Available online at www.sciencedirect.com ScienceDirect

Geochimica et Cosmochimica Acta 157 (2015) 56–85 www.elsevier.com/locate/gca

Petrology of igneous clasts in Northwest Africa 7034: Implications for the petrologic diversity of the martian crust

Alison R. Santos a,b,⇑, Carl B. Agee a,b, Francis M. McCubbin a,b, Charles K. Shearer a,b, Paul V. Burger a, Romain Tarte`se c, Mahesh Anand c,d

a Institute of Meteoritics, MSC03 2050, University of New Mexico, Albuquerque, NM 87131, USA b Department of Earth and Planetary Sciences, MSC03 2040, University of New Mexico, Albuquerque, NM 87131, USA c Planetary and Space Sciences, The Open University, Walton Hall, Milton Keynes MK7 6AA, UK d Department of Earth Sciences, The Natural History Museum, Cromwell Road, London SW7 5BD, UK

Received 5 August 2014; accepted in revised form 17 February 2015; available online 25 February 2015

Abstract

The martian meteorite Northwest Africa (NWA) 7034 was examined both petrographically and geochemically using sev- eral micro-beam techniques including electron probe microanalysis and secondary ion mass spectrometry. We have identified various clast types of igneous, sedimentary, and impact origin that occur within the breccia, and we define a classification scheme for these materials based on our observations, although our primary focus here is on the petrology of the igneous clasts. A number of different igneous clasts are present in this meteorite, and our study revealed the presence of at least four different igneous lithologies (basalt, basaltic andesite, trachyandesite, and an Fe, Ti, and P (FTP) rich lithology). These lithologies do not appear to be related by simple igneous processes such as fractional crystallization, indicating NWA 7034 is a polymict breccia that contains samples from several different igneous sources. The basalt lithologies are a good match for measured rock compositions from the martian surface, however more exotic lithologies (e.g., trachyandesite and FTP lithologies) show this meteorite contains previously unsampled rock types from Mars. These new rock types provide evi- dence for a much greater variety of igneous rocks within the martian crust than previously revealed by martian meteorites, and supports recent rover observations of lithologic diversity across the martian surface. Furthermore, the ancient ages for the lithologic components in NWA 7034 indicate Mars developed this lithologic diversity in the early stages of crust formation. Ó 2015 Elsevier Ltd. All rights reserved.

1. INTRODUCTION et al., 2013) and noble gas measurements that match mea- surements of the martian atmosphere (Cartwright et al., The martian meteorite Northwest Africa (NWA) 7034 2014). This meteorite is a breccia with a basaltic bulk com- (and pairings so far identified: NWA 7533, NWA 7475, position, initially classified as a “porphyritic basaltic mono- NWA 7906, NWA 7907, NWA 8114, NWA 8171) is unique mict breccia” by Agee et al. (2013). Agee et al. (2013) also from other martian meteorites in several ways, but its mar- identified a diverse set of clasts containing a wide variety tian origin was confirmed by pyroxene Fe/Mn ratios (Agee of textures (gabbros, quenched melts, and oxide rich reac- tion spherules). This initial classification of the meteorite was based on the bulk basaltic composition of the breccia and continuous pyroxene and feldspar compositions within ⇑ Corresponding author at: Institute of Meteoritics, MSC03 2050, University of New Mexico, Albuquerque, NM 87131, USA. the clasts and matrix. NWA 7533, however, was classified E-mail address: [email protected] (A.R. Santos). as a polymict breccia by Humayun et al. (2013), who http://dx.doi.org/10.1016/j.gca.2015.02.023 0016-7037/Ó 2015 Elsevier Ltd. All rights reserved. A.R. Santos et al. / Geochimica et Cosmochimica Acta 157 (2015) 56–85 57 defined a somewhat different set of clasts (plutonic lithic single thin sections only display a two-dimensional view clasts such as monzonites and norites, basalts, and impact of the materials). Additional petrographic data were col- melt clasts). Initially it would seem that the two studies lected from a total of 12 probe mounts and thin sections were examining very different portions of a very heteroge- that include the original 6 used for quantitative geochemical neous breccia; however the back-scattered electron images analysis. Images of all samples used for this study are found and other petrographic information reported in both stud- in Figs. 1–3, and in the Supplementary Material. ies indicate that there was some overlap in observed textu- ral features despite a lack of overlap in terminology (the 2.1. Electron probe microanalysis (EPMA) latter being related to differences in scale and methods of observation, as well as differences in interpretation). A EPMA was conducted using a JEOL JXA 8200 electron detailed study of all the various textural features and clast microprobe equipped with 5 wavelength dispersive spec- types has yet to be performed for NWA 7034 and its trometers in the IOM at UNM. Data were collected using pairings. both the JEOL manufacturer’s software and the Probe Initial studies of NWA 7034 determined that the mete- for EPMA (PFE) software from Probe Software, Inc. orite’s bulk composition coincides with the composition Each phase was analyzed using specific beam current and of the average martian crust determined from mission data spot size conditions, but a 15 kV accelerating voltage was (McSween et al., 2009; Taylor and McLennan, 2009; Agee used for all phases. Feldspars were analyzed using a et al., 2013). This bulk composition also matches some of 15 nA beam current and 5 lm spot diameter in an attempt the rocks and soils measured in Gusev Crater by the to minimize Na mobility during analysis. Potassium-rich Mars Exploration Rovers (MER) (Gellert et al., 2006; feldspar analyses were affected by this problem preferential- McSween et al., 2006a,b). In fact, this meteorite represents ly, as determined from both plots of Na X-ray count rate the strongest link between a martian meteorite and the geo- vs. time and mineral stoichiometry (low A site sums). chemistry of the martian surface determined by remote This problem was reconciled by adding Na2O to each ana- sensing (Boynton et al., 2007; McSween et al., 2009), lysis showing evidence of Na mobility until the analysis did although the nakhlites and chassignites have been linked not display normative quartz or corundum, based on the to some of the rocks analyzed in Gusev Crater by the methods of McCubbin and Nekvasil (2008). This typically MER Spirit (Filiberto, 2008; Nekvasil et al., 2009; provided a similar correction as when employing the McCubbin et al., 2013). In addition to the martian surface, time-zero correction for Na in the PFE software, however Agee et al. (2013) determined that NWA 7034 also has geo- this method did not provide an independent check on the chemical similarities linking it to other martian meteorites, stoichiometry. Pyroxene was analyzed using a 20 nA beam based on major element chemistry of feldspar and pyroxene current and 5 lm spot size. Iron was measured as Fe2+, and as well as Fe/Mn ratios of pyroxenes. the methods of Droop (1987) were used to calculate the Prior studies of NWA 7034 have demonstrated the amount of Fe3+ in pyroxene. Apatite was analyzed using importance of this sample to understanding the martian a 20 nA beam current, and the largest possible beam dia- crust (Agee et al., 2013; Humayun et al., 2013; Cartwright meter was used, although the smallest beam diameter we et al., 2014; Gattacceca et al., 2014; Nemchin et al., 2014; employed in the present study was approximately 3 lm. Muttik et al., 2014b); however there has yet to be a study Fluorine was analyzed using an LDE1 crystal and Cl was that has described in detail all of the various lithologic com- analyzed using a PET crystal. In order to account for ponents within NWA 7034. Therefore, the primary objec- changes in F and Cl X-ray count rates during analyses tive of the present study was to produce petrologic and (Stormer et al., 1993), a time dependent intensity correction textural descriptions and classifications for each of the was applied using the PFE software. These apatites were lithologic components in this meteorite, with a focus on not measured for their OH abundance, but the abundance describing the various igneous lithologies within the mete- of this component was computed stoichiometrically assum- orite and examining how they compare to other martian ing F + Cl + OH = 1 following McCubbin et al. (2010b). surface rocks and meteorites. These two goals were accom- Fe–Ti oxides were analyzed using a 30 nA beam current plished through detailed petrographic analysis in conjunc- with spot sizes down to 1 lm. The Fe2+/Fe3+ was deter- tion with micro-beam geochemical analysis using electron mined in Fe–Ti oxides by charge balance after performing probe microanalysis (EPMA) and secondary ion mass spec- a cation normalization assuming all other cations present trometry (SIMS). in the phase were analyzed. Different standards were employed for the analysis of each phase. Pyroxene stan- 2. METHODS dards included orthoclase (Si, Al, K), ilmenite (Ti, Fe), chromite (Cr), cobalt metal (Co), nickel metal (Ni), spes- Chemical and mineralogical data were collected on 6 sartine (Mn), augite (Mg, Ca), albite (Na), sphene (Ti), thin sections and probe mounts derived from slices of the and pyrope (Fe). Feldspar standards included orthoclase original main mass of NWA 7034 housed in the collection (Si, Al, K), chromite (Cr), albite (Na), augite (Mg, Fe), of the Institute of Meteoritics (IOM) at the University of spessartine (Mn), and anorthite (Ca). Apatite standards New Mexico (UNM). Some of these thin sections and included quartz (Si), rutile (Ti), pyrope (Al), chromite probe mounts are from parallel cuts into the same slice, (Cr), ilmenite (Fe), spessartine (Mn), olivine (Mg), and as such provide us with a three-dimensional view of Durango apatite (P, Ca), albite (Na), strontium fluoride some of the larger materials within the meteorite (whereas (F), sodalite (Cl), and barite (S). Fe–Ti oxide standards 58 A.R. Santos et al. / Geochimica et Cosmochimica Acta 157 (2015) 56–85

Fig. 1. PMgAl X-ray map of UNM Section 2,3. P-red, Mg-green, Al-blue, Fe–Ti oxides and sulfides are white to gray. (A) Large proto- breccia clast containing a distinct matrix from the bulk breccia, isolated mineral fragments of varying sizes, and igneous clasts. (B) Back scattered electron (BSE) image of Clast 76, a trachyandesite clast. (C) BSE image of Clast 74F, a trachyandesite clast. The coarse grained feldspar, pyroxene, and Fe–Ti oxide assemblage was not included in the study of this clast as it appears to represent a different lithology that is out of equilibrium with the surrounding fine grained clast material. (D) BSE image of Clast 73, a basalt clast. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) included chromite (Mg, Cr), quartz (Si), ilmenite (Ti, Fe), sections were carbon coated and examined with an electron vanadium metal (V), gahnite (Al, Zn), spessartine (Mn), beam current of 0.6 nA at an accelerating voltage of 20 kV. nickel metal (Ni), and cobalt metal (Co). All mineral ana- Magnification used was 300 and dwell time was set to lyses other than Fe–Ti oxides were used only if they showed 100 ms/pixel. Three thin sections were mapped using an good stoichiometry (stoichiometry parameters are listed in FEI Quanta 3D Field Emission Gun (FEG) SEM in the the Supplementary Materials), and Fe–Ti oxide analyses Department of Earth and Planetary Sciences at UNM. were checked against the tie-lines within the FeO–Fe2O3– These maps were collected with a 20 kV accelerating volt- TiO2 ternary. A bulk matrix composition was also deter- age and 0.85 nA beam current. Dwell times were 250 ls mined using EPMA (Figs. 17 and 20); the details of this per pixel, and eight frames were collected per map field. procedure and the full composition are reported in the Magnification was set to 300. These maps were assembled Supplementary Materials. using a stitching plugin in the Fiji software (Preibisch et al., 2009). The FEI Quanta 3D FEG SEM at UNM was also 2.2. Image collection used to collect high resolution BSE images of specific clasts and textures. This imaging was conducted using a 10 kV Back-scattered electron (BSE) images and individual X- accelerating voltage and 16 nA beam current. ray maps of whole polished thin sections and probe mounts Qualitative X-ray maps of specific regions within clasts were acquired using a Quanta 3D dual beam Focused Ion were collected using the JEOL JXA 8200 electron micro- Beam (FIB) Secondary Electron Microscope (SEM) at the probe in the IOM at UNM. Mapping was conducted using Open University, UK, fitted with an Oxford Instruments Probe Image software, part of the Probe for EPMA soft- INCA energy dispersive X-ray detector. For this purpose, ware suite. Maps were collected with a 15 kV accelerating A.R. Santos et al. / Geochimica et Cosmochimica Acta 157 (2015) 56–85 59

Fig. 2. PMgAl X-ray map of UNM Section 1B,2. P-red, Mg-green, Al-blue, Fe–Ti oxides and sulfides are white to gray. (A) Impact melt clast, a portion of which was studied by Udry et al. (2014b). Note the skeletal but large mineral grains, indicating a rapid cooling but not quenching. The clast boundary in contact with the bulk matrix is rounded with aligned mineral grains roughly parallel to the boundary. BSE image highlighting the texture of this clast is in Fig. 8. (B) Proto-breccia clast, again having a matrix distinct from the bulk breccia matrix. (C) Altered, devitrified melt clast. (D) BSE image of Clast 5, a basalt clast. (E) BSE image of Clast 6, a basalt clast. (F) BSE image of FTP Clast 1, a phosphate and oxide rich clast. Note the extreme textural difference between the basalt and FTP clasts. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) voltage, 50 nA beam current, 1 lm beam diameter, and a Bulk clast compositions were calculated using three 100 ms dwell time per analysis. K, Na, and Fe were mea- pieces of information: average mineral compositions for sured using WDS, while Mg was measured using EDS. the clast, mineral modes (vol. %), and mineral densities. Average mineral compositions were determined from 2.3. Calculation of clast modal mineralogy and bulk clast EPMA data. Mineral densities were determined using unit compositions cell parameters. The molar mass of the unit cell was deter- mined using the average mineral composition to account Modal mineralogies (vol. %) of clasts were determined for minor element substitutions, and the lattice volume by analyzing clast BSE images collected on the electron was calculated using values from Deer et al. (1992). microprobe using the program ImageJ. Image scales were Mineral densities were used to convert volume based set so that areas were calculated in sq. microns (lm2). mineral modes to mass based mineral modes. Next, the Pixel thresholds were then assigned to different phases amount of each oxide component in the bulk clast was within the clasts, and the percentage of the clast area occu- calculated using the following equation: pied by pixels within the thresholds was determined by the AllX phases program. Once the percentage of clast area occupied by X ðwt%Þ¼ x M ð1Þ each phase was determined, the values were normalized to oxide x phase A phase A A¼Phase 1 100% (cracks within the clasts were not counted, leaving some of the clast area unaccounted for). Some clasts were where X oxide xðwt%Þ is the mass percent of oxide component large enough to occur in multiple thin sections, and the x in the bulk clast composition in wt%, xphase A is the mass mineral modes of these clasts were combined to give a percent of oxide x in phase A, M phase A is the modal fraction more accurate mineral mode for these clasts. The modal (wt.) of phase A. This process is repeated for each oxide combination for multiple exposures of the same clast was component of interest to construct the bulk composition weighted based on the total surface area of each exposure. of each clast. Fe–Ti oxides and apatite were not able to 60 A.R. Santos et al. / Geochimica et Cosmochimica Acta 157 (2015) 56–85

Fig. 3. PMgAl X-ray map of UNM Section 3A,3. P-red, Mg-green, Al-blue, Fe–Ti oxides and sulfides shown in white to gray. (A) BSE image of Clast 1B, a parallel slice of Clast 1, a basaltic andesite clast. (B) An altered, devitrified melt clast. Close up BSE image of this clast is in Fig. 8. (C) BSE image of an FTP clast. This clast is one of few of this type to contain pyroxene. (D) Proto-breccia clast with a distinct matrix. (E–G) Lithic clasts containing matrix and larger mineral grains. These clasts do not appear very distinct from the bulk matrix, but they can be distinguished in BSE images. Further textural and chemical analysis is required to determine if these are altered melt clasts or proto-breccia clasts. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) be analyzed quantitatively in some clasts due to small grain 166Er, 174Yb, and 138Ba, as well as on the background size. For these clasts, the average mineral composition from position to determine instrument noise. Phosphate mea- their clast group was used in calculating their bulk surements included 40Ca in place of 30Si. Absolute con- composition, as these phases were relatively homogenous centrations of the trace elements were calculated by first among groups. Apatite composition for Clast 1 was deter- normalizing peak intensity to known SiO2 concentration mined from the average apatite composition for all clasts, (or CaO, in the case of the phosphates) as an internal cor- as apatite was the most uniform phase and there are no rection for ion yield, and then calibrating using the other clasts in a group with Clast 1 that contain apatite. empirical relationship between measured peak intensity and the (independently measured) known concentration 2.4. Secondary ion mass spectrometry within the standard. Reported results surpass the 2r cal- culated detection limit (defined as twice the standard Trace elements and rare earth elements (REEs) in apa- deviation of analysis for any given unknown). tite, pyroxene, and feldspar within igneous clasts were Calibration curves were customized to the phases being measured using a Cameca ims 4f secondary ion mass analyzed (i.e., Durango fluorapatite was used as a stan- spectrometer in the IOM at UNM. Samples were bom- dard for phosphates, plagioclase from Labrador, Canada barded with a primary beam of O- ions accelerated and the Moore County meteorite, composition given in through a 10 kV nominal potential. A 30 nA beam cur- Papike et al., 1997, were used as standards for feldspar, rent was used resulting in a 35 lm spot size. A 25 lm and a clinopyroxene megacryst from Kakanui, New aperture was used for some analyses as mineral grain sizes Zealand, composition given in Mason and Allen, 1973, within clasts were typically smaller than 35 lm. Secondary was used for pyroxene analyses). Complete REE patterns ions were filtered using a sample offset voltage of 75 V. could not be measured on all phases, so measured values Analyses consisted of 10 cycles of peak and background were extrapolated to infer the complete REE pattern (e.g., counting. Measured peaks for pyroxene and feldspar were the LREE slope of feldspar measurements was inferred 30Si, 88Sr, 139La, 140Ce, 146Nd, 147Sm, 151Eu, 153Eu, 163Dy, for the HREE slope in feldspars). A.R. Santos et al. / Geochimica et Cosmochimica Acta 157 (2015) 56–85 61

Table 1 Summary of characteristics of different clast types observed in this study. Number of observations indicates number of clasts used to establish distinguishing characteristics of the clast group. Clast type (number Distinguishing characteristics Reference of observations) sample Proto-breccia (4) Fine grained matrix surrounding coarser grained mineral fragments and other types Fig. 7 of clasts; distinct texturally or compositionally from bulk matrix. Matrix may contain different proportions and distributions of silicate phases, phosphates, and Fe–Ti oxides than bulk matrix Melt clasts (7) Contains devitrified glassy material or skeletal/plumose crystals, may contain relict Figs. 8 and grains. Devitrified mesostasis may contain numerous small Fe–Ti oxides, olivine 3B may be present as skeletal/plumose crystals, common plagioclase and pyroxene crystals Igneous clasts (30) Interlocking mineral grains lacking matrix material. This clast group is subdivided into four separate groups listed below Basalt clasts (19) Basalt bulk composition. Textures include subophitic (with either strongly anhedral Clasts 2, 5, 6, grains, subhedral grains, or irregular grains) and granulitic. DominatedP by plagioclase 73 and pyroxene with minor apatite and Fe–Ti oxides. Mg#s 53–62, Fe3+/ Fe 0.06–0.35 Trachyandesite clasts (3) Trachyandesite bulk composition. Contain poikilitic texture and subophitic texture with Clast 56 irregular grain boundaries. ContainP abundant K-feldspar typically poikilitically enclosing plagioclase. Mg#s 54–57, Fe3+/ Fe 0.08–0.29 Basaltic andesite clasts (2) Basaltic andesite bulk composition.P Granulitic texture. Very minor amounts of apatite and Clast 1 Fe–Ti oxides. Mg#s 57, 28, Fe3+/ Fe 0.03, 0.1 FTP clasts (6) Basaltic texture. Relatively large apatite and Fe–Ti oxide grains tending towards euhedral FTP clast 64 morphologies; frequently lackP pyroxene. P2O5 rich, Mg#52 when containing pyroxene, 10 when lacking pyroxene; Fe3+/ Fe 0.24–0.38

3. RESULTS based on chemical constraints and mineralogy. Figs. 1–3 show combined X-ray maps for three thin sections, with Lithologic domains within NWA 7034 were distin- examples from different lithologic domains outlined (maps guished on the basis of texture and mineralogy. The for all other thin sections and probe mounts are found in domains identified are as follows: bulk rock matrix, igneous the Supplementary Material). clasts, proto-breccia clasts, and melt clasts. The igneous clasts were further subdivided based on composition and 3.1.1. Bulk matrix mineralogy using igneous rock classification guidelines The bulk matrix domain consists of the interconnected, (i.e., Le Maitre, 1984, 2002; Le Bas and Streckeisen, 1991) fine-grained (0.1–5 lm), crystalline groundmass that holds approved by the International Union of Geological the meteorite together. The fine-grained portion of the Sciences (IUGS). We identified several rock types repre- groundmass consists of 0.1–1 lm sized pyroxene, plagio- sented as clasts within NWA 7034 including basalt, basaltic clase, phosphate, and Fe–Ti oxide phases, including a andesite, trachyandesite, and FTP clasts (phosphate and nanophase Fe-hydroxide (Muttik et al., 2014a). This mate- Fe–Ti oxide rich lithology like those described by Owens rial is holocrystalline (i.e., no glass or amorphous compo- and Dymek, 1992 that falls outside traditional IUGS vol- nent) and shows evidence of mild thermal annealing canic rock classification parameters). Importantly, we based on interlocking submicron grain boundaries that emphasize that these given rock names are not meant to meet at 120° angles (Muttik et al., 2014a). The bulk matrix imply a petrogenetic heritage, a priori, and are only meant domain is the primary host for coarser grained materials in to provide a description of the clasts, as stated in the IUGS NWA 7034 such as clasts and mineral fragments, which are classification schemes (e.g., Le Bas and Streckeisen, 1991; both suspended throughout the meteorite. Details of each Le Maitre, 2002). Table 1 summarizes the major observa- specific clast type that has been identified will be provided tions for each lithologic domain, and can be used as a guide in the following sections. The large single mineral fragments for classifying these materials. Petrographic descriptions of are pervasive throughout the meteorite; however they will each lithologic domain are given below, with detailed min- not be discussed in this work beyond the present section eralogy and petrology provided for the igneous clast group, because they lack the petrologic context required to proper- which is the primary focus of the present study. ly identify their source/origin. The larger mineral fragments range from >5 lm to a few millimeters in diameter, with 3.1. Textural description and mineralogy of various lithologic smaller fragments being more abundant. The larger frag- domains ments consist of the same minerals as the fine grained groundmass. Fig. 4 contains images of the bulk matrix Textural differences exist among the lithologic domains showing the size variability of mineral fragments. This within NWA 7034. These textural differences provide the domain was previously referred to as a fine grained basaltic basis for our classification scheme for this meteorite, and porphyry by Agee et al. (2013) and as interclast crystalline further subdivisions within each domain are determined matrix (ICM) by Humayun et al. (2013). 62 A.R. Santos et al. / Geochimica et Cosmochimica Acta 157 (2015) 56–85

Fig. 4. BSE images of different portions of typical NWA 7034 clast-bearing bulk matrix. (A) Contains less coarse grained clasts, but can still be considered poorly sorted. (B) Contains some larger mineral fragments and a small igneous clast in the lower right.

Fig. 5. BSE images illustrating the textural variation among the basalt igneous clasts. Px-pyroxene, pl-plagioclase, ap-apatite, ox-Fe–Ti oxides, ksp-alkali feldspar. (A) Clast 3 contains a subophitic texture with anhedral grains and microphenocrysts. (B) Clast 57 has a subophitic texture with anhedral grains and a more equal distribution in grain size (note: the large plagioclase on the right side of the clast was not included in the clast analysis so as not to bias the composition and mineralogy data). (C) Clast 78 has a subophitic texture with anhedral grains and small regions of alkali feldspar poikilitically enclosing rounded plagioclase. Many pyroxenes also display patchy Fe–Mg zoning. White regions following grain boundaries are remnant gold coating. (D) Clast 5 has a granulitic texture with grains that tend more towards euhedral and equant morphologies than in the other three clasts; note the triple junctions between silicate grains. Apatite is also larger in this clast. The dashed yellow line outlines a region of small Fe–Ti oxides mixed with pyroxene; this region was not considered in the clast analysis as it may be due to an open system process. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

3.1.2. Clasts Table 1. A textural description of each clast type is provid- The bulk matrix domain of NWA 7034 hosts a variety of ed below. polymineralic lithic fragments that we refer to collectively as clasts. The three main clast types identified based on tex- 3.1.2.1. Igneous clasts. Igneous clasts are clasts consisting of tural characteristics and mineralogy are igneous clasts, pro- interlocking mineral grains that lack matrix material (we to-breccia clasts (clasts deriving from a breccia(s) that are using the term “igneous” to denote holocrystalline existed prior to the assembly of the NWA 7034 parent material formed by crystallization of silicate melt). We rock), and melt clasts. A summary of these clast types as identified thirty igneous clasts in the present study, and they well as images are provided in Figs. 1–3 and 5–8 and appear to be the most abundant (numerically) clast type A.R. Santos et al. / Geochimica et Cosmochimica Acta 157 (2015) 56–85 63

Fig. 6. BSE images of Clast 1, a basaltic andesite clast with granulitic texture. Note the larger size of the silicates relative to basalt clasts, subhedral grains, and frequent triple junctions. B shows a close-up of the region outlined in A to illustrate the high Ca pyroxene lamellae (labeled cpx) present throughout this clast (low Ca pyroxene host labeled opx). The small apatite in B is the typical occurrence of apatite in this clast type. Labels are the same as in Fig. 1.

Fig. 7. BSE images of a proto-breccia clast. (A) Proto-breccia clast (outlined in yellow) shown in Fig. 1. (B and C) Close up views of the texture of the clast shown in A. These clasts contain a very fine grained, crystalline matrix, but also have larger mineral fragments suspended throughout. This clast matrix has a higher proportion of feldspar and pyroxene relative to the bulk matrix, with less phosphates and less pervasive Fe–Ti oxides. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) within NWA 7034. The clasts themselves vary in size from oxides, and in some cases contain alkali feldspar and minor 0.04 mm2 to 3mm2. While their edges are smooth and phases such as zircon, Fe-sulfide (occasionally altered), and rounded, the clasts vary in shape but tend to appear oblong rutile. Both subophitic (17 clasts) and granulitic (2 clasts) in the two dimensions visible in thin section. A wide variety textural subtypes are found among the basalt clasts of textures are exhibited by this clast group. We have sub- (Fig. 5). Further description of these subtypes is provided divided each of the igneous clast types based on their bulk below. Both Humayun et al. (2013) and Agee et al. (2013) composition and mineralogy, so that we can report the tex- recognized clasts with similar subophitic textures (called tures of each specific igneous clast type. A summary of the microbasalts by Humayun et al., 2013 and gabbro by details of the igneous clasts identified and described in the Agee et al., 2013). present study is provided in Table 2, which includes refer- Clasts exhibiting a subophitic texture (a list of clasts ences to igneous clast images provided in Figs. 1–3, 5, 6 along with their textural type is found in Table 2) contain and 14–16, and in Figs. S1–S10 in the Supplementary subhedral to anhedral plagioclase with an average grain size Materials. of 61 36 lm. Clast 6 displays irregular silicate grain 3.1.2.1.1. Basalt clasts. Basalt clasts were the most abun- boundaries, but the overall texture is consistent with the dant igneous clast type identified in this study, with a total subophitic nature of the other basalt clasts. Clasts 78 and number of nineteen individual basalt clasts comprising this 75 contain regions of poikilitic alkali feldspar, where single group. The basalt clasts range in size from 0.04 to 3 mm2 crystals of alkali feldspar enclose round plagioclase grains. and consist of pyroxene, plagioclase, apatite, and Fe–Ti Eleven clasts showing subophitic texture also appear to 64 A.R. Santos et al. / Geochimica et Cosmochimica Acta 157 (2015) 56–85

Fig. 8. BSE images of different melt clasts. (A) Altered, devitrified melt clast from Fig. 3B. The darker regions (pl) are clusters of small feldspathic crystals (plagioclase composition) while the lighter gray regions (gl/px) are devitrified mesostasis and skeletal pyroxene grains. The clast is rimmed by feldspathic stubs that are again of plagioclase composition. The large bright grains near the center of the clast (FeS) are Fe- sulfides. The dotted line shows a possible cluster of trapped pre-existing material that was incorporated into the melt clast. (B) Close up view of an altered, devitrified melt clast from UNM Section 2,2 showing clusters of small feldspathic crystals, devitrified glass, and skeletal pyroxene being altered to Fe–Ti oxides. (C) Close up view of the melt clast outlined in A of Fig. 2, note the skeletal grain shapes. Olivine in this clast is also being converted to Fe–Ti oxides. This melt clast appears to be much less altered than those in parts A and B. px-pyroxene, ol- olivine, gl-glass. contain microphenocrysts of plagioclase and pyroxene that 5 has a similar size distribution to the Fe–Ti oxides. The lar- reach between 300 and 400 lm in their longest dimension. gest apatite grains in Clast 59 are 28 23 lm. Pyroxene in subophitic textured clasts is also subhedral to 3.1.2.1.2. Basaltic andesite clasts. Two individual basaltic anhedral, with an average grain size of 54 31 lm. Some andesite clasts were identified in this study. Clast 1 is pyroxene grains in Clast 6 contain a high concentration 1.6 mm2 in size, and Clast 77 is 0.9 mm2 in size. of fine grained (less than 1 lm) Fe–Ti oxides that appear Minerals found in this clast group are plagioclase, pyrox- to be the result of a chemical reaction. Fe–Ti oxides in ene, and Fe–Ti oxides, with minor apatite also present in the subophitic clasts are typically anhedral grains on the Clast 1. Both clasts in this group have a granulitic texture order of 10 s of microns in size, and can also occur as inclu- (Fig. 6). The basaltic andesite clasts exhibit numerous sions within other mineral phases. Apatite is also typically 120° triple junctions between silicate grains and the pres- anhedral, grains are on the order of 10 s of microns in size, ence of coarser silicate phases. Plagioclase and pyroxene are and can occur as inclusions within silicate phases. Two subhedral to euhedral, with average grain sizes of clasts (Clasts 73 and 75) within the subophitic texture group 141 115 lm and 91 64 lm, respectively. Pyroxene in contain mostly subhedral to euhedral silicate mineral Clast 1 contains fine lamellae of high Ca pyroxene grains, and may represent a textural gradation between (Fig. 6B). Fe–Ti oxides occur as anhedral grains averaging the subophitic textured clasts and the granulitic textured 26 18 lm in size. Apatite is present in Clast 1 as <1 lm clasts. diameter inclusions within silicate phases that were identi- The granulitic textural subtype is characterized by fre- fied by Energy Dispersive Spectroscopy (EDS). quent 120° triple junctions between silicate phases and 3.1.2.1.3. Trachyandesite clasts. Three distinct clasts were is found in Clasts 5 and 59. Plagioclase is subhedral to euhe- identified as trachyandesite clasts in this study. The tra- dral and ranges in size from 29 26 lmto89 57 lm. chyandesite clasts range in size from 0.08 to 1 mm2, and Pyroxene is also subhedral to euhedral and ranges in size contain plagioclase, pyroxene, alkali feldspar, Fe–Ti oxides, from 27 25 lm to 115 59 lm. Fe–Ti oxides are subhe- and apatite. Poikilitic and subophitic textural subtypes are dral, and have a range of sizes in Clast 5, with some greater present among these clasts. than 50 lm in diameter and others less than 10 lm in dia- Two clasts in this group display a poikilitic texture, meter. The largest Fe–Ti oxide grains in Clast 59 are where alkali feldspar encloses rounded plagioclase and 24 23 lm. Apatite is subhedral to euhedral, sometimes pyroxene grains (Fig. 1B). Plagioclase is anhedral and occurring as inclusions within silicate phases, and in Clast rounded with an average grain size of 31 24 lm. A.R. Santos et al. / Geochimica et Cosmochimica Acta 157 (2015) 56–85 65

Pyroxene is also anhedral and has an average grain size of distinguished as a separate clast type. The matrix within 33 24 lm. Some subhedral Fe–Ti oxide grains are pre- these clasts has a different texture and/or phase abundances sent, but many are anhedral grains or small inclusions with- than the bulk matrix domain (e.g., Fig. 7). Clast boundaries in silicate phases. The largest Fe–Ti oxide phases are on the are sometimes more diffuse than those of igneous clasts. order of tens of microns in diameter. Apatite has similar Proto-breccia clasts tend to be larger than igneous clasts, petrographic characteristics as the Fe–Ti oxides, where ranging in size from 0.9 0.2 mm to 11 12 mm. grains can be subhedral or anhedral and also occur as inclu- Mineral fragments tend to be the same mineralogy as sions within silicate phases. The largest apatite grains are observed in the bulk rock matrix and include feldspar, approximately 30 lm in diameter. pyroxene, Fe–Ti oxides, and apatite. Figs. 1A, 2B, and 7 One trachyandesite clast has a subophitic texture show examples of this clast type. (Fig. 1C). Silicate phases within this clast have highly irre- gular mineral grain boundaries, similar to those seen in 3.1.2.3. Melt clasts. Melt clasts, many of which occur as basalt Clast 6. The silicate phases in the trachyandesite clast spherules, consist primarily of devitrified glass or devitrified tend to have higher aspect ratios than those in basalt Clast glass with acicular phenocrysts of olivine and pyroxene 6. Feldspar grains, both plagioclase and alkali feldspar, are with minor plagioclase and Fe–Ti oxides. Minerals formed subhedral to anhedral and 83 45 lm in size on average. from the cooled melt are skeletal or plumose. Typical min- Pyroxene is also subhedral to anhedral and has an average erals include pyroxene, feldspar, and olivine. Fe–Ti oxides grain size of 33 18 lm. Fe–Ti oxides tend to be elongate, are also present throughout the devitrified mesostasis and anhedral grains, the largest of which is 24 9 lm, although at the edges of the mineral grains that crystallized from one subhedral grain 44 43 lm in size is present. Apatite is the melt spherule. These clasts range in size from present as subhedral grains averaging 14 6 lm in size. 0.7 0.6 mm to 4 3 mm in the two dimensions measur- 3.1.2.1.4. FTP clasts. Six clasts were identified as FTP clasts able in thin section, however some have been found to con- in this study, and they range in size from 0.04 to 0.3 mm2, tinue through multiple parallel slices, making them some of which is within the low end of the size range defined by the the largest clasts in the meteorite. Most of these clasts are other igneous clasts. A clast of this type was previously rounded and have distinguishable but somewhat diffuse identified by Agee et al. (2013; see their Supplementary boundaries with the bulk rock matrix. Different examples Materials) as an “apatite–ilmenite–alkali feldspar cluster.” of this clast type are shown in Figs. 2A, 3B, and 8. At least FTP clasts are composed of plagioclase, apatite, and Fe– one of these clasts was identified by Agee et al. (2013) and Ti oxides (typically ilmenite), and sometimes include pyrox- Humayun et al. (2013) also identified clasts of this nature as ene, alkali feldspar, and Fe-sulfides. One FTP clast, FTP melt spherules and CLIMR (clast-laden impact melt rock) Clast 1, contains regions of what appears to be a mesosta- when they were interpreted to contain precursor materials. sis-like material consisting predominantly of acicular grains Udry et al. (2014b) analyzed a “vitrophyric clast” initially of silicates. FTP clasts typically display a basaltic texture identified in Agee et al. (2013) matching this textural that is distinct from the textures of the other igneous clast description in a thin section made from a parallel slice of types (Figs. 2F and 3C). Plagioclase grains are subhedral UNM Section 1B,2 (i.e., the clast analyzed by Udry et al., to euhedral and typically elongate, ranging in size from 2014b is a slice from the same clast that is shown in 459 139 lmto37 24 lm. Pyroxene, when present, is Figs. 2A and 8C) and determined it was formed due to anhedral and averages 62 63 lm in size. Fe–Ti oxides impact melting based on its high nickel content (1020 ppm). are present as euhedral, rectangular grains that are typically The textures within these clasts indicate they formed due much larger than the Fe–Ti oxides present in the other to rapid cooling of a melt, however further classification of igneous clast groups (average grain size of 77 29 lm). these clasts (i.e., impact versus volcanic derivation) is diffi- Regions of fine-grained oxides mixed with other phases cult as it is not possible to determine the origin of the melts such as pyroxene or apatite are found in all of the FTP from texture alone. Several tests and sets of criteria have clasts identified in this study. Apatite grains tend to be sub- been developed to distinguish between impact and volcanic hedral to euhedral and average 50 32 lm in size, which is origin of melt clasts in planetary regolith materials (e.g., also larger than the average apatite within the other igneous Delano, 1986; Fagan et al., 2013; Neal et al., 2015), howev- clast groups. er they are not appropriate for use in martian systems or cannot be applied specifically to the melt clasts within 3.1.2.2. Proto-breccia clasts. Clasts with non-igneous tex- NWA 7034. Suites of criteria, such as those proposed by tures are grouped under the term proto-breccia clasts. Delano and Livi (1981) and Delano (1986) for classifying These clasts typically consist of fine grained matrix sur- pristine lunar glasses, are not directly applicable to Mars, rounding coarser mineral fragments or other clast types as we do not have enough knowledge of the martian surface (e.g., basalt clasts). The matrix of this clast type is com- or mantle to know if the processes responsible for the lunar posed of fine grained, crystalline mineral phases, similar criteria operate the same way on Mars. Specifically, highly to the bulk matrix. This combination of coarse material siderophile element (HSE) criteria, such as the uniform and surrounded by fine material is typical of breccias, and they Mg-correlated Ni abundances discussed by Delano and Livi likely represent fragments of a breccia that existed prior to (1981) and Delano (1986) cannot be applied to martian the assembly and lithification of NWA 7034. Their matrix melt spherules in NWA 7034 at present because their sec- and grain boundaries are distinct from the interconnected ondary alteration history is unknown. If the materials were bulk matrix domain of NWA 7034, allowing them to be derived from an oxidized source (portions of the martian 66 A.R. Santos et al. / Geochimica et Cosmochimica Acta 157 (2015) 56–85

Table 2 List of igneous clasts identified in the present study including their clast type, texture, thin section where they were found, and images included in this paper. Clast name Clast type Texture Thin section Image 2 Basalt Subophitic 3A,2 Figs. S2, S10A 2B Basalt Subophitic 3A,3 Fig. S10B 3 Basalt Subophitic 3A,2 Fig. 5 4 Basalt Subophitic 1B,2 Fig. S10C 5 Basalt Granulitic 1B,2 Figs. 2 and 5 6 Basalt Subophitic (irregular) 1B,2 Fig. 2 31 Basalt Subophitic 3A,2 Fig. S10D 44 Basalt Subophitic 3A,2 Fig. S10E 55 Basalt Subophitic 3B,2 Fig. 14, Fig. S3 57 Basalt Subophitic 3B,2 Fig. 5, Fig. S3 59 Basalt Granulitic 3B,2 Fig. S10F 66 Basalt Subophitic 1B,2 Fig. S10G 67 Basalt Subophitic 1B,2 Fig. S10H 70 Basalt Subophitic 2,3 Fig. S10I 71 Basalt Subophitic 2,3 Fig. S10J 72 Basalt Subophitic 2,3 Fig. S10K 73 Basalt Subophitic 2,3 Fig. 1 75 Basalt Subophitic 2,3 Fig. S10L 78 Basalt Subophitic 2,2 Fig. 5, Fig. S1 80 Basalt Subophitic 2,3 Fig. S10M 1 Basaltic Andesite Granulitic 3A,2 Fig. 6, Fig. S2 1B Basaltic Andesite Granulitic 3A,3 Figs. 3 and 15 77 Basaltic Andesite Granulitic 2,3 Fig. S10N 56 Trachyandesite Poikilitic 3B,2 Figs. S10O, S3 74F Trachyandesite Subophitic (irregular) 2,3 Fig. 1 76 Trachyandesite Poikilitic 2,3 Figs. 1 and 16 FTP Clast 1 FTP Basaltic 1B,2 Fig. 2 FTP Clast 2 FTP Basaltic 1B,2 Fig. S10P FTP Clast 3 FTP Basaltic 3A,2 Fig. S10Q FTP Clast 12 FTP Basaltic 3A,3 Fig. S10R FTP Clast 15 FTP Basaltic BRI chips Fig. S10S FTP Clast 64 FTP Basaltic 1B,2 Fig. S10T

mantle are thought to be more oxidizing than the lunar indicator of impact origin. The fO2 of the martian mantle, mantle, e.g., Herd et al., 2002; Herd, 2003), it would be pos- which can influence HSE abundance in mantle source sible for them to have higher HSE abundances, but if the regions, is not well known. In light of a lack of appropriate oxidation occurred due to a post crystallization process, it criteria to distinguish martian impact and volcanic process- would not necessarily influence the HSE abundances. es, and the known and unknown factors that need further Many of the other tests to discern a volcanic origin are assessment, it is our opinion that these NWA 7034 melt based on olivine characteristics (e.g., Fagan et al., 2013; clasts should serve as a tool for assessing the differences Neal et al., 2015), however this mineral phase is rare to between martian impact melts and volcanic melts, and absent in NWA 7034 clasts. Parameters such as crystal size should be thoroughly studied instead of subjected to the distribution of plagioclase (e.g., Neal et al., 2015) likely can- same classification criteria as melt clasts from other plane- not be determined for a statistically significant number of tary bodies. grains due to the small size of the NWA 7034 clasts. Several known and unknown factors indicate further 3.2. Mineral chemistry of igneous clasts study of this meteorite is needed before formation processes for these melt clasts can be constrained and they can be 3.2.1. Basalt clasts classified as either impact or volcanic in origin. It is known Average mineral modes and average mineral composi- that the NWA 7034 parent rock was subjected to martian tions for each clast group are given in Tables 3 and 4, surface alteration processes during its history (Agee et al., respectively. Basalt clasts have plagioclase modal abun- 2013; Muttik et al., 2014b), and an assessment of how this dances between 33% and 61%, pyroxene modal abundances could affect the HSE abundances and distributions in the between 23% and 57%, Fe–Ti oxide modal abundances NWA 7034 materials has not been conducted. It is also between 4% and 9%, and apatite modal abundances known that at least twenty-three shergottites have been between 2% and 8%. Alkali feldspar was only found in measured to have chondritic HSE ratios (Brandon et al., two of these clasts, Clasts 75 and 78, and the modal abun- 2012), suggesting chondritic HSE ratios alone are not an dance of this phase is 8% and 17% respectively. A.R. Santos et al. / Geochimica et Cosmochimica Acta 157 (2015) 56–85 67

Table 3 Average mineral modes determined for each igneous clast type with 1r standard deviation in the last digit given in parentheses. Only two basaltic andesite clasts were considered, so the value for each clast is given as Clast 1, Clast 77. Mineral modes for all clasts are given in Table S5. Basalt clasts Trachyandesite clasts Basaltic andesite clasts FTP clasts Plagioclase 50(8) 37(9) 53, 70 50(20) Alkali feldspar 2(5) 23(8) 0 4(8) Low Ca pyroxene 20(20) 23(6) 0 10(20) High Ca pyroxene 10(20) 10(8) 44, 27 0 Apatite 5(2) 2.9(9) 1.4, 0 22(9) Magnetite 3(1) 2(3) 0.4, 0.7 4(3) Ilmenite 3(1) 1.1(9) 0.7, 2.3 20(10)

Plagioclase ranges in composition from (less than 3 lm) in many clasts to be analyzed quantitative- An19–55Ab44–74Or1–12; these minerals are unzoned (feldspar ly by EPMA. ternary compositions shown in Fig. 9). Examination of thin sections confirms this is the mineral plagioclase and not the 3.2.2. Basaltic andesite clasts shock-vitrified phase maskelynite. The few occurrences of Basaltic andesite clasts have 53% and 70% modal plagio- alkali feldspar (present in the poikilitic texture mentioned clase (for Clasts 1 and 77 respectively), 44% and 27% modal in Section 3.1.2.1.1) range from Or55–84Ab14–38An2–7. pyroxene, 1% and 3% modal Fe–Ti oxides, and 1% modal Pyroxene occurs as low Ca (Wo 6 5%), moderate Ca apatite (not present in Clast 77). (5% < Wo < 20%), and high Ca (Wo > 20%) pyroxene Plagioclase ranges in composition from (quadrilateral compositions shown in Fig. 10). The range An23–41Ab56–72Or2–5 (Fig. 9). Alkali feldspar occurs as small in Mg# (Mg/[Mg + Fe2+] 100 in mol%) for these miner- (less than 10 lm) patches adjacent to or surrounded by albitic als is between 47 and 85 (average of 63). Some pyroxenes in feldspar in Clast 1B (a parallel slice of Clast 1, Figs. 3A and these clasts display patchy Fe–Mg zoning. Clast 75 contains 15). Two feldspar grains were measured for REEs and trace a pyroxene grain with a small, euhedral core of higher En elements (Fig. 11, Tables 5 and S7). HREEs in these feldspars pyroxene (Figure S10L). Other clasts (Clast 73) contain were below detection limits, but La, Ce, and Nd could be grains or portions of grains with higher En content as well, measured. The average La/NdCI for the two grains is 2.39. but these are a minor occurrence, and on average, normal Eu was detected and displayed a large positive anomaly, igneous zoning is absent from this clast group. Three pyrox- however this value should be taken as an upper limit as there ene grains in a single basalt clast (Clast 78) were analyzed is a BaO interference with Eu. Sr was present in these grains at for REEs (Table 5, Table S7). These grains are low Ca 412 ppm, and Ba was present at 104 ppm. pyroxene, and have relatively flat REE patterns at around Pyroxene in Clast 1 has an Mg#59 throughout the 25CI (Fig. 11). The average CI normalized La/Yb ratio clast (Fig. 10), and fine lamellae of high Ca pyroxene are (La/YbCI) for these grains is 0.95, with an average La/ present in all pyroxene grains. Pyroxene near the domains SmCI of 1 and Dy/YbCI of 0.85. The REE patterns for these of alkali feldspar contains higher Al than typically seen in grains also have a negative Eu anomaly. NWA 7034, and has an aqueous alteration texture that is Magnetite is the most common Fe–Ti oxide (Fig. 12), distinct from the texture of other igneous clast pyroxene. although ilmenite is also present both as oxyexsolution The texture suggests these localized mineral features derive within larger magnetite grains and as individual grains from a secondary fluid process; this process had a limited (Fig. 13). Clasts 75 and 6 contain a high Cr oxide phase effect on the bulk clast composition based on the small area (20–50 wt% Cr2O3), which is likely an inverse spinel of alteration. Pyroxene in Clast 77 is high Ca pyroxene and containing 30–78% chromite component, 12–59% mag- has an Mg#31 throughout the clast; lamellae are absent netite component, 7–10% spinel component, and 1–4% from pyroxenes in this clast. Two pyroxene grains from ulvo¨spinel. Clast 1 were measured for REEs. The REE patterns are A single rounded olivine grain, approximately 10 lmin consistent with each other, and have a La/YbCI of 0.04 diameter, was found enclosed within a pyroxene in Clast 55 and 0.05. The HREE have a higher slope, with Dy/YbCI (Fig. 14). The olivine is Fo68 in composition (Table 4) and of 0.75, as opposed to the LREE with La/SmCI 0.09. the pyroxene enclosing it is En59Fs31. Pyroxene bordering The Eu anomaly is negative in both of these patterns. one side of the olivine is high Ca pyroxene. Small Fe–Ti Fe–Ti oxides are typically magnetite, but ilmenite is also oxides are present within and around this olivine. The tex- present. Clast 1 contains Cr-rich magnetite that ranges ture of this olivine grain is suggestive of resorption. This is from 18% to 35% chromite component, 5% to 10% spinel the only olivine observed within an igneous clast, although component, 2% to 5% ulvo¨spinel component, and 53% to some olivine has been identified in partially crystallized 75% magnetite component. Apatite in Clast 1 is too small devitrified melt clasts. to be analyzed quantitatively (present as micron scale inclu- Minor phases such as zircon and sulfides are typically sions in silicate minerals), but qualitative analysis using less than 10 lm in size. Apatites are Cl-rich, and too small EDS confirms it is also Cl-rich. 68 A.R. Santos et al. / Geochimica et Cosmochimica Acta 157 (2015) 56–85

Table 4 Average mineral compositions for each igneous clast type, values listed are oxide wt% with 1r standard deviation in the last decimal place given in parentheses. *-olivine composition is from a single grain in Clast 55. A complete listing of mineral analyses used in this study is located in the Tables S1–S4. Basalt clasts Basaltic andesite Plagioclase Alkali Low Ca High Ca Apatite Magnetite Ilmenite Olivine* Plagioclase (n = 314) feldspar pyroxene pyroxene (n = 52) (n = 79) (n = 23) (n = 29) (n =9) (n = 156) (n = 99)

SiO2 58(2) 65.1(4) 53.3(9) 52.2(7) 0.3(2) 0.2(2) 0.10(8) 36.82 59(2) TiO2 0 0 0.2(1) 0.4(2) 0.01(2) 3(3) 45(1) 0.03 0 Al2O3 26(2) 19.0(2) 0.5(3) 0.9(5) 0.1(1) 1(1) 0.2(1) 0.09 26(1) Cr2O3 0 0 0.11(8) 0.2(1) 5(10) 0.06(8) 0.04 0 Fe2O3 0.01(2) 0.04(4) 50(10) 13(5) FeO 0.6(2) 0.4(2) 22(3) 20(4) 0.7(2) 32(3) 32(3) 28.41 0.43(7) MnO 0 0 0.7(1) 0.6(1) 0.06(2) 0.3(3) 1.6(5) 0.56 0 CoO 0.05(1) 0.05(1) 0.10(5) 0.05(5) 0.04 NiO 0.04(3) 0.08(8) 0.25(7) 0.02(3) 0.10 MgO 0.03(6) 0.02(6) 22(2) 18(3) 0.1(1) 0.8(7) 4(1) 33.92 0.01(2) CaO 8(3) 0.8(3) 1.5(4) 6(5) 53.5(7) 0.09 7(1)

Na2O 6(1) 2.9(9) 0.01(2) 0.1(1) 0.16(4) 0 7.2(6) K2O 1(3) 12(2) 0.01(1) 0.01(3) 0.6(2) P2O5 40.5(5) H2O 0.24(7) F 0.7(2) Cl 4.6(4)

Basaltic andesite (con’t) Trachyandesite clasts Low Ca High Ca Magnetite Ilmenite Plagioclase Alkali feldspar pyroxene pyroxene (n = 12) (n =5) (n = 33) (n = 31) (n =3) (n = 16)

SiO2 52.6(2) 51(1) 0.1(2) 0.2(2) 62(2) 64.6(6) TiO2 0.32(2) 0.34(4) 3(4) 46.0(9) 0 0 Al2O3 0.48(4) 0.6(1) 2(1) 0.24(2) 24(1) 19.2(2) Cr2O3 0.13(2) 0.06(8) 7(9) 0 0 0 Fe2O3 0.03(3) 0.01(2) 50(10) 11(3) FeO 24.9(5) 24(1) 31(5) 34(1) 0.38(8) 0.4(1) MnO 0.59(4) 0.8(2) 0.5(4) 1.3(2) 0 0 CoO 0.05(1) 0.10(5) 0.03(4) NiO 0 0.25(3) 0.03(2) MgO 19.5(2) 12(6) 0.8(5) 3.6(5) 0.01(4) 0.01(1) CaO 2.1(2) 11(7) 5(1) 0.6(2)

Na2O 0.04(4) 0.15(8) 8.4(7) 2.2(5) K2O 0.01(2) 0.02(2) 0.7(1) 13.0(8) P2O5 H2O F Cl

Trachyandesite clasts (con’t) FTP clasts Low Ca High Ca Apatite Magnetite Ilmenite Plagioclase Alkali Low Ca pyroxene pyroxene (n =8) (n = 12) (n =9) (n = 33) feldspar pyroxene (n = 63) (n = 11) (n =2) (n =3)

SiO2 52.5(9) 53(1) 0.3(2) 0.2(2) 0.10(5) 59(2) 64.8(5) 53.4(4) TiO2 0.17(4) 0.31(9) 0 4(4) 43(2) 0.11(6) 0.10(1) 0.33(3) Al2O3 0.3(2) 0.7(2) 0.06(5) 1.0(7) 0.3(2) 26(1) 19.0(2) 0.4(1) Cr2O3 0.07(2) 0.2(1) 1(1) 0.1(1) 0 0 0.02(3) Fe2O3 0.01(2) 0.03(4) 53(8) 17(4) 0.03(3) FeO 25(4) 12(3) 0.7(2) 32(7) 31(3) 0.6(2) 0.5(1) 21(1) MnO 0.7(1) 0.31(1) 0.07(3) 0.2(2) 1.2(5) 0 0 0.54(3) CoO 0.06(1) 0.02(2) 0.10(5) 0.03(4) NiO 0.06(1) 0.03(2) 0.5(8) 0.05(4) MgO 20(3) 13(2) 0.08(6) 0.6(5) 3.9(6) 0.04(9) 0 22(1) A.R. Santos et al. / Geochimica et Cosmochimica Acta 157 (2015) 56–85 69

Table 4 (continued) Trachyandesite clasts (con’t) FTP clasts Low Ca High Ca Apatite Magnetite Ilmenite Plagioclase Alkali Low Ca pyroxene pyroxene (n =8) (n = 12) (n =9) (n = 33) feldspar pyroxene (n = 63) (n = 11) (n =2) (n =3) CaO 1.2(3) 21(2) 53.5(7) 7(1) 0.8(2) 1.7(4)

Na2O 0 0.26(5) 0.19(4) 7.3(7) 4(2) 0.02(3) K2O 0.02(2) 0.02(1) 0.6(2) 11(3) 0.04(1) P2O5 40.5(6) H2O 0.22(6) F 0.8(5) Cl 4(1) FTP clasts (con’t) Apatite (n = 12) Magnetite (n = 4) Ilmenite (n = 19)

SiO2 0.3(3) 0.4(5) 0.1(1) TiO2 0 6(3) 45(1) Al2O3 0.01(2) 1.8(2) 0.13(5) Cr2O3 2(2) 0.04(6) Fe2O3 49(6) 13(3) FeO 0.5(2) 34(3) 35(2) MnO 0.05(2) 0.3(2) 1.2(3) CoO 0.09(6) 0.03(5) NiO 0.28(1) 0.01(2) MgO 0.12(8) 0.7(2) 2.3(7) CaO 53.4(4)

Na2O 0.17(4) K2O P2O5 40.1(7) H2O 0.3(1) F 0.5(1) Cl 4.6(6)

Fig. 9. Feldspar ternary diagrams for the basalt clasts (A), basaltic andesite clasts (B), trachyandesite clasts (C), and FTP clasts (D). The feldspar compositional ranges of the four clast types overlap, although the clast types do not have the same ranges. 70 A.R. Santos et al. / Geochimica et Cosmochimica Acta 157 (2015) 56–85

Fig. 10. Pyroxene quadrilateral diagram for the different igneous clast groups. (A) Basalt clasts. Most individual clasts in this group contain a range in Mg contents in pyroxene. (B) Basaltic andesite clasts. Both of the clasts in this group have a high degree of Fe and Mg equilibration in their pyroxenes. (C) Trachyandesite clasts. These clasts show a similar range in pyroxene Mg number as the basalt clasts. (D) FTP clasts. The few pyroxenes analyzed from this clast group have low Ca contents.

Table 5 REE contents of minerals from different igneous clast groups determined using SIMS. Values given are ppm element with 1r standard deviations listed in parentheses after the appropriate digit. n = number of analyses. Pyroxene Apatite Feldspar Clast type Basalt n = 3 Basaltic andesite n = 2 Trachyandesite n = 1 FTP n = 2 Basaltic Andesite n =2 La (ppm) 6(2) 0.10(2) 0.696 34(9) 2.27(5) Ce 14(3) 0.49(6) 1.674 80(30) 3.8(2) Nd 11(2) 1.0(1) 1.080 50(20) 1.8(1) Sm 3.7(2) 0.68(6) 18(7) Eu 1(1) 0.086(0) 0.293 9(1) 4.446(9)a Dy 5.7(3) 1.81(2) 1.657 16(3) Er 4.1(5) 1.42(8) 1.566 8(2) Yb 4.6(3) 1.61(7) 2.812 7(1) a Upper limit of Eu concentration due to BaO interference. REE contents of individual minerals are listed in Table S7.

3.2.3. Trachyandesite clasts observed in the basalt clasts, but normal igneous zoning is Trachyandesite clasts contain modal abundances of pla- again absent. A single low Ca pyroxene was analyzed from gioclase ranging from 28% to 45%, pyroxene abundances this clast group (En49Fs48) for REEs, and the resulting CI ranging from 30% to 35%, alkali feldspar abundances rang- normalized REE pattern is shown in Fig. 11. The La/ ing from 16% to 32%, Fe–Ti oxide abundances ranging YbCI value is 0.2, indicating a LREE depleted pattern. from 2% to 5%, and apatite abundances ranging from 2% The La/NdCI for this pattern is 1.24 while Dy/YbCI is 0.39. to 4%. Apatite is also Cl-rich and often too small to analyze by This clast group contains a plagioclase compositional EPMA. Oxide phases tend to be magnetite but ilmenite is range that is distinct from the other igneous clasts in that also present. it does not reach An contents higher than 30% and some grains contain very low An content (An5–30Ab67–88Or3–7) 3.2.4. FTP clasts (Fig. 9). Alkali feldspar ranges in composition from FTP clasts are more variable in their mineral modes Or68–90Ab9–29An1–5. Qualitative X-ray maps of one tra- than the other clast groups. FTP clasts contain between chyandesite clast show a heterogeneous distribution of K 15% and 70% plagioclase. The clast containing pyroxene within the alkali feldspar (Fig. 16). contains 37% modal pyroxene. A single clast contains alkali Both high and low Ca pyroxene is present in these clasts, feldspar, which has a modal abundance of 19%. Apatite and Mg#s range from 48 to 87 (average of 59) (Fig. 10). ranges in abundance from 9% to 33%, and Fe–Ti oxides Core-rim zoning of Fe and Mg is found in some pyroxenes range from 7% to 38% of the mode. in these clasts. Zones contain diffuse boundaries and Mg Plagioclase ranges in composition from content increases going from the core to the rim of the An19–41Ab57–75Or2–6 (Fig. 9). When present, alkali feldspar grains (Fig. 16C and D). Some pyroxene grains with patchy has an average composition of Or64Ab32An4. Pyroxene is Fe–Mg zoning are also present, which is similar to textures low Ca in composition with an average Mg# of 65. Apatite A.R. Santos et al. / Geochimica et Cosmochimica Acta 157 (2015) 56–85 71

Fig. 11. CI normalized REE contents from different mineral grains acquired by SIMS (CI data from Anders and Grevesse, 1989). Dashed lines represent interpolated values. REE contents of similar phases measured in QUE 94201 (McSween et al., 1996), Shergotty (pyroxene and maskelynite from Wadhwa et al., 1994; apatite from Lundberg et al., 1988), and NWA 7034 bulk rock (Agee et al., 2013) are shown in each panel for comparison. (A) REE contents from pyroxene measured in basalt Clast 78 (black circles) and trachyandesite Clast 76 (green circles). The Wo content and Mg# of the measured pyroxenes are listed on the figure. Sm and Gd in Clast 76 were below the detection limit. (B) REE contents from pyroxene measured in basaltic andesite Clast 1 (red circles) along with pyroxene Wo content and Mg#. The NWA 7034 igneous clast pyroxenes have higher REE concentrations than those in the other martian meteorites, and the low Ca pyroxenes appear to be from a more LREE enriched magma than those from Shergotty based on LREE pattern slopes. (C) REE content of feldspar in basaltic andesite Clast 1. Eu content is an upper limit due to likely BaO interference. Dashed lines represent interpolated values. NWA 7034 values are higher than those for the other two martian meteorites, and the LREE trend in Clast 1 tends to follow that shown by the enriched shergottite Shergotty. (D) REE concentrations of two chlorapatite grains in FTP Clast 3. Dashed lines represent interpolated values. The small positive Eu anomaly in the FTP clast data is not confirmed in other NWA 7034 phosphates, but the overall apatite pattern is again close to that found in Shergotty apatite. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

is also Cl-rich in these clasts. REEs in apatite from one of the- clasts fall close to the picrite field, however it is more appro- se clasts have been measured, and these minerals have La/ priate to classify them as Fe-rich basalts as the absence of YbCI of 3.23 and a positive Eu anomaly (Fig. 11). Oxide phas- olivine is probably a result of the high abundance of ferric es tend to be ilmenite, but magnetite is also present in some iron. The bulk compositions for these clasts are mostly in clasts. This is opposite of the other igneous clast types which agreement with those reported by Humayun et al. (2013) tend to have more abundant magnetite (Fig. 12). for their CLIMR and microbasalt clasts (Humayun et al., 2013, Supplementary Table S1), although some slight varia- 3.3. Bulk composition and oxygen fugacity of igneous clasts tion does exist. Mg#s (Mg/[Mg + Fe2+] 100 in mol%) for this clast group range from 53 to 62, which is within the 3.3.1. Basalt clasts range for martian olivine-phyric shergottites, but higher A large set of clasts have bulk compositions that fall in than that for basaltic shergottites and lower than that for the basalt field of the Total Alkali vs. Silica (TAS) classifi- lherzoliticP shergottites (Taylor and McLennan, 2009). cation diagram of Cox et al. (1984) shown in Fig. 17; aver- Fe3+/ Fe (determined in the bulk clast composition calcu- age bulk compositions for each igneous clast group are lations, where bulk clast Fe3+ derives from Fe3+ calculated given in Table 6 (a full listing of clast compositions is found in Fe–Ti oxides and pyroxene) ranges from 0.06 to 0.35 in 3+ in Table S6). In other classification schemes, such as that of theseP clasts, which is substantially higher than the Fe / Le Maitre (1984), there are a few clasts whose bulk compo- Fe values for the most oxidized shergottites (Herd, sitions fall in the trachybasalt and tephrite fields. Some 2003; Herd et al., 2001, 2002). 72 A.R. Santos et al. / Geochimica et Cosmochimica Acta 157 (2015) 56–85

Fig. 12. FeO–TiO2–Fe2O3 diagrams showing the Fe–Ti oxide compositions in the basalt clasts (A), basaltic andesite clasts (B), trachyandesite clasts (C), and FTP clasts (D).

Fig. 13. BSE images of oxyexsolution of ilmenite from host magnetite in Clasts 73 (panel A) and 75 (panel B). Note the trellis type texture of the exsolved ilmenite, which is indicative of the oxyexsolution process (Haggerty, 1991). Mag-magnetite, ilm-ilmenite, pl-plagioclase, px- pyroxene, ap-apatite, ksp-alkali feldspar, Au-remnant gold coat.

Many of these clasts contain magnetite–ilmenite pairs, thermodynamic model of many of the tested pairs is poor, which were used with the QUILF program (Andersen thus pairs were only used (regardless of proximity of mag- et al., 1993) to determine fO2 in five clasts. Oxide pairs were netite and ilmenite grains) if they fell on or within the error first checked for equilibrium Mg–Mn distributions after envelope of the defined linear regression line for Mg–Mn Bacon and Hirschmann (1988). Attempts were made to equilibrated oxide pairs from Bacon and Hirschmann use oxide pairs where both oxide grains are in contact with (1988) and showed a good fit to the QUILF thermodynam- each other (i.e., showing petrographic evidence of equilibra- ic models (i.e., low “uncertainty” on calculated temperature tion), however the limited size and abundance of these and fO2). The calculated fO2 values ranged from DFMQ + 2 phases warranted use of nonadjacent pairs within some to DFMQ + 4 over temperature ranges of 611 °C–761 °C individual clasts. The goodness of fit to the QUILF (temperature “uncertainty” ranges from 9 to 64, A.R. Santos et al. / Geochimica et Cosmochimica Acta 157 (2015) 56–85 73

Fig. 14. (A) BSE image of basalt Clast 55. The only olivine found outside of a melt clast is present within a pyroxene grain in this clast. (B) BSE image showing the outlined region in A. Fe–Ti oxides, some of which are Cr-rich, are present along the olivine–pyroxene boundary. ol- olivine, px-pyroxene, cpx-high Ca pyroxene, pl-plagioclase, ap-apatite, ox-Fe–Ti oxide.

3.3.2. Basaltic andesite clasts Two clasts fall in the basaltic andesite field in the TAS classification diagram of Cox et al. (1984). No analogous compositions have been reported by previous authors from this meteorite or its pairs. The most significant feature of these clasts that sets them apart from the basalt clasts is the low abundance of Fe–Ti oxides and apatite; their min- eralogy is therefore dominated by plagioclase and pyrox- ene, leading to their elevated SiO2 contents relative to the basalt clasts. The two clasts comprising this group (Clast 1 and Clast 77) are extremely different from each other (see Section 3.2.2). The Mg#s for these two clasts are 57 (Clast 1) and 28 (Clast 77); the Clast 1 Mg# is within the range established by the basalt clasts, but Clast 77 has a very different Mg# due to the Ca and Fe rich nature of Fig. 15. BSE image of localized alteration in Clast 1B (parallel slice its pyroxene. Oxygen fugacity calculations could not be per- of Clast 1). px-pyroxene, pl-plagioclase, alb-albitic plagioclase, ksp- formed for either of these clasts as quantitative analyses alkali feldspar, ox-Fe–Ti oxide, alt-altered pyroxene region con- could not be obtained from any oxide pairs.P Due to the taining a texture suggestive of aqueous alteration. low abundance of oxide phases, the Fe3+/ Fe for these clasts are 0.03 and 0.1 for Clasts 1 and 77 respectively.

3.3.3. Trachyandesite clasts “uncertainty” in log fO2 values are from 0.1 to 0.8). These A subset of bulk clast compositions fall in the trachyan- temperatures indicate closure of the two oxide system below desite field of the TAS classification diagram of Cox et al. the melt solidus, so we cannot definitively determine (1984) (these clasts fall in the basaltic trachyandesite field whether or not the oxides are recording an fO2 that reflects of the classification scheme of Le Maitre, 1984). No previ- the melt from which they crystallized or a secondary oxida- ous work describes clasts similar to these, although alkali tion process that occurred during subsolidus re-equilibra- feldspar has been observed within NWA 7034 (Agee tion. Some clasts (e.g., Clasts 71, 73, 75) contain larger et al., 2013; Humayun et al., 2013). TrachyandesiteP clast magnetite grains with ilmenite lamellae having a trellis type Mg#s range from 54 to 57. The Fe3+/ Fe range for this texture, which is indicative of subsolidus oxyexsolution clast group is similar yet slightly smaller than that for the (Haggerty, 1991), although this texture is not present in basalt clasts, ranging from 0.08 to 0.29. all of the clasts. It is likely that the clasts did not all expe- The fO2 of Clast 76 was determined from magnetite–il- rience the same cooling/oxidation history, and it is difficult menite pairs using the same methods employed for the at the present time to fully decipher these differences. basalt clasts. The average calculated fO2 is DFMQ + 2 with Regardless of the process that imparted the current oxida- a temperature of 850 °C (temperature “uncertainty” is 56– tion state to these clasts, these relatively high fO2 values rep- 64, log fO2 “uncertainty” is 0.6–0.7). This temperature is resent a highly oxidized material present in the martian also below magmatic solidus temperatures, but there is no crust that was (and is) available to interact with other mar- visible evidence for oxyexsolution of ilmenite in this clast, tian magmas (e.g., as an oxidized crustal assimilant as dis- however discrete ilmenite grains may also form due to post cussed by Herd et al., 2002). crystallization re-equilibration of Fe–Ti oxide phases 74 A.R. Santos et al. / Geochimica et Cosmochimica Acta 157 (2015) 56–85

Fig. 16. (A) BSE image of trachyandesite Clast 76 (same as Fig. 1B); yellow dashed outlines correspond with the qualitative X-ray maps in B– D (B–D: warm colors indicate a higher relative concentration). (B) K (left) and Na (right) maps of region B. Note the uneven distribution of K and Na in the poikilitic alkali feldspar. (C and D) Mg (left) and Fe (right) maps of regions C and D. The large pyroxene clusters in each set of maps show an increase in Mg toward the outer rim of the grains, while higher Fe content is at the center of the grains. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

(Haggerty, 1991), so it is still uncertain what process imparted this calculated fO2 to the clast.

3.3.4. FTP clasts An estimated range for the bulk compositions of the FTP clast type was determined from the average mineral compositions for this clast type and three clasts represent- ing the range in modal mineralogy. The field for this esti- mated range is shown in Fig. 17. These clasts represent a unique lithology that has not been identified previously as a martian rock type. Due to the high abundance of non-sili- cate phases, these clasts fall outside of the fields on the Cox et al. (1984) classification scheme, and would fall in the foi- dite field of the Le Maitre (1984) classification scheme. However, the FTP clasts we have observed in this study are highly variable in their mineral modes, and this is likely due in large part to their small size, so we hesitate at this juncture to assign a formal name to this lithology, as it can- Fig. 17. Total alkali vs. silica (TAS) classification diagram from not be classified by the same methods as the other clast Cox et al. (1984) showing the four igneous clast types discussed types. In order to accurately classify this clast group, we from NWA 7034 along with an average bulk matrix composition. need a much larger sampling to determine what character- Red dashed line divides alkaline and subalkaline igneous rocks istics are actually representative of this group. Therefore, (Irvine and Baragar, 1971). Humphrey crystallization trends are foidite may not be an entirely appropriate name for these from McCubbin et al. (2008) and discussed in the text. Microbasalt clasts, but based on observed mineralogy, the known and CLIMR clasts from NWA 7533 measured by Humayun et al. (2013) are shown for comparison. While the basalt clasts are the lithologies that are most similar to this clast group are ter- most numerous, they show the greatest compositional variation. restrial FTP rocks such as nelsonites and oxide-apatite gab- FTP clast field estimate methods are discussed in the text. (For bronorites (described by Owens and Dymek, 1992). Mg#s interpretation of the references to color in this figure legend, the from bulk composition estimates are around 10 except for reader is referred to the web version of this article.) the clast containing pyroxene, which has an Mg# of 52. A.R. Santos et al. / Geochimica et Cosmochimica Acta 157 (2015) 56–85 75

Table 6 Average bulk compositions for the different igneous clast types with 1r standard deviation in the last digit given in parentheses. Values listed are wt% oxide. Basaltic andesite clast compositions listed are (Clast 1, Clast 77). Individual clast compositions are listed in Table S6. Basalt clasts Trachyandesite clasts Basaltic andesite clasts FTP clasts

SiO2 48(2) 54(1) 53.6, 53.5 31(8) TiO2 2.3(8) 1.0(5) 0.7, 1.9 8(2) Al2O3 12(2) 11.5(8) 12.8, 16.5 10(4) Cr2O3 0.3(3) 0.06(1) 0.2, 0 0.15(5) FeOtotal 14(3) 12(2) 13.3, 10.9 16(4) MnO 0.34(7) 0.27(3) 0.3, 0.4 0.30(8) CoO 0.02(1) 0.02(1) 0, 0.02 NiO 0.03(2) 0.05(4) 0, 0 0.02(1) MgO 9(2) 7.0(4) 9.7, 2.2 3(5) CaO 8(2) 6(2) 6.3, 10.1 15(5)

Na2O 3.0(5) 3.1(9) 3.2, 4.6 3(1) K2O 0.4(4) 3(1) 0.3, 0.4 0.7(8) P2O5 1.9(8) 1.2(4) 0.6, 0 9(3) H2O 0.01(1) 0.01 0, 0 0.07(3) F 0.03(1) 0.02 0.01, 0 0.12(5) ClP 0.2(1) 0.15(6) 0.07, 0 1.1(4) Fe3+/ Fe 0.20(7) 0.2(1) 0.03, 0.10 0.31(7)

P Fe3+/ Fe for the bulk composition estimates ranges from not involved in breccia formation on airless bodies (e.g., 0.24 to 0.38. Oxygen fugacity calculated from magnetite–il- Moon and asteroids). If an impact was involved in the col- menite pairs (using the same methods as for the previous lection and transport of the NWA 7034 materials, it did not clast types) ranges from DFMQ – 0.1 to DFMQ + 3 over leave evidence such as high pressure phases (e.g., ringwood- temperatures of 587 °C–706 °C (temperature “uncertainty” ite, bridgmanite, maskelynite, stishovite) or glassy matrix. ranges from 11 to 75, log fO2 “uncertainty” ranges from 0.2 to 1.1). Magnetite grains with trellis type ilmenite lamellae 4.2. Do the igneous clast compositions represent unaltered are also present in these clasts. This is the lowest tem- magma compositions? perature range for any of the igneous clast types, possibly indicating that more substantial subsolidus equilibration Before in depth studies of petrogenesis and source occurred in these clasts. The regions of fine grained oxides region characteristics can be performed on newly described present in these clasts may be evidence of a subsolidus pro- igneous rock compositions, it is important to establish the cess; further assessment of these regions is needed to deter- limits of interpretation for the compositions. For example, mine if it was an open system process, so the fO2’s of these if a sample shows strong evidence for substantial crystal clasts are the least likely to be magmatic. The high abun- accumulation, this must be considered when determining dance of phosphates within this lithology make it important the magma composition. Cumulus crystals have been in terms of martian petrology, as it represents a significant observed in many martian basalts (i.e., shergottites), and source of incompatible elements. so we attempt to determine whether or not the igneous clasts from this study have been affected by crystal 4. DISCUSSION accumulation. Textural characteristics can be used to assess processes 4.1. Appropriate petrologic classification of NWA 7034 such as crystal accumulation. Cumulus pyroxene grains have been identified in many other martian meteorites The results and observations from this study have shown (e.g., basalts Shergotty and Zagami, gabbro ALH 77005, NWA 7034 to contain a diverse set of lithologies, both in nakhlites and chassignites) on the basis of texture and min- terms of igneous and non-igneous materials, that cannot eral chemistry (Floran et al., 1978; McSween et al., 1979; have come from a single source (i.e., sedimentary, impact, Stolper and McSween, 1979; Treiman, 2005; Papike et al., and igneous sources are needed). The results of our study 2009), and each NWA 7034 clast type was examined for evi- support the classification of this meteorite by Humayun dence of this process. Textural evidence for cumulus crys- et al. (2013) as a polymict breccia. While determination of tals such as oriented pyroxene grains as seen in Shergotty the assembly mechanism for this breccia is beyond the and Zagami (Stolper and McSween, 1979) is not present scope of this study, our observations suggest the materials in any of the clasts. Core-rim mineral zoning, such as that within this rock were not physically collected by a single- seen in cumulate nakhlites (Bunch and Reid, 1975), is stage process (i.e., NWA 7034 does not represent a pure absent from all silicate phases in the various clast types. impact breccia). The martian surface was and is a dynamic The exception to this observation is the trachyandesite environment (e.g., Malin and Edgett, 2000a,b), so breccias Clast 76, which contains pyroxene with a diffuse zoning may be formed by the transport and collection of materials boundary and increasing Mg content towards the grain via many different agents such as wind and water that are rims, suggesting a change in conditions during 76 A.R. Santos et al. / Geochimica et Cosmochimica Acta 157 (2015) 56–85 crystallization of the pyroxene; this is not conclusive evi- Exploration Rovers (Tuff et al., 2013; Schmidt et al., dence for cumulus crystals, however, and could be due to 2013), and could also explain some of the observations of an increase in fO2 during crystallization. elevated abundances of the siderophile elements Co and Other tests for cumulus crystals use chemical con- Ni in various lithologic components of NWA 7034 (Agee straints, such as mineral/melt exchange equilibria. et al., 2013; Humayun et al., 2013). However, accurate assessment of cumulus minerals using this technique requires knowledge of parameters that are 4.3. Possible models for the origin of lithologic diversity not yet constrained for the NWA 7034 clasts. For example, based on initial observations exchange equilibria are dependent on composition, pres- sure, and temperature. The crystallization pressure of the Given that NWA 7034 contains many igneous-textured NWA 7034 clasts is not known, and temperatures recorded clasts that exhibit a wide range of compositions, it is neces- by most mineral thermometers are likely recording sub- sary to ask whether or not this range is likely generated solidus temperatures. The greatest unknown with regards from disparate magmatic source regions or processes such to composition is the melt Fe3+/Fe2+, as we cannot deter- as fractional crystallization that can generate diverse mine if the calculated fO2 for the clasts represents the oxida- igneous lithologies from a single magmatic source. In the tion state of a magma or an oxidation state acquired after following, we discuss petrologic scenarios to try and pro- crystallization. This ratio will heavily influence exchange vide first order constraints on the petrogenesis of the equilibria between ferromagnesian minerals (pyroxene in igneous clasts in NWA 7034. This discussion of petroge- the case of the NWA 7034 clasts) and the melt. As an exam- nesis will take place under the assumptions that (1) there 2+ ple of the effect of fO2, and as a result melt Mg/Fe are no subdivisions within the clast groups (e.g., only one 2+ (mol%), we tested howP much the melt Mg/Fe changes unit of basalt is represented by the basalt clast group) when shergottite Fe3+/ Fe was applied to two basalt and (2) all igneous clasts are genetically related. clasts, Clast 2 and ClastP 73 (near the low and high end of the range for clast Fe3+/ Fe, respectively). 4.3.1. Fractional crystallization Shergottite fO2 values from Herd et al. (2001) and Herd Fractional crystallization is known to produce ranges of (2003) were used for the shergottites Zagami and Sayh al magmatic liquid compositions from single source regions in UhaymirP (SaU) 005. These fO2’s were used to calculate igneous systems on Earth (e.g., Nekvasil et al., 2004; the Fe3+/ Fe for each shergottite bulk composition using Whitaker et al., 2007) and both isobaric and polybaric frac- 3+ Pthe method of Kress and Carmichael (1991). These Fe / tional crystallization processes have also been shown to Fe values were then applied to the FeOtotal for Clasts 2 produce similar liquid trends in martian compositions and 73. The original Mg/Fe2+ of these clasts are 1.44 and (e.g., Christensen et al., 2005; McSween et al., 2006a; 1.20 forP Clast 2 and 73, respectively. With the Zagami Nekvasil et al., 2007, 2009; McCubbin et al., 2008; Udry 3+ Fe / Fe, these valuesP become 1.30 and 0.93, while with et al., 2014a), and should be considered in the petrogenesis the SaU 005 Fe3+/ Fe they become 1.27 and 0.90. The of the NWA 7034 clasts. The trends of bulk compositions more oxidized clast, Clast 73, shows a greater change when of NWA 7034 clasts on the TAS diagram in general agree the lower fO2’s of shergottites are applied, and this change is with those produced by fractional crystallization. Liquid significant enough to show that better constraints are need- lines of descent similar to those seen in terrestrial silica- ed before mineral/melt exchange equilibria can be used to saturated alkalic suites, where an increase of alkalis is fol- test for cumulus grains, and the clast compositions present- lowed by a subsequent increase in silica content, are formed ed here should be utilized with this result in mind. by these processes (Nekvasil et al., 2004; Whitaker et al., The silica-poor bulk composition of the NWA 7034 2007). The general compositional trend between the basalt clasts initially appears to beP inconsistent with their NWA 7034 basalt and trachyandesite clasts agrees with this 3+ lack of olivine, however the Fe / Fe ratios of the bulk pattern (Fig. 17). Some of the lower SiO2, lower alkali clast compositions leads to a displacement of the olivine– basalt clasts follow the general trend of fractional crystal- orthopyroxene peritectic to lower SiO2 contents. lization where early plagioclase crystallization is not sup- Consequently, we attribute the lack of olivine to a high oxy- pressed, and the liquid SiO2 content decreases while gen fugacity in these martian magmas that facilitated the alkalis are not enriched (similar to the trend of anhydrous peritectic reaction of olivine reacting to form enstatitic isobaric and polybaric fractionation of the Humphrey pyroxene as the liquid becomes more SiO2 rich during crys- basalt in McCubbin et al., 2008 and Udry et al., 2014a, tallization (Andersen, 1915). This peritectic reaction was respectively). The trend in feldspar composition between proposed by Stolper and McSween (1979) to explain the the clasts would also suggest a fractional crystallization lack of olivine in intercumulus liquids in Shergotty and relationship (Fig. 18A) because the plagioclase in clast types Zagami, and it is also supported by our observation of a with higher silica (trachyandesite and basaltic andesite) remnant olivine core in basalt Clast 55 (Fig. 14). This pro- does not reach as high of an anorthite content as plagio- cess eliminates the need for post crystallization aqueous clase in clasts with lower silica (basalt). The FTP clasts alteration, followed by impact re-melting and crystalliza- do not follow this trend, however. tion as proposed by Humayun et al. (2014) to explain the Contrary to the previous observations, trends in pyrox- lack of olivine in NWA 7034. Furthermore, the presence ene mineral chemistry and Mg# between clast groups sug- of early oxidized magmas on Mars is supported by compo- gest that all of the NWA 7034 igneous clasts cannot be sitions of ancient surface rocks analyzed by the Mars related by fractional crystallization. Fig. 18B shows the A.R. Santos et al. / Geochimica et Cosmochimica Acta 157 (2015) 56–85 77

Fig. 18. (A) Feldspar compositions from the four igneous clast groups in NWA 7034 identified in this study (same as Fig. 9A–D) plotted with a trend in feldspar chemistry derived from high pressure experimental fractional crystallization of Nandewar volcanic rock compositions (purple arrow points in direction of increasing degree of evolution of liquid) from Nekvasil et al. (2004). (B) Pyroxene compositions from the four igneous clast groups in NWA 7034 (same as Fig. 10A–D) with two fractional crystallization trends showing change in pyroxene content with evolution of the melt. Purple arrow is from experimental crystallization of Humphrey composition from McCubbin et al. (2008) and orange arrows are from Nekvasil et al. (2004). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) pyroxene quadrilateral with clast pyroxene compositions fields, (3) relatively high concentrations of immiscibility and fractional crystallization trends from two different promoting elements, and (4) similar mineral compositions studies for comparison. Changes in pyroxene chemistry but different phase abundances among the clast groups. among the NWA 7034 clast groups do not follow either At this time and using these features, it cannot be proven of these trends. REEs in clast pyroxenes also indicate the or disproven whether liquid immiscibility was involved in clasts are not related by simple fractional crystallization. the genesis of these clasts, however there is a strong case REE abundances would be expected to increase with for a thorough investigation into the likelihood of this pro- decreasing pyroxene Mg# (and evolution of liquid the cess in the context of martian magmas. Therefore, we find it clasts crystallized from), but the REEs show no systematic appropriate to consider SLI as a possible process in the pet- variation with Mg# or clast SiO2 content. However, there is rogenetic history of the NWA 7034 clasts and provide fur- presently a very limited dataset of pyroxene REE contents ther discussion below. Much of the information regarding from these clasts and more data is required to fully support FTP rocks and SLI comes from terrestrial studies, and we this observation. The bulk clast Mg#s show very little hesitate to apply all of the information from these studies change from one clast group to another and are overlap- to our samples. For example, terrestrial FTP rocks often ping, which is not expected from a suite of rocks related occur on the outcrop scale in association with massif by fractional crystallization. In summary, the igneous clasts anorthosites, but the relationship between the two rock cannot be genetically linked by fractional or equilibrium types is not clear (Owens and Dymek, 1992). Since no direct crystallization under the assumptions posed at the begin- evidence for anorthosites in NWA 7034 was found, and no ning of this section, so multiple magmatic sources may be martian anorthosites have been identified on the surface required to explain the origin of the clast groups. Should with a high degree of certainty (e.g., Carter and Poulet, one or both of our assumptions prove false, it is possible 2013), we cannot infer the presence of anorthosites on there is a subset of clasts that are genetically related, but Mars based on the existence of the FTP clasts. the trends that would indicate this relationship are obscured The petrogenesis of terrestrial FTP rocks is still debated when all the clast data is considered together. and interpretations typically include at least one of the fol- lowing processes: SLI, fractional crystallization, and crystal 4.3.2. Silicate liquid immiscibility accumulation (discussed in Owens and Dymek, 1992 and Given the wide variation in SiO2 abundances among the references therein). The FTP clasts are likely not related various igneous clasts identified, we consider the possibility to the other igneous clasts in NWA 7034 by simple fraction- of silicate liquid immiscibility (SLI) to link some of the clast al crystallization, as discussed in Section 4.3.1 above. types identified in NWA 7034 because this process directly Consequently, it is reasonable to consider the possible gen- leads to large variations in SiO2 over limited spatial scales esis of the NWA 7034 FTP clasts by silicate liquid immisci- (Zirkel, 1873; Rosenbusch, 1887). SLI can prove difficult bility. A parent melt and conjugate liquid are also required to identify if direct evidence of both conjugate liquids in SLI, so we include all of the NWA 7034 igneous clasts in (e.g., melt inclusions) is not present (see discussion in this discussion as they may represent different stages in the Charlier et al., 2013). Features such as melt inclusions have evolution or unmixing of a melt. not been reported in NWA 7034; however, there are certain Fields defining two-liquid regions on several different features within and among some of the clast types examined ternary diagrams are shown in Fig. 19. The NWA 7034 in this study that are suggestive of liquid immiscibility, clasts fall within or close to the fields of immiscibility, including: (1) the presence of FTP lithologies, (2) proximity and while this alone cannot be taken as proof of this pro- of clast compositions to previously established two liquid cess, it warrants further investigation. Typical SLI 78 A.R. Santos et al. / Geochimica et Cosmochimica Acta 157 (2015) 56–85

Fig. 19. Diagrams showing fields of silicate liquid immiscibility determined from experiments using terrestrial basalt compositions from Charlier and Grove (2012) (A and B) and natural lunar samples from Shearer et al. (2001) (C). NWA 7034 basalt clasts (black dots), basaltic andesite clasts (red dots), trachyandesite clasts (green dots), and FTP clasts (cyan dots), as well as martian surface rocks (yellow triangles), could be products of some degree of SLI based on their proximity to the two-liquid fields. Martian surface rock compositions include Wishstone and Champagne (Usui et al., 2008), Jake_M (Stolper et al., 2013), Irvine and Backstay (McSween et al., 2006a), Adirondack, Humphrey, and Mazatzal (McSween et al., 2006b), and rocks from the Pathfinder site (Bru¨ckner et al., 2003). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) generates strongly distinct conjugate liquids, but Charlier The evidence listed above for the NWA 7034 clasts and and Grove (2012) show certain starting basalt compositions their similarities to other martian surface rocks (e.g., (those that cross the binodal near its closure point) can Wishstone class rocks, average martian crust; Usui et al., generate conjugate liquids that are not vastly different from 2008; McSween et al., 2009) suggest SLI could be a mechan- each other in composition. This would likely be the case for ism for lithologic diversity on Mars. However, the definitive several of the clasts in NWA 7034 and other martian rocks, identification of SLI without melt inclusions is difficult. as they lie near the closure point of the binodal, and would Furthermore, several variables (e.g., pressure, water con- make identifying conjugate liquids more difficult than in tent, starting basalt composition, see Charlier and Grove, other cases in the Solar System (e.g., for lunar rocks; 2012 and references therein) that influence the development Pernet-Fisher et al., 2014). of immiscibility are still unconstrained for these clasts. The presence of relatively high concentrations of P2O5, These two complications indicate experiments specific to TiO2, and alkalis, all found in the NWA 7034 igneous martian compositions, pressures, and fO2’s are needed to clasts, have been found to promote immiscibility in some fully assess the possibility that SLI was involved in the pet- basaltic liquids (Charlier and Grove, 2012). The FTP clasts rogenesis of these clasts or other martian rocks. contain particularly high concentrations of these elements, and would therefore be the most likely to have formed 4.4. Petrologic comparison of igneous clasts with other due to SLI. martian igneous rocks As shown in Figs. 9, 10 and 12, the mineral chemistry for all of the igneous clast types in NWA 7034 is overlap- Numerous igneous rocks from Mars, in the form of ping, at least to some degree. One of the major factors that meteorites and surface rocks analyzed remotely during imparts the different bulk clast compositions is the different Mars missions, have been analyzed and petrologically char- abundances of each mineral phase. Crystallizing emulsions acterized over the previous several decades (e.g., McSween of immiscible conjugate liquids in terrestrial intrusions have et al., 1999; Borg and Draper, 2003; McSween et al., been found to crystallize mineral phases of the same com- 2006a,b; McCubbin et al., 2013; Schmidt et al., 2014). position, but in different proportions (Charlier et al., These rocks make up the primary basis for our understand- 2011). This is again particularly true for the FTP clasts, ing of Mars’ bulk composition, primary mineralogy, and as they contain vastly different mineral modes from the thermal and magmatic evolution (e.g., Papike et al., 2009; other clast types, but still contain broadly the same phases. Taylor, 2013; Mezger et al., 2013; Grott et al., 2013). The A.R. Santos et al. / Geochimica et Cosmochimica Acta 157 (2015) 56–85 79 igneous clasts in NWA 7034 can provide additional insights planet’s history, given the ancient ages of 4.4 Ga for the into our understanding of Mars’ evolution, but first it is basaltic components in NWA 7034 (Humayun et al., important to compare and contrast each clast type with pre- 2013; Nyquist et al., 2013; Tarte`se et al., 2014; Yin et al., viously analyzed martian igneous rocks. The NWA 7034 2014). An early oxidized martian crust is also in agreement clasts represent igneous lithologies that have both similari- with recent estimates by Tuff et al. (2013) who compared ties and differences with known martian igneous rocks. The fO2 of martian surface rocks and meteorites as a function major distinguishingP features for the NWA 7034 clasts of age. Ancient zircons from Earth also provide evidence include Fe3+/ Fe ratios (0.03–0.38), REE abundances for early crustal oxidation (Trail et al., 2011), which is (enriched relative to shergottites), mineralogy, and bulk typically attributed to the presence of liquid water early compositions (Table S6). Each of these features is discussed in Earth’s history (i.e., Wilde et al., 2001; Mojzsis et al., below in the context of the martian crust. 2001). Consequently, an early oxidized martian crust is also consistent with the presence of liquid water very early in 4.4.1. Oxygen fugacity of martian crustal rocks Mars’ history (e.g., Nemchin et al., 2014). Sharp et al. The fO2 values determined using QUILF (Andersen (2013) attributed this early oxidation of planetary crusts et al., 1993) for each of the igneous clast groups in NWA to H2 loss, and Mars is likely to have followed a similar 7034 indicate they are more oxidized (DFMQ 0.1 to oxidation path to Earth during early planetary degassing DFMQ + 4) than typical martian igneous rocks due to its high volatile abundance relative to Earth (DFMQ 3.7 to DFMQ + 0.7; Herd, 2003, 2006; Schmidt (Dreibus and Wanke, 1985; McCubbin et al., 2010a, 2012; et al., 2013). The higher fO2 values calculated for the Gross et al., 2013a) (as opposed to the Moon which fol- NWA 7034 clasts reflect either a higher magmatic fO2 or lowed a different oxidation path due to its low volatile a post-magmatic process (i.e., slow-cooling, secondary ther- abundance relative to Earth, see Sharp et al., 2013). mal metamorphism, or aqueousP alteration) that altered the Additional work that investigates the precise timing of 3+ original magmatic Fe / Fe value. If the fO2’s are the oxidation within the clasts in NWA 7034 is required before result of post-magmatic alteration during slow cooling or any firm conclusions can be drawn regarding the oxidation thermal metamorphism, it does not necessarily require that state of the earliest martian crust from NWA 7034. the clasts formed from magmas with elevated oxygen fuga- city. However,P an entirely subsolidus mechanism for the 4.4.2. Rare earth element abundances in martian crustal elevated Fe3+/ Fe is not supported by the observed tex- rocks tures and mineralogy in the clasts (Figs. 1–3 and 5, REE contents of NWA 7034 igneous clast minerals Figs. S1, S5, S10). The clasts have igneous textures com- (pyroxene, plagioclase, and chlorapatite) were compared posed of many individual discrete subhedral pyroxene, pla- to REEs measured in the depleted shergottite QUE 94201 gioclase, apatite, and magnetite grains. Both silica and (McSween et al., 1996), the enriched shergottite Shergotty olivine are nearly absent from the phase assemblage, and (Lundberg et al., 1988; Wadhwa et al., 1994), and bulk rock at least one of these phases would be needed if a subsolidus NWA 7034 (Agee et al., 2013)(Fig. 11). All measured oxidation reaction occurred (i.e., Fe–Mg olivine reacting to NWA 7034 phases have greater REE concentrations than form Mg-rich pyroxene and magnetite or Fe–Mg pyroxene the two shergottite meteorites used for comparison (with reacting to form silica, magnetite, and Mg-rich pyroxene). the exception of the HREEs in apatite, which are roughly Furthermore, the pyroxene and magnetite grains are not equal among the three meteorites). Further, the CI normal- exclusively in contact with each other, which would be ized LREE trends (the slopes from La to Sm) of the NWA expected since both would be involved in any oxidation 7034 phases agree with those from the same phases in reaction given the phase assemblage of the clasts (i.e., only Shergotty, with the exception of the low Ca pyroxene in Fe-bearing phases). Based on the paucity of alteration tex- the basalt and trachyandesite clasts. These clast pyroxenes tures and low-T alteration minerals in the igneous clasts, we show a greater LREE enrichment than seen in Shergotty. rule out aqueous alteration as the primary cause for the Based on the measured REE values, it is likely the bulk elevated abundances of Fe3+ in the NWA 7034 clasts, REE pattern for these clasts is LREE enriched, similar to although some evidence of aqueous alteration does exist the enriched shergottites, and as seen in microbasalt clasts in NWA 7034 (Muttik etP al., 2014b). Consequently, we measured by Humayun et al. (2013). This observation interpret the elevated Fe3+/ Fe as a feature of the magmas agrees with the relatively high concentration of other that crystallizedP the clasts in NWA 7034. Importantly, the incompatible elements in these clasts (e.g., K, P, Ti) and elevated Fe3+/ Fe could indicate an oxidized mantle is likely indicating either derivation from or interaction source region that is unique compared to those defined by with a geochemically enriched source, as indicated by other martian meteorites and surface rocks or the parental Agee et al. (2013). magmas to the clasts interacted with a highly oxidized crus- tal component. P 4.4.3. Mineralogy of the martian crust The elevated Fe3+/ Fe values, whether they be from Precise mineral compositions have only been directly the mantle source or through crustal assimilation, lend fur- measured in martian meteorites (as opposed to surface ther support to the trend seen by Herd et al. (2002) showing rocks). Although the actual silicate mineral phases are gen- shergottite oxidation states correlate positively with degree erally the same among the clasts and other meteorites, pla- of geochemical enrichment. Furthermore, it would indicate gioclase in the NWA 7034 clasts are typically more Ab rich that the crust of Mars was oxidized very early in the than in shergottites (Papike et al., 2009). The plagioclase 80 A.R. Santos et al. / Geochimica et Cosmochimica Acta 157 (2015) 56–85

Fig. 20. Same diagram as in Fig. 17 with other martian rocks (surface and meteorite) added for comparison (Gellert et al., 2006; Usui et al., 2008; McSween et al., 2009; Schmidt et al., 2014). NWA 7034 basalt clasts and average bulk matrix most closely match compositions from Gusev Crater and the average martian crust determined by GRS. Basaltic andesite clasts fall in similar ranges as the TES field, while the trachyandesite and FTP clasts have no previously seen analogues. Some of the rocks seen in Gale Crater are even more alkali rich than clasts from NWA 7034. The SNC meteorites are not a match for any of the clast groups in this compositional space. compositions in the NWA 7034 clasts are similar to some of 2013; Gross et al., 2013b). Cr-rich spinel was commonly the alkali-rich feldspar compositions in the chassignites absent or only a very minor component in the clasts, and (Floran et al., 1978; Johnson et al., 1991; Nekvasil et al., magnetite is the most abundant oxide phase. A paucity of 2007; McCubbin and Nekvasil, 2008). The pyroxenes in Cr-rich spinel is atypical of martian igneous meteorites many of the shergottites and nakhlites typically have a and consistent with the parental magmas of the NWA higher Wo content than measured in many of the NWA 7034 clasts being oxidized, given the low solubility of Cr 7034 clasts, with the exception of orthopyroxenite ALH in silicate liquids with increasing fO2 (i.e., Hanson and 84001 (Berkley, 1987; Mittlefehldt, 1994; Papike et al., Jones, 1998; Bell et al., 2014). 2009) and some of the pyroxenes reported in the chassig- nites and opx-bearing nakhlites (Treiman, 2005; Nekvasil 4.4.4. Compositions of martian crustal rocks et al., 2007; McCubbin et al., 2013). NWA 7034 clast pyrox- The basalt clasts are the most similar to alkaline surface ene does not have the high Fe content seen in shergottites rocks measured by the MER Spirit in Gusev Crater (e.g., Smith and Hervig, 1979; Kring et al., 1996). (Fig. 20); specifically, the rock Irvine has a normative min- Extensive chemical zoning within pyroxene grains, exhibit- eralogy similar to the measured mineral modes for the ed by almost all shergottites and many nakhlites (e.g., basalt clasts (McSween et al., 2006a). The basalt clasts also Bunch and Reid, 1975; Steele and Smith, 1982; McSween fall within the compositional range established for the aver- et al., 1996; Day et al., 2006; Udry et al., 2012; Gross age martian crust by orbital GRS from Mars Odyssey on et al., 2013b), is absent from the NWA 7034 clasts. the TAS diagram (Boynton et al., 2007; McSween et al., However, the NWA 7034 clasts have greater compositional 2009), indicating these components of NWA 7034 are rep- variability among pyroxene grains within a single clast than resentative of crustal rocks from Mars (an extension of is seen in meteorites like Chassigny and ALH 84001 (Papike the conclusions of Agee et al., 2013). The modal mineralogy et al., 2009). These differences in pyroxene compositional of the pyroxene phyric (or basaltic) shergottites are also ranges indicate a difference in cooling history between other similar to the basalt clasts in NWA 7034, but they generally martian meteorites and the clasts in NWA 7034. In terms of contain more pyroxene than the basalt clasts (Mikouchi non-silicate minerals, the clasts in NWA 7034 lack merril- et al., 1996; Mikouchi, 2001; Papike et al., 2009). The high- lite, which is present in all martian meteorites with the er pyroxene:plagioclase ratios result in lower total alkali exception of the chassignites and nakhlites (Shearer et al., abundances in the pyroxene phyric shergottites (Fig. 20) 2011, 2014; McCubbin et al., 2013, 2014). However, the compared to the NWA 7034 basalt clasts, which is why they clasts do contain Cl-rich apatite, which occurs in all other are not as strong of a match as the Irvine class rocks ana- martian meteorites (e.g., Steele and Smith, 1982; Rubin lyzed by MER. et al., 2000; Barrat et al., 2002a,b; Imae and Ikeda, 2007; Basaltic andesite clasts show compositional similarities McCubbin and Nekvasil, 2008; McCubbin et al., 2012, with the Pathfinder sulfur free rock (Bru¨ckner et al., A.R. Santos et al. / Geochimica et Cosmochimica Acta 157 (2015) 56–85 81

2003) and global orbital Thermal Emission Spectrometer While the context of the NWA 7034 lithologies (located (TES) data from Mars Global Surveyor (McSween et al., within a breccia) makes it difficult to completely piece 2009). While the TES measurements are thought to repre- together their petrogenesis, from the present study we can sent compositions that are potentially affected by alteration say the following. It appears unlikely that all of the igneous (McSween et al., 2009), these clasts suggest that there may clasts identified in this study are genetically related. This be lithologies on Mars of this composition; however their leaves two possibilities to explain the relationships of the abundance in NWA 7034 is substantially subordinate to NWA 7034 igneous clasts: (1) the clasts are not all related the clasts of basaltic composition (roughly 1 basaltic ande- and derive from multiple, diverse sources, and identification site clast for every 10 basalt clasts). of related clasts requires trace element or isotopic data, or No other martian rocks are directly similar to the tra- (2) the clasts derive from the same source and their compo- chyandesite clasts, but they are similar to some Gale Crater sitional variation is a result of a complex combination of rocks in their high alkali contents (the NWA 7034 clasts have igneous processes (e.g., combined SLI and fractional crys- higher SiO2 at these alkali contents, however; Stolper et al., tallization, fractional crystallization–assimilation, etc.). 2013; Schmidt et al., 2014). Alkali feldspar, a major con- Further chemical and isotopic analyses are needed to deter- stituent of this clast type, has also been found in chassignites mine which of these cases is most likely. Regardless of the and nakhlites as the mineral feldspar or a maskelynitized mechanism for this diversity, the igneous clasts within this equivalent (e.g., Bunch and Reid, 1975; Floran et al., 1978; meteorite provide evidence for significant lithologic diversi- Nekvasil et al., 2007; McCubbin and Nekvasil, 2008), and ty within the ancient martian crust. Phosphorus rich it is present in some of the normative mineralogies deter- lithologies such as the Gusev Crater Wishstone class are mined for several Gale Crater rocks (Schmidt et al., 2014). likely not a unique occurrence, based on the presence of The FTP clasts are the most unique of the NWA 7034 the FTP clasts within NWA 7034. Mars has (or had at clast types for many reasons. The compositions of these one time) a reservoir that was enriched in REEs and poten- clasts are lower in silica than other known martian rocks tially had a relatively high fO2; this reservoir may be the and higher in phosphorus and titanium. Based on these same as the one sampled by the enriched shergottites. properties, they appear most similar to the Wishstone class Martian magmas may be capable of experiencing silicate rocks from Gusev Crater (Gellert et al., 2006), which are liquid immiscibility based on the observed enrichments in interpreted as phosphorus rich basalts (Usui et al., 2008). immiscibility inducing elements seen in some of the NWA Merrillite was determined to be the phosphate mineral pre- 7034 clasts, which has the potential to generate diverse sent in the Wishstone rocks (Usui et al., 2008), while Cl- lithologies from a single parent magma. All of the igneous apatite is the phosphate phase present in the FTP clasts processes potentially implicated by this study operated on in NWA 7034. Mineral abundances are difficult to compare Mars during its earliest crust building stages, but further due to the limited clast sampling, however the data from information from both the igneous clasts within NWA this study suggest the FTP clasts have a much greater abun- 7034 and rocks on the martian surface is needed to deter- dance of Fe–Ti oxides than Wishstone, and unlike mine if some of these processes (and sources) were tempo- Wishstone, ilmenite is far more abundant than magnetite. rally limited. If so, Mars could have experienced a period of vast igneous diversity during its early crust building, 4.4. Summary while igneous rocks from a more restricted compositional range were formed during the rest of its magmatic history. The evolved and exotic clast compositions within NWA 7034 can be added to the growing list of evidence for crustal ACKNOWLEDGEMENTS diversity on Mars, which already includes rocks such as the mugearite Jake_M and the Pathfinder andesites (McSween We would like to greatly thank J. Lewis for his assistance with et al., 1999; Bru¨ckner et al., 2003; Stolper et al., 2013), orbi- the FEG SEM mapping at UNM and M. Spilde for his assistance with mapping using the electron microprobe. The manuscript was tal data (McSween et al., 2009; Taylor et al., 2010), obser- improved by helpful reviews from R.H. Hewins, M. Humayun, vations of evolved melt pockets within martian meteorites and an anonymous reviewer, as well as additional comments by (Ikeda, 2005; Filiberto et al., 2014), and experimental stud- A.A. Nemchin (associate editor). This work was funded by the ies (Minitti and Rutherford, 2000; Dann et al., 2001; NASA Cosmochemistry grant NNX14AI23G to CBA, the Musselwhite et al., 2006; Monders et al., 2007; Filiberto NASA Mars Fundamental Research Program grant et al., 2008, 2010; McCubbin et al., 2008; Nekvasil et al., NNX13AG44G to FMM, and the NASA Cosmochemistry grant 2009). Furthermore, the clasts within NWA 7034 are NNX13AH85G to CKS. This work was partially supported by a ancient (4.4 Ga, Nyquist et al., 2013; Humayun et al., UK Science and Technology Facilities Council (STFC) research 2013; Tarte`se et al., 2014; Yin et al., 2014), suggesting that grant to MA (grant number ST/I001298/1). ARS acknowledges early martian magmatism generated a compositionally support from the NM Space Grant Consortium. diverse crust in the early stages of crust building. This ancient, diverse crust was then available throughout mar- tian history to weather and erode to become sediment/ APPENDIX A. SUPPLEMENTARY DATA sedimentary rocks, and to interact with subsequent martian magmas erupting through or becoming emplaced in the Supplementary data associated with this article can be crust, the martian atmosphere, and any fluids that may found, in the online version, at http://dx.doi.org/10.1016/ have been present. j.gca.2015.02.023. 82 A.R. Santos et al. / Geochimica et Cosmochimica Acta 157 (2015) 56–85

REFERENCES Charlier B., Namur O., Toplis M. J., Schiano P., Cluzel N., Higgins M. D. and Vander Auwera J. (2011) Large-scale silicate liquid Agee C. B., Wilson N. V., McCubbin F. M., Ziegler K., Polyak V. immiscibility during differentiation of tholeiitic basalt to granite J., Sharp Z. D., Asmerom Y., Nunn M. H., Shaheen R., and the origin of the Daly gap. Geology 39, 907–910. Thiemens M. H., Steele A., Fogel M. L., Bowden R., Charlier B., Namur O. and Grove T. L. (2013) Compositional and Glamoclija M., Zhang Z. S. and Elardo S. M. (2013) Unique kinetic controls on liquid immiscibility in ferrobasalt-rhyolite meteorite from early Amazonian Mars: water-rich basaltic volcanic and plutonic series. Geochim. Cosmochim. Acta 113, breccia Northwest Africa 7034. Science 339, 780–785. 79–93. Anders E. and Grevesse N. (1989) Abundances of the elements – Christensen P. R., McSween H. Y., Bandfield J. L., Ruff S. W., meteoritic and solar. Geochim. Cosmochim. Acta 53, Rogers A. D., Hamilton V. E., Gorelick N., Wyatt M. B., 197–214. Jakosky B. M., Kieffer H. H., Malin M. C. and Moersch J. E. Andersen O. (1915) The system anorthite–forsterite-silica. Am. J. (2005) Evidence for magmatic evolution and diversity on Mars Sci. 39, 407–454. from infrared observations. Nature 436, 504–509. Andersen D. J., Lindsley D. H. and Davidson P. M. (1993) QUILF Cox K. G., Bell J. D. and Pankhurst R. J. (1984) The Interpretation – a Pascal program to assess equilibria among Fe–Mg–Mn–Ti of Igneous Rocks. George Allen & Unwin, London. oxides, pyroxenes, olivine, and quartz. Comput. Geosci. 19, Dann J. C., Holzheid A. H., Grove T. L. and McSween H. Y. 1333–1350. (2001) Phase equilibria of the Shergotty meteorite: constraints Bacon C. R. and Hirschmann M. M. (1988) Mg/Mn partitioning as on pre-eruptive water contents of martian magmas and a test for equilibrium between coexisting Fe–Ti oxides. Am. fractional crystallization under hydrous conditions. Meteorit. Mineral. 73, 57–61. Planet. Sci. 36, 793–806. Barrat J. A., Gillet P., Sautter V., Jambon A., Javoy M., Gopel C., Day J. M. D., Taylor L. A., Floss C. and McSween H. Y. (2006) Lesourd M., Keller F. and Petit E. (2002a) Petrology and Petrology and chemistry of MIL 03346 and its significance in chemistry of the basaltic shergottite North West Africa 480. understanding the petrogenesis of nakhlites on Mars. Meteorit. Meteorit. Planet. Sci. 37, 487–499. Planet. Sci. 41, 581–606. Barrat J. A., Jambon A., Bohn M., Gillet P., Sautter V., Gopel C., Deer W. A., Howie R. A. and Zussman J. (1992) An Introduction to Lesourd M. and Keller F. (2002b) Petrology and chemistry of the Rock Forming Minerals. Prentice Hall, New York. the picritic shergottite North West Africa 1068 (NWA 1068). Delano J. W. (1986) Pristine lunar glasses – criteria, data, and Geochim. Cosmochim. Acta 66, 3505–3518. implications. J. Geophys. Res. Solid Earth Planets 91, D201– Bell A. S., Burger P. V., Le L., Shearer C. K., Papike J. J., Sutton D213. S. R., Newville M. and Jones J. (2014) XANES measurements Delano J. W. and Livi K. (1981) Lunar volcanic glasses and their of Cr valence in olivine and their applications to planetary constraints on mare petrogenesis. Geochim. Cosmochim. Acta basalts. Am. Mineral. 99, 1404–1412. 45, 2137–2149. Berkley, J. L. (1987) Petrology and compositional trends in five Dreibus G. and Wanke H. (1985) Mars, a volatile-rich planet. new antarctic diogenites. 18th Lunar Planet. Sci. Conf. 62–63. Meteoritics 20, 367–381. Borg L. E. and Draper D. S. (2003) A petrogenetic model for the Droop G. T. R. (1987) A general equation for estimating Fe-3+ origin and compositional variation of the martian basaltic concentrations in ferromagnesian silicates and oxides from meteorites. Meteorit. Planet. Sci. 38, 1713–1731. microprobe analyses, using stoichiometric criteria. Mineral. Boynton W. V., Taylor G. J., Evans L. G., Reedy R. C., Starr R., Mag. 51, 431–435. Janes D. M., Kerry K. E., Drake D. M., Kim K. J., Williams R. Fagan A. L., Neal C. R., Simonetti A., Donohue P. H. and M. S., Crombie M. K., Dohm J. M., Baker V., Metzger A. E., O’Sullivan K. M. (2013) Distinguishing between Apollo 14 Karunatillake S., Keller J. M., Newsom H. E., Arnold J. R., impact melt and pristine mare basalt samples by geochemical Bruckner J., Englert P. A. J., Gasnault O., Sprague A. L., and textural analyses of olivine. Geochim. Cosmochim. Acta Mitrofanov I., Squyres S. W., Trombka J. I., d’Uston L., 106, 429–445. Wanke H. and Hamara D. K. (2007) Concentration of H, Si, Filiberto J. (2008) Experimental constraints on the parental liquid Cl, K, Fe, and Th in the low- and mid-latitude regions of Mars. of the Chassigny meteorite: a possible link between the J. Geophys. Res. Planets 112, E12S99. Chassigny meteorite and a Martian Gusev basalt. Geochim. Brandon A. D., Puchtel I. S., Walker R. J., Day J. M. D., Irving A. Cosmochim. Acta 72, 690–701. J. and Taylor L. A. (2012) Evolution of the martian mantle Filiberto J., Treiman A. H. and Le L. (2008) Crystallization inferred from the Re-187–Os-187 isotope and highly siderophile experiments on a Gusev Adirondack basalt composition. element abundance systematics of shergottite meteorites. Meteorit. Planet. Sci. 43, 1137–1146. Geochim. Cosmochim. Acta 76, 206–235. Filiberto J., Dasgupta R., Kiefer W. S. and Treiman A. H. (2010) Bru¨ckner J., Dreibus G., Rieder R. and Wanke H. (2003) Refined High pressure, near-liquidus phase equilibria of the Home Plate data of alpha proton X-ray spectrometer analyses of soils and basalt Fastball and melting in the Martian mantle. Geophys. rocks at the Mars Pathfinder site: implications for surface Res. Lett. 37, L13201. chemistry. J. Geophys. Res. Planets 108, 8094. Filiberto J., Gross J., Trela J. and Ferre E. C. (2014) Gabbroic Bunch T. E. and Reid A. M. (1975) The Nakhlites, part I. Shergottite Northwest Africa 6963: an intrusive sample of Petrography and mineral chemistry. Meteoritics 10, 303–315. Mars. Am. Mineral. 99, 601–606. Carter J. and Poulet F. (2013) Ancient plutonic processes on Mars Floran R. J., Prinz M., Hlava P. F., Keil K., Nehru C. E. and inferred from the detection of possible anorthositic terrains. Hinthorne J. R. (1978) Chassigny meteorite – cumulate dunite Nat. Geosci. 6, 1008–1012. with hydrous amphibole-bearing melt inclusions. Geochim. Cartwright J. A., Ott U., Herrmann S. and Agee C. B. (2014) Cosmochim. Acta 42, 1213–1229. Modern atmospheric signatures in 4.4 Ga Martian meteorite Gattacceca J., Rochette P., Scorzelli R. B., Munayco P., Agee NWA 7034. Earth Planet. Sci. Lett. 400, 77–87. C., Quesnel Y., Cournede C. and Geissman J. (2014) Charlier B. and Grove T. L. (2012) Experiments on liquid Martian meteorites and Martian magnetic anomalies: a new immiscibility along tholeiitic liquid lines of descent. Contrib. perspective from NWA 7034. Geophys. Res. Lett. 41, 4859– Mineral. Petrol. 164, 27–44. 4864. A.R. Santos et al. / Geochimica et Cosmochimica Acta 157 (2015) 56–85 83

Gellert R., Rieder R., Bruckner J., Clark B. C., Dreibus G., that may represent bulk melt rather than a cumulate fraction. Klingelhofer G., Lugmair G., Ming D. W., Wanke H., Yen A., 27th Lunar Planet. Sci. Conf. 705–706. Zipfel J. and Squyres S. W. (2006) Alpha particle X-ray Le Bas M. J. and Streckeisen A. L. (1991) The IUGS systematics of spectrometer (APXS): results from Gusev Crater and calibra- igneous rocks. J. Geol. Soc. London 148, 825–833. tion report. J. Geophys. Res. Planets 111, E02S05. Le Maitre R. W. (1984) A proposal by the IUGS subcommission Gross J., Filiberto J. and Bell A. S. (2013a) Water in the martian on the systematics of igneous rocks for a chemical classification interior: evidence for terrestrial MORB mantle-like volatile of volcanic rocks based on the total alkali silica (TAS) diagram. contents from hydroxyl-rich apatite in olivine-phyric shergottite Austr. J. Earth Sci. 31, 243–255. NWA 6234. Earth Planet. Sci. Lett. 369, 120–128. Le Maitre R. W. (2002) Igneous Rocks: A Classification and Gross J., Filiberto J., Herd C. D. K., Daswani M. M., Schwenzer S. Glossary of Terms. Cambridge University Press, New York. P. and Treiman A. H. (2013b) Petrography, mineral chemistry, Lundberg L. L., Crozaz G., McKay G. and Zinner E. (1988) Rare- and crystallization history of olivine-phyric shergottite NWA earth element carriers in the Shergotty meteorite and implica- 6234: a new melt composition. Meteorit. Planet. Sci. 48, 854– tions for its chronology. Geochim. Cosmochim. Acta 52, 2147– 871. 2163. Grott M., Baratoux D., Hauber E., Sautter V., Mustard J., Malin M. C. and Edgett K. S. (2000a) Evidence for recent Gasnault O., Ruff S. W., Karato S. I., Debaille V., Knapmeyer groundwater seepage and surface runoff on Mars. Science 288, M., Sohl F., Van Hoolst T., Breuer D., Morschhauser A. and 2330–2335. Toplis M. J. (2013) Long-term evolution of the Martian crust– Malin M. C. and Edgett K. S. (2000b) Sedimentary rocks of early mantle system. Space Sci. Rev. 174, 49–111. Mars. Science 290, 1927–1937. Haggerty S. E. (1991) Oxide textures – a mini-atlas. Rev. Mineral. Mason B. and Allen R. O. (1973) Minor and trace elements in 25, 129–219. augite, hornblende, and pyrope megacrysts from Kakanui, New Hanson B. and Jones J. H. (1998) The systematics of Cr3+ and Zealand. NZ J. Geol. Geophys. 16(4), 935–947. Cr2+ partitioning between olivine and liquid in the presence of McCubbin F. M. and Nekvasil H. (2008) Maskelynite-hosted spinel. Am. Mineral. 83, 669–684. apatite in the Chassigny meteorite: insights into late-stage Herd C. D. K. (2003) The oxygen fugacity of olivine-phyric magmatic volatile evolution in martian magmas. Am. Mineral. martian basalts and the components within the mantle and 93, 676–684. crust of Mars. Meteorit. Planet. Sci. 38, 1793–1805. McCubbin F. M., Nekvasil H., Harrington A. D., Elardo S. M. Herd C. D. K. (2006) Insights into the redox history of the NWA and Lindsley D. H. (2008) Compositional diversity and 1068/1110 martian basalt from mineral equilibria and vanadi- stratification of the martian crust: inferences from crystalliza- um oxybarometry. Am. Mineral. 91, 1616–1627. tion experiments on the picrobasalt Humphrey from Gusev Herd C. D. K., Papike J. J. and Brearley A. J. (2001) Oxygen Crater, Mars. J. Geophys. Res. Planets 113, E11013. fugacity of martian basalts from electron microprobe oxygen McCubbin F. M., Smirnov A., Nekvasil H., Wang J. H., Hauri E. and TEM-EELS analyses of Fe–Ti oxides. Am. Mineral. 86, and Lindsley D. H. (2010a) Hydrous magmatism on Mars: a 1015–1024. source of water for the surface and subsurface during the Herd C. D. K., Borg L. E., Jones J. H. and Papike J. J. (2002) Amazonian. Earth Planet. Sci. Lett. 292, 132–138. Oxygen fugacity and geochemical variations in the martian McCubbin F. M., Steele A., Nekvasil H., Schnieders A., Rose T., basalts: implications for martian basalt petrogenesis and the Fries M., Carpenter P. K. and Jolliff B. L. (2010b) Detection of oxidation state of the upper mantle of Mars. Geochim. structurally bound hydroxyl in fluorapatite from Apollo mare Cosmochim. Acta 66, 2025–2036. basalt 15058,128 using TOF-SIMS. Am. Mineral. 95, 1141– Humayun M., Nemchin A., Zanda B., Hewins R. H., Grange 1150. M., Kennedy A., Lorand J. P., Gopel C., Fieni C., Pont S. McCubbin F. M., Hauri E. H., Elardo S. M., Vander Kaaden K. and Deldicque D. (2013) Origin and age of the earliest E., Wang J. H. and Shearer C. K. (2012) Hydrous melting of Martian crust from meteorite NWA 7533. Nature 503, 513– the martian mantle produced both depleted and enriched 516. shergottites. Geology 40, 683–686. Humayun, M., Hewins, R. H., Lorand, J. P. and Zanda, B. (2014) McCubbin F. M., Elardo S. M., Shearer C. K., Smirnov A., Hauri Weathering and impact melting determined the mineralogy of E. H. and Draper D. S. (2013) A petrogenetic model for the the early martian crust preserved in Northwest Africa 7533. comagmatic origin of chassignites and nakhlites: inferences 45th Lunar Planet. Sci. Conf. #1880. from chlorine-rich minerals, petrology, and geochemistry. Ikeda Y. (2005) Magmatic inclusions in martian meteorites. Meteorit. Planet. Sci. 48, 819–853. Antarct. Meteor. Res. 18, 170–187. McCubbin F. M., Shearer C. K., Burger P. V., Hauri E. H., Wang Imae N. and Ikeda Y. (2007) Petrology of the Miller Range 03346 J. H., Elardo S. M. and Papike J. J. (2014) Volatile abundances nakhlite in comparison with the Yamato-000593 nakhlite. of coexisting merrillite and apatite in the martian meteorite Meteorit. Planet. Sci. 42, 171–184. Shergotty: implications for merrillite in hydrous magmas. Am. Irvine T. N. and Baragar W. R. A. (1971) Guide to chemical Mineral. 99, 1347–1354. classification of common volcanic rocks. Can. J. Earth Sci. 8, McSween H. Y., Stolper E. M., Taylor L. A., Muntean R. A., 523–548. Okelley G. D., Eldridge J. S., Biswas S., Ngo H. T. and Johnson M. C., Rutherford M. J. and Hess P. C. (1991) Chassigny Lipschutz M. E. (1979) Petrogenetic relationship between Allan petrogenesis – melt compositions, intensive parameters, and Hills 77005 and other achondrites. Earth Planet. Sci. Lett. 45, water contents of martian (questionable) magmas. Geochim. 275–284. Cosmochim. Acta 55, 349–366. McSween H. Y., Eisenhour D. D., Taylor L. A., Wadhwa M. and Kress V. C. and Carmichael I. S. E. (1991) The compressibility of Crozaz G. (1996) QUE94201 shergottite: crystallization of a silicate liquids containing Fe2O3 and the effect of composition, Martian basaltic magma. Geochim. Cosmochim. Acta 60, 4563– temperature, oxygen fugacity and pressure on their redox states. 4569. Contrib. Mineral. Petrol. 108, 82–92. McSween H. Y., Murchie S. L., Crisp J. A., Bridges N. T., Kring, D. A., Gleason, J. D., Hill, D. H., Jull, A. J. T. and Anderson R. C., Bell J. F., Britt D. T., Bruckner J., Dreibus G., Boynton, W. V. (1996) QUE94201, a new Martian meteorite Economou T., Ghosh A., Golombek M. P., Greenwood J. P., 84 A.R. Santos et al. / Geochimica et Cosmochimica Acta 157 (2015) 56–85

Johnson R., Moore H. J., Morris R. V., Parker T. J., Rieder R., saturated alkalic suites: an experimental study. J. Petrol. 45, Singer R. and Wanke H. (1999) Chemical, multispectral, and 693–721. textural constraints on the composition and origin of rocks at Nekvasil H., Filiberto J., McCubbin F. M. and Lindsley D. H. the Mars Pathfinder landing site. J. Geophys. Res. Planets 104, (2007) Alkalic parental magmas for chassignites? Meteorit. 8679–8715. Planet. Sci. 42, 979–992. McSween H. Y., Ruff S. W., Morris R. V., Bell J. F., Herkenhoff Nekvasil H., McCubbin F. M., Harrington A., Elardo S. and K., Gellert R., Stockstill K. R., Tornabene L. L., Squyres S. W., Lindsley D. H. (2009) Linking the Chassigny meteorite and the Crisp J. A., Christensen P. R., McCoy T. J., Mittlefehldt D. W. Martian surface rock Backstay: insights into igneous crustal and Schmidt M. (2006a) Alkaline volcanic rocks from the differentiation processes on Mars. Meteorit. Planet. Sci. 44, Columbia Hills, Gusev Crater, Mars. J. Geophys. Res. Planets 853–869. 111, 15. Nemchin A. A., Humayun M., Whitehouse M. J., Hewins R. H., McSween H. Y., Wyatt M. B., Gellert R., Bell J. F., Morris R. V., Lorand J. P., Kennedy A., Grange M., Zanda B., Fieni C. and Herkenhoff K. E., Crumpler L. S., Milam K. A., Stockstill K. Deldicque D. (2014) Record of the ancient martian hydrosphere R., Tornabene L. L., Arvidson R. E., Bartlett P., Blaney D., and atmosphere preserved in zircon from a martian meteorite. Cabrol N. A., Christensen P. R., Clark B. C., Crisp J. A., Des Nat. Geosci. 7, 638–642. Marais D. J., Economou T., Farmer J. D., Farrand W., Ghosh Nyquist, L. E., Shih, C. Y., Peng, Z. X. and Agee, C. B. (2013) A., Golombek M., Gorevan S., Greeley R., Hamilton V. E., NWA 7034 martian breccia: Disturbed Rb-Sr systematics, Johnson J. R., Joliff B. L., Klingelhofer G., Knudson A. T., preliminary 4.4 Ga Sm-Nd age. 76th Annu. Meteoritical Soc. McLennan S., Ming D., Moersch J. E., Rieder R., Ruff S. W., Meeting. #5318. Schroder C., de Souza P. A., Squyres S. W., Wanke H., Wang Owens B. E. and Dymek R. F. (1992) Fe–Ti–P-rich rocks and A., Yen A. and Zipfel J. (2006b) Characterization and massif anorthosite – problems of interpretation illustrated from petrologic interpretation of olivine-rich basalts at Gusev the Labrieville and St-Urbain plutons, Quebec. Can. Mineral. Crater, Mars. J. Geophys. Res. Planets 111, 17. 30, 163–190. McSween H. Y., Taylor G. J. and Wyatt M. B. (2009) Elemental Papike J. J., Fowler G. W. and Shearer C. K. (1997) Evolution of composition of the Martian crust. Science 324, 736–739. the lunar crust: SIMS study of plagioclase from ferroan Mezger K., Debaille V. and Kleine T. (2013) Core formation and anorthosites. Geochim. Cosmochim. Acta 61, 2343–2350. mantle differentiation on Mars. Space Sci. Rev. 174, 27–48. Papike J. J., Karner J. M., Shearer C. K. and Burger P. V. (2009) Mikouchi T. (2001) Mineralogical similarities and differences Silicate mineralogy of martian meteorites. Geochim. between the Los Angeles basaltic shergottites and the Asuka- Cosmochim. Acta 73, 7443–7485. 881757 lunar mare meteorite. Antarct. Meteor. Res. 14, 1–20. Pernet-Fisher J. F., Howarth G. H., Liu Y., Chen Y. and Taylor L. Mikouchi, T., Miyamoto, M. and McKay, G. A. (1996) A. (2014) Estimating the lunar mantle water budget from Mineralogy and petrology of a new Antarctic shergottite phosphates: complications associated with silicate-liquid-im- QUE94201: a coarse-grained basalt with unusual pyroxene miscibility. Geochim. Cosmochim. Acta 144, 326–341. zoning. 27th Lunar Planet. Sci. Conf. 879–880. Preibisch S., Saalfeld S. and Tomancak P. (2009) Globally optimal Minitti M. E. and Rutherford M. J. (2000) Genesis of the Mars stitching of tiled 3D microscopic image acquisitions. Pathfinder “sulfur-free” rock from SNC parental liquids. Bioinformatics 25, 1463–1465. Geochim. Cosmochim. Acta 64, 2535–2547. Rosenbusch H. (1887). . Mittlefehldt D. W. (1994) ALH84001, a cumulate orthopyroxenite Rubin A. E., Warren P. H., Greenwood J. P., Verish R. S., Leshin member of the martian meteorite clan. Meteoritics 29, 214–221. L. A., Hervig R. L., Clayton R. N. and Mayeda T. K. (2000) Mojzsis S. J., Harrison T. M. and Pidgeon R. T. (2001) Oxygen- Los Angeles: the most differentiated basaltic martian meteorite. isotope evidence from ancient zircons for liquid water at the Geology 28, 1011–1014. Earth’s surface 4300 Myr ago. Nature 409, 178–181. Schmidt M. E., Schrader C. M. and McCoy T. J. (2013) The Monders A. G., Medard E. and Grove T. L. (2007) Phase primary fO(2) of basalts examined by the Spirit rover in Gusev equilibrium investigations of the Adirondack class basalts from Crater, Mars: evidence for multiple redox states in the martian the Gusev plains, Gusev Crater, Mars. Meteorit. Planet. Sci. 42, interior. Earth Planet. Sci. Lett. 384, 198–208. 131–148. Schmidt M. E., Campbell J. L., Gellert R., Perret G. M., Treiman Musselwhite D. S., Dalton H. A., Kiefer W. S. and Treiman A. H. A. H., Blaney D. L., Olilla A., Calef, III, F. J., Edgar L., Elliot (2006) Experimental petrology of the basaltic shergottite B. E., Grotzinger J., Hurowitz J., King P. L., Minitti M. E., Yamato-980459: implications for the thermal structure of the Sautter V., Stack K., Berger J. A., Bridges J. C., Ehlmann B. L., Martian mantle. Meteorit. Planet. Sci. 41, 1271–1290. Forni O., Leshin L. A., Lewis K. W., McLennan S. M., Ming Muttik, N., Keller, L. P., Agee, C. B., McCubbin, F. M., Santos, D. W., Newsom H., Pradler I., Squyres S. W., Stolper E. M., A. R. and Rahman, Z. (2014a) A TEM Investigation of the Thompson L., VanBommel S. and Wiens R. C. (2014) Fine-Grained Matrix of the Martian Basaltic Breccia NWA Geochemical diversity in first rocks examined by the Curiosity 7034. 45th Lunar Planet. Sci. Conf. #2763. Rover in Gale Crater: evidence for and significance of an alkali Muttik N., McCubbin F. M., Keller L. P., Santos A. R., and volatile-rich igneous source. J. Geophys. Res. Planets 119, McCutcheon W. A., Provencio P. P., Rahman Z., Shearer C. 64–81.

K., Boyce J. W. and Agee C. B. (2014b) Inventory of H2O in the Sharp Z. D., McCubbin F. M. and Shearer C. K. (2013) A ancient martian regolith from Northwest Africa 7034: the hydrogen-based oxidation mechanism relevant to planetary important role of Fe oxides. Geophys. Res. Lett.. formation. Earth Planet. Sci. Lett. 380, 88–97. Neal C. R., Donohue P. H., Fagan A. L., O’Sullivan K., Oshrin J. Shearer C. K., Papike J. J. and Spilde M. N. (2001) Trace-element and Roberts S. (2015) Distinguishing between basalts produced partitioning between immiscible lunar melts: an example from by endogenic volcanism and impact processes: a non-destruc- naturally occurring lunar melt inclusions. Am. Mineral. 86, tive method using quantitative petrography of lunar basaltic 238–246. samples. Geochim. Cosmochim. Acta 148, 62–80. Shearer C. K., Papike J. J., Burger P. V., Sutton S. R., McCubbin Nekvasil H., Dondolini A., Horn J., Filiberto J., Long H. and F. M. and Newville M. (2011) Direct determination of Lindsley D. H. (2004) The origin and evolution of silica- europium valence state by XANES in extraterrestrial merrillite: A.R. Santos et al. / Geochimica et Cosmochimica Acta 157 (2015) 56–85 85

implications for REE crystal chemistry and martian magma- Treiman A. H. (2005) The nakhlite meteorites: augite-rich igneous tism. Am. Mineral. 96, 1418–1421. rocks from Mars. Chem. Erde Geochem. 65, 203–270. Shearer C. K., Burger P. V., Papike J. J., McCubbin F. M. and Bell Tuff J., Wade J. and Wood B. J. (2013) Volcanism on Mars A. S. (2014) Crystal chemistry of merrillite from martian controlled by early oxidation of the upper mantle. Nature 498, meteorites: mineralogical recorders of magmatic processes and 342–345. planetary differentiation. Meteorit. Planet. Sci.. http:// Udry A., McSween H. Y., Lecumberri-Sanchez P. and Bodnar R. dx.doi.org/10.1111/maps.12355. J. (2012) Paired nakhlites MIL 090030, 090032, 090136, and Smith J. V. and Hervig R. L. (1979) Shergotty meteorite – 03346: insights into the Miller Range parent meteorite. mineralogy, petrography and minor elements. Meteoritics 14, Meteorit. Planet. Sci. 47, 1575–1589. 121–142. Udry A., Balta J. B. and McSween H. Y. (2014a) Exploring Steele I. M. and Smith J. V. (1982) Petrography and mineralogy of fractionation models for Martian magmas. J. Geophys. Res. 2 basalts and olivine–pyroxene-spinel fragments in achondrite Planets 119, 1–18. EETA79001. J. Geophys. Res. 87, A375–A384. Udry A., Lunning N. G., McSween H. Y. and Bodnar R. J. (2014b) Stolper E. and McSween H. Y. (1979) Petrology and origin of the Petrogenesis of a vitrophyre in the martian meteorite breccia shergottite meteorites. Geochim. Cosmochim. Acta 43, 1475– NWA 7034. Geochim. Cosmochim. Acta 141, 281–293. 1498. Usui T., McSween H. Y. and Clark B. C. (2008) Petrogenesis of Stolper E. M., Baker M. B., Newcombe M. E., Schmidt M. E., high-phosphorous Wishstone Class rocks in Gusev Crater, Treiman A. H., Cousin A., Dyar M. D., Fisk M. R., Gellert R., Mars. J. Geophys. Res. Planets 113, 13. King P. L., Leshin L., Maurice S., McLennan S. M., Minitti M. Wadhwa M., McSween H. Y. and Crozaz G. (1994) Petrogenesis of E., Perrett G., Rowland S., Sautter V. and Wiens R. C. (2013) shergottite meteorites inferred from minor and trace-element The petrochemistry of Jake_M: a Martian mugearite. Science microdistributions. Geochim. Cosmochim. Acta 58, 4213–4229. 341. http://dx.doi.org/10.1126/science.1239463. Whitaker M. L., Nekvasil H., Lindsley D. H. and Difrancesco N. J. Stormer J. C., Pierson M. L. and Tacker R. C. (1993) Variation of (2007) The role of pressure in producing compositional F-X-ray and Cl-X-ray intensity due to anisotropic diffusion in diversity in intraplate basaltic magmas. J. Petrol. 48, 365–393. apatite during electron-microprobe analysis. Am. Mineral. 78, Wilde S. A., Valley J. W., Peck W. H. and Graham C. M. (2001) 641–648. Evidence from detrital zircons for the existence of continental Tarte`se, R., Anand, M., McCubbin, F. M., Santos, A. R. and crust and oceans on the Earth 4.4 Gyr ago. Nature 409, 175– Delhaye, T. (2014) Zircons in Northwest Africa 7034: recorders 178. of crustal evolution on Mars. 45th Lunar Planet. Sci. Conf. Yin, Q.-Z., McCubbin, F. M., Zhou, Q., Santos, A. R., Tarte`se, R., #2020. Li, X., Li, Q., Liu, Y., Tang, G., Boyce, J. W., Lin, Y., Yang, Taylor G. J. (2013) The bulk composition of Mars. Chem. Erde W., Zhang, J., Hao, J., Elardo, S. M., Shearer, C. K., Rowland, Geochem. 73, 401–420. D. J., Lerche, M. and Agee, C. B. (2014) An Earth-like Taylor S. R. and McLennan S. M. (2009) Planetary Crusts: Their beginning for ancient Mars indicated by alkali-rich volcanism Composition, Origin, and Evolution. Cambridge University at 4.4 Ga. 45th Lunar Planet. Sci. Conf. #1320. Press, Cambridge. Zirkel F. (1873) Die Mikroskopische Beschaffenheit der Mineralien Taylor G. J., Martel L. M. V., Karunatillake S., Gasnault O. and und Gesteine. Wilhelm Englemann, Leipzig, Germany. Boynton W. V. (2010) Mapping Mars geochemically. Geology 38, 183–186. Associate editor: Alexander Nemchin Trail D., Watson E. B. and Tailby N. D. (2011) The oxidation state of Hadean magmas and implications for early Earth’s atmo- sphere. Nature 480, 79–U238.