Carbonate assemblages in Cold Bokkeveld CM chondrite reveal complex parent body evolution

The paragenesis of carbonates in the Cold Bokkeveld CM chondrite is determined from a detailed petrographic, chemical, spectroscopic, and isotopic study of nine associations of carbonates (aragonite, calcite, and dolomite) with other secondary minerals that occur within the meteorite. Our study reveals the existence of carbonates displaying petrographic features that are distinct from those of type 1 and type 2 carbonates commonly observed in CM2 meteorites. These include carbonates interstitial to octahedral magnetite crystals, for which a new designation of “type 1c” is suggested. The O isotopic values of dolomite (δ18O ranging from +21.1 to +25.8‰ and Δ17O from −4.9 to −4.0‰) are similar to those measured in dolomites from other CM chondrites. The presence of complex carbonates with a CaCO3 core and Mg‐enriched rim implies several generations of fluids and/or their evolving composition on the CM parent body(ies). Petrographic characteristics indicate at least six stages of potentially overlapping carbonate and phyllosilicate formation events. We show that type 1 and type 2 calcite have distinct Raman spectral characteristics. Type 1 calcite is characterized by very broad peaks, whereas type 2 calcite displays narrow peaks similar to those of typical abiotic terrestrial calcite, suggesting high crystallinity. A carbonate Raman spectrum showing features characteristic of both aragonite and calcite likely documents an aragonite‐calcite phase transition. Raman spectroscopy also reveals the presence of organic matter in the majority of carbonates. This indicates that organic carbon was mobilized by aqueous fluids for extended periods.


INTRODUCTION
Carbonaceous chondrite meteorites represent some of the most primitive materials formed in the early solar system.CM chondrites are the most abundant and diverse group of carbonaceous chondrites and experienced various degrees of brecciation, aqueous alteration, and metamorphic heating.Some moderately altered CM2 chondrites, including Cold Bokkeveld, Jbilet Winselwan, and Nogoya, are highly complex impact breccias that contain lithic clasts and mineral fragments showing various degrees of aqueous alteration (Bischoff et al. 2017;Lentfort et al. 2020;Zolensky et al. 2020).Their sub-mm-scale heterogeneity and petrographic features point toward major preaccretionary aqueous alteration on small precursor planetesimals, followed by large-scale parent body aqueous alteration (Bischoff 1998;Bischoff et al. 2017).In contrast, the lower degree of inter-and intrasample heterogeneity in highly altered CM1 chondrites presumably attests to more extensive parent body alteration overprinting much of the inherited, preaccretionary alteration features.
Petrographic characteristics revealed up to two carbonate-forming events operating on the parent body of Cold Bokkeveld (Lee et al. 2014); the oxygen isotopic compositions of the carbonates have been employed to determine their formation temperatures.Carbonate clumped isotope thermometry yielded carbonate formation temperatures for separate Cold Bokkeveld aliquots of ~26, ~35, and ~71 °C (Guo and Eiler 2007), consistent with the relatively low temperature alteration of CM2 chondrites inferred from C and O isotopic compositions (e.g., Alexander et al. 2015).However, the novel method of Verdier-Paoletti et al. (2017) gave much higher carbonate formation temperatures ranging from 97.9 to 225.3 °C.Clumped isotopes have also been used to infer basic conditions with a minimum pH value of 11.7 during carbonate formation (Guo and Eiler 2007).
Despite the broad literature on CM chondritic carbonate minerals, the paragenetic sequence of carbonates and associated phases on the CM chondrite parent body is not fully understood.We present a combined petrographic, chemical, spectroscopic, and isotopic study of nine carbonate assemblages in Cold Bokkeveld in order to determine the alteration processes associated with their formation and to reconstruct their formation sequence.These carbonates are found in association with pyrrhotite (Fe [1− x ] S [x = 0-0.17]),pentlandite ([Fe,Ni]  O 4 ), and phyllosilicates.

Scanning Electron Microscopy
A FEI Quanta 650F instrument equipped with two Bruker XFlash 6|30 silicon drift EDX detectors and a Deben Centaurus CL detector with a wide range (185-850 nm) scintillator at the Department of Earth Sciences, University of Cambridge was employed for secondary electron (SE), back-scattered electron (BSE), and CL imaging, and for qualitative chemical analysis using energy-dispersive spectroscopy (EDS).All analyses were done on a carbon coated thin section under high-vacuum conditions.SE and BSE images were collected using an accelerating voltage of 20 keV and a working distance of 13 mm.CL images were obtained using an accelerating voltage of 5 keV, a working distance of 11.4 mm, and a dwell time of 100 μs.EDS element maps were collected using an accelerating voltage of 20 keV, a working distance of 13 mm, and an acquisition time of 10-20 min.

Electron Probe Microanalysis
A CAMECA SX-100 microprobe with five wavelength dispersive spectrometers at the Department of Earth Sciences, University of Cambridge was used for quantitative chemical analysis.
Carbonates were analyzed using a beam diameter of 2 μm.For Na and Ca analysis, an accelerating voltage of 15 keV and a beam current of 4 nA was used, while for Mg, Si, Cr, Mn, Fe, Ni, and Sr, an accelerating voltage of 15 keV and a beam current of 20 nA was used.Standards included jadeite for Na, periclase for Mg, diopside for Si and Ca, Cr-metal for Cr, Mn-metal for Mn, fayalite for Fe, Ni-metal for Ni, and celestine for Sr.Typical detection limits (in wt%) were Na (0.08), Mg (0.01), Si (0.02), Ca (0.23), Cr (0.03), Mn (0.04), Fe (0.04), Ni (0.06), and Sr (0.04).The amount of CO 2 was calculated assuming stoichiometric compositions of carbonates containing all Mg, Ca, Fe, Mn, and Sr detected.

Raman Spectroscopy
Raman spectra were collected across the 100-1800 cm −1 spectral range using a confocal LabRAM 300 (Horiba Jobin Yvon) Raman spectrometer with 300 mm focal length objective, at the Department of Earth Sciences, University of Cambridge.A holographic grating of 1800 g mm −1 coupled to a Peltier front-illuminated CCD detector (1024 × 256 pixels in size) enabled a spectral resolution of ~1 cm −1 .The excitation line at 532.05 nm was produced by a Ventus 532 laser source (Laser Quantum) operated at 100 mW at the source and focused on the sample using an Olympus LMPLFLN 50× long working distance objective.The spot size was ~1 μm and for each spot, five accumulations of spectra, each with 20 s exposure time, were obtained, resulting in a total acquisition time of 100 s.
All Raman spectra were treated by PeakFit software.For each Raman spectrum, the baseline was subtracted, and peak features were determined by least squares fitting to Voigt profiles for each Raman peak/ band.Peak positions were calibrated against the measured excitation of a Ne light emission reference spectrum (Saloman and Sansonetti 2004).

Secondary Ion Mass Spectroscopy
Oxygen isotopic composition was measured by a CAMECA Nano Secondary Ion Mass Spectrometry (NanoSIMS) 50L instrument at the School of Physical Sciences, The Open University.The analyses were conducted by rastering a focused Cs + primary ion beam with a current of ∼50 pA over the areas of interest.Before analysis, a current of ∼150 pA was used for implanting Cs + ions evenly and pre-sputtering the gold coating and possible surface contamination across the areas of interest.Secondary ions were collected simultaneously, with 16   16 O were monitored to avoid contributions from inclusions and surrounding phases.The mass resolving power was set to >10,000 (CAMECA definition, based on the measured peak width between 10% and 90% of the peak), which was sufficient to resolve the interference of 16 OH from the 17 O peak.An electron gun was used for charge compensation.Crack-and contamination-free areas were chosen for NanoSIMS analysis.The measured areas had dimensions of 5 × 5 µm, while pre-sputtering was done on areas of 7 × 7 µm.Analysis times were ~6 min with ~2 min pre-sputtering.The total count of 16 O in a single measurement was ~9 × 10 9 .Oxygen isotope variations are expressed in delta notation, i.e., in ‰ units relative to a bracketing standard.For δ 17 O and δ 18 O (‰ versus standard mean ocean water): Cold Bokkeveld carbonates The departure from the terrestrial fractionation line is quantified using Δ 17 O or 17 O-excess: Instrumental mass fractionation (IMF) effects were calibrated with measurements, before and after measuring the unknown samples, of a dolomite standard provided by the Natural History Museum, London.Sample BM1982,189 (from near Legananam Pot, Dungiven, County Londonderry, UK) has a composition of δ 18 O = 20.0AE 1.0‰ and inferred δ 17 O = 10.4AE 1.6‰.An additional IMF correction was applied to the Cold Bokkeveld samples to account for the discrepancy in the Fe + Mn concentration compared to the abundance in the standard.The detailed calibration for the matrix effect for varying Fe in dolomite-ankerite solution series has been reported by Śliwi ński et al. (2016), and is essentially identical to an earlier study undertaken using the Open University NanoSIMS (Tomkinson 2013).While neither study considered the effects of Mn in isolation, the effects of Mn are considered comparable to that of Fe, and therefore to account for the high abundance of Mn in the samples the correction was applied using the approach of Śliwi ński et al. ( 2016) using abundance of Fe + Mn instead of just Fe.The standard used has an Fe# (= [Fe + Mn]/[Fe + Mn+Mg]) of 0.12 while the samples range from 0.17 to 0.24.The mean value of 0.19 results in an additional matrix correction of 1.5‰ for δ 18 O following the calibration of Śliwi ński et al. (2016).And while the specific composition of the individual spots analyzed by NanoSIMS is unknown, the observed range in Fe# is small, resulting in a variation of matrix correction of only AE0.5‰.Given the uncertainties in the measurements and the range of oxygen isotopic compositions, the additional corrections for Fe + Mn variation in the samples are negligible and only the correction for the average composition has been applied.As matrix effects are entirely mass dependent, the matrix effect on the δ 17 O values was applied in a mass-dependent manner.
The 2σ uncertainties for the NanoSIMS measurements were generated by combining internal errors for each analysis with the standard deviation of the mean of the corresponding standards.

RESULTS
Natural History Museum, London thin section Cold Bokkeveld P5449 from sample BM.1727 was searched for the presence of carbonate minerals.Nine carbonate assemblages showing different petrographic characteristics were systematically studied.In these, three carbonate minerals were identified by the combination of SEM, EPMA, and Raman spectroscopy: aragonite, calcite, and dolomite.

Petrography and Chemistry
Here, we provide a petrographic description of the individual carbonate assemblages.These originate from a number of clasts showing different degrees of aqueous alteration.Results of chemical analysis of carbonates and those of the associated sulfides are summarized in Tables 1 and 2, respectively.
Assemblage 1 is a globular ~25 × 30 µm object with well-defined boundaries embedded in a phyllosilicate and tochilinite-rich matrix (Figs.1a and 1b).In this object, P-bearing (0.28 wt%, n = 1) pentlandite is uniformly distributed in interstitial carbonate.The carbonate object has a speckled appearance in reflected light as opposed to the highly reflective nature of type 1 and type 2 carbonates (see below).The object has a CaCO 3 (likely calcitic) core and Mg-enriched (likely dolomitic) rim.
Assemblage 3 is a globular ~100 × 130 μm object embedded in a phyllosilicate-rich matrix (Figs.1e and  1f).It comprises a pyrrhotite crystal with a pentlandite rim, which is partially replaced by calcite along the pyrrhotite/pentlandite boundary.Some of the pyrrhotite that is in direct contact with pentlandite and calcite is altered to magnetite.The morphology of the pyrrhotite/ pentlandite grain (uniformly oriented crystal remnants at the pyrrhotite/pentlandite boundary) indicates that its partial dissolution had a crystallographic control.
Assemblage 4 comprises two globular calcite crystals with tochilinite rims and an aragonite crystal, embedded in a phyllosilicate-rich matrix (Figs.2a and  2b).While the larger calcite crystal is relatively pure and inclusion-free, the other contains numerous submicrometer sulfide inclusions visible on the Ni element map.Chemical analysis of this calcite crystal shows the presence of Cr, Fe, and Ni (Table 1).The globular shape of calcite crystals (resembling a chondrule) and the presence of micrometer-size sulfide inclusions is consistent with type 2a calcite (Lee et al. 2014).Abundant type 2 carbonate occurring as pseudomorphs replacing chondrule silicate minerals was also reported in the Nogoya CM2 chondrite (Benedix et al. 2003).2c and 2d).The tochilinite in Fig. 2c is in contact with pentlandite.These euhedral calcite crystals represent type 1a calcite formed by cementation of fluid-filled pore space (Lee et al. 2014).After the growth of these into the pores, the remaining porosity between the calcite crystals and walls was filled with tochilinite (Lee et al. 2014).
Assemblage 7 consists of subhedral calcite crystals in close proximity to a pentlandite grain, embedded in a phyllosilicate-rich matrix (Figs.2e and 2f).These subhedral crystals represent type 1b calcite formed by cementation of fluid-filled pore space (Lee et al. 2014).In contrast to 1a calcite, they lack tochilinite and/or phyllosilicate rim (Lee et al. 2014).
Assemblage 8 contains a ~1.1 × 0.6 mm calcite-rich chondrule pseudomorph, in which type 1a calcite and dolomite are often in contact with one another and with tabular pyrrhotite crystals (Fig. 3).The clast also contains abundant phyllosilicates and minor pentlandite.Whereas the central part of the assemblage is richer in calcite grains, there are more dolomite crystals at the edges.Dolomite also forms rims around calcite crystals.The large dolomite crystals show zoning visible in CL images (Figs.3e and 3f).Dolomite has a Ca/Mg molar ratio ranging from to 0.9 to 1.1 and is rich in Mn and Fe (MnO: 3.41-5.18wt%, FeO: 2.36-3.76wt%, n = 6).Assemblage 9 consists of octahedral magnetite crystals with interstitial calcite and phyllosilicate (Fig. 4).Magnetite is dispersed in some highly altered regions of Cold Bokkeveld matrix (Barber 1981).It may form by the alteration of pyrrhotite; magnetite crystals often pseudomorph pyrrhotite in carbonaceous chondrites (Zolensky et al. 2002;Gounelle and Zolensky 2014).Here, no obvious pseudomorphs were observed, but the lack of larger sulfide crystals in the magnetiterich parts of the thin section further supports this interpretation.
Aside from carbonate minerals, other Ca-rich assemblages include calcium-aluminum-rich inclusions (CAIs) (Fig. 5).The numerous CAIs found in the studied thin section are predominantly spinel-pyroxene CAIs, the most abundant group of CAIs in Cold Bokkeveld (Greenwood et al. 1994).All observed CAIs have prominent Ca-rich pyroxene rims.They are spineldominated and often contain perovskite crystals.In one case, a corundum crystal and a perovskite crystal are found with interstitial spinel.Although calcium carbonate is reported to exist in 8% of the CAIs of Cold Bokkeveld (Greenwood et al. 1994), we have not been able to locate any in CAIs.

Raman Spectroscopy
Raman spectroscopy is a fast and robust method to distinguish between carbonate minerals and to detect organic carbon in the structures of carbonates.The latter results from the ability of calcite (and other carbonates) to adsorb organic matter onto their surfaces while they are growing (e.g., Suess 1970).A brief overview of Raman-active modes of carbonate minerals and organic carbon is provided in Appendix S1 in supporting information.
Raman spectra of carbonates in assemblages 1 and 2 could not be collected because of their sensitivity to the Raman laser.Even lowering the laser power to 1% of the power used for carbonates in the other assemblages caused instantaneous burning of the samples.Representative Raman spectra of carbonates from the other assemblages in Cold Bokkeveld as well as from terrestrial reference samples are presented in Fig. 6.The Raman spectrum of the CaCO 3 grain in assemblage 4, next to type 2 calcite grains, shows four broad peaks at 144, 172, 194, and 265 cm −1 (Fig. 6).Peaks at 172 and 194 cm −1 can be assigned to aragonite, while those at 144 and 265 cm −1 to calcite, indicating an aragonite-calcite phase mixture.Aragonite in contact with calcite in Cold Bokkeveld was described previously (Barber 1981).The spectrum of aragonite from assemblage 4 also contains broad, low-intensity G and D bands at 1390 and 1588 cm −1 , respectively, corresponding to organic carbon.Spectra of calcite were collected from six of the assemblages (Table 3, Fig. 6).In order to compare the characteristics of Raman spectra of meteoritic and terrestrial calcite, spectra of nine calcites from several terrestrial geological settings were also collected.These include calcites precipitated from hydrothermal fluids in cavities within basalts from Bulhary and Hajnáčka, Slovakia; prismatic calcite collected from a cavity within an outcrop of serpentinized peridotite in a mélange in the Othris ophiolite, Greece; calcite representing bulk marble from Tuhár, Slovakia; calcite in a cavity of marble from Ružiná, Slovakia; calcites in limestone from Gemerská Ves and Margecany, Slovakia; calcite from a speleothem from the Pes ¸tera S ¸ugȃu, Romania; and a large single crystal calcite from D úhová priepas ť cave, Slovakia.Despite having different origins and formation mechanisms, the frequencies of the six calcite peaks were relatively uniform, ranging 156-157, 281-282, 711-712, 1085, 1434-1435, and 1748-1750 cm −1 .In contrast, the peak positions of calcite from Cold Bokkeveld show significant variation (Figs.7a-c).The range for external modes, involving bonds between Ca 2+ cations and the CO 3 2− groups, exceeds 10 cm −1 .Of the analyzed meteoritic calcites, the peaks of type 2 calcite in assemblage 4 plot the closest to the terrestrial ones.Type 1 calcite grains have peak positions at significantly lower frequencies and calcite in assemblage 3 at even lower frequencies (Figs.7a-c).The positions of Raman peaks corresponding to external modes (measured at 142 and 259 cm −1 ) in calcite in assemblage 3 are lower than those of any (including exotic) naturally occurring terrestrial rhombohedral carbonate measured using the same instrumental  Cold Bokkeveld carbonates conditions (Farsang et al. 2018(Farsang et al. , 2021)).Raman spectra of most of the meteoritic calcites contain broad G and D bands, indicating the presence of organic carbon in their structure (Table 3).The intensity of these peaks relative to the highest intensity ν 1 carbonate peak and relative to each other varies greatly, suggesting large differences in organic carbon content and speciation (Fig. 6).
The meteoritic and terrestrial calcites also differ in the line widths of their Raman modes, with the full width half maximum (FWHM) of peaks broader in the Cold Bokkeveld assemblages than that in terrestrial carbonates by up to ~30 cm −1 .The only exception is type 2 calcite in assemblage 4 that shows comparable peak widths with terrestrial carbonates.We also note that for the two external modes and the highest intensity ν 1 internal mode, there is a strong negative correlation (R 2 = 0.80, 0.97, and 0.87, respectively) between the peak position and FWHM of the peak (Figs.7d-f).
The Raman spectrum of dolomite in assemblage 8 exhibits all the peaks characteristic of dolomite except the lowest intensity 229 cm −1 peak.As for calcite, the peak positions of Cold Bokkeveld dolomite are lower (except the very low-intensity 882 cm −1 peak) and FWHM values higher than that of the terrestrial reference sample.The G and D bands indicative of organic carbon are broad and of low intensity, significantly lower than in calcite in the same assemblage.

Oxygen Isotopes
The presence of large dolomite grains in assemblage 8 with cavity and impurity-free regions allowed the measurement of 17   (Benedix et al. 2003;Lindgren et al. 2017), ranging from −4.9 to −4.0‰.

Crystallinity of Carbonates
The dark speckled appearance of carbonates in assemblage 1 and 2, in contrast to the bright appearance of other carbonates in reflected light, and their behavior in the laser beam suggest that these carbonates are microcrystalline and rich in opaque, probably organic material-rich particles that absorb optically and result in excessive heating from the Raman laser.
In assemblage 4, where a grain consisting of an aragonite/calcite phase mixture and crystals of type 2 calcite are also present, aragonite may be a primary phase and type 2 calcite a secondary replacement as suggested by Lee et al. (2014).If this is the case, the replacement does not involve the full dissolution of aragonite followed by the precipitation of calcite.Instead, the bimineralic grain may document a gradual aragonite-calcite phase transition.This phase transition has been observed experimentally at temperatures ranging from 385 to 468 °C (Antao and Hassan 2010; Cold Bokkeveld carbonates Parker et al. 2010;Farsang et al. 2018) and the Raman spectra collected upon the transition exhibit similar features to those seen here (Farsang et al. 2021).The transition is typically inhibited kinetically by the large associated volume change (from ~34 to ~37 cm 3 mol −1 as aragonite transforms to calcite), but the presence of aqueous fluids can significantly accelerate the kinetics and lower the effective activation energy for the phase transition, such that it may then occur at much lower temperatures (Bischoff and Fyfe 1968).In the presence of water, at 108 °C, the transition requires hours to days to take place, while at ambient temperatures it would require months to years (Bischoff and Fyfe 1968).Partial substitution of smaller cations (e.g., Mg 2+ ) for Ca 2+ is often responsible for an increase in frequency of Raman or infrared modes in calcite.For instance, the ν 1 peak positions of calcites in some CM2 chondrites have been found to be 2-3 cm −1 higher than in pure terrestrial calcites, a feature attributed to significant levels of cations other than Ca 2+ (Chan et al. 2017).Generally, the smaller the cation, the shorter the interatomic distances, resulting in shifts to higher frequency (Krishnamurti 1956).In this study, though, we see the opposite effect, i.e., some chondritic calcite peaks appear shifted to significantly lower frequency positions (by ~1 to 20 cm −1 in calcite in assemblage 3) than in the pure, highly crystalline abiotic terrestrial calcites.It is not certain what causes the red shift of these peaks.Substitution of large cations (e.g., K + ) for Ca 2+ could lead to the observed decrease in frequency, but chemical analysis showed no enrichment by an element that would have an ion bigger than Ca 2+ .A large shift of peaks corresponding to external modes (>10 cm −1 ) requires significant and measurable substitution.Previous work has shown that the peak positions in calcites with abundant organic material are very similar to those of organic-carbon-free calcites and differ by no more than ~1 cm −1 (Chan et al. 2017), so the presence of organic carbon cannot be invoked as a reasonable explanation either.Therefore, we speculate that the extreme shifts in peak positions of calcite in assemblage 3 may be explained by changes in local environment during the Raman spectroscopic analysis rather than compositional changes.We note that calcite in assemblage 3 is almost entirely surrounded by pyrrhotite, pentlandite, and magnetite, i.e., phases that have much higher thermal conductivity than phyllosilicates, in which the other calcite grains occur.Consequently, the Raman laser may heat up the entire assemblage, resulting in lower measured frequency values (Farsang et al. 2018).
Peak widths of chondritic calcites are generally much larger than those of terrestrial carbonates.The only exception is type 2 calcite in assemblage 4, implying high crystallinity for type 2 calcite.Broadened peaks generally indicate either inhomogeneous and extensive substitution by a cation of a different size or an increase in local strain, possibly related to lower crystallinity, smaller effective correlated structure, and the abundance of structural defects.The lower peak positions and higher FWHM values in Cold Bokkeveld dolomite than that in the terrestrial reference sample also indicate the presence of impurities and/or lower crystallinity, consistent with significant Mn-and Fe-enrichment.

Carbonate Formation Conditions
Oxygen isotopes in carbonate minerals can be used to infer the properties of the aqueous fluid from which they originated, the temperatures at which they formed, and the degree of aqueous alteration on the CM parent body (Johnson and Prinz 1993;Riciputi et al. 1994;Benedix et al. 2003;Guo and Eiler 2007;Lee et al. 2013Lee et al. , 2014;;Lindgren et al. 2017;Verdier-Paoletti et al. 2017;Telus et al. 2019).In the presence of aqueous Cold Bokkeveld carbonates fluid, significant oxygen isotope diffusional exchange between the fluid and calcite (and other carbonates) occurs at moderate temperatures of 400-500 °C (Farver 1994).Since these temperatures were probably not reached in the CM parent body(ies) (see below), the oxygen isotopic composition of the carbonates is related to that of the original fluid and has not been subject to subsequent re-equilibration and can therefore provide valuable information on the evolution and chemistry of the fluid from which the carbonates precipitated.
Variations in O isotopic composition between carbonate grains can imply different formation temperatures, sources of fluid, or evolving fluid composition for CM chondrites (Chan et al. 2017).
Our measurements reveal that dolomite has δ 18 O values intermediate between those of type 1 and type 2 calcite, supporting the idea of dolomite precipitation occurring between the main stages of type 1 and type 2 calcite formation (Lindgren et al. 2017).However, it should be pointed out that the O isotopic values of type 1 and type 2 calcite and dolomite may overlap with one another (Fig. 8) and it is possible that any of these phases form from fluids of very similar composition.Indeed, similar Δ 17 O values measured in calcite and dolomite were previously invoked to imply that these two phases precipitated from the same fluid on CM parent bodies (Benedix et al. 2003).However, our dolomite Δ 17 O values are typically lower than those measured in calcite with comparable δ 18 O (Fig. 8a) (Benedix et al. 2003;Lindgren et al. 2017), suggesting precipitation from two fluids.Furthermore, the contact of calcite and dolomite in single carbonate grains in assemblage 8 (Fig. 3), with Fe/Ca and Mn/Ca ratios systematically higher in dolomite than coexisting calcite, point toward disequilibrium between these phases and distinct precipitation events (Riciputi et al. 1994).Based on the model prediction for the variation in carbonate isotopic composition as a function of formation temperature of Alexander et al. (2015), our measured δ 18 O values suggest dolomite precipitation temperatures of ~70-100 °C.Given the lack of δ 13 C data, these temperatures should be treated as first approximation.The activation of fluids and subsequent dolomite precipitation in CM chondrites was suggested to form in response to a transient heating event and/or a brief increase in fluid Mg/ Ca ratios (Lindgren et al. 2017).Weight% levels of Mn in carbonates observed in this study and by Riciputi et al. (1994) suggest that dolomite concentrates this element.

Paragenetic Sequence of the CM Chondrite Parent Body(ies)
The CM chondrite parent body(ies) underwent extensive aqueous alteration as shown by the presence of secondary minerals (e.g., McSween 1979;Brearley [2006] and references therein).The diversity and distribution of lithic clasts and mineral fragments showing various degrees of aqueous alteration (Bischoff et al. 2017;Lentfort et al. 2020) indicate that some alteration may have already taken place on precursor planetesimals, shortly before and/or while they accreted to their parent body(ies).Alternatively, lithic clasts and mineral fragments showing various degrees of aqueous alteration could be assembled from pieces of the same accreted parent body, although it is difficult to envisage a parent body brecciation process responsible for such a fine-scale distribution of these clasts and fragments.It is therefore possible that different lithic clasts and mineral fragments record different aqueous alteration events (e.g., triggered by impacts), which either did not take place on the parent body or operated only locally on the parent body.In our reconstruction of the paragenetic sequence, we assume that much of the aqueous alteration happened on the finally accreted parent body, while recognizing that alternative scenarios may be possible.
The first carbonate minerals crystallized early on the CM chondrite parent body; e.g., 53 Mn-53 Cr radiometric dating of carbonates from four different CM chondrites yielded a formation age range of 4563.4+0.4/-0.5 Ma (Fujiya et al. 2012).The different stages of carbonate and serpentine formation almost certainly overlapped.Evidence was found in this study for calcite predating serpentine (and tochilinite) formation (see below), as also seen in other studies (Lee et al. 2013;Marrocchi et al. 2014;Fujiya et al. 2015;Verdier-Paoletti et al. 2017) and postdating the surrounding phyllosilicates (Barber 1981).These observations can be explained by the work of Lee et al. (2014), who suggest that carbonates in the CM2.5-CM2.2chondrites experienced four distinct stages of mineralization including (1) the formation of aragonite, calcite, and dolomite by cementation of pore space; (2) the formation of tochilinite and Fe-rich serpentine replacing the edges of pore-filling carbonates; (3) the formation of Mg-rich serpentine replacing pore-filling carbonates; and finally, (4) calcite replacing dolomite and remaining anhydrous silicates (Lee et al. 2014).
Observations of aragonite apparently being replaced by calcite (Fig. 2a) and tochilinite attacking calcite grains (Figs. 1c,1d,2c,and 2d) support this model.However, based on our petrographic observations, we propose two additional mineral forming events.In assemblage 9, octahedral magnetite crystals likely replace sulfide minerals (Fig. 4).The lack of S-bearing alteration products (e.g., tochilinite) in the vicinity of thousands of magnetite crystals indicates that, at least locally, the formation of interstitial calcite postdates the sulfur mobilization (and S-bearing secondary mineral formation) event.This is in contrast with type 1a or 1b calcite precipitating prior to tochilinite formation.We therefore propose a new type of calcite, type 1c.Furthermore, the fine-grained tochilinite rim indicates that at least one additional sulfur mobilization/tochilinite formation event occurred even after type 2 calcite formation (Fig. 2a).A hightemperature event, such as an impact, could have provided the heat necessary not only for late-stage calcite precipitation, resulting in type 2 calcite (Tyra et al. 2012), but also to initiate the spectroscopically documented aragonite-calcite phase transition also enhanced by the presence of aqueous fluids (Fig. 6).Isotopic measurements indicate that carbonates on the Cold Bokkeveld parent body formed roughly at the same time or equilibrated at the same temperature as serpentines (Verdier-Paoletti et al. 2017).Based on our observations, we propose at least six stages of potentially overlapping carbonate and phyllosilicate formation events (Fig. 9).
Carbonates in assemblages 1, 2, and 3 do not show the characteristics typical of type 1 or type 2 carbonates.Those in assemblages 1 and 2 are interstitial to pentlandite and share unusual optical features (Figs.1a and 1c).The complex nature of carbonate in assemblage 1 (a CaCO 3 core with an Mg-enriched rim) and tochilinite replacing calcite in assemblage 2 indicate that this kind of carbonates formed at the very early stages of aqueous activity on the CM parent body.There are two possible formation scenarios for the carbonate grain in assemblage 1 (Figs.1a and 1b): either it was originally calcic and exchanged cations with Mg-rich fluid or it precipitated from a fluid that changed from being Ca to Mg rich.Complex carbonates with Ca-carbonate-dominated cores have been observed before in the QUE 93005 CM2 chondrite and, regardless of which of the two formation scenarios outlined operated, indicate that the Ca-carbonates formed earlier (Rubin et al. 2007).In assemblage 3, part of the pyrrhotite grain with pentlandite rim clearly dissolved (Figs.1e and 1f).The lack of S-bearing alteration products (such as tochilinite) in its vicinity indicates that the dissolution of sulfide and the Cold Bokkeveld carbonates precipitation of the spectroscopically distinct calcite also took place very early on the CM parent body.
The presence of organic carbon in most of the analyzed carbonate grains (Fig. 6) indicates that organic carbon was mobilized by aqueous fluids for extended periods.For comparison, disordered organics were present in only ~25% of analyzed CM chondritic carbonate grains (Chan et al. 2017).The large variability in G and D band positions (Table 3) suggests that organic carbon modified by distinct heating processes has been incorporated into the carbonate minerals (Quirico et al. 2018).
The diverse compositions and petrographic features of carbonates and associated phases indicate precipitation from locally formed, physically restricted aqueous fluids that only penetrated through small regions.This interpretation is consistent with highly variable O isotopic compositions in carbonates (e.g., Telus et al. 2019).Different fluid compositions, in turn, may arise from the heterogeneity of accreted ices and mineral assemblages in Cold Bokkeveld clasts that range from CM2.7 to CM2.1 (Bischoff et al. 2017;Lentfort et al. 2020).The intrasample mineralogical heterogeneity in different CM chondrites, including Cold Bokkeveld, reflects either sampling from different source regions within the CM parent body(ies) or from different precursor planetesimals, possibly following their earlystage fracturing in the protoplanetary disk.Impacts and collisions with water-or ice-rich bodies (or water-or icerich dust clouds) could have led to the introduction of water and/or heating events on the CM parent body(ies), followed by alteration and mineral forming events that only a subset of CM chondrites exhibit.

CONCLUSIONS
Our study reveals the existence of carbonates that display petrographic features quite distinct from those of type 1 and type 2 carbonates.These include carbonates interstitial to octahedral magnetite crystals, for which a new designation of "type 1c" is suggested.Abundant dolomite with high Mn contents has been found.The O isotopic values of dolomite (δ 18 O ranging from 21.1 to 25.8‰ and Δ 17 O from −4.9 to −4.0‰) are similar to those measured in dolomites from other CM chondrites.The presence of complex carbonates with a CaCO 3 core and Mg-enriched rim implies several generations of fluids and/or their evolving composition on the CM parent body(ies).Petrographic characteristics indicate at least six stages of potentially overlapping carbonate and phyllosilicate formation events, probably enhanced by heat generated due to collisional events.We show that type 1 and type 2 calcite have distinct spectral characteristics (peak positions and widths) and can be distinguished using Raman spectroscopy.Type 1 calcite displays very broad Raman peaks, whereas type 2 calcite yields Raman spectra with very narrow peak widths, similar to those of terrestrial calcite, suggesting very high crystallinity.We report Raman spectral evidence for an aragonitecalcite phase transition.Raman spectroscopy also revealed the presence of organic matter in the majority of carbonates.This indicates that organic carbon was mobilized by aqueous fluids for extended periods.

Fig. 6 .
Fig. 6.Raman spectra of carbonates in Cold Bokkeveld CM chondrite.Peak positions corresponding to carbonate modes are marked.The wide bands around 1355 and 1575 cm −1 represent the D and G bands of carbon, respectively, and indicate the presence of organic material in the carbonates.Raman spectra of terrestrial aragonite, dolomite (Farsang et al. 2018), and calcite were taken at 25 °C.

Fig. 7
Fig. 7. a-c) Raman shifts in calcite in six different assemblages in Cold Bokkeveld CM chondrite and nine terrestrial samples.d-f) FWHM as a function of peak position of the external (translational, T and librational, L) modes and highest intensity ν 1 internal mode.

Fig. 8 .
Fig. 8. Three isotope plots of (a) carbonate grains in Cold Bokkeveld and (b) dolomite grains in CM and C chondrites.Errors are 2σ.The solid line in each plot is the terrestrial fractionation line, defined as δ 17 O = 0.52 × δ 18 O.

Table 2 .
Chemical composition of sulfides associated with carbonates in Cold Bokkeveld CM chondrite.
All values are in wt%.b.d.l.= below detection limit

Table 1 .
Chemical composition of carbonates in Cold Bokkeveld CM chondrite.
All values are in wt%.b.d.l.= below detection limit Cold Bokkeveld carbonates Assemblages 5 and 6 comprise euhedral calcite crystals surrounded by tochilinite, embedded in a phyllosilicate-rich matrix (Figs. O and 18 O isotopes in this phase.The oxygen isotopic compositions of dolomite grains we have measured, as well as previous measurements conducted on dolomite grains in CM and C chondrites and other carbonate phases in Cold Bokkeveld, are reported in Table 4.While δ 18 O values for dolomite reported here are very similar to those previously reported for Cold Bokkeveld (Benedix et al. 2003) and ALH 84049 (Tyra et al. 2016) CM chondrites, and the Sutter's Mill C chondrite (Jenniskens et al. 2012), the Δ 17 O values are slightly lower.Compared to calcite, our dolomite δ 18 O values are between those typical of type 1 and type 2 calcite that have average δ 18 O of ~32-36 and 21-24‰, respectively (Lindgren et al. 2017), whereas Δ 17 O values of dolomite are typically lower

Table 3 .
Peak positions and corresponding full width half maxima of the calcite and organic carbon bands.

Table 4 .
Oxygen isotopic compositions of dolomite grains in CM and C chondrites and other carbonate phases in Cold Bokkeveld.
All values are in ‰.All relative to standard mean ocean water (SMOW).* The cited ranges refer to dolomite grains not containing calcite.