Twelve samples belonging to the chassignite and nakhlite subgroups of Martian meteorites were investigated using a variety of micro-beam analytical techniques to gain insight into the petrogenesis of these two meteorite classes. There are a striking number of geochemical similarities between the chassignites and nakhlites, including mineralogy and petrology, crystallization age, cosmic-ray exposure age, and radiogenic isotopic compositions. However, there are also geochemical differences, namely in trace element systematics of pyroxenes, that have led some authors to conclude that the nakhlites are comagmatic with each other, but not comagmatic with the chassignites. On the basis of data presented here, we propose a model in which these differences can be reconciled by the addition of an exogenous Cl-rich fluid to the chassignite-nakhlite magma body shortly after the formation of the cumulate horizon that was sampled by the Chassigny meteorite. This model is supported by the textural and chemical associations of the volatile-bearing minerals apatite, amphibole, and biotite, which record a history starting with the addition of a Cl- and LREE-enriched fluid to the magma body. As the magma continued to crystallize, it eventually reached chloride saturation and degassed a Cl-rich fluid phase. Depending on the provenance of the Cl-rich fluid, this model could explain how the chassignites and nakhlites originated from an LREE-depleted source, yet all exhibit LREE-enriched bulk-rock patterns. Additionally, the model explains the range in oxygen fugacity that is recorded by the chassignites and nakhlites because eventual exsolution and loss of Cl-rich fluid phases near the end of crystallization of the nakhlite sequence leads to auto-oxidation of the magma body due to the preferential partitioning of Fe2+ into the fluid phase.
Together, the chassignites and nakhlites make up what are in all likelihood ten individual samples of cumulate igneous rocks from Mars (Floran et al. 1978; Treiman 2005; Beck et al. 2006; Mikouchi et al. 2012). They comprise two of the three classic subgroups of the SNC (shergottite-nakhlite-chassignite) Martian meteorite clan, to which the petrologically distinct Allan Hills (ALH) 84001 (ancient orthopyroxenite cumulate; Romanek et al. 1994) and Northwest Africa (NWA) 7034 (basaltic breccia; Agee et al. 2013) have more recently been added. The mineralogy of the chassignites and nakhlites originally led to their classification as distinct meteorite subgroups (McSween and Treiman 1998). The two chassignites, Chassigny and NWA 2737, are cumulate dunites with minor chromite, orthopyroxene, and other phases, whereas the nakhlite clan (Nakhla, Governador Valadares, Lafayette, Miller Range [MIL] 03346 and pairings, NWA 817, NWA 998, NWA 5790, and Yamato-000593/749/802) are cumulate clinopyroxenites that contain a host of minor phases and highly variable proportions of mesostasis (Floran et al. 1978; Treiman 2005; Beck et al. 2006; Mikouchi et al. 2012). Primarily due to the notable textural differences among them, much of the previous work conducted on the origin of the nakhlites has focused on whether they can reasonably be inferred to have been derived from a single thick lava flow or shallow magma chamber (Lentz et al. 1999; Mikouchi et al. 2003, 2012; Imae et al. 2005; Treiman 2005; Day et al. 2006; Treiman and Irving 2008; Hallis and Taylor 2011). A comagmatic origin for the nakhlites is also supported by concordant crystallization ages, ejection ages, and isotopic compositions (Jagoutz 1996; Shih et al. 1998, 1999, 2006; Brandon et al. 2000; Nyquist et al. 2001; Treiman 2005; Caro et al. 2008; Carlson and Boyet 2009; Korochantseva et al. 2011). Interestingly, although Wadhwa and Crozaz (1995) argued on the basis of trace element ratios in pyroxenes that Chassigny and the nakhlites could not reasonably be linked to the same magmatic system, the crystallization ages, ejection ages, trace element compositions, and initial isotopic compositions of Chassigny and NWA 2737 are remarkably similar to the nakhlites, a compelling observation that has largely been overlooked when models of chassignite and nakhlite petrogenesis are formulated.
Another intriguing feature of both the chassignites and nakhlites is the diversity of volatile-bearing mineral phases they contain. Specifically, the nakhlites and chassignites host kaersutite, Ti-biotite, apatite, chlorian ferrohornblende, and potassic chlorohastingsite (Johnson et al. 1991; McSween and Harvey 1993; Watson et al. 1994; Righter et al. 2002; Sautter et al. 2006; McCubbin and Nekvasil 2008; Filiberto and Treiman 2009b; McCubbin et al. 2009, 2010a). Understanding the origin of these volatile-bearing minerals is important for understanding not only the petrogenesis of the rocks themselves, but also the behavior of magmatic volatiles in Martian magmatic systems. Mars is often described as a volatile-rich planet (Dreibus and Wanke 1985; Halliday et al. 2001; McLennan 2003; Albarede 2009), and this richness in volatiles is often extended to the magmatic volatiles (McSween et al. 2001; Taylor et al. 2010). On Earth, magmatic volatiles (i.e., H2O, F, Cl, C-species, and S-species) play an important role in the physicochemical processes that control thermal stabilities of minerals and melts, magma eruptive processes, and transportation of economically important metals. However, the abundances and roles of magmatic volatile elements may differ between Earth and Mars. For example, volatiles in the C-O-H system (O2, H2O, H2, CO2, CO, CH4) dominate the volatile budget of the silicate Earth (Jambon 1994; Mysen et al. 2009), whereas carbon-species may play a significantly smaller role in Martian magmatic systems due to the low oxygen fugacity of the Martian mantle (i.e., estimated to be 1–2 log units above the iron-wustite (IW) equilibrium; see Herd and Papike 2000; Wadhwa 2001, 2008; Herd et al. 2002; Herd 2003, 2006a, 2008; McCanta et al. 2004; Shearer et al. 2006). In contrast, the present day terrestrial mantle is estimated to be within 1–2 log units of the fayalite-magnetite-quartz (FMQ) equilibrium (Bryndzia and Wood 1990; Luth et al. 1990; Wood et al. 1990; Canil 2002; Kelley and Cottrell 2009). The lower oxygen fugacity in the Martian mantle stabilizes graphite at the expense of C-O species with a corresponding net decrease in CO32− solubility in any partial melts (Hirschmann and Withers 2008; Stanley et al. 2011, 2012), although there are some C-O-H species present in Martian magmas, and they have been implicated in the production of macro molecular carbon and polyaromatic hydrocarbons (Steele et al. 2012). Consequently, other volatile elements like the halogens or H-species may play a more active role in Martian magmatic systems than carbon. In fact, it has been postulated that chlorine is the dominant volatile in Martian magmatic systems (Filiberto and Treiman 2009a; Patiño Douce et al. 2011), and other studies have discussed the importance of chlorine along with water in Martian magmas (Lentz et al. 2001; McSween et al. 2001; McSween 2006; McCubbin and Nekvasil 2008; McCubbin et al. 2008b, 2009, 2010a, 2012; Nekvasil et al. 2009; Taylor et al. 2010; Hallis et al. 2012; Usui et al. 2012).
Chlorine enrichment on the Martian surface has been well documented by the Gamma ray spectrometer that was onboard the 2001 Mars Odyssey orbiter, which indicated average chlorine concentrations ranging from 2000 ppm to greater than 8000 ppm Cl (Boynton et al. 2007; Taylor et al. 2010). Chlorine has also been discovered in the form of chloride deposits in the southern highlands of Mars (Osterloo et al. 2008, 2010; Glotch et al. 2010), as perchlorate within soil in the northern hemisphere by the Phoenix Lander (Hecht et al. 2009) and as a significant component in the rocks and soils analyzed by both of the Mars Exploration Rovers (MER) (Gellert et al. 2004, 2006). Most recently, concentrated brines were implicated as causing recurring slope lineae observed on equator-facing slopes in the southern hemisphere of Mars between 48 °S and 32°S (McEwen et al. 2011). Consistent with the orbiter and lander evidence for elevated chlorine on Mars, studies of the Martian meteorites have revealed a Cl-rich mineralogy. Chlorine-rich apatites are reported in nearly all Martian meteorites (Bunch and Reid 1975; Floran et al. 1978; Meyer 1998; Mojzsis and Arrhenius 1998; Bridges et al. 2001; Greenwood 2005; Imae et al. 2005; Treiman 2005; Beck et al. 2006; Patiño Douce and Roden 2006; Treiman et al. 2007; McCubbin and Nekvasil 2008; Treiman and Irving 2008; Filiberto and Treiman 2009b; Channon et al. 2011; Patiño Douce et al. 2011), and Cl-rich amphibole has been reported in the nakhlite MIL 03346 (Sautter et al. 2006; McCubbin et al. 2009). On Earth, monovalent anion components in apatites and amphiboles are dominantly fluorine and/or hydroxyl (Morrison 1991; Piccoli and Candela 2002; Patiño Douce and Roden 2006; Martin 2007; Schumacher 2007), and elevated chlorine contents in silicates and phosphates are typically attributed to formation by Cl-rich hydrothermal fluids (Vanko 1986; Boudreau and McCallum 1989, 1992; Boudreau 1993; Meurer and Boudreau 1996; Mazdab 2003).
Much of the focus on the Cl-rich mineralogy of the Martian meteorites comes from studies of the nakhlite and chassignite meteorites, which have a broader array of magmatic volatile-bearing minerals than do the shergottites. Some have argued that the apparent Cl-enrichment in these meteorites is due to secondary alteration or assimilation of volatile-rich materials (Bridges et al. 2001; Greenwood 2005; Sautter et al. 2006), whereas others implicated elevated chlorine contents in the Martian mantle to account for the chlorine enrichment (Filiberto and Treiman 2009a; Patiño Douce et al. 2011). There is a clear link between chlorine-rich hydrothermal fluids and magmatic activity in Chassigny and MIL 03346 (McCubbin and Nekvasil 2008; McCubbin et al. 2009), but it is presently unclear whether the Cl-rich fluid was exogenous or endogenous to the magmatic system.
In the present study, we focus primarily on the volatile-bearing mineralogy of the nakhlite and chassignite meteorites to determine the role that fluorine, chlorine, and water have played in their petrogenetic histories. We primarily investigate the volatile-bearing mineral apatite; however, compositions of coexisting amphibole and biotite, as well as pyroxenes of various textural occurrences, are reported and discussed. Using our data sets and the extensive literature on the chassignites and nakhlites, we address outstanding questions concerning their petrogenesis: What is the origin of their diverse array of volatile-bearing minerals? Are open-system processes required to explain their origin? Do pyroxenes in both meteorite groups rule out a comagmatic origin? Are the overlapping ages and isotopic compositions of the chassignites and nakhlites coincidence, or can both meteorite groups be explained as a single comagmatic unit? Lastly, we propose a comprehensive model that addresses the multiple mineralogical, petrologic, and isotopic constraints on the origin of both the chassignites and nakhlites.
We have collected electron probe microanalytical (EPMA) data on apatites from the chassignite NWA 2737, which we combined with previously published apatite data from Chassigny (McCubbin and Nekvasil 2008). Additionally, we collected EPMA data on apatites from several nakhlites, including Nakhla, Governador Valadares, Lafayette, MIL 03346, MIL 090030, MIL 090032, MIL 090136, NWA 817, NWA 998, and NWA 5790. We compiled amphibole compositions from MIL 03346 and Chassigny (Sautter et al. 2006; McCubbin et al. 2009, 2010a), and we collected EPMA data for amphibole in MIL 090030, MIL 090032, MIL 090136, and NWA 5790. We also collected SIMS data for F, Cl, and OH from a single Ti-biotite grain in an olivine-hosted melt inclusion within the Chassigny meteorite, which we combined with previously published analyses of Ti-biotite (Johnson et al. 1991; Righter et al. 2002). We collected EPMA data of Cl-bearing maskelynite in olivine-hosted melt inclusions within the Chassigny meteorite, and we collected SIMS data for F, Cl, and OH from a single patch of maskelynite in an olivine-hosted melt inclusion in Chassigny. We combined our maskelynite analyses with previously published analyses of Cl-H2O-bearing maskelynite (Varela et al. 2000; Boctor et al. 2003). The list of meteorites investigated, as well as the volatile-bearing phases present and their respective textural occurrence, is presented in Table 1.
Table 1. List of Martian meteorites investigated in the present study along with the volatile-bearing minerals and their textural occurrence in each sample
Volatile-bearing minerals present and textural occurrence
C = possible cumulus phase; IC = intercumulus region; MI = melt-inclusion hosted; MH = mineral-hosted.
F-Cl apatite 40–60 mol% Cl in X-site; F-rich apatite <40 mol% Cl in X-site; Cl-rich apatite >60 mol% Cl in X-site.
Apatite was identified; however, it was too small to analyze.
Apatites, amphiboles, and pyroxenes in the nakhlite and chassignite meteorites were analyzed using a JEOL JXA 8200 electron microprobe in the Institute of Meteoritics at the University of New Mexico. An accelerating voltage of 15 kV and a nominal beam current of 25 nA were used during each analysis, following the procedure of McCubbin et al. (2010c). We analyzed the elements Si, Ce, Y, Fe, Mg, Ca, Na, P, F, and Cl in apatite and Si, Ti, Al, Cr, Mg, Fe, Mn, Ca, Na, K, F, and Cl in amphibole and pyroxenes. Fluorine was analyzed using a light-element LDE1 detector crystal, and Cl was analyzed using a PET detector crystal. The standards were as follows: Ca, P, F (natural fluorapatite from India Ap020 from McCubbin et al. 2012); F (synthetic SrF2, which was used both as an F standard and as an additional check on the F standardization from Ap020); Ca in amphibole only (synthetic diopside); Cl (sodalite, with scapolite check standard); Mn (spessartine); Mg (enstatite); Na, Al (albite); Ce (CePO4); Y (YPO4); Si, Fe (almandine); Cr (chromite); Ti (titanite); K (orthoclase). To reduce or eliminate electron beam damage, we used a 10 μm spot for standardization and 2–10 μm diameter beams for analysis of apatite grains in all the Martian samples. Tests on Durango apatite show that count rates for F are consistent for spot sizes down to about 2 μm (McCubbin et al. 2010c, 2011).
Hydroxyl cannot be measured directly by the EPMA technique; however, a missing component in the X-site of apatite can be calculated on the basis of stoichiometry. If both F and Cl are analyzed with sufficient accuracy (McCubbin et al. 2008a), this missing component can be attributed to some combination of the anions OH−, O2−, CO32−, S2−, Br−, and I− and/or structural vacancies (Pan and Fleet 2002) and/or structural H2O (Mason et al. 2009). The most likely constituent for this missing component in terrestrial and lunar igneous systems is OH− (Piccoli and Candela 2002; Boyce et al. 2010; McCubbin et al. 2010b; Greenwood et al. 2011), and secondary ion mass spectrometry measurements of OH− in Martian apatites indicate the same (Leshin 2000; Boctor et al. 2003; Guan et al. 2003; Greenwood et al. 2008; McCubbin et al. 2012).
Stormer et al. (1993) documented that fluorine and chlorine X-ray count rates change with time during electron microprobe analysis of apatite as a function of crystallographic orientation. Goldoff et al. (2012) presented a method to minimize changes in count rate for F and Cl; however, the apatites analyzed in the present study were too small to apply this technique. Accordingly, we monitored the apatite analyses for time-dependent count rates and discovered that our fluorine count rates were not always constant during the course of an analysis. Chlorine count rates, on the other hand, were found to be constant for all of our analyses, consistent with the findings of McCubbin et al. (2010c, 2011). We were able to monitor the fluorine X-ray counts in real time with the chart recorder function in the JEOL software. However, we could not use this information to construct count rate versus time plots, which are required for correcting fluorine X-ray count-rate variation. Therefore, any analyses that displayed highly variable fluorine X-ray count rates from our samples were rejected. Even with this criterion, some of the apatites contained more than their stoichiometric amount of F + Cl. In these cases, we were able to obtain only a minimum Cl:F ratio by assuming 1 − Cl = F, after McCubbin et al. (2010c, 2011). This calculation is justified because we did not observe variability in the chlorine X-ray count rates during any of the apatite analyses in the present study.
Secondary Ion Mass Spectrometry (SIMS)
The measurements of F, H2O (as OH−), Cl, and S were performed on a single Ti-biotite crystal and a single patch of maskelynite from the Chassigny meteorite using a Cameca 6f ion microprobe at the Department of Terrestrial Magnetism, Washington, DC using the procedure of Hauri et al. (2002) and McCubbin et al. (2010a). The focused (5–10 nA) 10 kV Cs+ primary ion beam was rastered on the sample to a 25 by 25 μm area. The secondary ion beam was extracted at −5 kV from a 5 μm diameter portion of the rastered area with a field aperture. An electron flood gun (−5 kV) was used to compensate for charge build-up in the analysis area. A mass resolution of approximately 6000 was tuned to eliminate the mass interferences between 17[OH] and 17O. Standardization on multiple basaltic glass compositions with a wide range of volatile contents was performed at the beginning of the session. Information concerning the compositions of the standards used (1833-1, 1846-12, 1833-11, WOK28-3, 519-4-1, 1654-3, B330, B333, B366) have been published previously (Deloule et al. 1995; Hauri et al. 2002, 2006). Each analysis lasted about 15 min.
Field Emission Scanning Electron Microscopy (FE-SEM)
Backscattered electron (BSE) images were acquired using either a JEOL JXA 8200 electron microprobe (University of New Mexico) or a JEOL JSM 6500F scanning electron microscope with a liquid-N2-cooled sapphire Si(Li) EDS detector (EDAX) using a 15 kV operating voltage (Geophysical Laboratory, Carnegie Institution, Washington, DC). The EDS and BSE images were used for two purposes in this study. The first was to identify the volatile-bearing minerals and textural relationship of those minerals with the surrounding phases in each of the meteorites investigated. The second was to characterize the composition, crystal morphology, and shape of the Ti-biotite grain that was to be analyzed by SIMS, so it could be positively identified for SIMS analysis in the 6f instrument. Importantly, electron beam exposure to the Ti-biotite was limited as much as possible to minimize potential dehydrogenation of the Ti-biotite crystal, as has been documented to occur for some amphiboles (Wagner et al. 2008).
Mineralogy and Petrology of the Chassignites and Nakhlites
The chassignites are cumulus igneous rocks from Mars that consist of two samples, Chassigny and Northwest Africa (NWA) 2737. These meteorites are both cumulate dunites (Mason et al. 1975; Floran et al. 1978; Wadhwa and Crozaz 1995; McSween and Treiman 1998; Beck et al. 2006; Nekvasil et al. 2007; Treiman et al. 2007) with cumulus olivine and chromite. Interstitial to the cumulus grains of olivine and chromite are low-Ca and high-Ca pyroxene, alkali-maskelynite, Cl-rich apatite, silica, baddeleyite, ilmenite, pyrrhotite, and pyrite (Mason et al. 1975; Floran et al. 1978; Johnson et al. 1991; Greenwood et al. 2000b; Beck et al. 2006; Nekvasil et al. 2007; Treiman et al. 2007; McCubbin and Nekvasil 2008; McCubbin et al. 2010a). Chassigny also contains maskelynite (i.e., plagioclase), but this phase is not present in the intercumulus regions of NWA 2737 (He et al. 2013).
Some of the cumulus olivines have polyphase holocrystalline inclusions (referred to hereafter as melt inclusions) ranging in diameter from <10 μm to >350 μm. The phases present in the melt inclusions include augite, low-Ca pyroxene, baddeleyite, kaersutite, pyrrhotite, Al-rich chromite, pentlandite, Ti-biotite, F-rich apatite, alkali maskelynite, silica, and ilmenite (Floran et al. 1978; Johnson et al. 1991; Watson et al. 1994; Greenwood et al. 2000b; Varela et al. 2000; Righter et al. 2002; Beck et al. 2006; Monkawa et al. 2006; Nekvasil et al. 2007; Treiman et al. 2007; McCubbin and Nekvasil 2008; McCubbin et al. 2010a). Similar to the intercumulus regions, Chassigny also contains maskelynite (i.e., plagioclase) within its olivine-hosted melt inclusions, but this phase is not present in the olivine-hosted melt inclusions of NWA 2737 (He et al. 2013). Chassigny also contains a low-temperature alteration mineral assemblage consisting of calcium carbonate, magnesite, and gypsum that appears unrelated to the igneous activity (Wentworth and Gooding 1994; Bridges et al. 2001).
The nakhlite meteorites consist of thirteen individual samples, Nakhla, Lafayette, Governador Valadares, NWA 817, NWA 998, NWA 5790, MIL 03346, MIL 090030, MIL 090032, MIL 090136, Yamato (Y) 000593, Y 000749, and Y 000802. The three Yamato samples are believed to be paired with each other (Imae et al. 2005), and the four Miller range samples are probably paired with one another (Hallis and Taylor 2011; Udry et al. 2012). Thus, the collection probably consists of eight unique specimens. The nakhlite meteorites are cumulate clinopyroxenites with cumulus clinopyroxene and olivine grains (Bunch and Reid 1975; Harvey and McSween 1992b; Wadhwa and Crozaz 1995; Dyar et al. 2005; Imae et al. 2005; Treiman 2005; Day et al. 2006; Imae and Ikeda 2007; Treiman and Irving 2008; Hallis and Taylor 2011). NWA 998 also has rare cumulus orthopyroxene that constitutes approximately 2% of the meteorite (Treiman 2005; Treiman and Irving 2008), and this is the only nakhlite to date for which cumulus orthopyroxene has been reported. Among the nakhlites, there are highly variable amounts of mesostasis between the cumulus grains (i.e., 7–34%; Mikouchi et al. 2012), and the crystallinity of the mesostasis is also variable (Lentz et al. 1999; Treiman 2005). The phases present in the mesostasis include skeletal titanomagnetite, fayalitic olivine, chromite, pyrrhotite, silica, apatite, zircon, baddeleyite, pyrite, chalcopyrite, potassic feldspar, and Fe-rich silicate glass (Bunch and Reid 1975; Greenwood et al. 2000a; Imae et al. 2005; Treiman 2005; Aoudjehane et al. 2006; Day et al. 2006; Treiman and Irving 2008).
Some of the cumulus clinopyroxenes and olivines contain complex polyphase melt inclusions that range in diameter from <10 μm to approximately 350 μm. These partially crystallized melt inclusions are reported to contain a large array of phases that include Al-Ti-rich augite, titanomagnetite, ilmenite, pigeonite, fayalitic olivine, Ti-Al spinel, pyrrhotite, cristobalite, tridymite, apatite, alkali feldspar, chalcopyrite, glass, Cl-rich amphibole, hematite, goethite, and jarosite (Treiman 1986, 1990, 1993; Harvey and McSween 1992a; Sautter et al. 2002, 2006; Dyar et al. 2005; Imae et al. 2005; Aoudjehane et al. 2006; Day et al. 2006; Imae and Ikeda 2007; McCubbin et al. 2009). However, not all of these phases are present in every inclusion, and some of the melt-inclusion minerals are specific to individual meteorites.
The nakhlites also have low-temperature hydrous alteration phases that are likely Martian in origin. These minerals include Cl-poor smectite, iddingsite, iron oxy-hydroxides, halite, siderite, epsomite, anhydrite, gypsum, jarosite, and laihunite (Reid and Bunch 1975; Gooding et al. 1991; Treiman et al. 1993; Romanek et al. 1998; Bridges and Grady 1999, 2000; Bridges et al. 2001; Gillet et al. 2002; Treiman 2005; Herd 2006b; Sautter et al. 2006; Vicenzi et al. 2007; Noguchi et al. 2009; Changela and Bridges 2010; Bridges and Schwenzer 2012).
Although our primary focus in the present study was volatile-bearing minerals, we also investigated the mineralogy and textural association of the other minerals within the chassignites and nakhlites so that we could place the volatile-bearing mineralogy into a broader petrologic context. Our findings for the mineralogy of the nakhlites and chassignites were broadly consistent with the many previous investigations of these meteorites, with a few important exceptions that will be presented below. In addition, we collected EPMA data on intercumulus and melt inclusion-hosted pyroxenes from Chassigny and cumulus pyroxenes from Nakhla, Governador Valadares, MIL 03346, NWA 998, and Lafayette (Tables S10–S18).
Our investigation of the chassignites was broadly consistent with previous reports of the petrology, mineralogy, and petrography of these meteorites. Consequently, we focus here only on the textural occurrence and geochemistry of the volatile-bearing mineral phases.
Textural Occurrence of Apatite in Chassignites
Both of the chassignites investigated in the present study contained apatite. Apatite is present in the late-stage mesostasis and is typically hosted by maskelynite or alkali-maskelynite. It is also present within olivine-hosted polyphase melt inclusions in both chassignites (consistent with the findings of McCubbin and Nekvasil 2008; see Figs. 1a and 1b for NWA 2737). Apatite grain sizes in the chassignites range from submicrometer to 40 μm in the shortest dimension and grain shapes varied from euhedral to anhedral (Figs. 1a and 1b). Apatite is the only phosphate present within the chassignites, making them one of the only achondrite groups from the solar system that does not contain the volatile-free phosphate merrillite (Jolliff et al. 1993, 2006; Hughes et al. 2006, 2008; Patiño Douce and Roden 2006; Shearer et al. 2011).
Crystal Chemistry of F, Cl, and OH in Chassignite Apatites
We did not analyze any new apatite grains in the Chassigny meteorite because our group has previously studied the apatite chemistry in Chassigny (McCubbin and Nekvasil 2008). Therefore, the only chassignite for which we analyzed apatite in the present study is NWA 2737. Apatite in NWA 2737 varied in composition primarily along the F-Cl join within the F-Cl-OH ternary. A representative electron microprobe analysis of apatite from NWA 2737 is presented in Table 2, and a complete data set of all apatite analyses from NWA 2737 is available in Table S1. The volatile contents of apatites from both are presented in the F-Cl-OH ternary diagram in Fig. 2a.
Table 2. Electron microprobe analyses of selected intercumulus apatites from nakhlites and chassignites
Apatites from the Chassigny meteorite have been previously investigated by McCubbin and Nekvasil (2008), but we reiterate the results of that study here for direct comparison with NWA 2737. The apatite grains in Chassigny have two compositional populations that correspond to the different textural occurrences (McCubbin and Nekvasil 2008). With respect to the apatite X-site, apatites within olivine-hosted melt inclusions in Chassigny have chlorine abundances ranging from about 5 to 24 mol%, fluorine abundances ranging from about 76 to 95 mol%, and missing component abundances ranging from 0 to 6 mol% (McCubbin and Nekvasil 2008). Intercumulus apatite grains in Chassigny have chlorine abundances ranging from about 34 to 63 mol%, fluorine abundances ranging from about 33 to 63 mol%, and missing component abundances ranging from 0 to 11 mol% (McCubbin and Nekvasil 2008). Only intercumulus apatite grains in NWA 2737 were analyzed by EPMA (due to the small size of apatites within the melt inclusions), and those apatites have chlorine abundances ranging from about 26 to 44 mol%, fluorine abundances ranging from about 51 to 74 mol%, and missing component abundances ranging from 0 to 12 mol%. We also conducted several semiquantitative energy dispersive spectroscopic analyses of apatites within NWA 2737, and we determined that the Cl peak of the apatites in olivine-hosted melt inclusions were smaller than the Cl peaks of intercumulus apatites (determined by semiquantitative EDS analysis), indicating the same bimodal chemistry of apatite that was observed in the Chassigny meteorite.
Textural Occurrence of Volatile-Bearing Silicates in Chassignites
Kaersutite and Ti-biotite have been previously reported in both chassignites (Johnson et al. 1991; Watson et al. 1994; Righter et al. 2002; Boctor et al. 2003; Monkawa et al. 2006; McCubbin and Nekvasil 2008; McCubbin et al. 2010a; He et al. 2013). In addition, Cl-H2O-bearing maskelynite has been reported in the Chassigny meteorite (Boctor et al. 2003; McCubbin and Nekvasil 2008). All three volatile-bearing silicate phases occur within olivine-hosted melt inclusions, but none appear in the intercumulus regions of the chassignites. We did not observe any volatile-bearing silicates in NWA 2737 in the present study, although they have been reported in NWA 2737 by others (Beck et al. 2006; He et al. 2013), so we assume that the limited amount of NWA 2737 sample in our study led to a sampling bias. Consequently, we report observations made from the Chassigny meteorite.
Kaersutite is typically subhedral to anhedral, ranging in size from <5 μm to approximately 50 μm (Johnson et al. 1991; McCubbin et al. 2010a). The kaersutite typically coexists with high-Ca and low-Ca pyroxene as well as maskelynite. The Ti-biotite is typically subhedral, approximately 10–15 μm in size, and is rarer than the kaersutite (Fig. 3a), but it is typically associated with the same minerals as the kaersutite. The Cl-H2O-bearing maskelynite (Fig. 3b) typically has feldspar stoichiometry and was reported to result from fluid inclusions, initially trapped within feldspar, that dissolved into the diaplectic glass during maskelynitization of the feldspar (McCubbin and Nekvasil 2008). The Cl-rich maskelynite typically fills the space interstitial to the other phases within the olivine-hosted melt inclusions (see Table 4).
Crystal Chemistry of Volatile-Bearing Silicate Minerals in Chassigny
Kaersutite from the Chassigny meteorite has been previously investigated by McCubbin et al. (2010a), but we will briefly present the results of that study here for direct comparison with the other volatile-bearing silicates in Chassigny. Kaersutite in the Chassigny meteorite has been analyzed previously by EPMA for major elements (Floran et al. 1978; Johnson et al. 1991; McCubbin et al. 2010a), by X-ray absorption near-edge spectroscopy (XANES) for Fe3+/∑Fe (Monkawa et al. 2006), and by SIMS for F, Cl, and OH− (Watson et al. 1994; McCubbin et al. 2010a). A complete crystal chemical model for the Chassigny kaersutite has been presented and proper stoichiometry attained using the data collected in the above studies (McCubbin et al. 2010a). Kaersutite is a Ti-rich amphibole that can have up to one monovalent structural formula unit (sfu) in the O(3) site. The O(3) site consists of two structural formula units, but one of those units consists of oxygen to charge balance the elevated Ti4+ contents in the kaersutite structure (Leake 1978; Popp and Bryndzia 1992; Popp et al. 1995, 2006; Leake et al. 1997, 2004; Burke and Leake 2005). The O(3) site of the kaersutite in the Chassigny melt inclusions is apportioned as follows: OH− occupies approximately 28% of the site (0.56 sfu OH−), F− occupies approximately 12% of the site (0.24 sfu F), Cl− occupies 1.5% of the site (0.03 sfu Cl), and O2− occupies approximately 58.5% of the site (1.17 sfu O) (McCubbin et al. 2010a).
The Ti-biotite in the Chassigny meteorite has been analyzed previously by EPMA for major elements plus fluorine and chlorine (2.08–2.3 wt% F, and 0.4 wt% Cl; Johnson et al. 1991; Righter et al. 2002) and by SIMS for OH− (0.5 wt% H2O; Watson et al. 1994). However, OH−, F, and Cl had not been simultaneously analyzed for Chassigny biotite in a single analysis spot; therefore, we analyzed a Ti-biotite from Chassigny for F, Cl, and OH− simultaneously using SIMS. Only one Ti-biotite grain was encountered during the present study, and that Ti-biotite contains 1.24 (3) wt% H2O, 0.72 (2) wt% F, and 0.03(0) wt% Cl, which deviates significantly from the other values that have been reported (parenthetical numbers represent the error in the last digit of the value and were calculated using the 2-sigma standard deviations of the slope uncertainty for the regression lines used in the SIMS calibrations). Despite the deviation in absolute values, the molar sum of F, Cl, and OH in both sets of analyses for the Ti-biotites are similar (difference of <0.04 sfu), indicating the same amount of oxy-component in the biotite. Because the amount of oxy-component in biotite is structurally controlled by the Ti-content (Henry et al. 2005), the range of reported volatile abundances may be representative of the compositional range of volatiles in the Chassigny Ti-biotites. We conducted the SIMS analysis of Ti-biotite before conducting EPMA analyses so that we could limit the beam exposure to the Ti-biotite prior to SIMS analysis. Subsequent to SIMS analysis, we inspected the SIMS pit to verify that we did not drill completely through the Ti-biotite grain, which we did not. However, after polishing and recoating, the grain was gone, so we could not get an EPMA analysis of the Ti-biotite grain we analyzed by SIMS. The molar proportions of F, Cl, and OH− in both the kaersutite and Ti-biotites from the Chassigny melt inclusions have been plotted in the F-Cl-OH ternary diagram in Fig. 4.
The Cl-H2O-bearing maskelynite present in olivine-hosted melt inclusions in the Chassigny meteorite have been previously analyzed by SIMS for H2O (Boctor et al. 2003) and EPMA for Cl (Varela et al. 2000). Our EPMA analyses of the Cl-bearing maskelynite will be presented in Table 4. The SIMS analysis of a single Cl-H2O-bearing maskelynite from a kaersutite-bearing melt inclusion shows 0.74 (2) wt% H2O, 0.0439 (2) wt% F, and 0.247(1) wt% Cl. These results are broadly consistent with the results of Boctor et al. (2003), Varela et al. (2000), and those presented in Table 4, although our water content is slightly higher than those reported previously. The Cl-H2O-bearing maskelynite analyzed in the present study ranged from 1500–2500 ppm Cl, 0–400 ppm F, and 7400 ppm H2O.
Although our investigation of the nakhlites was broadly consistent with previous reports of the major petrology, mineralogy, and petrography of these meteorites, we report here the first occurrence of cumulus orthopyroxene in the Lafayette meteorite (Table S10). In addition, apatite is the only phosphate that is present within the nakhlites, as there is no report of merrillite in the literature, and no merrillite was observed during the present study. Other than these two new observations, we focus here on the textural occurrence and geochemistry of the volatile-bearing mineral phases in the nakhlites.
Texture and Crystal Chemistry of Volatile-Bearing Minerals in the Nakhlites
Apatite crystal morphology and grain size vary substantially from one sample to the next, although apatite compositions primarily varied along the F-Cl join within the F-Cl-OH ternary. These findings are broadly consistent with previous reports of nakhlite apatites (Bunch and Reid 1975; Treiman 2005; Treiman and Irving 2008). Apatite is present in all nakhlite samples investigated and typically occurs both in melt inclusions and in the mesostasis between cumulus grains. Only intercumulus apatite grains were large enough to be analyzed by EPMA. Selected electron microprobe analyses of apatites from the nakhlites are presented in Table 2, and complete data tables of all nakhlite apatite analyses from the present study are available in Tables S2–S8. The volatile contents of all the nakhlite apatites analyzed in the present study are presented in the F-Cl-OH ternary diagram in Fig. 2b. All of the apatites in the MIL nakhlites, NWA 817, and NWA 5790 were too small to analyze without significant overlap from surrounding phases, so all the analyses from these samples included overlap with a groundmass glass component. The groundmass glass was subtracted from each of the analyses using the same method described in detail by McCubbin et al. (2009). Importantly, the mesostasis material in each of the meteorites did not contain measurable fluorine or chlorine, so the F:Cl ratio of each apatite analysis was not affected by this phase subtraction.
Cl-rich amphiboles are the other volatile-bearing magmatic mineral in the nakhlites, but they are not present in every sample. When they are present, they exclusively occur in mineral-hosted melt inclusions. These findings are broadly consistent with previous reports of nakhlite amphiboles (Sautter et al. 2006; McCubbin et al. 2009). Selected electron microprobe analyses of amphiboles from the nakhlites are presented in Table 3, and complete data tables of all nakhlite amphibole analyses from the present study are available in Table S9. The volatile contents of all the nakhlite amphiboles analyzed in the present study are presented in the F-Cl-OH ternary diagram in Fig. 4.
Table 3. Selected electron microprobe analyses of Cl-rich amphiboles in the MIL nakhlites and NWA 5790
All amphibole analyses from the present study, along with calculated structural formulae and amphibole name are available online in Table S9.
Apatites are subhedral to anhedral and are typically associated with late-stage mesostasis consisting of silica, maskelynite, and alkali-maskelynite (Fig. 1c). Apatites range from submicrometer to approximately 6 μm in the shortest dimension. Chlorine abundances in the apatites range from about 8 to 78 mol%, fluorine abundances range from about 17 to 90 mol%, and missing component abundances range from 0 to 33 mol%. Furthermore, the fluorine and chlorine abundances of apatites in Governador Valadares are highly variable, and apatite compositions plot fairly continuously along the F-Cl join. Apatite was the only volatile-bearing igneous mineral identified in Governador Valadares.
Apatites are subhedral to anhedral, and they are typically associated with maskelynite and alkali-maskelynite without silica. Apatites are occasionally included within maskelynite patches (Fig. 1d). The apatites typically range from submicrometer to approximately 15 μm in the shortest dimension. Apatites within Lafayette have chlorine abundances ranging from about 30 to 88 mol%, fluorine abundances ranging from about 11 to 71 mol%, and missing component abundances ranging from 0 to 13 mol%. In addition, the fluorine and chlorine abundances of apatites in Lafayette are highly variable, and apatite compositions plot fairly continuously along the F-Cl join. Apatite was the only volatile-bearing igneous mineral identified in Lafayette.
Apatites are subhedral to anhedral and are typically associated with late-stage mesostasis consisting of silica, maskelynite, and alkali-maskelynite (Fig. 1e). The apatites range from submicrometer to approximately 10 μm in the shortest dimension. Apatites have chlorine abundances ranging from about 24 to 75 mol%, fluorine abundances ranging from about 12 to 76 mol%, and missing component abundances ranging from 0 to 14 mol%. The fluorine and chlorine abundances of apatites in Nakhla are highly variable, and apatite compositions plot fairly continuously along the F-Cl join. Apatite was the only volatile-bearing igneous mineral identified in Nakhla.
Apatites range in shape from euhedral to anhedral and range in size from approximately 5 μm to 50 μm in the shortest dimension. Apatites are typically associated with maskelynite and titanomagnetite (Fig. 1f), although there is also alkali-maskelynite in contact with some apatites. Apatites are also observed as inclusions within large (>80 μm) titanomagnetite grains (Fig. 1g), and some of them are closely associated with cumulus clinopyroxene and orthopyroxene. Apatites in NWA 998 have chlorine abundances ranging from about 37 to 54 mol%, fluorine abundances ranging from 29 to 55 mol%, and missing component abundances ranging from 0 to 30 mol%. Apatites from this sample showed little variation along the F-Cl join, and most of the variation was within the F-OH join of the apatites. Apatite was the only volatile-bearing igneous mineral identified in NWA 998.
Paired MIL Nakhlites (MIL 03346, MIL 090030, MIL 090032, and MIL 090136)
Apatites are highly acicular and typically subhedral to anhedral. The grains occur in the glassy mesostasis of the meteorites and are typically 2 μm or less in the shortest dimension with aspect ratios ranging from 10:1 to 50:1 (Fig. 1h). The mesostasis minerals associated with apatite include silica, fayalitic olivine, and an Fe-rich glass that may represent a fine intergrowth of maskelynite, alkali maskelynite, and submicrometer fayalitic olivines. Subsequent to the subtraction of the glass component from the apatite analyses from the MIL nakhlites, we determined that the apatites have chlorine abundances ranging from about 9 to 27 mol% of the X-site, fluorine abundances ranging from about 73 to 91 mol% of the X-site, and no detectable missing component in the X-site that could be attributed to OH−. Apatites from the MIL nakhlites showed little variation along the F-Cl join, and the position of the apatite compositional cluster is located close to the fluorapatite apex of the ternary.
The Cl-rich amphiboles in the MIL nakhlites occur within clinopyroxene-hosted and olivine-hosted melt inclusions as anhedral to euhedral crystals that range in size from <5 μm to approximately 10 μm. However, in the present study, amphibole is observed exclusively in clinopyroxene-hosted melt inclusions. In addition, the Cl-rich amphibole is typically associated with magnetite to titanomagnetite, pyrrhotite, and an Fe-Si rich glass (Figs. 3c and 3d), although it has also been found in association with hydrothermal jarosite and hematite in MIL 03346 (McCubbin et al. 2009). The composition of the Cl-rich amphiboles within the olivine-hosted and pyroxene-hosted melt inclusions in MIL 03346 have been discussed in detail by McCubbin et al. (2009) and Sautter et al. (2006). Furthermore, the compositions of the amphibole obtained in the present study are broadly consistent with analyses of Cl-rich amphibole from MIL 03346 (Sautter et al. 2006; McCubbin et al. 2009). Selected EPMA analyses from the present study are presented in Table 3, and all the amphibole analyses from the MIL nakhlites collected during the present study, along with the calculated structural formulae and IMA name, are available in Table S9.
Apatites are highly acicular and typically subhedral to anhedral. The grains occur in the glassy mesostasis of the meteorites and are typically 2 μm or less in the shortest dimension with aspect ratios ranging from 10:1 to 50:1. The mesostasis minerals associated with apatite include silica, fayalitic olivine, and an Fe-rich glass that may represent a fine intergrowth of maskelynite, alkali maskelynite, and submicrometer fayalitic olivines. Subsequent to the subtraction of the glass component from the apatite analyses from NWA 817, we determined that the apatites have chlorine abundances ranging from 15 to 29 mol%, fluorine abundances ranging from about 71 to 85 mol%, and missing component abundances ranging from 0 to 6 mol%. Apatites from NWA 817 showed little variation along the F-Cl join, and the position of the apatite compositional cluster is located close to the fluorapatite apex of the ternary. Apatite was the only volatile-bearing igneous mineral identified in NWA 817.
Apatites are highly acicular and typically subhedral to anhedral. The grains occur in the glassy mesostasis of the meteorites and are typically 2 μm or less in the shortest dimension with aspect ratios ranging from 10:1 to 50:1 (Fig. 1i). The mesostasis minerals associated with apatite include silica, fayalitic olivine, and an Fe-rich glass that may represent a fine intergrowth of maskelynite, alkali maskelynite, and submicrometer fayalitic olivines. Subsequent to the subtraction of the glass component from the apatite analyses, we determined that apatites in NWA 5790 have chlorine abundances ranging from 12 to 26 mol%, fluorine abundances ranging from about 74 to 84 mol%, and missing component abundances ranging from 0 to 7 mol%. Apatites from NWA 5790 showed little variation along the F-Cl join, and the position of the apatite compositional cluster is located close to the fluorapatite apex of the ternary.
The Cl-rich amphiboles in NWA 5790 occur within clinopyroxene-hosted melt inclusions as anhedral crystals that range in size from <5 μm to approximately 10 μm. The Cl-rich amphibole is typically associated with titanomagnetite, pyrrhotite, and an Fe-Si rich glass (Figs. 3e and 3f). The compositions of the Cl-rich amphiboles within the pyroxene-hosted melt inclusions in NWA 5790 are broadly consistent with analyses of Cl-rich amphibole from MIL 03346 (Sautter et al. 2006; McCubbin et al. 2009). A selected EPMA analysis of amphibole from NWA 5790 is presented in Table 3, and all the amphibole analyses from this sample, along with the calculated structural formulae and IMA name, are available in Table S9.
Previous investigations of the nakhlites and chassignites have generally considered them as two petrologically distinct subgroups of Martian meteorites. Although open-system processes have been called upon for the chassignite crystallization history (Wadhwa and Crozaz 1995; McCubbin and Nekvasil 2008), closed-system evolution of the nakhlite magmas is typically called upon (Wadhwa and Crozaz 1995; Treiman 2005; Day et al. 2006; Udry et al. 2012). In the subsequent discussion, we develop a methodology to understand apatite volatile chemistry in the nakhlites and chassignites based on experimentally determined behavior of apatite-melt partitioning. This partitioning data, coupled with an understanding of chloride solubility in silicate liquids, are then used with textural observations to develop the hydrothermal history of the chassignite and nakhlite magmatic system. We then present links between the volatile chemistry of phases from both the chassignites and nakhlites, and we point out other geochemical and textural similarities that support the chassignites and nakhlites as one comagmatic subgroup of Martian meteorites. We then present a model for crystallization of the chassignite-nakhlite magma body as a shallow sill or intrusion and explain many of the sequential variations observed within the cumulate pile in the context of the hydrothermal fluid history recorded in the minerals.
Apatite/Silicate-Melt Partitioning of F, Cl, and OH and Melt Cl− Solubility
To use the volatile concentrations of the mineral apatite to precisely determine the abundances of volatiles in coexisting silicate melt or fluids, thermodynamic models for the apatite solid solution and for the apatite components in multicomponent silicate melts are required. Although some thermodynamic models for apatite have been developed (Candela 1986; Tacker and Stormer 1989, 1993; Zhu and Sverjensky 1991; Hovis and Harlov 2010), they are incomplete. Furthermore, a complete mixing model for apatite components in silicate melts and fluids is not available. However, several experimental studies have determined the apatite-melt and apatite-fluid partitioning behavior of F and Cl in terrestrial and Martian systems (Zhu and Sverjensky 1991; Brenan 1993; Mathez and Webster 2005; McCubbin et al. 2008b; Webster et al. 2009; Huh et al. 2011; Vander Kaaden et al. 2012), and although these are likely to be compositionally dependent (Mathez and Webster 2005; Webster et al. 2009; Vander Kaaden et al. 2012), they can be used to give a first-order approximation of the relative abundances of F and Cl in a melt or fluid that co-existed with apatite. It is important to note that F, Cl, and OH are essential structural constituents in the mineral apatite (Hughes et al. 1989, 1990; Hughes and Rakovan 2002; Pan and Fleet 2002; Piccoli and Candela 2002), so the partitioning behavior of the volatiles should deviate substantially from Henrian behavior of trace elements. Similar to the partitioning behavior of Fe and Mg between olivine and silicate liquids (Roeder and Emslie 1970; Toplis 2005; Filiberto and Dasgupta 2011), the partitioning behavior of F, Cl, and OH between apatite and silicate melt are best described by an exchange coefficient (Equation (1)) (Webster et al. 2009; Huh et al. 2011; Vander Kaaden et al. 2012)
where is the wt% F in the melt, is the wt% F in the apatite, is the wt% Cl in the melt, and is the wt% Cl in apatite. These exchange coefficients change as a function of melt composition (Webster et al. 2009; Vander Kaaden et al. 2012), but they may also change as a function of pressure, temperature, apatite composition, and oxygen fugacity. Therefore, the available data on apatite-melt partitioning should be considered very limited. From the data that are available, it is clear that at pressures ≤1.0 GPa and temperatures ≤ 1050 °C, fluorine is preferred in apatite over chlorine, and chlorine is preferred over hydroxyl (Mathez and Webster 2005; McCubbin et al. 2008b, 2013; Webster et al. 2009; Huh et al. 2011; Vander Kaaden et al. 2012). Furthermore, for the F-Cl apatite system that is most relevant to nakhlites and chassignites, for basaltic liquids (Vander Kaaden et al. 2012) and for rhyodacitic liquids (Webster et al. 2009). Vander Kaaden et al. (2012) also reported and for the basaltic system they investigated.
Relative Abundances of F, Cl, and OH From Apatite in Basaltic Systems
Using the apatite-melt partition coefficients for basaltic systems from Mathez and Webster (2005), McCubbin et al. (2011) constructed a diagram for determining the relative F, Cl, and H2O abundances of an apatite-saturated basaltic melt, based on the fluorine, chlorine, and hydroxyl contents of the apatite crystallized from that melt. Now that additional apatite-melt partitioning data have become available, we have modified that diagram (Fig. 5a) to reflect all of the available apatite-melt partitioning data in basaltic systems (Mathez and Webster 2005; McCubbin et al. 2008b; Vander Kaaden et al. 2012). McCubbin et al. (2011) also added an additional boundary that represents the upper limit of chlorine that can be incorporated into an apatite from a chloride-undersaturated basaltic melt, based on the upper limit of chloride solubility in basalts determined by Webster and De Vivo (2002) (approximately 4.5 wt%). However, more recent apatite-melt partitioning data indicate that chlorapatite is a stable magmatic phase, and there is no fluid-only field near the Cl-apex of the apatite ternary (Vander Kaaden et al. 2012). Consequently, we do not include any fluid-only fields within Fig. 5a, but importantly, this does not preclude the possibility that the apatite in the nakhlites and chassignites formed in the presence of a fluid phase.
Relative Abundances of F, Cl, and OH From Apatite in Rhyolitic Systems
Much of the apatite in the present study coexists with phases common in evolved magmas, including 2-feldspars (i.e., plagioclase + alkali-feldspar) and silica; therefore, basaltic systems may not always be applicable to interpreting the apatite volatile contents of the chassignites and nakhlites. Consequently, we have modified the diagram in Fig. 5a specifically for rhyolitic-rhyodacitic systems (Fig. 5b). To construct this diagram, we used the apatite-melt exchange coefficients for F and Cl in rhyolitic to rhyodacitic melt compositions determined by Webster et al. (2009) (i.e., ). Webster et al. (2009) did not report the apatite-melt partitioning of H2O, although it was present within their experiments. Furthermore, based on the information reported by Webster et al. (2009), the DH2Oapat/rhyolite−rhyodacite was << 1 for all of the experiments, which is consistent with the values reported by Vander Kaaden et al. (2012) and Huh et al. (2011). Furthermore, Huh et al. (2011) investigated a Cl-free, F- and H2O-bearing andesitic haplobasalt and found similar values to those reported by Vander Kaaden et al. (2012) for basaltic compositions (i.e., ) so we decided to use this value for Fig. 5b. Because much of the change in apatite composition for the nakhlites and chassignites occurs primarily along the F-Cl join, the second diagram is used primarily to illustrate the shift in apatite F-Cl-OH ternary space for Cl-dominated melts in more evolved silica-rich systems.
Modeled Chloride Solubility in Proposed Chassignite and Nakhlite Liquids
Chloride solubility in silicate liquids changes largely as a function of composition, and it decreases with increasing silica content (Webster 1992, 1997, 2004; Webster and Rebbert 1998; Webster et al. 1999; Webster and De Vivo 2002; Aiuppa et al. 2009). Webster and De Vivo (2002) have shown that typical basaltic liquids can have a maximum of about 4.5 wt% Cl, while rhyolitic magmas have less than 0.5 wt%. To provide a first-order idea of the Cl concentrations required to reach chloride saturation in chassignite and nakhlite magmas, we have modeled the chloride solubility of several liquid compositions that are relevant to these systems. The Gusev basalts Humphrey and Backstay (analyzed by the MER Spirit; Gellert et al. 2006; McSween et al. 2006) have been proposed as possible parent liquid compositions for the Chassigny melt inclusions (Filiberto 2008; Nekvasil et al. 2009). Humphrey requires approximately 4.16 wt% Cl and Backstay, which is more evolved, requires 2.65 wt% Cl to saturate in a chloride-rich fluid phase. As discussed above, the originally basaltic parental liquids of the chassignites and nakhlites evolved to fairly silica-rich compositions based on the association of apatite with 2-feldspars (i.e., plagioclase + alkali-feldspar) and silica in many of the samples studied; therefore, it is important to understand how the chloride-solubility in the chassignite and nakhlite liquids changed as the melt evolved. Consequently, we have computed the chloride solubility for the intercumulus glass in the nakhlite MIL 03346 using the glass composition in Sautter et al. (2006) and the Cl-solubility model of Webster and De Vivo (2002). This glass would be chloride-saturated at approximately 1.36 wt% Cl, which is appreciably lower than the chloride-saturation values for the proposed basaltic parental liquids.
Magmatic Versus Hydrothermal Origin for Phases in the Chassignites and Nakhlites
Textural and Chemical Similarities Among Apatite Populations in the Chassignites and Nakhlites
Apatite data from the present study have been plotted on the relative volatile abundance diagrams from Figs. 5a and 5b (Figs. 6a–d). Although apatites in the chassignites are fairly similar to each other, apatite in the nakhlites can be grouped into three distinct chemical/textural populations. The first population consists of apatites from a single meteorite, NWA 998. Apatites from NWA 998 do not vary greatly in F:Cl ratio and cluster near the middle of the F-Cl join. Furthermore, the apatites are generally coarser grained than other nakhlite apatites and occur as mineral inclusions in titanomagnetites. Apatites from Nakhla, Lafayette, and Governador Valadares make up the second population. Apatites from these samples vary greatly in F:Cl ratio and have similar crystal shapes, sizes, and surrounding mineralogy that typically includes at least two feldspars (i.e., plagioclase and alkali-feldspar) and, in two out of three cases, silica. Apatites from the MIL nakhlites, NWA 817, and NWA 5790 make up the third population. Apatites from these samples do not vary greatly in F:Cl ratio and tend to be F-rich. The apatites in these samples are highly acicular and have the same surrounding phases that include fayalitic olivine, Fe-rich glass, two feldspars, and silica. Because it is difficult to determine definitively whether the apatite equilibrated with a basaltic or rhyolitic liquid, we have plotted all the data on both diagrams from Figs. 5a and 5b for comparison, and the relevance of each is discussed below.
Magmatic Versus Hydrothermal Apatite in the Chassignites
McCubbin and Nekvasil (2008) reported that addition of an exogenous Cl-rich fluid to the intercumulus regions of the Chassigny meteorite above 700 °C is required to explain a number of observations, including the Cl-enrichment of intercumulus apatites compared to the F-rich apatite compositions in coexisting olivine-hosted melt inclusions. Using the relative volatile abundance diagrams from Fig. 5, which were not available during the McCubbin and Nekvasil (2008) study, we will be able to provide further insight into any fluids present within both textural regimes (intercumulus and melt inclusion) of Chassigny. The apatites in both textural regimes are typically associated with two feldspars and sometimes, silica. Consequently, the rhyolitic relative volatile abundance diagram is most relevant for this system, although the actual silicate liquid may have been less evolved than rhyolite. All of the intercumulus apatites from Chassigny and NWA 2737 require formation and/or equilibration with a melt or fluid dominated by Cl over F (Fig. 6d). However, some of the apatite compositions within the olivine-hosted melt inclusions equilibrated with a melt that had more F than Cl, although many compositions equilibrated with a Cl-rich melt or fluid (Fig. 6d). It is important to note that the melt inclusion liquid did saturate in a Cl-bearing fluid phase based on the elevated H2O and Cl contents of coexisting maskelynite (Varela et al. 2000; Boctor et al. 2003; McCubbin and Nekvasil 2008; McCubbin et al. 2010a; this study Table 4) that is beyond what is typically incorporated into the feldspar crystal structure (Nakano et al. 2001; Johnson and Rossman 2004; Johnson 2006). It seems that NWA 2737 did not experience Cl-enrichment in the intercumulus regions to the extreme degree recorded by Chassigny, although semiquantitative energy dispersive spectroscopy of melt inclusion apatites showed a smaller Cl-peak than intercumulus apatites, indicating that Cl-enrichment of the intercumulus regions of NWA 2737 did occur.
Table 4. Electron microprobe analyses of Cl-bearing alkali-maskelynite in olivine-hosted melt inclusions in Chassigny
Although the presence of other Cl-rich minerals in the intercumulus regions of the chassignite meteorites may seem to be a prerequisite for Cl-rich metasomatism, the Mg-rich nature of the cumulus dunites (Fo68–78; Floran et al. 1978; Beck et al. 2006) would have prevented the formation of some Cl-rich silicates. Specifically, Mg-rich amphiboles and micas exclude Cl from their structure due to the large ionic radius of the chloride ion (referred to as Mg-Cl avoidance principle; Munoz 1984; Morrison 1991). Therefore, chloride-dominated fluids alone would not stabilize these phases. Additionally, sodalite is not present within the assemblage; however, based on the intercumulus mineral assemblage (presence of two feldspars and silica), the nepheline activity of the system was likely less than 1 (however, no corundum is present in the assemblage), which would inhibit sodalite formation (Sharp et al. 1989) even if the activity of NaCl was 1.
Magmatic versus Hydrothermal Apatite in the Nakhlites
Apatites in the nakhlites were divided into three populations based on similarities in texture and compositions shared among some of the nakhlite meteorites. Population 1 consists of apatites from NWA 998. Apatites within this meteorite display limited variability in F-Cl ratio, are rich in both F and Cl, and have a detectable missing component that can be attributed to OH−. Although some apatites in NWA 998 are in contact with alkali-rich (evolved) feldspars, they do not appear to have formed with the evolved feldspar in the late-stage pockets interstitial to the cumulus phases. In fact, apatites in this meteorite are texturally similar and of similar size to the cumulus pyroxene phases in the rock, and some of the apatite grains may be cumulus. Furthermore, apatites occur as mineral inclusions within titanomagnetite, indicating apatite may have formed relatively early in the paragenetic sequence of the parent liquid. Therefore, the basaltic relative volatile abundance diagram from Fig. 5 is likely more relevant for apatite population 1. Population 1 apatites plot within the fields indicating more Cl and H2O than F in the silicate liquid from which apatites formed (Fig. 6a). However, the elevated missing components of apatite in NWA 998 may not be due to hydroxyl because SIMS analyses from NWA 998 for H2O indicate only 0.07–0.11 wt% H2O (Channon et al. 2011). Importantly, there is no detectable difference between the volatile abundances of apatites included within titanomagnetite grains and those that are not mineral-hosted. This observation is evidence that the apatites of subpopulation 1 are magmatic as opposed to hydrothermal because mineral-hosted apatites would probably be shielded from a secondary metasomatic fluid (e.g., McCubbin and Nekvasil 2008). Importantly, even if population 1 is magmatic in origin, the elevated chlorine contents could be the result of secondary processes involving a Cl-rich fluid interacting with the parent magma prior to apatite saturation. Regardless of which apatite formation process occurred, elevated Cl relative to F and H2O in NWA 998 was either a characteristic of the parental magma or was a characteristic acquired early before extensive crystallization of the parent magma occurred.
Population 2 consists of apatites from Nakhla, Governador Valadares, and Lafayette, which display extreme variability in F-Cl ratio that cannot be explained by any typical igneous processes (Meurer and Boudreau 1996), as fractionation of volatile-free silicates and oxides will not appreciably change the F-Cl ratio of an evolving magma. Furthermore, most of the apatite compositions require equilibration with a melt dominated by Cl over F and H2O regardless of whether they formed from rhyolitic or basaltic systems (Figs. 6a and 6b). The variability in apatite compositions can be attributed to a number of secondary processes ranging from incomplete equilibration with metasomatic fluids to spinodal decomposition of F-Cl apatite during metasomatism, which has been observed in some pegmatitic systems on Earth (White et al. 2005). However, because apatite occurs all along the F-Cl binary in the F-Cl-OH ternary system, we rule out spinodal decomposition of the apatites as a cause for the variability. The large variability in F-Cl ratio of population 2 indicates that the F-Cl ratio of the fluid or melt from which they formed was variable over the time span of apatite formation. The most likely mechanism that would cause a change in F-Cl ratio is either open-system degassing, which will cause preferential loss of Cl to the fluid and retention of F in the melt (Aiuppa et al. 2009; Ustunisik et al. 2011), or fluid addition which could drive the ratio in either direction. However, given the higher affinity for Cl in a fluid phase, fluid addition would probably drive the apatite compositions to more Cl-rich values. If degassing occurred during the formation of population 2, then the most Cl-rich apatites would have formed prior to loss of a Cl-rich fluid, and the more F-rich apatites would have formed as the Cl-rich fluid was lost. If the apatites formed during fluid addition to the system, the more F-rich apatites would likely be the earliest formed apatites with the Cl-rich compositions forming subsequent to fluid addition. Discriminating between these two mechanisms cannot be done based on textural and chemical data alone; therefore, it will be discussed later in the context of viable petrogenetic models for the chassignites and nakhlites.
Population 3 consists of apatites from the MIL nakhlites, NWA 817, and NWA 5790. Apatites within these meteorites display more limited variability in F-Cl ratio and the apatite compositions are grouped closer to the F-apex in the F-Cl-OH ternary system (Fig. 6). All of the apatites within population 3 are associated with two feldspars and silica. Furthermore, the apatites are associated with groundmass Fe-rich silicate glass that has a chloride solubility of 1.36 wt% Cl (calculated from the model presented in Webster and De Vivo 2002). Therefore, we determined that the rhyolitic relative volatile abundance diagram (from Fig. 5b) is most appropriate for interpreting population 3 apatites. Nearly all of the apatite data from population 3 plot within the field indicating a melt with elevated Cl compared to F and H2O (Fig. 6b). The mesostasis glass in the population 3 nakhlites does not have chlorine or fluorine contents measurable by EPMA, so they probably degassed before complete solidification. Recent experimental evidence suggests that F, Cl, and H2O will differentially degas at low pressure, and this process would have the effect of increasing the fluorine contents of magmatic apatites during degassing due to earlier loss of H2O and Cl (Ustunisik et al. 2011). Consequently, apatite probably began to crystallize as the silicate melt was degassing, and the earliest formed apatites are represented by the most Cl-rich apatite. These apatites probably formed from a chloride-saturated melt. As the melt continued to degas and crystallize, the apatites then evolved to more F-rich compositions represented on the F-Cl-OH ternary diagram (Fig. 6b).
Volatile-Bearing Silicates and the Timing of Cl-Enrichment in Chassignites and Nakhlites
Amphibole and mica exclusively occur within mineral-hosted polyphase melt inclusions in the chassignite and nakhlite meteorites (Floran et al. 1978; Johnson et al. 1991; Treiman 2005; Sautter et al. 2006; McCubbin and Nekvasil 2008; McCubbin et al. 2009). This observation may provide important insight into the volatile histories of these meteorites because melt inclusions typically behave more like closed systems than do the intercumulus regions (Veksler 2006; Webster 2006; McCubbin and Nekvasil 2008). However, as discussed below, there are some important exceptions to closed-system behavior of melt inclusions when magmatic volatile loads are elevated (McCubbin and Nekvasil 2008; McCubbin et al. 2009).
Timing of Cl-Enrichment for the Chassignites
McCubbin and Nekvasil (2008) noted the subtle differences in mineral assemblage between the olivine-hosted melt inclusions and the intercumulus regions of the Chassigny meteorite. The differences they identified were in the F- and Cl-contents of the apatites, the absence of detectable Cl in the intercumulus maskelynite, the absence of kaersutite and Ti-biotite in the intercumulus regions, and the differences in ternary components of feldspar in the two textural regimes. Furthermore, they noted much lower closure temperatures (by approximately 200 °C) for pyroxene and for feldspar (computed using QUILF and SOLVCALC respectively, Andersen et al. 1993; Wen and Nekvasil 1994) in the melt inclusion assemblages, which they attributed to volatile-retention in the melt inclusions and H2O-loss in the intercumulus regions. They then postulated that the exogenous Cl-rich fluid that chlorinated the apatite must have had a lower water activity than the intercumulus melt, resulting in dehydration of the intercumulus melt. This process would destabilize kaersutite and Ti-biotite, causing them either not to form or react out of the intercumulus regions in Chassigny. The melt inclusions would have been largely shielded from this process, allowing the hydrous phases to form. Consequently, it can be inferred that an exogenous Cl-rich fluid was introduced to the cumulus horizon sampled by Chassigny subsequent to the formation and accumulation of the cumulus olivine, but before or during the early stages of intercumulus melt crystallization.
Timing of Cl-Enrichment for the Nakhlites
Only the nakhlites from population 3 have volatile-bearing silicates in their mineral-hosted melt inclusions. The amphiboles in the melt inclusions are typically very Cl-rich (i.e., >5 wt% Cl); on Earth, amphiboles with similarly elevated Cl contents are typically associated with alkali-halide metasomatism (Mazdab 2003). Apatites from these rocks show evidence of having formed from a chloride-saturated magma, which is consistent with the inferred elevated volatile load of liquid trapped as melt inclusions in cumulus olivine and pyroxene. Sautter et al. (2006) postulated that the Cl-rich amphibole was the result of assimilation of Cl-rich soil in the magma. However, such a process requires the magma to be at low pressure before entrapment of melt inclusion liquids. This requirement would make it difficult to account for some of the more OH-rich amphibole that has been analyzed in the melt inclusions because H2O solubility in silicate liquids at low pressure is very low (Burnham 1994; McMillan 1994). Furthermore, the elevated Cl-contents could just as easily be explained by other exogenous, or even potentially endogenous, sources as has been noted for magmatic Cl-enrichment of terrestrial layered mafic intrusions (Mathez et al. 1985; Boudreau and McCallum 1992; Boudreau 1993; Cawthorn 1994; Mathez 1995; Meurer and Boudreau 1996; Willmore et al. 2000; Glebovitsky et al. 2001). More details on this process of Cl-enrichment in terrestrial layered mafic intrusions can be found in McCubbin and Nekvasil (2008). Further support against soil assimilation for the source of Cl-enrichment and late oxidation of the nakhlite magma is provided by isotopic data on MIL 03346, which shows isotopic equilibrium between cumulus clinopyroxene and mesostasis at the collective nakhlite crystallization age (Shih et al. 2006).
The Cl-rich amphibole is likely present only within the melt inclusions because it formed from a supercritical fluid that would have rapidly escaped if not trapped within the melt inclusion. In fact, McCubbin et al. (2009) reported that many of the melt inclusions in MIL03346 may have ruptured due to the elevated volatile load and the increase in volume associated with a fluid–vapor phase transition. The Cl-rich amphiboles appear in melt inclusions in the Mg-rich cores of olivine (Fo42 to Fo21; Sautter et al. 2006), as well as in the more evolved portions of melt inclusions in cumulus pyroxenes. Thus, the Cl-enrichment of these systems probably occurred early, before or during olivine crystallization of the cumulus layer(s) sampled by the population 3 nakhlites. Furthermore, given the elevated volatile load of the amphibole-bearing inclusions, the cumulus grains could have trapped the Cl-rich fluid that was degassing from the nakhlite magma body. Cl-rich amphibole was not observed in NWA 817 melt inclusions; however, the apatite compositions in NWA 817 are identical to those of the other subpopulation 3 apatites that indicated chloride saturation at the time of apatite crystallization. This observation could be evidence that the Cl-rich fluid was exogenous to the magmatic system and did not interact with NWA 817 until late in the crystallization sequence; however, it must be noted that the absence of Cl-rich amphibole could also be explained by extensive melt inclusion rupture in this sample. We studied only a single section of NWA 817 for the work reported here, so it is possible that future studies of NWA 817 could yet find Cl-rich amphibole in some of the inclusions.
Petrogenetic Model for the Nakhlites and Chassignites as a Comagnetic Unit
Within error, all of the chassignites and nakhlites yield 40Ar-39Ar, 146Sm-142Nd, 147Sm-143Nd, 187Re-187Os, and/or 87Rb-87Sr ages of approximately 1.3 Ga, and this age has been interpreted as an igneous crystallization age (Brandon et al. 2000; Nyquist et al. 2001; Treiman 2005; Korochantseva et al. 2011). Furthermore, the nakhlites and chassignites have similarly depleted radiogenic isotopic compositions (Jagoutz 1996; Shih et al. 1998, 1999, 2006; Nyquist et al. 2001; Caro et al. 2008; Carlson and Boyet 2009) that are distinct from those of depleted shergottites (Nyquist et al. 1991; Borg et al. 1997, 2003; Borg and Draper 2003; Symes et al. 2008). The calculated present day 147Sm/144Nd and 87Rb/86Sr ratios for the nakhlite and chassignite sources span a narrow range of 0.233–0.239 and 0.074–0.078, respectively, and overlap significantly (Fig. 7), indicating the chassignites and nakhlites formed by partial melting of a nearly identical source that was LREE depleted (Wadhwa and Crozaz 1995; Jagoutz 1996; Nyquist et al. 2001; Caro et al. 2008; Carlson and Boyet 2009; Debaille et al. 2009). However, somewhat paradoxically, given their depleted initial 143Nd/144Nd ratios, the two meteorite groups have similar LREE-enriched bulk-rock REE patterns (Wadhwa and Crozaz 1995; Treiman 2005; Treiman and Irving 2008) (Fig. 8a). This similarity indicates that both groups either formed by low degrees of partial melting in the source or were enriched in LREEs subsequent to partial melting in the Martian mantle, although the LREE source (e.g., Cl-rich fluid) must also come from an isotopically depleted source. Additionally, all of the nakhlites and chassignites have similar cosmic ray exposure ages (approximately 11 Ma), indicating that they may have been launched during the same impact on Mars (Eugster et al. 2002; Eugster 2003; Marti and Mathew 2004; Bogard and Garrison 2008; Park et al. 2008; Korochantseva et al. 2011). These similarities have led to the conclusion that the nakhlites and chassignites came from very similar magmatic source regions in the Martian mantle (Wadhwa and Crozaz 1995; Brandon et al. 2000; Carlson and Boyet 2009; Debaille et al. 2009).
Numerous authors have used the above information, along with other geochemical and textural information, to suggest that the nakhlites are comagmatic and were part of the same cumulate pile (e.g., Harvey and McSween 1992; Wadhwa and Crozaz 1995; Lentz et al. 1999; Mikouchi et al. 2003, 2012; Treiman 2005; Day et al. 2006; Imae and Ikeda 2007; Korochantseva et al. 2011). Despite all the similarities between the nakhlites and chassignites, there are a few geochemical differences between the two meteorite classes that have led some authors to conclude that the chassignites are not comagmatic with the nakhlites (Wadhwa and Crozaz 1995). Using the results presented here, we propose a model that explains many of these differences and is consistent with the chassignites being comagmatic with the nakhlites. Specifically, we can reconcile many of the differences between the two meteorite classes with the addition of an exogenous Cl-rich fluid to the magma body at the time of formation of the cumulus horizon sampled by the Chassigny meteorite, as proposed previously by McCubbin and Nekvasil (2008). It is important to note that the addition of both exogenous and endogenous Cl-rich fluids to crystallizing magma bodies has also been proposed in some terrestrial layered mafic intrusions (Mathez et al. 1985; Boudreau and McCallum 1992; Boudreau 1993; Cawthorn 1994; Mathez 1995; Meurer and Boudreau 1996; Willmore et al. 2000; Glebovitsky et al. 2001).
Reconciling the Geochemical Differences between the Chassignites and Nakhlites
Despite the strong similarities between the nakhlites and chassignites, there are some differences in pyroxene trace element ratios that have led some to conclude that the nakhlites and chassignites are not comagmatic. Wadhwa and Crozaz (1995) showed that the Ti/Ce and Ti/Y ratios in Chassigny pyroxenes are approximately 400 and 270, respectively, whereas the same ratios in Nakhla, Governador Valadares, and Lafayette pyroxenes are approximately 1000 and 500, respectively. The differences in the ratios between the two meteorite classes are primarily a reflection of a drop in the concentrations of Ce and Y rather than an increase in Ti (Wadhwa and Crozaz 1995). Wadhwa and Crozaz (1995) argued that those differences could not be explained by fractional crystallization of a single magma, and consequently Chassigny was not comagmatic with the nakhlites. However, other factors must first be considered before this conclusion is reached. The pyroxenes in Chassigny are not cumulus, but intercumulus grains (Nekvasil et al. 2007; McCubbin and Nekvasil 2008) that cannot be directly compared with cumulus grains from other portions of the supposed common magma body because the mineral compositions from the two textural regimes can evolve along unique paths (which has been shown for terrestrial layered mafic intrusions; e.g., Sparks et al. 1985; Barnes 1986; Toplis et al. 2008; Humphreys 2009). Additionally, gain and/or loss of Cl-rich fluids to the magma system can change Ce and Y contents dramatically beyond what can be typically attributed to fractional or equilibrium crystallization (Lieftink et al. 1994; Stalder et al. 1998; Mayanovic et al. 2009).
Wadhwa and Crozaz (1995) also cited Ti/Al ratios in pyroxenes as evidence against a comagmatic origin; however, these ratios are not inconsistent with such an origin. Pyroxenes in Chassigny and in Nakhla, Governador Valadares, and Lafayette all have Ti/Al ratios ≤1:4 (Fig. 9). Because Ti and Al enter the pyroxene crystal structure together as a coupled substitution, Ti/Al ratios in pyroxene typically vary between 1:4 and 1:2, depending on the relative importance of the Al-Al substitution during basalt crystallization (see Bence and Papike 1972). Overlapping Ti/Al ratios of approximately 1:4 are indicative of a plagioclase undersaturated magma. Many of the pyroxenes in Fig. 9 display Ti/Al ratios <1:4 because the pyroxenes have elevated Cr contents (up to 7700 ppm Cr for Chassigny and up to 3300 ppm Cr for the nakhlites), which is accommodated in the pyroxene via a Cr-Al substitution. In fact, the higher Ti/Al ratios in Chassigny pyroxene reported by Wadhwa and Crozaz (1995) as evidence against a comagmatic origin for the chassignites and nakhlites can be largely explained by the higher Cr abundances in Chassigny pyroxene, which is not surprising, given the abundance of cumulus chromite in Chassigny and the absence of cumulus chromite in the three nakhlites. Thus, the Ti/Al ratios between Chassigny pyroxene and Nakhla, Governador Valadares, and Lafayette pyroxene are not inconsistent with a comagmatic origin.
Rare earth element patterns in pyroxene have also been used to argue against a comagmatic origin for the chassignites and nakhlites. Chondrite-normalized REE patterns in augite cores in Chassigny are LREE depleted and not parallel to REE patterns of corresponding augite rims. However, the core-rim pairs in cumulus augites from the nakhlites are both LREE enriched and largely parallel (as illustrated in Fig. 8b). The addition of a Cl-rich fluid during crystallization of the Chassigny intercumulus regions would act as both a chlorine source and an LREE source to the magma body because Cl-rich fluids are commonly enriched in LREE due to the effect of Cl on fluid/melt partitioning of LREE under hydrothermal conditions (e.g., Lieftink et al. 1994; Stalder et al. 1998; Mayanovic et al. 2009). If this event affected the entire magma body simultaneously, it would have occurred largely before the crystallization of pyroxene that formed the cumulate horizons that were sampled by the nakhlites. Therefore, the LREE-depleted augite cores in Chassigny can be explained by formation prior to contamination with an exogenous fluid with the LREE-rich rims forming after the addition of the Cl-rich fluid. Furthermore, the pyroxenes that compose the nakhlites would be consistently LREE-enriched because the addition of the fluid would have occurred prior to their crystallization. To illustrate the timing of LREE addition in Fig. 8b, the rim value for each rare earth element in a given core-rim pair was divided by the corresponding value from the core analysis. Next a regression line was calculated through the normalized pattern for each sample and the slope of that line was calculated along with the error on the slope. The slopes were then plotted as a function of relative depth in the cumulate pile, and a dotted line was used to represent the position of core-rim pairs with parallel REE patterns. Deviation from zero indicates variable LREE or HREE abundances during augite crystallization. This process reconciles the chassignites and nakhlites forming from LREE-depleted source regions and having LREE-enriched bulk rock patterns as long as the exogenous Cl-rich fluid originated from a depleted source on Mars. Furthermore, it removes the constraint that the nakhlites and chassignites needed to form by low degrees of partial melting, although it does not exclude that possibility.
Cl-Rich Fluid Addition Recorded by Volatile-Bearing Minerals
The process of early Cl-rich fluid addition to the magma body is also consistent with the two populations of apatites present in the chassignites and the three populations of apatites that are observed in the nakhlites. Prior to the addition of a Cl-rich fluid to the magma body, the melt inclusions were trapped within cumulus olivine sampled by the chassignites. These melt inclusions went on to form F-rich apatites that co-existed with H2O-bearing amphibole and biotite (McCubbin and Nekvasil 2008; McCubbin et al. 2010a). As the intercumulus liquid in Chassigny crystallized, an exogenous Cl-rich fluid interacted with and dissolved into the magma body. This fluid chlorinated and dehydrated the intercumulus melt while leaving the sealed melt inclusions within olivine hosts untouched (McCubbin and Nekvasil 2008). This process produced Cl-rich apatites in the intercumulus regions of the chassignites and F-rich apatites in the melt inclusions (McCubbin and Nekvasil 2008). Furthermore, the dehydration of the melt provides a reason for the absence of amphibole and biotite from the intercumulus regions in the chassignites and nakhlites (McCubbin and Nekvasil 2008; McCubbin et al. 2009). This bimodal apatite chemistry as evidence for open-system conditions is further supported by the presence of single apatite chemistry between olivine-hosted melt inclusions and intercumulus apatite in lunar troctolite 76535 (Elardo et al. 2012).
The addition of Cl-rich fluid to the magma body at the time of chassignite cumulate layer crystallization could have affected the magma body as a whole, and hence any subsequently crystallizing lithologies. Furthermore, the elevated Cl load would continue to increase as the magma body crystallized because Cl is highly incompatible. The population 1 apatites, represented by NWA 998, would have formed from this Cl-rich liquid, and the apatites in NWA 998 are indeed more Cl-rich than intercumulus apatite in Chassigny and NWA 2737. Moreover, the NWA 998 meteorite is interpreted to be at the bottom of the nakhlite cumulate pile (Mikouchi et al. 2012), indicating the cumulus phases present probably formed before the phases in the other nakhlites. The lack of variability in the F:Cl ratio in comparison with nakhlite apatite population 2 indicates that the magma had not yet reached chloride saturation at the time of apatite population 1 crystallization. At the time of population 2 apatite crystallization, represented by Nakhla, Governador Valadares, and Lafayette, the magma body was likely either at chloride saturation or reached chloride saturation slightly after the onset of apatite crystallization, which is indicated by the large nonmagmatic variability (Meurer and Boudreau 1996) of apatites within population 2 (Figs. 2 and 6). The apatites probably started out similar in composition to those in population 1. As crystallization proceeded and chloride saturation in the melt was reached, the Cl contents of population 2 increased and exceeded the maximum Cl contents of population 1. As the Cl-rich fluid was lost from the system, the magmatic apatite compositions continued to evolve to more F-rich compositions, giving the tremendous range in F:Cl ratio that is exhibited by population 2 apatites. This degassing trend has also been demonstrated experimentally for apatites in lunar melt compositions (Ustunisik et al. 2011). Apatite population 3, represented by the MIL nakhlites, NWA 817, and NWA 5790, would then consist of apatites that formed after much of the Cl was lost from the evolving magmatic system. In fact, Cl-rich fluid loss was probably occurring at or slightly after the time of formation of the cumulus augites and olivines in the meteorites that host population 3 apatites because all but NWA 817 has Cl-rich inclusions that contain a variety of Cl-rich amphiboles that are indicative of highly saline hydrothermal activity.
The Nakhlite Cumulate Pile and the Position of the Chassignites
Based on the amount of mesostasis present, the inferred cooling histories of the meteorites, and the Fe:Mg ratios of the ferromagnesian phases (Table 5), the stratigraphy of the pile relative to the nakhlite samples analyzed to date is as follows: NWA 998, Lafayette, Governador Valadares, Nakhla, Yamato 000593/749/802, NWA 817, MIL 03346/090030/090032/090136, NWA 5790 (Lentz et al. 1999; Treiman 2005; Day et al. 2006; Treiman and Irving 2008; Mikouchi et al. 2012). Based on the same criteria, Chassigny would have originated significantly below NWA 998, and NWA 2737 would originate slightly below Chassigny if nakhlites and chassignites are comagmatic. Based on experimental work that used proposed parental liquids to the Chassigny melt inclusions (Nekvasil et al. 2007, 2009; Filiberto 2008; Filiberto et al. 2008; McCubbin et al. 2008b; Filiberto and Treiman 2009a), there is likely a pigeonite–olivine cumulus horizon, which has not yet been sampled, stratigraphically above the cumulus horizon that was sampled by the Chassigny meteorite and below the horizon sampled by NWA 998. A sample of this putative horizon, should one ultimately be discovered, would help further tie the chassignites and nakhlites to a single igneous event. The envisioned process of Cl-rich fluid addition as well as the complete cumulus pile and corresponding meteorites that sample each horizon is presented in Fig. 10.
Table 5. Stratigraphy of the chassignite-nakhlite cumulate pile along with textural and geochemical information for each meteorite
Fe-rich olivine composition[1–5]
Fe-rich pyroxene composition (Mg#)[1–5]
Oxygen fugacity (ΔFMQ) [2,4, 6–9]
Numbers in brackets correspond to the following sources of data for each section in the table:  Mikouchi et al. (2012),  Treiman et al. (2007),  Mason et al. (1975),  Beck et al. (2006),  Floran et al. (1978),  Szymanski et al. (2010),  Treiman and Irving (2008),  Noguchi et al. (2009),  Treiman (2005).
−1 to 0
−1.1 to −0.6
−1.25 to −0.5
Cl-Rich Fluids and Oxygen Fugacity of the Chassignites and Nakhlites
The chassignites and nakhlites record a fairly narrow range in oxygen fugacity that is typically within 1.5 log units of the fayalite-magnetite-quartz (FMQ) oxygen buffer (Dyar et al. 2005; Treiman et al. 2007; Szymanski et al. 2010). Oxygen fugacity estimates for each of the nakhlites and chassignites are presented in Table 5, and their values have some correlation with their position within the cumulate pile. Overall, the chassignites and nakhlites are oxidized when compared with the shergottites, which span a range of oxygen fugacities from FMQ –3.5 to –1 (Herd and Papike 2000; Wadhwa 2001, 2008; Herd et al. 2002; Goodrich et al. 2003; Herd 2003, 2006a, 2008; McCanta et al. 2004). Questions have been raised as to whether the magmatic source region of the nakhlites and chassignites is oxidized or whether the oxidation occurred as a secondary process during crystallization (Righter et al. 2008; Szymanski et al. 2010). Given the depleted isotopic compositions of the chassignites and nakhlites, it could be assumed that their magmatic source region was reduced if the depleted shergottites are used as a comparative analog (Herd et al. 2002). Furthermore, there are several lines of evidence that indicate that the oxidation state of the chassignite and nakhlite magmas was increasing during crystallization. Specifically, redox gradients are recorded in several mineral phases in the nakhlites (Domeneghetti et al. 2006, 2007; Treiman and Irving 2008) and the sulfide–sulfate transition is recorded in MIL 03346 melt inclusions that host pyrrhotite and hydrothermally precipitated jarosite (McCubbin et al. 2009). In fact, estimates of the oxygen fugacity of the earlier stages of crystallization from the mineral phases yield consistently lower oxygen fugacity (near FMQ -2) for several nakhlites including MIL 03346, NWA 817, and NWA 998 (Treiman 2005; Domeneghetti et al. 2006, 2007; Rutherford and Hammer 2008), which overlaps with the reducing values recorded by the depleted shergottites (Herd et al. 2002; Wadhwa 2001, 2008). Sautter et al. (2006) speculated that the source of oxidation (and Cl) for MIL 03346 could be Martian soil; however, this explanation is not consistent with a hydrous component in the MIL nakhlite amphiboles. During crystallization, the Fe2+/∑Fe decreases because Fe3+ is incompatible (Ghiorso and Sack 1995), and this process may be partially responsible for the increase in calculated oxygen fugacity during crystallization of the nakhlites (as suggested by Treiman 2005). In addition to this process, we propose that the late-stage oxidation was enhanced by the exsolution and loss of a Cl-rich fluid phase. When Cl-rich fluids exsolve from magmatic systems, they preferentially partition Fe2+ into the fluid, which enriches the residual melt in Fe3+ and has the effect of auto-oxidizing the magma body (Bell and Simon 2011). Although a maximum change of about 1 log unit fO2 is anticipated by the process of Cl-rich fluid loss from the magma (Bell and Simon 2011), the effect on calculated fO2 could be greater due to the effect of Cl− on the activity coefficient of Fe in the silicate liquid prior to degassing. These combined processes could explain the high oxygen fugacities of all the nakhlites and chassignites. Furthermore, the most oxidized samples are nearest to the top of the cumulus pile, which would be the most affected by Cl-loss (and hence Fe2+ loss) from the silicate liquid (see Table 5). Chassigny, NWA 2737, and NWA 998, which are the most reduced of the chassignites and nakhlites at FMQ −1.25 to −1 (Treiman et al. 2007), would be least affected by this process because these samples were near the bottom of the sampled cumulus pile and largely solidified by the time the magma body reached chloride saturation.
Emplacement of Chassignites and Nakhlites: Layered Intrusion/Sill or a Lava Flow?
The origin of the chassignites and nakhlites has been a subject of debate for many years (Treiman 1986, 1990, 2005; McSween and Treiman 1998; Lentz et al. 1999; Day et al. 2006; Sautter et al. 2006, 2012; Nekvasil et al. 2007, 2009; McCubbin and Nekvasil 2008; McCubbin et al. 2009). One of the most important topics of debate is the mode of emplacement for these rocks. The lowest and most slowly cooled members of the group seem to require formation as a slowly cooled intrusion at pressure or a very thick flow (Lentz et al. 1999; Treiman 2005; Day et al. 2006; Nekvasil et al. 2007), while the vitrophyric textures of the uppermost members require rapid cooling and almost certainly made up part of a lava flow on the Martian surface (Day et al. 2006; Hallis and Taylor 2011). Given the volatile-rich history presented in the present study, we propose a hybrid model for chassignite-nakhlite emplacement, although we note here that it is largely a modified version of the model proposed previously by Day et al. (2006). The lowermost members of the unit probably formed in a slowly cooled relatively shallow magma chamber. Although claims have been made that the melt inclusion assemblage of Chassigny can only form at pressures of approximately 0.9 GPa (Nekvasil et al. 2007, 2009), we note that higher pressures in the olivine-hosted melt inclusions in comparison with the intercumulus regions of Chassigny could have been attained through overpressure of the olivine host due to its mechanical strength (Gaetani et al. 2012; Veksler 2006). As the magma in the chamber continued to cool and crystallize, it eventually reached volatile saturation. At this point, the escaping fluid provided a driver for eruption onto the surface. Volatile exsolution was concomitant with lava eruption, which is supported by the record of fluid saturation during the formation of cumulus minerals in NWA 817, the MIL nakhlites, and NWA 5790. These evolved, Fe-rich lavas, which entrained cumulus clinopyroxene and olivine crystals from the top of the cumulate pile, would have erupted and cooled relatively quickly and do not require residence times in a thick flow. Moreover, these lavas would likely be subject to different secondary alteration environments than the cumulate pile, which could explain differences in secondary alteration minerals between the different cumulate rocks (Reid and Bunch 1975; Gooding et al. 1991; Treiman et al. 1993; Wentworth and Gooding 1994; Romanek et al. 1998; Bridges and Grady 1999, 2000; Bridges et al. 2001; Gillet et al. 2002; Treiman 2005; Herd 2006b; Sautter et al. 2006; Vicenzi et al. 2007; Noguchi et al. 2009; Changela and Bridges 2010; Bridges and Schwenzer 2012).
We have conducted a survey of volatile-bearing minerals in the nakhlites and chassignites, and propose a new model for their comagmatic petrogenesis. The nakhlites and chassignites have a chlorine-rich mineralogy that can be largely explained by the interaction between magma and Cl-rich fluids during crystallization. Furthermore, the addition of this fluid relaxes many of the previous constraints used to rule out the nakhlites and chassignites being comagmatic rocks. On the basis of inferred stratigraphy of the cumulus pile along with major, minor, and trace element distributions among phases within the two meteorite classes, we envision the following petrogenetic model for the chassignites and nakhlites:
The depleted Martian mantle (similar to the depleted shergottite source region) partially melts under reducing (≤FMQ −2) conditions. The partial melt then separates from the source and rises until stalling in the shallow crust (with or without an initial stage of fractionation, perhaps at the base of the Martian crust).
The magma body begins to cool and crystallizes olivine and chromite, which then settle to the bottom of the magma body forming the cumulus horizons sampled by the chassignites. Subsequent to settling of the horizon sampled by Chassigny, and during crystallization of intercumulus augite in Chassigny, a Cl-rich, LREE-enriched exogenous fluid from an isotopically depleted source interacts with the magma body and leaves the residual melt elevated in chlorine, depleted in water, and enriched in LREE.
The magma body continues to crystallize and forms the cumulus layer sampled by NWA 998 consisting primarily of clinopyroxene, orthopyroxene, and olivine ± cumulus apatite and Ti-magnetite. At this point, the evolving magma is approaching chloride saturation. However, before this layer forms, a pigeonite-rich cumulate layer currently absent from the meteorite collection probably formed stratigraphically between Chassigny and NWA 998.
The magma body continues to crystallize clinopyroxene and olivine, and soon reaches chloride saturation. The exsolved, low-density fluid migrates away from the accumulated pile, preferentially removing more Fe2+ than Fe3+ and increasing oxygen fugacity of the cumulus pile toward FMQ. The formation and subsequent departure of a Cl-rich fluid are recorded by the very large variation in apatite compositions that range from nearly endmember chlorapatite to nearly endmember fluorapatite. This fluid, once exsolved, may have also provided a mechanism for eruption of lava onto the Martian surface from the magma chamber.
Near the top of the magma body, just after Cl-saturation, some of the cumulus olivine and clinopyroxene traps the high Cl fluid/melt in melt inclusions, which later form Cl-rich amphibole. However, this fluid is continuously filtering out of the magma system and Cl-rich amphibole is not stabilized in the intercumulus regions of the magma chamber. As the Cl-rich fluid leaves the upper portions of the cumulus pile, it continues to extract proportionally more Fe2+ than Fe3+, raising the oxygen fugacity of the upper portions of the magma body to FMQ +1.5. As the Cl-rich fluid escapes, some of the residual magma from the chamber erupts onto the Martian surface, bringing with it entrained cumulus minerals from the top of the cumulus pile. These lavas cool relatively quickly, giving the vitrophyric texture that is common to many of the upper members of the nakhlite series.
Subsequent to the cessation of igneous activity, the crystalline magma body and the lava flows are altered by low-temperature aqueous fluids, leaving behind many of the minor oxides, sulfates and carbonates that have been previously identified. Approximately 1.3 Ga after their formation (approximately 11 Ma before the present), the crystalline intrusion is struck with an impactor that ejects lithologic fragments into interplanetary space, some of which eventually land on Earth.
We thank the reviewers Lydia Hallis, Justin Filiberto, and James Day for very insightful and helpful comments that greatly improved the quality of the manuscript. We also thank Christine Floss for all of her effort as Associate Editor. We thank Hanna Nekvasil, Donald Lindsley, Aaron Bell, and James Papike for helpful discussions on the topic of this manuscript. This work was funded by NASA Cosmochemistry grants NNX11AG76G to F. M. M. and NNX10AI77G to C. K. S. Additionally, S. M. E. gratefully acknowledges support from NASA Earth and Space Science Fellow NNX12AO15H, Cosmochemistry grant NNX08AH81G to D. S. D., and a graduate fellowship from the New Mexico Space Grant Consortium during this study.