We have sampled sulfide grains from one pristine CM2 chondrite (Yamato [Y-] 791198), one thermally metamorphosed CM2 chondrite (Y-793321), and two anomalous, metamorphosed CM/CI-like chondrites (Y-86720 and Belgica [B-] 7904) by the focused ion beam (FIB) technique and studied them by analytical transmission electron microscopy (TEM). Our study aims at exploring the potential of sulfide assemblages and microstructures to decipher processes and conditions of chondrite petrogenesis. Complex exsolution textures of pyrrhotite (crystallographic NC-type with N ≈ 6), troilite, and pentlandite occur in grains of Y-791198 and Y-793321. Additionally, polycrystalline 4C-pyrrhotite-pentlandite-magnetite aggregates occur in Y-791198, pointing to diverse conditions of gas–solid interactions in the solar nebula. Coarser exsolution textures of Y-793321 grains indicate higher long-term average temperatures in the <100 °C range compared to Y-791198 and other CM chondrites. Sulfide mineralogy of Y-86720 and B-7904 is dominated by aggregates of pure troilite and metal, indicating metamorphic equilibration at sulfur fugacities (fS2) of the iron-troilite buffer. Absence of magnetite in equilibrium with sulfide and metal in Y-86720 indicates higher peak temperatures compared with B-7904, in which coexistence of troilite, metal, and magnetite constrains metamorphic temperature to less than 570 °C. NC-pyrrhotite occurs in both meteorites as nm-wide rims on troilite grains and, together with frequent anhydrite, indicates a retrograde metamorphic stage at higher fS2 slightly above the fayalite-magnetite-quartz-pyrrhotite buffer. Fine-grained troilite-olivine intergrowths in both meteorites suggest the pre-metamorphic presence of tochilinite-serpentine interlayer phases, pointing to mineralogical CM affinity. Pseudomorphs after euhedral pyrrhotite crystals in Y-86720 in turn suggest CI affinity as do previously published O isotopic data of both meteorites.
CM chondrites are one of the cosmochemically most primitive meteorite groups, but at the same time, they are probably one of the mineralogically most complex rock types. Most characteristically, they are rich in fine-grained matrix material and experienced extensive aqueous alteration (McSween 1979; Zolensky et al. 1997), documented by the presence of phyllosilicates (typically Fe-rich serpentines; e.g., Bunch and Chang 1980), tochilinite (a complex Fe-rich hydroxide/sulfide interlayer mineral; Zolensky 1987), and Ca and Ca-Mg carbonates (e.g., Johnson and Prinz 1993). The low-temperature mineral assemblage (typically formed at temperatures significantly <100 °C; Guo and Eiler 2007; Zolensky et al. 1993) indicates an episode of exposure of primary, anhydrous minerals to a water-rich environment. Although most alteration probably occurred on the CM parent body by hydrothermal fluids (e.g., Kerridge and Bunch 1979; Brearley 2003), there is debated evidence that some hydration could have occurred in the solar nebula prior to the accretion of the parent body or its predecessors (Metzler et al. 1992; Bischoff 1998; Brearley 2003; Ciesla et al. 2003).
The origin of sulfide minerals in hydrothermally altered chondrites represents a similar issue: troilite (stoichiometric FeS), pyrrhotite (nonstoichiometric Fe1-xS), and pentlandite ([Fe,Ni]9S8) can potentially form by gas–solid interactions and (specifically the latter two) by hydrothermal parent-body alteration. A distinction is not always straightforward. It has been shown in theory and experiment that troilite is the first Fe-rich sulfide to form from a cooling solar gas (Lauretta et al. 1996, 1997). Pentlandite, pyrrhotite, and an apparently intermediate phase in CM2 chondrites have been discussed to be products of parent-body alteration of pre-existing sulfides, tochilinite, or metal (e.g., Bunch and Chang 1980; Tomeoka and Buseck 1985; Hanowski and Brearley 2001; Zolensky and Le 2003; Chokai et al. 2004; Bullock et al. 2007). Zolensky and Thomas (1995) showed that pyrrhotite is abundant in interplanetary dust particles (IDPs) and suggested its formation via extended sulfidation of troilite in the solar nebula. Some pentlandite in IDPs, however, might have formed through aqueous alteration. Zolensky et al. (1996) described pyrrhotite and pentlandite as products of extreme hydrothermal alteration in CM1 clasts of the Kaidun meteorite. Boctor et al. (2002), Brearley and Martinez (2010), Brearley (2010), and Maldonado and Brearley (2011) proposed that pyrrhotite- and pentlandite-bearing grains in typical CM2 chondrites formed in the solar nebula, either through sulfidation of metal or during chondrule melting.
Some CM and CM/CI-like chondrites experienced thermal metamorphism that partially led to the dehydration of phyllosilicates. Particularly among the meteorites collected in Antarctica by Japanese NIPR field parties (Yamato and Belgica recovery sites), a large fraction of CM- and CI-like chondrites displays thermo-metamorphic overprints, which have sparked extensive studies of these rocks (e.g., Akai 1988; Ikeda 1992 and references therein). The number of CM and CM/CI-like chondrites known to have been thermally affected is steadily growing (e.g., Nakato et al. 2011), indicating that metamorphism and dehydration might be a relatively common process in the evolution of hydrous asteroids. Due to the complex phase relationships and thermochemistries of Fe and Fe,Ni sulfides, their petrogenetic record within these meteorites is interesting and in the process of being tapped (e.g., Kimura et al. 2011). However, the study of sulfides remains challenging for at least two reasons: First, the small sizes and small-scale intergrowths of different sulfide phases and, second, the relatively unconstrained nature and history of sulfides in the premetamorphic chondrites.
In this contribution, we focus on the sulfide mineralogies of Antarctic CM and CM/CI-like chondrites that experienced very different degrees of thermal metamorphism, from none to severe. We overcome the challenge of small size by using focused ion beam (FIB) preparation and analytical transmission electron microscopy (TEM). The aim is to decipher the phase assemblages and microstructures of Fe sulfides, in particular those of pyrrhotites, to obtain constraints on key parameters of primary formation, secondary alteration, and metamorphism, including sulfur and oxygen fugacities, metamorphic temperatures, and cooling rates.
Previous Work and Constraints on Metamorphic Histories
Prominent members of the anomalous group of metamorphosed CM/CI-like meteorites are Y-82162, Y-86720, and B-7904 (often referred to as the Belgica grouplet). Besides mineralogical evidence for anomalous thermal histories, these rocks also display petrographic, mineralogical, geochemical, and isotopic characteristics that place them in an uncertain relationship with both CM and CI chondrite groups (Bischoff and Metzler 1991; Ikeda 1992). Oxygen isotopic compositions discussed by Clayton and Mayeda (1999) are very similar, but distinct from typical CM and CI chondrites, although a genetic link to the CI group may be possible through mass-dependent oxygen isotope fractionation during dehydration. Y-793321 is an atypical CM2 chondrite that has experienced metamorphic heating, leading to the breakdown of tochilinite and partial dehydration of phyllosilicates (Nakamura 2006). Contrary to the Belgica grouplet, Y-793321 shows O isotopic ratios much closer to the CM trend and could have originated from a rather weakly hydrated CM precursor (Clayton and Mayeda 1999). Trace element geochemistry (Paul and Lipschutz 1990; Yamamoto and Nakamura 1990) places Y-86720 in proximity to the CM group or between CM and CI, and B-7904 also in proximity to the CM group. The trace element composition of Y-793321 is reported to be consistent with typical CM2 chondrites (Tonui et al. 2002). The Belgica grouplet chondrites have experienced metamorphic heating in the range of about 500–800 °C. Y-793321 was heated to a much lesser degree. A compilation of estimated metamorphic temperatures is given in Table 1. A large uncertainty factor in these estimates arises from the unknown duration of heating, because many criteria rely on strongly kinetically controlled processes, such as phyllosilicate dehydration and migration of mobile trace elements. Nakato et al. (2008), for example, estimated the heating time of B-7904 in the order of 10–1000 days at 700 °C and 1–100 h at 890 °C.
Table 1. Compilation of reported metamorphic temperatures for Y-793321 and the Belgica grouplet chondrites
Heating stage (numerals I to IV) of Nakamura (2005), metamorphic category (letters A to C) of Kimura et al. (2011).
1: Tomeoka et al. (1989), 2: Akai (1990), 3: Paul and Lipschutz (1990), 4: Zolensky et al. (1991), 5: Akai (1992), 6: Zolensky et al. (1993), 7: Lipschutz et al. (1999), 8: Tonui et al. (2002), 9: Nakamura (2005), 10: Nakamura (2006), 11: Nakato et al. (2008).
We have included Y-791198 in our study, because this meteorite is regarded as one of the most primitive CM chondrites known. Textural and chemical evidence indicates that this meteorite is a primary accretionary rock that experienced only a limited degree of parent-body aqueous alteration (Metzler et al. 1992; Rubin et al. 2007; Chizmadia and Brearley 2008), no thermal metamorphism (e.g., Naraoka et al. 2004), and no impact-induced brecciation (Metzler et al. 1992).
Pyrrhotites in the Fe(+Ni)-S System
Despite its simple composition, pyrrhotite is a remarkably complex mineral, demanding here a short overview of its mineralogy. Pyrrhotite, being Fe1-xS with x typically between 0 and 0.125, is a nonstoichiometric iron sulfide with Fe-site vacancies. Relevant pyrrhotite compositions occupy the Fe-S system between 46 and 50 atom% Fe and form a (omission) solid solution series between Fe0.875S and stoichiometric FeS (troilite). At temperatures above approximately 310 °C, the Fe vacancies are randomly disordered in the hexagonal, NiAs-based lattice (1C-pyrrhotite; Wang and Salveson 2005). It has been shown by several studies that the Fe/S ratio and the corresponding x value of solid, disordered 1C-pyrrhotite depends on temperature and sulfur fugacity (fS2; e.g., Rau 1976; Toulmin and Barton 1964) as shown in Fig. 1a. Hence, at any given temperature above approximately 310 °C, a specific pyrrhotite composition is in equilibrium with gaseous S2 of determined fugacity. The pyrrhotite field is bounded by the reactions:
Equation (1) describes the iron-troilite buffer (IT), which fixes fS2 as function of temperature. Equation (2) is the pyrite-pyrrhotite buffer. The addition of Ni into the hexagonal monosulfide solid solution (MSS, Fe1-xS–Ni1-xS) at 600 °C increases the metal/sulfur (M/S) ratio at given fS2 and decreases the metal deficiency x (Kosyakov et al. 2003; Raghavan 2004). For example, the equilibrium log fS2 for approximately x = 0.08 increases from −5 for Fe0.92S to −4 for (Fe0.55Ni0.37)S. In the presence of suitable redox couples, fS2 is constrained by oxygen fugacity fO2 and vice versa, e.g., through the reaction with magnetite (Whitney 1984):
These relationships allow the formulation of fS2 − fO2 buffers as for example the fayalite-magnetite-quartz-pyrrhotite (FMQ-Po) buffer of Eggler and Lorand (1993) shown in Fig. 1b.
The formation of pyrrhotite in the solar nebula was attributed by Kerridge (1976) and Zolensky and Thomas (1995) to extended sulfidation of previously formed troilite by H2S. In a simple system, this can be expressed by the reactions:
The reactions are more complex in nature, because the initial metal contains Ni and other components, leading to potentially complex reaction paths (Lauretta et al. 1997). As outlined above and shown in Fig. 1a, the compositional variable x of nonstoichiometric pyrrhotite depends on T and fS2, the latter being determined by T, P, and the H2S/H2 ratio. Notably, the pressure dependence of the Fe1-xS + S2 equilibrium is rather small (Toulmin and Barton 1964). This suggests that pyrrhotite formation by exchange with a gas phase can, in principle, be attributed to fS2 at values higher than the IT buffer. Hence, pyrrhotite will not form at conditions when troilite, Fe,Ni metal, and ambient gas are in equilibrium. At temperatures below approximately 310 °C, the Fe-S system within the pyrrhotite compositional range contains several one- and two-phase fields, which arise from (partial) ordering of Fe vacancies and formation of complex superstructures (Fig. 2). The most common and best-established pyrrhotite structure at ambient conditions is monoclinic 4C-pyrrhotite (Fe0.875S or Fe7S8), in which the vacancy ordering quadruples the NiAs-based c lattice dimension (Bertaut 1953; Powell et al. 2004). In terrestrial samples, 4C-pyrrhotite is typically associated with NC-pyrrhotites, in which N takes nonintegral values between 4 and 5, due to aperiodic structural modulations, and is positively correlated with the Fe/S ratio (Morimoto et al. 1975b; Harries et al. 2011). At Fe/S close to unity, 1C-pyrrhotite can persist down to temperatures of approximately 100 °C before vacancies order into superstructures. In this compositional range, troilite is a commonly observed exsolution phase coexisting with NC-pyrrhotites. Experimental studies by Novikov et al. (1977) showed that the structural evolution of coexisting pyrrhotite is rather complicated and proceeds in at least two stages. NC-pyrrhotite coexisting with troilite at room temperature shows either nonintegral N values of around 5.5 (~Fe0.91S or Fe10S11; Harries et al. 2011; Morimoto et al. 1975b) or an integral value of 6 (Fe0.92S or Fe11S12; Nakazawa and Morimoto 1970; Morimoto et al. 1975a; Becker et al. 2010a;). Experimental results by Nakazawa and Morimoto (1970) shown in Fig. 2 suggest that during cooling of NC-pyrrhotite, the N value increases continuously, and some of the nonintegral NC-pyrrhotites might be metastable due to sluggish ordering kinetics. Hence, some of the observed assemblages might not represent equilibrium states at room temperature, but rather frozen-in stages of their cooling history. The influence of Ni contents on the phase relationships of pyrrhotite below approximately 310 °C is not well known, but generally, Ni concentrations are low (typically less than 0.3 atom%; Vaughan and Craig 1978; Etschmann et al. 2004) due to exsolution of pentlandite, (Fe,Ni)9S8. The latter is the typical carrier of Ni in exsolved terrestrial MSS. The onset of pentlandite exsolution depends on the M/S ratio and Ni content of the MSS and commences between 610 and 300 °C (Naldrett et al. 1967; Kelly and Vaughan 1983; Etschmann et al. 2004).
Sample Preparation and Analytical Methods
We have selected fresh interior fragments of Yamato-791198 (,97), Yamato-793321 (,101), Belgica-7904 (,114), and Yamato-86720 (,86) to avoid weathering effects and thermal overprint from atmospheric passage. Fragments were epoxy mounted and carefully polished. Care was taken not to heat the samples above room temperature to avoid any potential deterioration of the original sulfide microstructures.
For microprobe analysis, a JEOL JXA-8200 equipped with five wavelength dispersive spectrometers was used and operated at 20 kV/20 nA and focused probe size (counting 30 s on peak, 15 s on background). Reference material for Fe and S calibration was Ni-free 4C-pyrrhotite. Ni and Co were calibrated using pure metals. SEM imaging and FIB sample preparation for TEM study were accomplished using a FEI Quanta3D field-emission FIB-SEM. For excavation of trenches around the targeted TEM foil sites, beam currents from 50 nA down to 3 nA at 30 keV ion energy were used. Cleaning before lift-out was applied using a 30 keV/1 nA beam. The remaining 0.5 to 1 μm thick lamellae were then cut free, extracted with an in situ micromanipulator, and attached to a posttype copper TEM grid. Final thinning was applied using beam currents of typically 300 pA, 100 pA, and 50 (or 30) pA at 30 keV energy. To minimize Ga implantation and ion beam-induced structural damage, the last polishing step usually involved a beam energy of 5 keV and a current of 80 pA or lower. Under these conditions, the stopping depth of Ga+ ions lies in the order of a few nm at grazing beam incidence (e.g., Rubanov and Munroe 2004). Heating effects are negligible in case of thermally well conducting sulfides as discussed by Wozniakiewicz et al. (2011). For TEM observations, we used a Philips CM20 FEG operated at 200 kV and equipped with a Thermo Noran HPGe EDX detector. Images and SAED patterns were recorded on conventional electron imaging film. EDX measurements and X-ray maps were acquired in scanning TEM (STEM) mode. EDX spectra were evaluated semiquantitatively using factory-calibrated k-factors.
SEM and EPMA Observations
Y-791198 shows an unbrecciated texture with abundant coarse-grained ferromagnesian objects mantled by fine-grained rims (FGRs; Figs. 3a and 3e). Calcite is abundant and based on a 2 × 2 mm EPMA X-ray map comprises about 1.5% (area) of the rock. Accessory oxide phases are predominantly chromite. The fine-grained material of the rims and interstitial material contains scarce, rounded Fe,Ni metal grains of typically 1–3 μm size, attesting to a low degree of aqueous alteration. Tochilinite-serpentine intergrowths (“PCP”) are frequent in the interstices between densely packed mantled objects.
Coarse-grained sulfides (>8 μm, excluding tochilinite) comprise about 0.5% of the meteorite by area and are in places also mantled by FGRs. The vast majority of sulfide grains are Fe,Ni sulfides with a small contribution of P- and Cr-bearing sulfides occurring together with minor schreibersite or barringerite (Nazarov et al. 2009). BSE images (Figs. 4a–c) and EPMA (Table 2) show that most of the Fe,Ni sulfides are intergrowths of a low-Ni Fe sulfide with fine-grained, exsolved pentlandite. The fine-grained nature of the intergrowth did not allow the separate analysis of these phases due to beam overlap; Table 2 lists approximate bulk values. The average M/S ratio of the mixtures is 0.991 ± 0.019 and, hence, practically at unity. Many grains have distinct pentlandite rims (Fig. 4c), and some grains contain larger domains or rims of coarse pentlandite, which could be analyzed separately by EPMA. This pentlandite shows variable molar Fe/(Ni + Co) ratios between 1.21 and 1.81 (average 1.59 ± 0.22; Table 2) and may have a slight metal deficiency (average M/S = 1.108 ± 0.021) compared with the ideal (Fe,Ni)9S8 composition (M/S = 1.125).
Table 2. EPMA analyses of sulfides and metals
n (n grains)
Sul = low-Ni Fe sulfide (pyrrhotite, troilite, or both), Pn = pentlandite, Sul(+Pn) = mixed measurements, M/S = metal/sulfur ratio, n = No. of measurements, n grains = No. of grains measured. Standard deviations of least significant digits in parentheses.
n (n grains)
Fe/(Ni + Co)
On the basis of our observations, we distinguish two types of sulfide grains in Y-791198. Grains of the first type are the compact sulfide grains described above. In the following, we term these “M-type,” because they mostly consist of only one, large monocrystalline domain. Within these single crystal domains, multiple sulfide phases may occur as inclusions. These phases are coherently or semicoherently intergrown and show well-defined crystallographic orientation relationships with the surrounding sulfide (see below). The second type consists of granular aggregates of μm to sub-μm-sized crystallites of pyrrhotite/troilite and pentlandite (Fig. 4d). In the following, we call these “P-type,” because they are polycrystalline and often porous. The constituting crystallites (typically <2 μm in size) are randomly oriented and do not show crystallographic orientation relationships. P-type sulfide aggregates often show concentric structures and contain abundant interstitial material. Because of this texture, reliable chemical analysis by EPMA was generally not possible.
Y-793321 is texturally very similar to Y-791198, but FGRs are much less developed and the meteorite is brecciated (Figs. 3b and 3f). Compared with Y-791198, serpentine and tochilinite form larger and more pure aggregates, and serpentine is distinctly zoned in Fe-Mg content. CAIs, albeit highly altered to phyllosilicates and Ca-phosphate, are distinctly more frequent than in Y-791198. Calcite occurs at approximately 0.8% by area and is less abundant than in Y-791198 (Figs. 3i and 3j). Magnetite is present and occurs as single grains and rare framboids.
The sulfide content is approximately 0.5% by area and similar to that of Y-791198, but sulfide grains are smaller on average. As in Y-791198, a small fraction of sulfides is comprised of P,Cr-bearing Fe,Ni sulfides (plus schreibersite/barringerite), but the majority of grains are common Fe,Ni sulfides of the M-type. Well-developed P-type aggregates appear to be absent. Also in Y-793321, small-scale pentlandite exsolution in the low-Ni, M-type sulfide is common and larger, segregated pentlandite areas exist as well. The approximate bulk composition of the fine-scale intergrowths of low-Ni Fe sulfide and pentlandite (Table 2) is mostly similar to Y-791198 and has an average M/S of 0.979 ± 0.021, being close to unity. Backscattered electron images indicate that the textures of intergrown Fe sulfide and pentlandite are similar to those observed in Y-791198 (Figs. 4e and 4f), but distinctly coarser. One of the grains subsequently studied by FIB-TEM showed clear signs of replacement by a Fe-rich fibrous mineral (Fig. 4e; see below). Unfortunately, only one pentlandite grain was found large and clean enough for analysis. This grain's composition shows a Fe/(Ni + Co) ratio of 1.35 ± 0.01 and agrees well with the compositions observed in Y-791198. Its M/S ratio corresponds to ideal pentlandite stoichiometry (Table 2).
In BSE images, B-7904 displays a visual appearance similar to typical CM chondrites, but shows a more uniformly developed fine-grained texture between coarse components (Figs. 3c and 3g). Typical phyllosilicate aggregates as seen in CM chondrites appear to be rare and carbonates in well-developed grains and tochilinite are absent (however, patchy aragonite was found by TEM, see below). Irregular aggregates of Ca sulfate are relatively frequent and also Ca phosphates (Cl-rich apatite and merrillite) occur, forming Ca-rich patches (Figs. 3g and 3k). Rare aggregates of Mg,Al spinel grains indicate the presence of relict CAIs. The sulfide fraction >8 μm is approximately 1.1–1.3% by area and strongly dispersed (Fig. 3k), showing grain diameters mostly below 20 μm. Sulfides are clearly more abundant than in Y-791198 and Y-793321. Sulfide grains have highly irregular shapes, are mostly polycrystalline, often porous, and are frequently accompanied by Fe,Ni metal grains (Fig. 5a). Intimate intergrowths between sulfide and minute Mg silicate grains (identified as olivine by TEM, see below) are abundant (Fig. 5b). A peculiar component observed are large (several tens of μm), porous, and sometimes spherical aggregates of Fe oxide, Fe,Ni metal, and Fe sulfide (Fig. 5c). P,Cr-bearing Fe,Ni sulfide grains are absent. The majority of sulfide grains is Ni-poor and shows no evidence for exsolved pentlandite. Their M/S ratio is at unity (0.990 ± 0.013; Table 2) and consistent with stoichiometric troilite. Only a few separate grains of pentlandite have been found, which appear to be Fe-rich. The measured Fe/(Ni + Co) ratios of a large grain and a small matrix grain are approximately 2.4 and 3.5, respectively (Table 2). These values are consistent with a Fe/Ni ratio of approximately 3.0 based on the analysis by Kimura et al. (2011), but their positions in the Fe-Ni-S ternary indicate mixtures between troilite and pentlandite (see below).
Y-86720 is, compared with B-7904, more uniform in texture and shows less abundance of large ferromagnesian objects/chondrules (Figs. 3d and 3h). As pointed out by Tomeoka et al. (1989), its matrix is strongly depleted in Fe compared with B-7904 and the other CM chondrites (Fig. 3h). SEM images show hardly any large phyllosilicate and no tochilinite aggregates. Large carbonate grains are absent, but the Ca X-ray distribution map shows Ca-rich “halos” surrounding cavities that are partially filled by sulfides (Figs. 3h and 3l). These halos might represent mobilized Ca of former carbonate grains. Accessory phases are Ca sulfate, Ca phosphate (mostly merrilite), chromite, ilmenite, and abundant metal grains. Despite an intensive search, we did not find magnetite in our sample of Y-86720, which confirms the observations of Tomeoka et al. (1989).
In Y-86720, sulfides >8 μm comprise approximately 2.0–2.4% by area and occur in two modes: small, irregular grains typically <20 μm in diameter and relatively large, partially or almost fully euhedral, platelet-shaped grains of up to 100 μm size (Fig. 5d and 5e). The euhedral sulfide grains are generally not associated with the Ca halos described above. We did not observe pentlandite, neither as exsolved phase nor as separate grains. The morphologically distinct sulfide grains show no significant difference in composition and have M/S ratios indiscernible from unity (0.998 ± 0.007 for large grains, 0.995 ± 0.006 for small, irregular grains) and very low Ni contents (<0.2 wt%; Table 2), consistent with troilite. Similar to B-7904, the fine-grained sulfides are often intimately intergrown with minute Mg-silicate grains (Figs. 5e and 5f). Many of the sulfide grains, including the euhedral ones, are associated with Fe,Ni metal grains (Fig. 5d), which show large variations in Ni concentrations (Table 2), ranging from very Ni-poor kamacite to Ni-rich taenite.
Compositions in the Fe-Ni-S Ternary
In a ternary Fe-Ni-S diagram (Fig. 6a), including the phase fields and tie lines at 200 °C after Craig (1973), the compositions of coarse pentlandite in Y-791198 fall at the end of tie lines connecting to Ni-rich MSS. These MSS compositions are not observed among the low-Ni sulfides and almost all analyses cluster closely near the edge between the FeS and Fe7S8 compositions. The same applies for the low-Ni sulfide compositions observed in Y-793321, which show a slight trend toward pentlandite compositions owing to mixed analyses and slightly coarser pentlandite blebs in Y-793321. Because the MMS field is expected to shrink with decreasing temperatures below 200 °C (Craig 1973), the MMS will become poorer in Ni, and coexisting pentlandite will become richer in Ni. This indicates that the coexisting Fe,Ni sulfides observed in Y-791198 have equilibrated at temperatures well below 200 °C as is often observed in terrestrial samples, where very Ni-poor pyrrhotite occurs abundantly next to pentlandite (e.g., Etschmann et al. 2004). Unfortunately, it is virtually impossible to obtain the bulk composition of individual sulfide aggregates from the compositions of the constituting phases, because the coarse-grained nature of segregated pentlandite renders estimates of its volume fraction highly uncertain.
The alleged pentlandite grains measured in B-7904 (Fig. 6b) plot in the Fe-Ni-S ternary between Fe-rich pentlandite and the troilite composition, where all the Ni-poor Fe sulfides of B-7904 and Y-86720 cluster. Probably, these intermediate pentlandite-like compositions represent small-scale mixtures of troilite and Fe-rich pentlandite. The pentlandite mixing line in B-7904 nonetheless appears to point to a Fe-rich pentlandite composition, with Fe/Ni of approximately 1.9–2.0. According to Kaneda et al. (1986), such Fe-rich pentlandite compositions might point to low fS2 during equilibration with coexisting MSS.
We have prepared FIB foils of two pentlandite-bearing M-type grains (98F01, 98F02; Fig 4b) and one P-type aggregate of microcrystals (98F06; Fig. 4d). The internal structure of compact M-type grains consists of nanoscale, flame-like, and crystallographically coherent intergrowths of troilite and twinned NC-type pyrrhotite, associated with sub-μm-sized, rod-shaped pentlandite grains (Figs. 7a–c; Fig. 8a). Sample 98F02 shows a thick rim of pure pentlandite enclosing the NC-pyrrhotite-troilite-pentlandite intergrowths (Fig. 7a). Flame-like troilite-pyrrhotite lamellae in both M-type grains are subparallel to (001) and have typical widths in the range of 80–200 nm. Troilite and pyrrhotite share a common orientation of their sulfur sublattice and are well distinguishable in NiAs SAED patterns due to their different superstructures (Fig. 8e). Table 3 lists N values determined from SAED patterns using the length ratios of inverse lattice vectors c* and q (Harries et al. 2011; Figs. 8a and 8e). In case of grain 98F01, N values are consistent with 6C-pyrrhotite (ideally N = 6), but the values of grain 98F02 (5.63 ± 0.13 and 5.74 ± 0.12), obtained from two different zone axes of the same crystal, are hardly consistent with an integral NC superstructure and indicate a superstructure intermediate between 5.5 and 6 °C. However, no splitting of reflections expected for a nonintegral superstructure (Harries et al. 2011) could be observed in 98F02 due to the rather broad spots. This is most likely due to the small domain size and intense twinning (see below), leading to somewhat disordered vacancy arrangements and possible disappearance of higher order superstructure reflections (being responsible for apparent splitting).
Standard deviations of least significant digits in parentheses.
Two axes given in case of twinning.
Each pair of c* and q values was obtained from the same SAED pattern. The given uncertainty relates to the measurement precision determined by measuring 10–15 spot distances, but accuracy of absolute values is not known due to lack of an internal standard and errors in sample height, camera length, etc. The ratio of c* and q is accurate within given uncertainty, as both c* and q are affected by the same systematic errors.
98F01 M-type grain
 + 
98F01 M-type grain
 + 
98F01 M-type grain
 + [1–10]
98F02 M-type grain
 + [1–10]
98F02 M-type grain
 + 
98F06 matrix grain
98F06 P-type grain
 + 
21F01 M-type grain
 + [1–10]
21F01 M-type grain
 + 
21F01 M-type grain
 + 
21F05 M-type grain
 + 
21F05 M-type grain
 + 
04F02 troilite rim
20F01 troilite rim
20F01 troilite rim
20F02 troilite rim
20F02 troilite rim
SAED patterns show that the NC-pyrrhotite is intimately twinned by 60° or 120° rotation about its c axis, resulting in superposition of two different zone axis patterns (Figs. 8b–d; due to the reduced symmetry of NC-pyrrhotite by vacancy ordering, the diffraction patterns of equivalent zone axes of the fundamental hexagonal NiAs-type structure are not equivalent any more in terms of the superstructure reflections). The orientation relationship between low-Ni Fe sulfides and pentlandite is equivalent to the one described by Francis et al. (1976) with respect to the NiAs-type fundamental structure of troilite and pyrrhotite (Fig. 8g): (111)Pn||(001)NiAs, , . The NC-pyrrhotite-troilite-pentlandite association occurs regardless of whether sulfide grains are rimmed by pentlandite (98F02) or not (98F01) and occurs also in direct contact with the hydrous, fine-grained rim material (Fig. 7b). The observed sulfide assemblage appears identical to the associations in CM2 chondrites described by Brearley and Martinez (2010) and Maldonado and Brearley (2011). Attached to the outer margin of the pentlandite-rimmed grain 98F02, we have found an 800 nm-sized subgrain of Fe-rich Zn sulfide, which appears to be in an epitactic relationship with the host grain.
The microcrystalline aggregate 98F06 consists of individual grains of abundant pentlandite, magnetite, and minor low-Ni Fe sulfide (Figs. 9a, 9b, and 9d). The quality of the only Fe sulfide grain in a suitable orientation (with an accessible zone axis perpendicular to the c direction) was not sufficient to record a very clean SAED pattern, but the presence of superstructure reflections in a NiAs equivalent zone axis clearly excludes troilite and indicates pyrrhotite (Fig. 9c). The triplets of superstructure reflections seen along the 00 l lattice row between fundamental NiAs reflections are very similar to twinned 4C-pyrrhotite (Fe7S8) (Pósfai et al. 2000; Berger et al. 2011) and could indicate a very Fe-deficient pyrrhotite. The N value determined as 3.66 ± 0.33 (1s; large uncertainty due to broadness of spots) is consistent with 4C-pyrrhotite. The majority of interstitial material between sulfide and metal grains is rich in Ca phosphate, likely merrillite based on elevated Na concentrations shown by EDX. An isolated 600 nm sulfide grain in the adjacent matrix (Fig. 9e) was found to have a suitable crystallographic orientation and the obtained SAED pattern (Fig. 9f) indicates a barely twinned NC-pyrrhotite with an N value of 5.48 ± 0.13, being consistent with 5.5C-pyrrhotite. STEM-EDX elemental mapping indicates that other small sulfide grains in the matrix are in approximately equal amounts Ni-rich (likely pentlandite) and Ni-poor (likely pyrrhotite).
FIB lamellae taken from two M-type grains (21F01, 21F05; Figs. 4e and 4f) show coherent intergrowths of troilite, NC-pyrrhotite, and pentlandite similar to Y-791198. However, compared with Y-791198, the intergrowths in both grains of Y-793321 are much coarser, and lamellae and pentlandite blebs are much better defined (Fig. 7d and 7e), positively affecting the quality of SAED patterns (Figs. 8b, 8c, 8f, and 8h). N values determined are in all cases consistent with 6C-pyrrhotite (Table 3). The orientation relationship between pyrrhotite/troilite and pentlandite is the same as in Y-791198 (Fig. 8h). Typical widths of lamellae and pentlandite precipitates are on the order of 300–500 nm. Also here, pyrrhotite is intensely twinned, but, contrary to Y791198, shows intensity variations of the twin components in SAED patterns when the sample is moved (Figs. 8b and 8c), indicating a coarser domain structure (which unfortunately is extremely difficult to image, because of the small distances between superstructure reflections). EDX analysis of pentlandite lamellae indicates no obvious internal zoning.
Grain 21F01 clearly shows signs of replacement by a fibrous mineral (Figs. 4e, 7f). STEM dark field imaging reveals that the fibrous material consists of numerous nanocrystalline domains (Fig. 10a). SAED shows spotty ring patterns with d-values corresponding to a spinel-structured mineral, most closely matching magnetite (Fig. 10c). Additional reflections in the SAED pattern point to the presence of a Fe sulfide (either troilite or pyrrhotite, which are hardly discernible in this case). EDX indicates that the crystallites are composed of Fe, O, and S, plus subordinate amounts of Mg and Si. The S content varies strongly regardless of the distance to the interface with the original sulfide, supporting the presence of a two-phase mixture of magnetite and Fe sulfide. These findings suggest that the original fibrous mineral has been replaced by these newly formed phases. The composition and morphology of this magnetite–sulfide assemblage strongly suggest that the original phase was a relatively pure tochilinite, which broke down during thermal metamorphism. The original tochilinite did not contain substantial interlayered serpentine, because Si and Mg concentrations in the replacement products are low. The occurrence indicates that tochilinite replaced the original sulfide grain during aqueous alteration. EDX X-ray maps show a significant enrichment of chromium up to 5–10 wt% in the nanocrystalline material close to the sulfide interface (Fig. 10b). It is not clear where Cr resides, but probably it is incorporated into the magnetite. The Cr enrichment close to the reactive interface suggests a local transport of Cr by a fluid during the aqueous alteration or metamorphism, because Cr is not detectable by EDX in the adjacent Fe,Ni sulfides. Potentially, the alteration of Cr-bearing Fe,Ni metal (Palmer and Lauretta 2011) in the surroundings of the grain could be a source of mobile Cr.
The fine-grained portion of Y-793321 (21F06) contains monophase pentlandite grains of variable size (several μm to less than 1 μm) without associated Fe sulfides. They are accompanied by magnetite, clinoenstatite, Ca-rich clinopyroxene, Fe-rich serpentine, and a calcium sulfate. The latter shows a fibrous texture and is difficult to identify unequivocally based on the electron diffraction patterns. Most probably, the original phase was gypsum, which partially dehydrated during sample preparation. The sulfate could be an oxidation product of sulfides during aqueous alteration or terrestrial weathering. However, no obvious Fe-bearing oxidation products such as FeOOH polymorphs or ferrihydrite have been found in the FIB samples.
In B-7904, we have sampled four sites by FIB: One sulfide-oxide-metal assemblage (04F01, Fig. 5c), and three sulfide aggregates (04F02; 04F03, Fig. 5b; 04F04). The sulfide-oxide-metal assemblage consists of 1–3 μm-sized grains of magnetite, troilite, and kamacite (Fig. 11a; all identified by SAED). In close association with kamacite, small grains of an Ni-rich alloy occasionally occur (its structure could not be established, but likely it is taenite or awaruite). No pyrrhotite was found. Large fractions of the interstitial space are empty pore space, but some pores are filled by anhydrite, which is well crystallized and yields clear zone axis diffraction patterns (Fig. 11b). Although evaporitic contamination in Antarctic meteorites is a concern (Losiak and Velbel 2011), the contaminant phases appear to be generally hydrous (Velbel 1988), and the good quality of the diffraction pattern rules out vacuum dehydration of pre-existing gypsum during sample preparation or observation (see remarks on Ca sulfate in Y-793321). Veins of fibrous anhydrite are present in the Antarctic CO3 chondrite ALH 77307 and could be terrestrial contamination due to the limited degree of preterrestrial aqueous alteration in this meteorite (Brearley 1993). In our samples, however, no indication of veining could be found.
The sampled sulfide aggregates consist of pure, polycrystalline troilite without evidence for exsolution of pyrrhotite. Only at the margins of grain aggregates in 04F02 weak, additional superstructure reflections of pyrrhotite could be detected in SAED patterns (Fig. 12a). The reflections are broad, but roughly consistent with 6C-pyrrhotite (N = 6.14 ± 0.14). A rather compact sulfide grain sampled in 04F04 (Fig. 11c) is composed of polycrystalline troilite without any detectable pyrrhotite or pentlandite. TEM-EDX indicates the presence of few and small Ni-poor Fe metal grains. The internal texture shows abundant triple junctions and subgrain boundaries. In sample 04F03 (Figs. 11d and 11e), compact, polycrystalline troilite grades into fine-grained troilite, which is intimately intergrown with abundant, sub-μm-sized grains of Fe-rich olivine (approximately Fa24 based on TEM-EDX). In this intergrowth, polycrystalline troilite shows frequent 120° (triple junction) grain boundaries. Rarely, approximately 1 μm-sized Ni-poor, but fairly Co-rich Fe metal grains occur (approximately 3 atom% Ni, approximately 8 atom% Co). The matrix portion sampled in 04F02 and 04F03 shows SAED ring patterns consistent with a composition dominated by sub-μm-sized olivine. Ca sulfate, most likely anhydrite, occurs as μm-sized, polycrystalline patches isolated within the matrix material. Embedded in silicates and anhydrite, sub-μm-sized grains of a very Ni-rich alloy occur (about equal amounts of Fe and Ni, approximately 2 atom% Co). In close proximity to the sulfide aggregates, a μm-sized patch of Ca carbonate was found and identified as aragonite by SAED. Alongside the margin of the sulfide aggregate sampled in 04F03, small pentlandite grains occur, but most Ni appears to be present as Ni-rich alloy grains within the matrix.
From Y-86720, we have obtained two FIB samples (20F01; 20F02, Fig. 5d) of euhedral, platelet-shaped sulfide crystals with adjacent matrix and one sample of a crystal with adjacent, fine-grained sulfide-silicate intergrowths (20F03, Fig. 5e). Imaging and SAED patterns indicate that the euhedral crystals are internally pure troilite. The troilite crystals contain frequent subgrain boundaries and inclusions of Fe,Ni metal (Figs. 11f and 11g), indicating that they are rather pseudomorphs than euhedral single crystals. In case of 20F01 and 20F02, narrow, less than 100 nm-wide rims of NC-pyrrhotite occur at the external margins of the troilite (Figs. 11h and 12b). The N values determined for the rims vary consistently around 5.7 (Table 3). Although uncertainties of individual N values are relatively large, the observations point to rims that have slightly lower Fe contents than those in B-7904. The fine-grained sulfide-silicate intergrowths in 20F03 were identified to be a mixture of troilite and olivine, similar to what is observed in B-7904. Grain sizes in these aggregates are typically on the order of a few 100 nm (Fig. 11i). TEM-EDX shows that the olivine is Fe-rich (Fa36–40) and has a molar Fe/Mn ratio of about 44. Many of these intergrowths show ameboid textures with poorly defined boundaries, suggesting a solid-state formation from a local precursor. The grain size of olivine in these aggregates is larger than that of the matrix and the composition is more Fe-rich (see below). We did not find the ferrihydrite-like material associated with or replacing troilite as described by Tomeoka et al. (1989). None of the troilite grains measured by TEM-EDX has a detectable Ni content (i.e., <0.5 wt%). The metal grain attached to the euhedral crystal sampled in 20F02 is kamacite (approximately 6 atom% Ni, hardly any Co) and was found to contain a high density of dislocations (Fig. 11g).
The sampled matrix portions show fibrous or scaly textures at the sub-μm scale (Fig. 11f), typical of phyllosilicates. However, SAED patterns obtained from such areas display spotty rings that can be indexed almost completely as olivine. TEM-EDX yields a (Mg + Fe)/Si ratio of approximately 2.1 and a Fe/(Fe + Mg) of approximately 0.24, consistent with ferroan olivine. In addition, the matrix contains larger, 1–3 μm-sized grains and grain aggregates of anhydrite, merrillite, and Ca,Mg carbonate. Similar to B-7904, the crystallinity of the anhydrite is very good (Fig. 11f). Based on SAED patterns, the carbonate appears to be a mixture of calcite and dolomite (or Mg-rich calcite with smaller d values compared with pure calcite). Metal, also found in direct association with anhydrite, ranges from Ni-poor (3–4 atom%) to Ni-rich (up to 25 atom%) compositions.
Sulfide Formation by Nebular Condensation Processes
The exsolution of pentlandite in pyrrhotite/troilite observed in Y-791198 and Y-793321 strongly indicates a formation of the precursor MSS at temperatures in excess of the MSS-pentlandite solvus, hence, above a minimum of 300 °C. The two modes of occurrence of pentlandite in the compact M-type sulfide grains, as sub-μm-sized blebs within pyrrhotite/troilite and as monomineralic rims, suggest two generations of pentlandite: A first generation forming the rims exsolved when the temperature dropped below the MSS-pentlandite solvus between approximately 300 and 610 °C (depending on bulk MSS composition). The grain then consisted of a Ni-poor 1C-pyrrhotite core with M/S ≤ 1 and a pentlandite rim with M/S > 1. In experiments and terrestrial pyrrhotite–pentlandite assemblages, the preferred nucleation of pentlandite along grain boundaries or cracks is well documented (e.g., Hawley and Haw 1957; Naldrett et al. 1967; Kelly and Vaughan 1983). Upon further cooling, the 1C-pyrrhotite exsolved troilite when it crossed the troilite-pyrrhotite solvus between 140 and 100 °C (Fig. 2). At this point, the second generation of pentlandite probably exsolved in the form of small blebs due to decreased solubility of Ni in the evolving pyrrhotite–troilite assemblage. During cooling below 100 °C, the 1C-pyrrhotite lamellae finally formed NC-type pyrrhotite through the ordering of Fe-site vacancies and additional formation of troilite. This scenario suggests a slow or strongly decelerating cooling from above 300 °C to below 100 °C, which allowed the exsolution assemblages to form (cf. Boctor et al. 2002; Brearley and Martinez 2010).
The temperature path recorded by the sulfide assemblage in Y-791198 is clearly inconsistent with the temperatures reached during hydrothermal alteration, which ranged well below 100 °C (Guo and Eiler 2007). Clear contacts between the sub-μm NC-pyrrhotite-troilite-pentlandite intergrowths and the surrounding hydrous matrix without pentlandite rims even at the nm-scale render an aqueous formation of pentlandite highly unlikely. Even in case the pentlandite rims were formed by aqueous processes, the formation temperature of the internal troilite-pyrrhotite intergrowths would still exceed the established CM2 alteration temperatures. Because Y-791198 is one of the least aqueously altered CM2 chondrites (Metzler et al. 1992; Rubin et al. 2007; Chizmadia and Brearley 2008), indicating a combination of lower alteration temperatures, lower water/rock ratios, and a shorter duration of the aqueous activity than typical for CM2 chondrites, an aqueous formation of pentlandite rims during processing under intense hydrothermal conditions can be effectively ruled out. Furthermore, there is no evidence that Y-791198 experienced any significant thermometamorphic overprint at temperatures above the pentlandite solvus. In general, the formation of the observed sulfide assemblages by thermal metamorphism is very unlikely, because such a process would require unrealistic conditions with respect to the precursor phases and diffusion distances: Starting from dispersed Ni-poor troilite and Fe,Ni metal grains, as observed in ordinary chondrites, the high-temperature formation of a Ni-rich MSS with M/S of about unity would require a long distance transport of metals (several 10 s or 100 s of μm) and a high availability of sulfur. However, Y-791198 does not show any petrographic evidence for thorough metamorphic equilibration and mass transport (cf. Chizmadia and Brearley 2008). An alternative scenario would require separate but clustered Fe sulfides and pentlandite grains before metamorphism, which then would turn into an MSS. In Y-791198, we have found such clusters of microcrystals as P-type sulfides (sample 98F06), but they exist in close proximity to M-type NC-pyrrhotite-troilite-pentlandite grains. It is highly unlikely that the fine-grained assemblages would have persisted in sub-mm-distance to other grain aggregates that homogenized during the same hypothetic metamorphic overprint. Hence, as aqueous or metamorphic formation of the M-type NC-pyrrhotite-troilite-pentlandite grains can be excluded, the origin of their precursor MSS must predate the accretion into the CM parent body. This supports the suggestions made by Boctor et al. (2002), Brearley and Martinez (2010), and Brearley (2010) that these sulfides are primary products of the solar nebula. Dai and Bradley (2001) described 6C-pyrrhotite from anhydrous IDPs, but suggested that pyrrhotite might have formed from an unknown, spinel-type sulfide during atmospheric entry and heating. There has been no subsequent verification of the existence of the supposed precursor, and our results indicate that pyrrhotite in anhydrous IDPs might indeed be a primary phase. This parallels the interpretation of the origin of pyrrhotite in IDPs by Zolensky and Thomas (1995), who suggested extended H2S sulfidation of primary troilite, exfoliated from metal grains, to form Fe-deficient pyrrhotite. Lauretta et al. (1996, 1997, 1998) showed that troilite is the first sulfide to form on metal of chondritic composition. However, sustained formation of troilite and eventual formation of Fe-deficient pyrrhotite will only occur if the sulfur fugacity remains above the IT buffer. There are several possible scenarios: (1) The gas–solid mixture equilibrates before Fe metal is consumed. In this case, the sulfide should be stoichiometric troilite. (2) H2S/H2 and corresponding fS2 remain above the IT buffer values while metal is completely converted. Troilite will then react to pyrrhotite. These two options obviously require a complete, “global” equilibrium (and sufficient temperature). (3) Pyrrhotite might form through local equilibrium at high H2S/H2 and fS2, either on the outside of thickly sulfide-coated metal grains (Lauretta et al. 1997) or from exfoliated troilite flakes (Zolensky and Thomas 1995). (4) Other components and phases such as gaseous H2O and iron oxides might have participated in sulfidation reactions, possibly documented by the 4C-pyrrhotite plus magnetite found in the P-type aggregate of Y-791198.
4C-pyrrhotite in the P-type sulfide in Y-791198 probably points to a distinct formation environment in the solar nebula, probably connected to more oxidizing conditions based on the presence of abundant magnetite. Its formation via aqueous processes on the CM parent body appears unlikely, as Fe sulfides were probably not stable phases under these conditions as will be discussed in the next section. Ciesla et al. (2003) pointed out that in the wake of shock waves in ice-rich nebular environments, the high vapor pressure of water would rapidly oxidize metallic iron and hydrate forsterite. If at the same time the fugacities of H2S and S2 were high, the formation of highly Fe-deficient 4C-pyrrhotite (x = 0.125) plus magnetite might have occurred in such a setting. In Fig. 1b, the intersection of the 4C isopleth with the pyrite-pyrrhotite buffer curve yields a minimum temperature of approximately 450–500 °C for this process (below this temperature, pyrite would be stable). At this temperature, fO2 would have been around −22 log units (approximately 8 log units above the magnetite-iron-buffer at any temperature) and fS2 around −5 log units.
In the case of M-type grains, the actual bulk M/S and Fe/Ni ratios of the homogeneous MSS precursors before pentlandite exsolution are interesting parameters to assess nebular conditions such as fS2 and fH2S, estimate the onset temperature of pentlandite exsolution, or discuss the role of kinetic effects as described by Lauretta et al. (1997). Unfortunately, the highly heterogeneous distribution of pentlandite in the grains after exsolution renders the determination of the bulk composition exceedingly difficult. On the contrary, the crystallographic parameters of pyrrhotites can be determined very precisely. The N values determined for pyrrhotites in Y-791198 appear to be consistently smaller than 6.0, while values determined for Y-793321 pyrrhotite are close to 6.0. The lower values of Y-791198 between N = 5.5 and 6.0 might indicate an incomplete equilibration between exsolved troilite and the coexisting NC-pyrrhotite, similar to many terrestrial samples. Conversely, the values close to 6.0 in Y-793321 possibly point to a much slower cooling rate in the <100 °C range. However, rather than reflecting nebular processes, these differences are more likely related to different parent-body thermal histories of Y-791198 and Y-793321 (see below).
Parent-Body Aqueous Alteration Versus Nebular Processes
As outlined above, the texture of M-type grains is clearly incompatible with low-temperature aqueous formation. The case for the P-type microcrystalline cluster of pentlandite and 4C-pyrrhotite (Y-791198, FIB section 98F06) is less clear. The presence of 4C-pyrrhotite in association with magnetite in this aggregate indicates an oxidized assemblage, which could be expected to form in a low-temperature alteration process. 4C-pyrrhotite is typically found in many terrestrial hydrothermal metal deposits and its presence in highly altered CI chondrites and cometary dust has been described by Berger et al. (2011) based on TEM observations. In addition, its occurrence in CI chondrites can be inferred from EPMA analyses by Kerridge et al. (1979) and Bullock et al. (2005), which show sulfur enrichments consistent with Fe7S8. The corrosion and replacement seen in the grain sampled in 21F01 of Y-793321 indicate that high M/S sulfides (with M/S higher than Fe7S8) have been unstable during aqueous alteration and probably transformed into a tochilinite-like phase. The instability of Ni-poor pyrrhotite during progressive alteration of CM chondrites has been described previously (e.g., Rubin et al. 2007) and similar corrosion has been observed by Brearley (2011) in the Mighei CM2 chondrite, who described replacement by a fibrous oxysulfide. A contrasting behavior of 4C-pyrrhotite, being formed, and 6C-pyrrhotite/troilite, being dissolved and replaced, would be highly enigmatic. In CI chondrites, the alteration conditions reached higher temperatures and were probably more oxidizing than in CM2 chondrites (for these Eh ≤−0.67 V; Guo and Eiler 2007), allowing magnetite to form instead of tochilinite. Hence, the local formation of 4C-pyrrhotite in section 98F06 might reflect spatially confined oxidizing conditions. Unfortunately, very little is known about the low-temperature stability relationships among various pyrrhotite variants and even less about the, likely strongly kinetically controlled, precipitation behavior in low-temperature aqueous systems (including potentially forming phases like smythite and mackinawite). Despite very similar compositions and structures, laboratory studies show differences in reactive behavior of different pyrrhotites (e.g. Belzile et al. 2004; Becker et al. 2010b; Harries et al. 2013). However, these differences, observed in oxidizing environments (Eh > 0), are rather due to distinct kinetics than reflecting different thermodynamic stabilities. Whether 4C-pyrrhotite could be stable relative to 6C-pyrrhotite or troilite in basic and reducing aqueous solutions of the CM parent body cannot be answered unambiguously, but it appears unlikely. Moreover, we did not find overgrowths of 4C-pyrrhotite on M-type grains in contact with the matrix, even on the nm-scale, suggesting that the formation of 4C-pyrrhotite on the CM parent body is unlikely to have occurred. Hence, we favor a nebular formation of the 4C-pyrrhotite-pentlandite-magnetite assemblage in the P-type aggregate as outlined above.
Low-Grade Parent-Body Metamorphism
With regard to the sulfide assemblages, the least thermally altered sample Y-793321 shows only very subtle signs of metamorphism. The corrosion product of the large sulfide grain in section 21F01 of Y-793321 was likely tochilinite that transformed into a mixture of Fe sulfide and magnetite. The onset of this reaction on laboratory time scales occurs at around 245 °C (Fuchs et al. 1973). The upper temperature limit may be constrained by the compact sulfide assemblages. It appears that the exsolved pentlandite has never been completely resorbed into a MSS, and, hence, it was not heated above the solvus between 300 and 610 °C. After a complete homogenization of the sulfide, it could be expected that the exsolving pentlandite would, at least partially, nucleate at the crystal margins as observed in Y-791198 and terrestrial samples (Kelly and Vaughan 1983). However, the grain in section 21F03 of Y-793321 does not show a pentlandite rim, indicating that it probably did not rehomogenize after aqueous corrosion had taken place.
Given these observations and data reported by Nakamura (2005, 2006), the peak metamorphic temperatures of Y-793321 surely crossed the troilite solvus occurring at a maximum temperature of 140 °C. As pointed out in the introduction, 6C-pyrrhotite does not form immediately at the start of troilite exsolution, but forms from 1C-pyrrhotite through ordering of Fe-site vacancies at temperatures below 100 °C. Given that many terrestrial troilite-bearing pyrrhotites, and maybe pyrrhotites in Y-791198 as well, appear to have been kinetically frozen in the process of low-temperature equilibration (showing non-integral NC structures), sulfides in Y-793321 obviously equilibrated well within the stability field of 6C-pyrrhotite + troilite. The considerably coarser internal textures of compact M-type sulfide grains in Y-793321 compared with Y-791198 indicate different thermal histories and point toward prolonged low-temperature thermal annealing of Y-793321 below 100 °C. Condit et al. (1974) showed that diffusion of Fe in pyrrhotite is rather fast. At temperatures below 100 °C, the driving force that leads to vacancy ordering is expected to slow down diffusion (unfortunately no data exist in this temperature range), but still diffusive transport can reasonably be assumed to be fast relative to other geological materials such as silicates. Hence, on the basis of data of Condit et al. (1974), we assume that diffusion on the sub-μm scale should be possible even at low temperatures around 0 °C during many hundreds of millions of years. There are two relevant scenarios that could account for the observed differences in sulfide textures of Y-791198 and Y-793321:
First, Nakamura (2006) identified Y-793321 as a regolith breccia and suggested short duration-heating by impacts during near-surface residence. For dark, chondritic material, the equilibrium temperature of a small-sized asteroid or asteroidal surface should be in the range of −50 to −100 °C in the main asteroid belt at 2.8 AU (Butler 1966) and colder at larger distance or during lower radiation output of the young sun. To achieve prolonged low-temperature annealing and coarsening of the pyrrhotite–troilite mixture, it would have been necessary to sustain regolith temperatures above the expected equilibrium temperature in the Main Belt by a suitably high rate of impacts. At the same time, peak heating during episodic impact events must not have exceeded temperatures above the respective solvi, to avoid partial rehomogenization or eradication of the existing exsolution textures.
In a second scenario, Y-793321 could have experienced higher long-term average temperatures than Y-791198 by receiving stronger radiation heating through relatively stable orbital parameters that brought the parent asteroid closer to the sun during much of its lifetime. At average solar distances in the vicinity of the orbits of Earth and Mars, the equilibrium temperature of a dark chondrite is expected to be around 0 °C and, hence, long-term residence in this region could be responsible for the different textural evolution of Y-793321 sulfides.
As pointed out by Nakamura (2006), based on the absence of Fe-Mg zoning in chondrule silicates, the peak heating of Y-793321, up to perhaps 500 °C, was likely transitional and not due to prolonged radiogenic heat production. Hence, heating must have occurred while the material resided close to the asteroid's surface. Heating by impacts is transient with relatively rapid cooling rates (Wittmann et al. 2010) and is mostly regarded as insufficient to cause long-term, large-scale heating of asteroidal bodies (Keil et al. 1997). On the other hand, solar heating up to the breakdown temperatures of phyllosilicates would require close perihelion distances of <0.1 AU (Nakamura 2005), which appears incompatible with a long-term survivable orbit. This suggests that impact-generated heat is most likely the explanation for the high-temperature metamorphic overprint, which is supported by shock metamorphic features (planar fractures, Nakamura 2006) observed in Y-793321. However, despite the fact that the low thermal diffusivity of regolith material might increase the retention of impact-generated heat, it appears questionable whether the rate of repeated impacts could have been high enough to keep the long-term average regolith temperature in a reasonable range to allow coarsening of sulfides in Y-793321 relative to Y-791198. A long-term residence of the Y-793321 parent body in the vicinity of the orbits of Earth and Mars and corresponding higher surface equilibrium temperatures appear to be better explanations for this phenomenon.
High-Grade Parent-Body Metamorphism
The troilite–metal assemblages of B-7904 and Y-86720 are unlike the sulfides and metals observed in CM chondrites studied here and reported elsewhere. In terms of sulfide mineralogy, the euhedral, metal-bearing troilite pseudomorphs seen in Y-86720 are the most telling evidence for a strong metamorphic overprint of these rocks. We interpret them to represent the replacement of former euhedral pyrrhotite crystals, similar to those known from CI chondrites. Assuming the original pyrrhotite to have been Fe7S8 (x = 0.125) as observed in CI chondrites (Berger et al. 2011), the following reactions describe the process:
The reactions are driven to the product sides by low fS2 and the troilite–metal assemblages in B-7904 and Y-86720 essentially constrain the sulfur fugacity at peak metamorphic temperatures to that of the IT buffer. As outlined in the introduction, the stable coexistence of Fe oxides and pyrrhotite couples fS2 and fO2. Figure 1b shows fO2 isobars for coexisting magnetite and pyrrhotite (Whitney 1984). Isobars for other coexisting Fe oxide minerals (including wüstite) can be calculated accordingly. In fS2 − T space, the intersections of isobars of equal fO2 for two pairs of pyrrhotite–oxide assemblages define the curves of joint fS2 − fO2 buffers, along which both fugacities are fixed at given temperature. The iron-magnetite-pyrrhotite (IM-Po) and iron-wüstite-pyrrhotite (IW-Po) buffers coincide with the IT buffer, because the sulfide in equilibrium with metal is always stoichiometric troilite. The magnetite-wüstite-pyrrhotite (MW-Po) buffer becomes relevant at temperatures above 570 °C. Although wüstite is not expected to be a stable phase in association with magnesian silicates and will likely dissolve in them (depending on SiO2 activity), the buffer reaction can approximately constrain the metamorphic temperature at which magnetite becomes unstable along the IT buffer line. For unity activity of FeO, the MW-Po line intersects the IT line at an invariant point at approximately 572 °C, corresponding to an fS2 of −12.7 to −13.1 log units (depending on the source of fS2 data; Toulmin and Barton 1964; Rau 1976) and an fO2 of −25.8 log units. FeO activities of less than one strongly shift the intersection toward lower temperatures (e.g., approximately 400 °C for aFeO = 0.6).
Texturally, especially with regard to sulfide morphology, Y-86720 bears much resemblance to CI chondrites, which carry abundant magnetite. Even if the precursor mineralogy was more CM-like, magnetite would have been a ubiquitous accessory phase. If we interpret the complete lack of magnetite in Y-86720 to be the result of reduction to FeO and assume that both Y-86720 and B-7904 initially had similar FeO and SiO2 activities in their silicate portions, then the following scenario arises: Y-86720 has experienced peak metamorphic temperatures above the stability limit of magnetite, while B-7904 was subjected to lower temperature at conditions in favor of magnetite conservation. This constrains the peak metamorphic temperature of B-7904 to less than 570 °C and suggests fS2 and fO2 to have been below those values given above for the invariant point at which troilite, iron, magnetite, and wüstite coexist. The temperature estimate compares well with the range derived by Zolensky et al. (1991), but is lower than the temperatures suggested by Akai (1990, 1992).
The 6C-like pyrrhotite occurring as rims on troilite in B-7904 and Y-86720 is apparently not in equilibrium with the coexisting metal, as Fe-deficiency in NiAs-structured sulfide would not be allowed by the low fS2 that led to the formation of Fe,Ni metal. Figure 1a shows that a composition of Fe0.917S (x = 0.083), corresponding to 6C-pyrrhotite, forms at fS2 of about 5–8 log units above the IT buffer. However, as any composition formed between the x = 0 (troilite) and x = 0.083 isopleths would decompose into troilite and 6C-pyrrhotite below 100 °C, it cannot be ruled out that the Fe-deficient troilite rims formed in this intermediate region. Clearly, the rims formed at elevated fS2 relative to the IT buffer, at which the bulk of the sulfides equilibrated. We interpret the formation of rims to represent a late-stage, low-temperature reaction during cooling when fS2 was no longer in equilibrium with the troilite–metal assemblage. At this stage, low amounts of postmetamorphic, hydrous fluids might have been involved in the alteration as documented by heat-sensitive minerals such as calcite and aragonite. Thermodynamic calculations show that anhydrite is not a stable phase under conditions of troilite stability due to low fS2. Figure 1b shows fS2 isopleths for an assemblage involving pyrrhotite, magnetite, anhydrite, and CaO (dissolved in silicates). The location of isopleths is given by the intersections of fO2 isobars of Equation 3 and corresponding isobars of the reaction:
At aCaO = 1 the isopleth is located close to the FMQ-Po buffer of Eggler and Lorand (1993), lying just below the x = 0.083 isopleth of 6C-pyrrhotite. Lower CaO activities shift the anhydrite isopleth toward higher fS2, into the field where 4C-pyrrhotite (x = 0.125) would be expected to form. Hence, at relatively high CaO activity (possibly documented by calcite/aragonite formation) and fS2 just above the FMQ-Po buffer, conditions would have been favorable for forming both the Fe-deficient rims on troilite and the anhydrite in B-7904 and Y-86720 (provided the carbonates and anhydrite are indeed preterrestrial). These observations indicate a two-stage evolution of the observed assemblages—a peak metamorphic stage and a retrograde, maybe hydrous, stage.
Relations Among Metamorphosed and Pristine CM/CI Chondrites
The oxygen isotopic compositions of B-7904 and Y-86720 are distinct from CI and CM chondrites, but could be related to CI chondrites through mass-dependent fractionation (Mayeda and Clayton 1998). Clayton and Mayeda (1999) suggested an isotopic model involving mixing between anhydrous rock and a water reservoir and point out possible genetic relationships between CI and CM chondrites. These could have been formed by mixing along a common or very similar mixing line, but were subsequently processed at different temperatures and water/rock ratios. The relationships are exemplified by the existence of anomalous, transitional CM/CI-like members, such as the meteorites Bells and Essebi, which not only have transitional oxygen isotopic compositions, but also contain magnetite in abundances similar to those of CI chondrites (Rowe et al. 1994; Clayton and Mayeda 1999).
From the perspective of sulfide mineralogy, the premetamorphic precursors of B-7904 and Y-86720 might have been similar transitional cases. Assuming that the 17O- and 18O-rich compositions are due to dehydration and evolution along an expected slope 0.5 line, the precursor probably had CI-like oxygen isotopic ratios. On the other hand, the fine-grained troilite-olivine intergrowths observed in Y-86720 and B-7904 suggest the former presence of interlayered tochilinite-serpentine aggregates, which broke down thermally into troilite, as suggested by Tomeoka et al. (1989), and additionally formed olivine from the silicate fraction. Because the sub-μm-sized, highly dispersed olivine grains are unlikely to have formed as a result of exsolution, and ameboid aggregate shapes exclude crystallization from a sulfide-silicate melt, there are few viable alternatives other than assuming an initially uniform, sulfide- and silicate-rich mixture. The possibility that sulfide infiltrated the olivine-rich matrix during the retrograde stage appears possible but remote, because this sulfide should have been pyrrhotite with little textural equilibration. Instead, we find texturally fairly equilibrated troilite with abundant triple junctions and relatively large olivine grains unlike those in the matrix (e.g., Fig. 5f). The inferred presence of tochilinite indicates that the precursor mineralogies of both meteorites might have been indeed CM-like, as sulfide-bearing interlayer minerals are generally absent in CI chondrites. The preservation of primary silicates in B-7904 and Y-86720 also texturally supports a rather weakly altered, CM-like premetamorphic mineralogy, while preserved euhedral sulfide morphologies in Y-86720 indicate CI similarities.
The sulfide mineralogies of CM and CI chondrites and their anomalous relatives prove to be very diverse and record complex evolutionary histories. In CM2 chondrites Y-791198 and Y-793321, the NC-pyrrhotite-troilite-pentlandite association found in M-type grains indicates high-temperature formation from a once homogeneous MSS. This is inconsistent with parent-body aqueous alteration and strongly points to processing in the solar nebula. An assessment of nebular conditions that led to the formation of the MSS precursors, instead of the usual Ni-poor troilite, is difficult due to the lack of knowledge about the bulk composition of now exsolved and highly heterogeneous sulfide grains. 4C-pyrrhotite in association with magnetite and pentlandite in P-type sulfide aggregates possibly points to a relatively oxidizing nebular environment during sulfidation and formation of these peculiar aggregates.
The microstructure of complex sulfide grains can further add information on low-temperature cooling rates and thermal evolution, suggesting that Y-793321 experienced higher long-term average temperatures than Y-791198 in the <100 °C temperature regime. TEM studies in combination with kinetic modeling of the exsolution process could provide better constraints on actual cooling rates. However, while data on diffusivity in disordered, high-temperature pyrrhotites are available, data on low-temperature diffusion rates in ordered pyrrhotites are still needed for such calculations. Also, detailed knowledge about low-temperature phase relations in the Fe-Ni-S system is desirable, specifically addressing the relationships between different pyrrhotite variants below 100 °C and the solubility of Ni in them. Our observations of two-stage pentlandite exsolution in Y-791198 sulfides suggest a decrease in Ni solubility during the troilite exsolution and/or the subsequent 1C- to NC-pyrrhotite ordering. Sub-μm pentlandite exsolution in Fe-rich pyrrhotite (x < 0.083) might therefore not necessarily constrain high-temperature processing near the classical pentlandite solvus, but rather point to processes in the 100–140 °C range.
Corrosion of NC-pyrrhotite, troilite, and pentlandite in M-type grains of Y-793321 is scarce, but indicative of instability of these phases during aqueous alteration. The corrosion product likely was tochilinite, which converted to magnetite plus troilite during the thermal overprint on Y-793321.
The highly metamorphosed Belgica grouplet meteorites Y-86720 and B-7904 show troilite as the dominant sulfide phase with no textural and mineralogical resemblance to sulfides in pristine or weakly metamorphosed CM2 chondrites. In both meteorites, troilite coexists with Fe,Ni metal, indicating peak metamorphism at fS2 of the IT buffer under strongly reducing conditions. The absence of magnetite in Y-86720 suggests conditions that crossed into the wüstite field and indicates a higher metamorphic grade than B-7904, in which magnetite is preserved. The metamorphic conditions of B-7904 can be constrained to T < 570 °C, fS2 < −13.1 and fO2 < −25.8 log units. Troilite in Y-86720 and B-7904 shows rims of 6C-like pyrrhotite, and the matrix of both meteorites contains anhydrite. Both phases are not stable under the conditions of the IT buffer and must have formed during cooling after the metamorphic event.
Texturally, the Belgica grouplet meteorites are very diverse, but observations on sulfide mineralogies link them to each other and to CM and CI chondrites. In Y-86720 euhedral, platelet-shaped troilite pseudomorphs occur akin to pyrrhotite crystals seen in CI chondrites. In Y-86720 and B-7904, extremely fine-grained and closely intergrown troilite–olivine mixtures occur, which most probably originated from the breakdown of tochilinite-serpentine intergrowths typically present in CM chondrites. Especially in case of Y-86720, these observations imply an intermediate mineralogical composition of the precursor rock between CM and CI chondrites.
We gratefully acknowledge sample loan by the National Institute of Polar Research (NIPR) of Japan and financial support by the Leibniz program of the German Research Foundation (DFG; LA 830/14-1 to F.L.) and the ENB program of the Bavarian State Ministry of Sciences, Research and the Arts (to D.H.). Reviews and editorial comments by M. Pósfai, T. Zega, and A. Brearley helped to improve the manuscript and are gratefully acknowledged.