of anisotropies/heterogeneities in the distribution of presolar O-rich dust in the formation region of the Sun, depending on how well mixed the precursor material of the prestellar core was. And since oxygen-rich and carbonaceous dust form during different evolutionary stages of AGB stars, they might as well experience differing degrees of mixing in the ISM. Moreover, silicates (and oxides) must not necessarily form around every evolved star, while model calculations indicate that the vast majority of low-mass AGB stars undergo C-rich episodes with dust formation. This could hypothetically result in localized "clumps" or "bubbles" with O-rich and C-rich dust. With greatly enhanced statistics, future studies might be able to monitor differences in the presolar grain concentrations of samples with different petrologic subtypes. The presolar SiC abundances of our CM samples do not differ significantly from the literature values obtained by ion imaging (Table 3; Fig. 10), supporting the idea of a roughly homogeneous distribution of SiC in primitive solar system materials (Davidson et al. 2014a). However, a potential bias might be introduced by the applied method of SiC detection by targeting 28 Sihotspots recorded during the O isotope analyses (see the Methods section). Smaller grains might have been missed; thus, the abundance might have been underestimated, but the preferential detection of larger grains could have an opposing effect by leading to overestimating the true abundances. Presolar silicate grains are considered to be more susceptible to aqueous alteration than the more refractory Al-oxide grains (Floss and Stadermann 2009a;Leitner et al. 2012a); thus, the presolar silicate abundances as well as the presolar silicate/oxide (sil./ ox.) ratios of primitive materials can provide information on the degree of alteration they experienced. While high sil./ox. ratios would thus indicate relatively unaltered material, lower ratios would be found for more aqueously altered host material, due to preferential destruction of presolar silicate grains in comparison to the more refractory species, such as Alrich oxides or SiC. In the vast majority of cases, the number of presolar oxide grains found in individual meteorites is very low (typically <5), resulting in relative errors of >50% for the respective sil./ox. ratios, making them practically useless to monitor none but the most extreme differences in the degrees of aqueous alteration between individual meteorites (i.e., in cases where very low sil./ox. ratios would be observed alongside sufficiently high ratios, depending on the respective statistics). The presolar silicate/oxide ratio of the CM chondrites studied here stands out, because it is the lowest ratio observed so far (sil./ox. = 1.5), only comparable with the ratio of~2 found for the CR2 chondrite NWA 852 (Leitner et al. 2012a). Surprisingly, the presolar O-rich grain inventories of the ICM of other CR2 chondrites and the hydrated lithic clasts in the CH/CB chondrites (Leitner et al. , 2018Koch and Floss 2017;Nittler et al. 2019) are dominated by silicates, although the presolar grain abundances are similar to the CM chondrites studied here (Fig. 11). Both the CM and CR chondrites are generally classified as petrologic type 2, but several classification schemes have been proposed to define suitable petrologic subtypes. The studies by Alexander et al. (2013) and Howard et al. (2015) both introduce a petrologic subtype scale that allows such an intergroup comparison for the CMs and CRs, based on the bulk water/OH H content. For the CM2 chondrites of our study, petrologic subtypes range from 1.5 to 1.7 on these scales (for Maribo and Sutter's Mill, no subtype has been determined yet). For the CR2 chondrites, the values are typically higher, ranging from 2.4 to 2.8 (with the exception of Al Rais with 2.0), and for the C3ungrouped LEW 85332 (which belongs to the metal-rich chondrites of the CR clan and whose presolar grain inventory was studied by Leitner et al. 2018), 2.3 is calculated. These results illustrate the general trend quite well (CM2s are more altered than CR2s, the finegrained clasts in the CH and CB chondrites and in LEW 85332 are also of petrologic types 1-2). However, there are some issues with the subtypes assigned to the CR chondrites; Renazzo is more severely altered than most of the Antarctic CR2s and has an average ICM abundance of <6 ppm for O-rich presolar grains (Fig. 9), but has a higher petrologic classification than QUE 99177 with an average ICM abundance of 177 ppm for O-rich presolar grains. There is a discussion of these issues in Alexander et al. (2013) and Leitner et al. (2016), and one of the major reasons for these discrepancies might be the fact that aqueous alteration on the respective parent bodies did not occur homogeneously, both in spatial distribution and temperature. Another major reason could be that assigning bulk petrologic types to these rocks is difficult, since both CM and CR chondrites are highly brecciated (Bischoff et al. 2006(Bischoff et al. , 2017. While different degrees of aqueous alteration would explain decreasing presolar silicate abundances in combination with decreasing sil./ox. ratios, we found abundances similar to those found for the ICM of CR2s and the matrix Fig. 10. Abundances of presolar SiC grains in parts per million (ppm) for the CM chondrites and Sutter's Mill from this study. For comparison, literature data for CM and CR chondrites, clast material from the metal-rich chondrites of the CR clan, Orgueil (CI1), and a set of ungrouped carbonaceous chondrites are shown (Marhas et al. 2006;Zhao et al. 2011Zhao et al. , 2013Zhao et al. , 2014Floss and Stadermann 2012;Leitner et al. 2012aLeitner et al. , 2018Davidson et al. 2014aDavidson et al. , 2015Verdier-Paoletti et al. 2019a; the "CR fg (fine-grained) average" was calculated with data from Floss andStadermann 2005, 2009b;Nguyen et al. 2010;Leitner et al. 2012c;Davidson et al. 2014aDavidson et al. , 2015. Data are from ion imaging surveys except for the star symbols (abundances inferred from noble gas analyses, Huss et al. 2003;Haack et al. 2012) and the white square symbol (data from stepped combustion measurements, Newton et al. 1995). All errors are 1r. FGR: fine-grained rims. (Color figure can be viewed at wileyonlinelibra ry.com.) clasts in the metal-rich chondrites together, but with different sil./ox. ratios.
This could indicate a heterogeneous distribution of presolar silicates and oxides within the carbonaceous chondrite forming regions. For a clear conclusion, improved statistics are needed (especially on the presolar oxide abundances). Poor statistics for presolar oxides is a critical issue, but if we focus on the few primitive chondrites with better constrained sil./ox. ratios (NWA 852, Adelaide, Acfer 094, DOM 08006; Vollmer et al. 2009;Floss and Stadermann 2012;Leitner et al. 2012a;Nittler et al. 2018), and combine the results from multiple studies of the most primitive CR and CO chondrites (Fig. 9), we find a relatively large spread between the individual data points. Due to the large statistical errors of some of the sil./ox. ratios considered here, the ratios are comparable to one another within 2 sigma-errors. Thus, it is still possible that with improved statistics, it will be possible to show that there is no significant variation in the sil./ox. ratios of different meteorite groups or individual meteorites. This would indicate a homogeneous distribution of presolar silicates and oxides in the protosolar nebula, that is, the circumstellar dust component had been well mixed in the ISM. However, if further investigations established more significant differences between individual meteorites or groups of meteorites, a potential reason for such heterogeneities is discussed the following paragraph.
An observational study by Karovicova et al. (2013) reported different dust shell structures around four Orich AGB stars: two stars displayed only features of Aloxide dust, one star featured a silicate dust shell, and the fourth star had both Al-oxide and silicate shells. Of course, the statistics are not (yet) significant, but it might be a first hint for a larger initial variation of circumstellar sil./ox. ratios, and therefore, the distributions of presolar silicates and oxides in the ISM and also the protosolar cloud might have been less homogeneous than originally anticipated. In a previous study (Leitner et al. 2012a), we estimated a presolar sil./ ox. ratio for the presolar dust component in the following way: We assumed (as a first-order approximation) that (during the formation of O-rich dust) almost all Si from AGB stars (which contributed 90% of the O-rich presolar dust to the protosolar nebula) condenses into silicate dust, while essentially all Al forms oxide dust. SiC and other carbonaceous presolar phases can be neglected at this point, since they form in another stellar evolutionary phase. If we choose pyroxene as silicate proxy and alumina (Al 2 O 3 ) as proxy for oxides, and further consider solar abundances for Al and Si (Asplund et al. 2009, we find a sil./ox. ratio for AGB star dust of Si/(Al/2) = 2 9 Si/Al = 23. Of course, the sil./ox. ratio of SN dust has to be considered, too. For a more accurate estimate, we investigated CCSN models from Rauscher et al. (2002) and Pignatari et al. (2016) with stellar masses of 15, 20, 25, 32, and 60 M • and calculated the ratios in the same way. The sil./ox. ratios range from 34 to 230, and as a consequence, we find lower limits for the "protosolar cloud initial" (pci) of 24 (with 10% SN contribution) and 25 (with 20% SN contribution; Fig. 11). However, this first-order approximation might be too simple by adopting a single presolar sil./ox. ratio, and the dust in the protosolar cloud might have consisted therefore of several heterogeneously distributed populations with different ratios.
We also observed a slightly larger average grain size (~330 nm) than for the presolar silicates and oxides in Fig. 11. Presolar silicate/oxide ratios (calculated from the respective presolar grain abundances) plotted versus the presolar O-anomalous grain abundances for the CM2 chondrites of this study (average). For comparison, data for the average from a set of CR2 chondrites  and "least altered" CR chondrites (Floss and Stadermann 2009a;Nguyen et al. 2010;Zhao et al. 2013;Leitner et al. 2016) are shown, as well as an average value for CO3.0 chondrites (Nguyen et al. 2010;Bose et al. 2012;Haenecour et al. 2018 (Table 4), we only considered reference data from studies conducted with the "standard" NanoSIMS settings (spatial resolution ≥100 nm). The recent investigations of Hoppe et al. (2015Hoppe et al. ( , 2017) used a higher spatial resolution (<100 nm), detecting a multitude of smaller (d < 150 nm) grains that are typically missed in lower resolution mode, and thus are not directly comparable with the results from other studies. Additional uncertainty arises from the fact that the diameters for many grains have been determined from NanoSIMS ion images, since different laboratories and researchers might well apply differing methods or criteria when estimating the size of a presolar grain. Unfortunately, we currently see no way to eliminate this potential systematical error. However, we have at least the possibility to use an "internal" calibration to compare the grain sizes of the presolar O-rich grains from the CM chondrites with data from studies conducted at the MPIC (Vollmer et al. 2009;Leitner et al. 2012aLeitner et al. , 2016Leitner et al. , 2018Kod olany ı et al. 2014). The average grain sizes from these "calibration studies" do not differ significantly from the averages found in other studies (Table 4) and are all smaller than the average value calculated for the CM chondrite grains. Moreover, many reported grain sizes are obtained from SEM or AES measurements and thus are more precise than the NanoSIMS-derived. Thus, the best way to get a "reliable" average grain size might be by integrating over a large number of grains from different studies. The larger average grain size of silicates from this study could be attributed to preferential destruction of smaller grains by progressing aqueous alteration. Al-oxides are generally considered to be largely unaffected by aqueous processes. The majority of presolar Al 2 O 3 grains are (thermally stable) a-Al 2 O 3 (corundum), but, for example, amorphous alumina has also been observed (Stroud et al. 2004). To our knowledge, no investigations have been conducted on a-Al 2 O 3 , but it has been shown that contact of water with (metastable) c-Al 2 O 3 leads to hydration/hydroxylation and, subsequently, to formation of Al-oxihydroxides and Alhydroxides at the Al 2 O 3 -surface; moreover, the pH value of the liquid has a strong influence on the speed Table 4. Average grain sizes for both the presolar O-rich and silicon carbide grains identified in this study, together with average grain sizes for other chondrites calculated from literature data (with the exception of the data from Davidson et al. (2014a); here, the average grain sizes and standard deviations were directly adapted from the source). For each value, the standard deviation and the number of grains used to calculate the average are given.  (2009b) of the respective reactions (e.g., Lef evre et al. 2002). Of course, we do not have exact information on the fluid interaction processes affecting the fine-grained CM constituents; however, this could be a possible scenario for the destruction of (smaller) amorphous Al-oxides. Some CM chondrites, including Jbilet Winselwan, Murchison, and Murray appear to have experienced thermal metamorphism to a certain degree (Nagashima et al. 2005;Tonui et al. 2014;Schrader and Davidson 2017;King et al. 2019), which could also have contributed to the preferential destruction of smaller grains. Reported peak metamorphic temperatures are relatively low: T < 220°C-230°C for Murray and Murchison (Schrader and Davidson 2017), and up to 400°C-500°C for some lithologies in Jbilet Winselwan (King et al. 2019). However, these temperatures are typically determined by investigating chondrules and other similarly large mineral grains; the thermal effects and peak temperatures experienced by the fine-grained matrix (which hosts the presolar grains) might be much stronger: Bland et al. (2014) showed in their study that low-energy impacts have the potential to cause strong temperature increases of even more than 1000 K in localized areas in the fine-grained fraction of chondrites, while larger and compact constituents like chondrules would act as heat sinks and not be affected in the same way as the matrix. Evidence for multiple of such lowenergy impacts has been found in CM chondrites (Lindgren et al. 2015); thus, presolar grain destruction by these events is a possible scenario that might be responsible for the observed presolar silicate and oxide abundances.
If we do a first-order estimate of the "prealteration" presolar silicate abundance, similar to the calculation made by Leitner et al. (2012a), but remembering that the scenario is likely more complicated, we find an estimated concentration of 170 (+130/-90) ppm in the CMs, much lower than the estimate of~800 ppm made for NWA 852 (CR2). We calculated these "pre-alteration" abundances by using the respective sil./ox. ratios and the "(pci)" ratio estimated above as calibration factors for the presolar silicate abundances: ([sil./ox.] pci /[sil./ox] meteorite )9 (presolar silicate abundance) meteorite . However, if we consider the different average grain sizes, the abundance must be adjusted: Amari et al. (1994) found that the grain size distribution of presolar SiC and graphites from grain separates follow a log-normal distribution and drew the conclusion that this was the preferred distribution law for grain formation in circumstellar envelopes, possibly reflecting a two-stage process of grain evolution by evaporation and recondensation. However, it has to be taken into account that the heavy processing necessary to produce these grain separates could have led to the destruction of smaller grains, and thus, the observed distribution might not be representative. Therefore, we apply here the MRN model distribution found for interstellar grains (Mathis et al. 1977), where the abundance scales with a À3.5 ("a" being the grain radius) to the presolar silicates and oxides. We find that 260 nm grains are predicted to be 2.3 times more abundant than 330 nm grains. With this correction factor, the estimate for presolar silicates becomes 400 (+300/-200) ppm, which overlaps, within error limits, with the~800 (AE340) ppm for NWA 852, and lies also in the range observed for primitive IDPs Busemann et al. 2009;Davidson et al. 2012).
The size distributions of the SiC grains in the CMs agree within error limits with the size distributions observed in other studies for Murchison, Sutter's Mill, metal-rich chondrites from the CR clan, and Orgueil (CI; Table 4). The average from the anomalous CM Bells is significantly larger (770 AE 155 nm), but this is addressed in the original study by Davidson et al. (2014a). For the CR chondrites, we find a difference between the average determined by Davidson et al. (2014a) from 114 grains, and the average calculated from a set of other studies (Floss and Stadermann 2009b;Nguyen et al. 2010;Leitner et al. 2012a;Zhao et al. 2013) Davidson et al. (2014a) range from 300 AE 65 nm to 490 AE 120 nm. Despite this relative scatter of the average values, the size ranges still overlap. Currently, we cannot resolve if the observed differences are indicative of a small difference in the respective grain sizes, or are mainly the result of different approaches of determining grain sizes, as discussed earlier in this manuscript. At least, our method of targeting 28 Si-hotspots could have introduced a bias toward larger SiC grains, since smaller ones (whose 28 Si intensities might be not as pronounced the signal from matrix silicates) might have been missed. However, this does not necessarily refute size-sorting of grains taking place in the solar nebula, since the presolar SiCs seem to have slightly larger average sizes than the silicates and oxides, and thus simply were not affected. However, it is interesting to note that for the various SiC grain separates that were studied over the years, a general difference between the average grain sizes of SiC from Murchison and other primitive meteorites was observed (Huss et al. 1997;Russell et al. 1997;Zinner et al. 2006). At a first glance, this stands in contradiction to our observations for the SiC in CM chondrites, and also to the findings of Davidson et al. (2014a) for SiC from the Murchison meteorite. However, this discrepancy can be explained by the still limited statistics on SiC from in situ investigations compared to the grains isolated from grain separates and the fact that for in situ ion imaging studies, typically, fine-grained areas are targeted, so that larger SiCs in coarser grained regions might be missed. Davidson et al. (2014a) conducted their study on presolar SiC contained in extracted insoluble organic matter, and not on meteorite thin sections. Therefore, such a potential bias should not have affected their measurements. However, if we apply again the MRN model distribution from Mathis et al. (1977), we find that for 16 SiC grains of~330 nm (as reported by Davidson et al. 2014a),~0.33 grains of 1 lm would be expected statistically. From this number, it can easily be seen that the grain statistics needs to be improved significantly for proper conclusions.

SUMMARY AND CONCLUSIONS
1. We conducted an investigation of the presolar grain (silicates, oxides, and SiC) inventories in the FGRs of a set of CM chondrites and the CM-related carbonaceous chondrite Sutter's Mill. Presolar Orich grains (silicates and oxides) are significantly less abundant (18 AE 5 ppm) in the fine-grained rims of the CM chondrites than in the fine-grained rims of CR2 chondrites (72 AE 13 ppm; Leitner et al. 2016) and also the fine-grained components of the most pristine chondrites (CO3.0, minimally altered CR2, Acfer 094; Nguyen et al. 2007Nguyen et al. , 2010Floss and Stadermann 2009a;Vollmer et al. 2009;Bose et al. 2010aBose et al. , 2012Zhao et al. 2013;Leitner et al. 2016;Haenecour et al. 2018;Nittler et al. 2018). The concentrations are, however, compatible with those determined for the interchondrule matrix material of a set of CR2 chondrites (Floss and Stadermann 2005;Leitner et al. 2016;Koch and Floss 2017) and hydrated lithic clasts from Acfer 182 (CH3), Isheyevo (CH/CB3), and LEW 85332 (C3-ungr.; Leitner et al. 2018). The hydrated state of the FGR material in CM CCs suggests destruction of a significant fraction of the original O-rich presolar grain population, indicating a degree of alteration similar to more severely altered CRs like Renazzo, and the hydrated clasts in the CH/CB chondrites. 2. The CM chondrites have the lowest presolar silicate/oxide ratio (~1.5) observed so far for primitive solar system materials. This may result from the destruction of presolar silicates by more severe aqueous alteration, but the lack of clear correlation between sil./ox. ratios and presolar Orich grain abundances could indicate an inherently heterogeneous distribution of silicate and oxide stardust in the protosolar cloud. This is in contrast with the relatively homogeneous distribution of presolar SiC inferred by Davidson et al. (2014a). 3. There is some evidence for slightly larger average grain sizes of presolar silicates and oxides in the CM chondrites compared to the grain inventories of other chondrites. In contrast, for the SiC population, we observed no significant difference from the average grain sizes of other chondrites, although the data appear to indicate a larger scatter than observed for the O-rich grains. Better statistics are needed, but the size difference could be indicative of a heterogeneous size distribution among the silicate and oxide stardust in the solar system. Alternatively, the larger grain sizes could result from preferential destruction of smaller grains by thermal metamorphism (and potentially aqueous alteration, depending on the chemical environment) experienced by the CM chondrites, while the SiC grains were less affected, due to higher thermal stability. 4. The range of Fe-contents observed in four presolar silicate grains corresponds with the range observed for the CR and CO3.0 chondrites, and the ungrouped minimally altered C2-chondrite Acfer 094. One large silicate from Jbilet Winselwan is crystalline, and is likely akermanite (Mg-rich melilite), or an akermanite-diopside composite. 5. The presence of presolar dust in the FGRs of CM chondrites studied here gives evidence for a nebular origin of the host rims, similar to the FGRs observed in CR ) and CO chondrites (Haenecour et al. 2018).
Research in Japan for providing the Y-791198 sample. We appreciate the constructive reviews by Ann Nguyen, Maitrayee Bose, a third anonymous reviewer, and the associate editor Larry Nittler, which have improved the quality of the manuscript. This research has made use of NASA's Astrophysics Data System.