Meteoritics & Planetary Science 52, Nr 1, 89–124 (2017) doi: 10.1111/maps.12740

Regolith breccia Northwest Africa 7533: Mineralogy and petrology with implications for early Mars

Roger H. HEWINS1,2*, Brigitte ZANDA1,3, Munir HUMAYUN4, Alexander NEMCHIN5,6, Jean-Pierre LORAND7, Sylvain PONT1, Damien DELDICQUE8, Jeremy J.BELLUCCI5, Pierre BECK9, Hugues LEROUX10, Maya MARINOVA11, Laurent REMUSAT1, Christa GOPEL€ 12, Eric LEWIN9, Marion GRANGE6, Allen KENNEDY6, and Martin J. WHITEHOUSE5

1Institut de Mineralogie, de Physique des Materiaux, et de Cosmochimie (IMPMC), Sorbonne Universite, Museum National d’Histoire Naturelle, UPMC Universite Paris 06, UMR CNRS 7590, IRD UMR 206, 75005 Paris, France 2Department of Earth and Planetary Sciences, Rutgers University, Piscataway, New Jersey 08854, USA 3IMCCE, Observatoire de Paris, 77 Av. Denfert Rochereau, CNRS UMR 8028, Paris F-75014, France 4Department of Earth, Ocean & Atmospheric Science and National High Magnetic Field Laboratory, Florida State University, Tallahassee, Florida 32310, USA 5Department of Geosciences, Swedish Museum of Natural History, Stockholm SE-104 05, Sweden 6Department of Applied Geology, Curtin University, Perth, Washington 6845, Australia 7Laboratoire de Planetologie et Geodynamique, Universite de Nantes, Nantes 44322, France 8Laboratoire de Geologie, Ecole Normale Superieure, Paris CEDEX 5 75231, France. 9UJF-Grenoble 1/CNRS-INSU, Institut de Planetologie et d’Astrophysique de Grenoble (IPAG) UMR 5274, Grenoble F-38041, France 10Unite Materiaux et Transformations, University of Lille & CNRS, Villeneuve d’Ascq F-59655, France 11Institut Chevreul, University of Lille & CNRS, Villeneuve dAscq F-59655, France 12Institut de Physique du Globe de Paris, Sorbonne Paris Cite, Univ Paris Diderot, UMR 7154 CNRS, Paris F-75005, France *Corresponding author. E-mail: [email protected] (Received 27 November 2015; revision accepted 02 September 2016)

Abstract–Northwest Africa 7533, a polymict Martian breccia, consists of fine-grained clast- laden melt particles and microcrystalline matrix. While both melt and matrix contain medium-grained noritic-monzonitic material and crystal clasts, the matrix also contains lithic clasts with zoned pigeonite and augite plus two feldspars, microbasaltic clasts, vitrophyric and microcrystalline spherules, and shards. The clast-laden melt rocks contain clump-like aggregates of orthopyroxene surrounded by aureoles of plagioclase. Some shards of vesicular melt rocks resemble the pyroxene-plagioclase clump-aureole structures. Submicron size matrix grains show some triple junctions, but most are irregular with high intergranular porosity. The noritic-monzonitic rocks contain exsolved pyroxenes and perthitic intergrowths, and cooled more slowly than rocks with zoned-pyroxene or fine grain size. Noritic material contains orthopyroxene or inverted pigeonite, augite, calcic to intermediate plagioclase, and chromite to Cr-bearing magnetite; monzonitic clasts contain augite, sodic plagioclase, K feldspar, Ti-bearing magnetite, ilmenite, chlorapatite, and zircon. These feldspathic rocks show similarities to some rocks at Gale Crater like Black Trout, Mara, and Jake M. The most magnesian orthopyroxene clasts are close to ALH 84001 orthopyroxene in composition. All these materials are enriched in siderophile elements, indicating impact melting and incorporation of a projectile component, except for Ni-poor pyroxene clasts which are from pristine rocks. Clast-laden melt rocks, spherules, shards, and siderophile element contents indicate formation of NWA 7533 as a regolith breccia. The zircons, mainly derived from monzonitic (melt) rocks, crystallized at 4.43 Æ 0.03 Ga (Humayun et al. 2013) and a 147Sm-143Nd isochron for NWA 7034 yielding 4.42 Æ 0.07 Ga (Nyquist et al. 2016) defines the crystallization age of all its igneous portions. The zircon from the monzonitic rocks has a higher D17O than other Martian

89 © The Meteoritical Society, 2016. 90 R. H. Hewins et al.

meteorites explained in part by assimilation of regolith materials enriched during surface alteration (Nemchin et al. 2014). This record of protolith interaction with atmosphere- hydrosphere during regolith formation before melting demonstrates a thin atmosphere, a wet early surface environment on Mars, and an evolved crust likely to have contaminated younger extrusive rocks. The latest events recorded when the breccia was on Mars are resetting of apatite, much feldspar and some zircons at 1.35–1.4 Ga (Bellucci et al. 2015), and formation of Ni-bearing pyrite veins during or shortly after this disturbance (Lorand et al. 2015).

INTRODUCTION analyses represent pristine or altered volcanic rocks, or volcanic debris in sedimentary rocks, or impact rocks Since the Viking mission (1976), a vast amount of (Ashley and Delaney 1999; McSween et al. 2003; information has been obtained on Mars. Its geology Schultz and Mustard 2004). Curiosity is currently and geochemistry have been studied using orbiters, making many fascinating observations at Gale Crater. rovers, and the SNC meteorites. Orbiting visible/ These include descriptions of polymict fragmentary two- infrared spectrometers, for example, OMEGA/MEx, feldspar-rich rocks of either sedimentary or impact have led to the production of global maps showing the origin (Mangold et al. 2013), and analyses of a rock distribution of olivine, pyroxene, hydrated minerals, and with the chemistry of mugearite, that is, basaltic trachy- ferric iron oxides (Mustard et al. 2005; Poulet et al. andesite (Gellert et al. 2013; Stolper et al. 2013; 2005, 2007; Ody et al. 2012). Infrared detection of Schmidt et al. 2014). Sautter et al. (2014) have plagioclase and its variability has clarified rock types: described many feldspar-rich samples, and have TES spectra suggested that the crust of Mars was not deconvolved the data using major-element ratios to purely basaltic (Bandfield et al. 2000; McSween et al. show that orthopyroxene, augite, plagioclase, and 2003), and CRISM spectra (Carter and Poulet 2013; orthoclase are present. Wray et al. 2013) indicated plagioclase-rich rocks Martian meteorites have been subjected to very showing that magmatic evolution must have been complete analysis, particularly of isotope abundances, complex. Oxides of ferric iron are dominant in the which cannot be measured by rovers, and their parent spectra for the northern hemisphere while abundant magmas yield information on the mantle of Mars (e.g., orthopyroxene is found only in the southern heavily Debaille et al. 2007). They might constitute an inverse cratered uplands of Noachian age. Maps of secondary ground truth for rover analyses where sample minerals have given rise to the concept of changing preparation and instrument calibration may be difficult. climate and alteration style with time (Bibring et al. However, the SNC meteorites differ geochemically in 2006), with the more abundant evidence of (less acid) some respects from Martian surface materials analyzed water in the Noachian (phyllosian alteration), making in situ (e.g., McSween et al. 2003). Matches between this period the most favorable for the development of shergottite and surface spectral signatures have been life. More recently, higher resolution spectral data have reported for certain regions (Ody et al. 2015). However, given detailed pictures of local geologic history the largest group of Martian meteorites, the shergottites, (Murchie et al. 2009). Carter et al. (2013) reported has very young radiometric ages, whereas basalts occurrences of 10 hydrated minerals, including the mineralogically similar to shergottites emplaced in Gusef discovery of epidote, giving evidence for hydrothermal Crater in the Hesperian date from ~3.65 Ga based on activity. The residual heat associated with large impact crater counting (Greeley et al. 2005; McSween et al. craters is likely to have generated hydrothermal systems, 2006). Mineral isochrons for the shergottites give very possibly accompanied by freezing of crater lakes, with young ages, typically <200 Ma (Nyquist et al. 2001; interesting biochemical possibilities (Newsom 1980; Borg et al. 2005; Bouvier et al. 2008; Niihara et al. Osinski et al. 2001, 2013; Abramov and Kring 2005). 2012), and they may therefore come from such very Rovers have provided in situ bulk rock analyses on restricted regions. There has been controversy over the Mars (e.g., McSween et al. 2009; Sautter et al. 2016). formation ages of shergottites (Bouvier et al. 2008) but Basaltic andesite was found in Ares Vallis by the young ages have recently been confirmed by dating Zr- Pathfinder mission (McSween et al. 1999, 2003) and rich phases in shergottites (Niihara et al. 2012; Moser picritic basalt at Gusev by Spirit (Gellert et al. 2004). et al. 2013). Based on crater counting on high-resolution These rocks are usually interpreted as volcanic in images, volcanic calderas on Mars vary greatly in age nature, but there is always the question of whether the (Robbins et al. 2011). Volcanism continued until a little Martian regolith breccia Northwest Africa 7533 91 over a hundred million years ago in a few volcanoes like microscopy, scanning electron microscopy (SEM), and Olympus Mons. Only the ancient orthopyroxenite ALH electron probe microanalyzer (EPMA). They are in part 84001, dated at 4.1 Ga (Bouvier et al. 2009; Lapen et al. serial polished sections, usually too thick for identifying 2010), has been a candidate to represent the Noachian of the same object in two sections. Backscattered electron the cratered southern uplands. Ody et al. (2015) discuss (BSE) maps and images of selected clasts were made at possible occurrences of similar orthopyroxene-bearing MNHN using a Tescan VEGA II LSU SEM in rocks on the Noachian surface. These cratered highlands conventional mode (mainly 15 keV and <20 nA), and make up nearly half the surface of Mars, and most of minerals were characterized with an SD3 (Bruker) EDS the largest impact structures occur in the oldest part of it detector. A Zeiss Σigma field emission SEM with (Tanaka et al. 2014). It is surprising that ancient GEMINI technology using a SmartSEMÒ interface and Martian meteorites are not more common and, in view an Oxford 50 mm2 detector was used at Laboratoire de of the extent of the early heavy bombardment on Mars, Geologie, Ecole Normale Superieure, Paris; high- that breccias have not been prominent among Martian resolution imaging was done on polished sections 7533- meteorites. This may be related to the efficiency of 1 and -2 at 15 keV and 8.4 nA. X-ray maps were made ejection of weakly lithified material (Hartmann and on both SEM instruments. All quantitative mineral Barlow 2006). analyses were made by wavelength-dispersive spectrometry The first Martian breccia meteorite identified was ontheCamecaSXFiveelectronmicroprobeatthe Northwest Africa 7034 (Agee et al. 2012, 2013; Ziegler Universite Paris VI, generally using 15 keV and 10 nA. et al. 2013; Cartwright et al. 2014) and shortly For analysis of Ni in silicates, we used 20 kV and afterward we obtained Northwest Africa 7533 (Hewins 300 nA, counting for 100 s on three WDS simultaneously et al. 2013a; Humayun et al. 2013), the subject of this (Hewins et al. 2014). San Carlos, Chassigny, and Eagle paper. These meteorites, and NWA 7475, NWA 7906, Station olivines and Tatahouine pyroxene (Jarosewich NWA 7907, NWA 8114, NWA 8171 (Korotev et al. et al. 1980; Smith et al. 1983; Barrat et al. 1999; 2013; Wittmann et al. 2013a, 2015; Hofmann et al. El Goresy et al. 1999; De Hoog et al. 2010), were 2014; MacArthur et al. 2015) are paired. These breccia analyzed as internal standards. The detection limit, meteorites are very different from the SNC meteorites, based on count rates on peak and background above as are known Martian surface rocks (McSween et al. and below the peak, is a probability function considering 2003; Stolper et al. 2013; Sautter et al. 2014) and, the risk of treating positive as zero, and vice versa, at because of the diverse clast suite, can potentially yield the 95% confidence level, and is 24 ppm Ni in our much information on the early crust of Mars. They pyroxene. form an important window on a previously unsampled Powder diffraction data were collected on a Bruker and probably major lithologic unit of Mars. A first D2 Phaser diffractometer equipped with a Lynxeye impression of NWA 7533, with which we concurred, CCD detector. X-rays were generated by a copper was that it was “fiendishly complex” (McSween 2013). cathode (k = 1.54060 A À 30 kV/10 mA) and the Ka However, we are now inclined to agree with Simonds radiations isolated with a Cu Kb filter. The sample et al. (1977) that “breccias aren’t so bad after all.” holder rotation speed was 20 rotations/min and a NWA 7533 has yielded evidence of formation in a numeric filter was applied to the data set in order to regolith, of the nature of early crustal reservoirs, of minimize the iron fluorescence contribution. zircon crystallization at 4.428 Ga, and of atmosphere- Transmission infrared spectra were measured on hydrosphere-surface interaction before this time 7533-LM in an environmental cell with a Brucker (Hewins et al. 2013a, 2013b; Humayun et al. 2013; Hyperion FTIR spectrometer in the 500–4000 cmÀ1 Nemchin et al. 2014). We therefore present here an range with a 4 cmÀ1 resolution. Diffuse visible and near account of the petrologic components in this polymict infrared reflectance spectra were measured using the breccia in the context of the new discoveries stemming spectro-photo-goniometer available at IPAG. from this important find. Three fine-grained matrix areas were extracted as focused ion beam (FIB) sections (~100 nm thick) using SAMPLES AND METHODS an FEI Strata DB 235 at IEMN at the University of Lille, France. They were studied by transmission A 17 g sample of NWA 7533 was cut without using electron microscopy (TEM) with an FEI Tecnai G2-20 any lubricating fluids (no water) at the Museum Twin (LaB6, 200 kV) and a Philips CM30 (LaB6, National d’Histoire Naturelle (MNHN). Seven polished 300 kV) at the electron microscopy center of the sections (7533-1 to -7) and one doubly polished University of Lille, France. Both instruments use an transparent thick section (7533-LM, ≥50 lm, energy dispersive spectrometer (EDS) for elemental unmounted) were prepared and studied using optical microanalysis. 92 R. H. Hewins et al.

D/H ratios of pyrite alteration products were communication), are indistinguishable from NWA 7034 analyzed on 7533-LM on the NanoSIMS installed at values (Agee et al. 2013). This material is very different MNHN Paris by Lorand et al. (2015). A 30 pA Cs+ from the SNC meteorites (Agee et al. 2013; Hewins primary beam was used to collect HÀ and DÀ et al. 2013a, 2013b; Humayun et al. 2013), as are many secondary ions in multicollection mode on a Martian surface rocks analyzed by rovers (McSween 20 9 20 lm² surface area at a raster speed of 1 ms/pixel. et al. 2009; Stolper et al. 2013; Sautter et al. 2014). Instrumental fractionation was corrected using a terrestrial Noble gas isotope data show the presence of a goethite sample with a D/H isotopic composition of trapped gas component in NWA 7034 like the modern À150&. Martian atmosphere (Cartwright et al. 2014) and this is One polished section 7533-3 was studied by laser the clearest isotopic evidence for a Martian origin for ablation inductively coupled plasma–mass spectrometry these regolith breccias. The oxygen isotope data are (LA-ICP-MS) for elemental abundances. Analyses were complex: an orthopyroxene mineral separate exhibited performed with a New Wave UP193FX (193 nm) D17Oof0.33&, indistinguishable from SNC compositions excimer laser ablation system coupled to a Thermo (Ziegler et al. 2013), confirming a Martian origin. Element XR at the Plasma Analytical Facility, FSU. A However, there is an overall trend of increasing D17Owith total of 73 elements, including all major elements increasing d18O in NWA 7034 separates, toward the (Humayun et al. 2010), S and Cl, were determined composition of known Martian alteration phases, and simultaneously in low-resolution mode using methods Nemchin et al. (2014) found D17OvaluesinNWA7533 previously described (Humayun et al. 2007; Gaboardi zircons ranging from within error of the Mars and Humayun 2009). Standardization utilized 6 MPI- Fractionation Line (MFL) to +2.16 Æ 0.56&. We discuss DING glasses, 3 USGS silicate glasses, NIST SRM 610, below the implications for the Martian atmosphere and SRM 1263a steel, the iron meteorites North Chile regolith. The FeO/MnO ratio for all pyroxenes in NWA (Filomena) and Hoba, and a pyrite (for S). Beam spots 7533 is ~34, except for those in clast-laden melt rocks (see were 50–150 lm ablated at 50 Hz, using either spot or below), like that of ALH 84001 pyroxene, and this is also a raster mode. Martian signature. Zr-rich phases were located by X-ray mapping of The bulk breccia composition is equivalent to alkali 7533-4 and analyzed by SIMS. U-Pb isotope analyses of basalt formed by 4% partial melting of primitive mantle zircon were performed on a sensitive high-resolution ion (Humayun et al. 2013) recycled into a suite of rocks in microprobe (SHRIMP II) at Curtin University, as this early crust. It cannot be said to be simply basaltic, described in Humayun et al. (2013). Oxygen isotope in that it contains pyroxenitic, noritic, and monzonitic ratios were obtained using a Cameca IMS1480 ion material. It is not simply related to the SNC meteorites: microprobe at the Swedish Museum of Natural History they contain olivine, except for the most fractionated (Nemchin et al. 2014) using 1 nA Cs+ primary beam shergottites, whereas olivine is almost entirely absent with an impact energy of 20 kV and operating at a from the breccia. SNC are volcanic and subvolcanic mass resolution sufficient to resolve 17OÀ1 from 16O1HÀ rocks, but NWA 7533 contains material deposited on (M/DM = 7500). The U-Pb apatite analyses were the surface, such as melt spherules, and lithic clasts of performed on the same instrument with a À13 kV rocks crystallized at depths possibly up to kilometers. molecular oxygen beam and operating at a mass Noritic clasts in the breccia contain inverted pigeonite resolution (M/DM) of 5400 (Bellucci et al. 2015). The (whereas pigeonite is widely preserved in SNC) with Pb isotopic compositions of plagioclase and alkali exsolution lamellae many microns thick, compared to À feldspar were determined using a 9 nA O2 primary only hundreds of nanometers in SNC pyroxenes. The beam and a mass resolution of 4860 (M/DM). breccia contains higher concentrations of siderophile elements than the SNC, for the same Mg content PROVENANCE OF THE METEORITE (Humayun et al. 2013), and its sulfides are of hydrothermal origin (pyrite) rather than magmatic The NWA 7533 breccia has mineralogical, chemical, origin (pyrrhotite) in the SNC (Lorand et al. 2015). and isotopic properties consistent with a Martian origin While the SNC are relatively young rocks (Amazonian), and similar to those of NWA 7034 and NWA 7475 this black and white breccia NWA 7533 is a major (Agee et al. 2013; Hewins et al. 2013a, 2013b; Humayun repository of geochemical information on early Mars, et al. 2013; Nemchin et al. 2014; Wittmann et al. 2015). along with ALH 84001, and some of the rocks analyzed The bulk oxygen isotopic composition by IR-laser by Curiosity at Gale Crater. As it is very diverse fluorination/mass spectrometry are d18O = 5.92; lithologically, we sketch our view of the breccia and its d17O = 3.65; D17O = 0.57 (all per mil) and the magnetic components in Fig. 1. We present below observations susceptibility, log X = 4.45 (Gattacceca, personal and data first for the matrix, followed by sections on Martian regolith breccia Northwest Africa 7533 93

Fig. 1. Schematic of components of NWA 7533 breccia; R = clast-laden melt rock, S = spherule, B = microbasalt, sd = sedimentary, Z = zoned-pyroxene, N = noritic, M = monzonitic, X = crystal clast, O = orthopyroxene, A = augite, A-P = augite- pigeonite, P = plagioclase, and K = K feldspar.

Fig. 2. Mg Ka image of polished section NWA 7533-2. Clast- spherules and shards, lithic and crystal clasts. The lithic laden melt rock bodies (CLMR) are darker (more Fe rich) material is described in a sequence of increasing grain than breccia matrix. Feldspar-rich spherule (S1) and size: clast-laden melt rocks, microbasaltic clasts, clasts sedimentary clast (SED) near top center. Microbasalt (MB), with zoned pyroxenes, and noritic- monzonitic clasts. noritic (N), and monzonitic (M) clasts are indicated. Pyroxene clasts (and olivine in partial spherule S2) are light gray, feldspar is black. THE BRECCIA COMPONENTS AND MATRIX

The breccia is complex with a variety of crystal matrix is crystalline and annealed, rather than a clastic clasts, inclusions, and crystal types, contained in breccia aggregate, and it was studied in detail by field emission matrix and in aphanitic melt rock bodies. We discuss (FE) SEM in sections 7533-1 and -2. It is composed the matrix here and subsequently present results by mainly of anhedral micron-sized plagioclase feldspar, lithological type of object. A cut slab is shown in with pyroxene grains down to a few hundred nm Fig. S1 in supporting information and an Mg Ka X-ray surrounding and embedded in feldspar (Figs. 3a and S2 map of a polished section (7533-2) in Fig. 2, in which in supporting information). The pyroxene is somewhat feldspars are dark gray and pyroxenes are medium to clustered into tiny aggregates (Fig. S2d) between light gray. Both show spherules, melt rock bodies, lithic plagioclase grains. Fe-rich phases, especially magnetite clasts, and crystal clasts. Figure 2 shows a feldspar-rich and probably maghemite, may be symplectitic or lacy, melt spherule 3 mm in diameter and a spherule and are irregularly distributed in the matrix, especially in fragment with olivine and pyroxene dendrites, estimated pyroxene. Microprobe analyses of minerals in the matrix at about 6 mm in former diameter, as well as smaller are of low quality because of the small grain size although crystal and monzonitic and microbasaltic lithic clasts. they are sufficient to identify pyroxenes and calcic The hosts of the smaller clasts are twofold, clast-laden plagioclase similar to those in the melt rocks. Analytical aphanitic bodies and matrix, which are hard to transmission electron microscope (ATEM EDS) analyses distinguish at this scale, although breccia matrix tends are shown in Table S1 in supporting information. For to be brighter, that is, more Mg-rich (Fig. 2). The plagioclase, they suffer from low cation totals. Since Ca aphanitic bodies, interpreted below as clast-laden melt correlates well with Si and Al, but Na does not, likely due rocks with fine-grained igneous textures (~10 lm to migration under the TEM beam, endmembers were crystals), and other clasts are embedded in very fine- calculated using Si and Al atoms per formula unit (a.f.u.) grained (~1 lm) compact crystalline breccia matrix. to define An, for example, An% = 100*(Al a.f.u À1), K to The melt rock bodies, spherules, and lithic clasts define Or, and obtaining Ab by difference. The matrix occur with crystal clasts down to at least 5 lm, all also contains some vugs (Lorand et al. 2015; Leroux supported by a matrix with 200 nm to 2 lm grains. This et al. 2015) associated with pyrite veins. 94 R. H. Hewins et al.

organized along regular subgrain boundaries or deformation bands. Fe-oxides, for which the electron diffraction patterns are compatible with magnetite and maghemite, are distributed throughout the matrix. The grain size ranges from 1 lm to a few nm. Large Fe- oxides are located in the pyroxene-feldspar grain boundaries, including triple junctions, and the smaller ones may occur within the pyroxene grain interiors. Most of the grain boundaries are unusually thick (10– 20 nm) and filled with a low-density amorphous material very sensitive to the electron beam (Fig. S2). The composition of the grain boundaries is close to those of the adjacent minerals but enrichments in SiO2, Al2O3, CaO, and Cl were frequently detected (Leroux et al. 2015). The amorphous state could be the result of beam damage and, because of a smectite-like composition, the high e-beam sensitivity could be due to the presence of water. The amorphous material may be related to the decomposition of pyroxene into silica and Fe-oxide that has been identified in pyroxene clasts (Leroux et al. 2016), but another plausible explanation is alteration in hot desert conditions (Leroux et al. 2015). HRTEM analyses of the matrix of NWA 7034 also showed minor phyllosilicates (<50 nm) on pyroxene grain boundaries (Muttik et al. 2014a; LPSC 2763), but the identification of phyllosilicates was not confirmed in this study.

FINE-GRAINED CLASTS

Clast-Laden Impact Melt Rock Fig. 3. Breccia matrix. a) BSE image showing crystal clasts in a recrystallized matrix with interlocking submicron pyroxene The largest objects (~1 cm) in NWA 7533 are (light gray) and plagioclase grains (dark gray), the pyroxene smooth or angular fine-grained melt bodies (Fig. 2), granules being embedded in plagioclase grain margins; and (b) TEM bright-field image of the matrix with granoblastic containing lithic and crystal fragments. They include texture, 120° contacts, and porosity on grain boundaries. oval and aerodynamically shaped bodies (Hewins et al. 2013a; Humayun et al. 2013), which are particles of clast-laden melt rocks (Grieve et al. 1974; Simonds The matrix was also studied by TEM on FIB 1975). Similar material, clast-rich amoeboid vitrophyre, sections (Figs. 3b and S2c). It is a granoblastic is the dominant component in NWA 7475 (Wittmann assemblage with low-Ca pyroxene and feldspar grains et al. 2015), and the “plumose” groundmass with clasts with an average grain size close to 300 nm (Leroux described in NWA 7034 (Agee et al. 2013) is probably et al. 2015). Although some well-equilibrated triple the same. A croissant-shaped body in 7533-1 appears to junctions are observed, as for NWA 7034 (Muttik et al. be sculpted by transport while partially molten 2014a), most grains have preserved an irregular (Humayun et al. 2013). The melt rock is a fine-grained morphology. Intergranular porosity is also high, in plagioclase -pyroxene igneous rock mainly with particular, at triple junctions (Figs. 3b and S2c). The subophitic texture and the aspect ratio of the low-Ca pyroxenes range from En64 to En74 and contain plagioclase laths (mostly 1~2 lm wide) is ~10. Where a high density of planar defects on (100) corresponding pyroxene grains are parallel to plagioclase laths, the to thin alternating domains of clino- and ortho- texture is more fasciculate. The textures of the melt pyroxene. Most of the feldspar grains are plagioclase rock are shown in Figs. 4a–d and S3 in supporting (from An14 to An57) with minor alkali feldspar. Three information. Finer grained melt rocks have abundant grains of Cl-rich apatite >1 lm in size are present in the small crystal clasts (Fig. 4a), angular to more rounded FIB section and one of them contains dislocations than in breccia matrix. Coarser grain size (~30 lm long) Martian regolith breccia Northwest Africa 7533 95

Fig. 4. Clast-laden melt rock. a) BSE image of subophitic groundmass of ~10 lm plagioclase (dark gray) and pyroxene (light gray), containing 10–200 lm crystal clasts (x) and ~100 lm pyroxene clumps (cl). b) Sharp contact between less fine-grained melt rock (~20 lm plagioclase) and enclosed monzonitic clast with perthitic feldspar (pe), augite (ag), and chlorapatite (ap). c) Orthopyroxene clump with calcite in cracks and dendrites in plagioclase aureole perpendicular to clump surface. High-Z phases include magnetite and pyrite. d) Si Ka image of clumps and aureoles in melt rock in NWA 7533-2 (lower left edge of Fig. 2). Up to about 30 aureoles can be distinguished. See also Figs. S3 and S4 in supporting information. is associated with fewer crystal clasts and the presence RGB composite images (Fig. S4a) extend into the of lithic clasts (Fig. 4b). Angular lithic clasts in the melt plagioclase aureoles, in which plagioclase laths extend rock are up to ~1 mm long. into the groundmass. The pyroxene dendrites may be The finer grained melt rocks are remarkable in attached to the pyroxene of the groundmass. A Ca Ka containing ellipsoidal aggregates or clusters of map demonstrates that all the fractures in the clumps orthopyroxene granules or dendrites (with embedded (dark in Figs. 4c and 4d) are filled with calcite magnetite, chromite, ilmenite, or pyrite) surrounded by (Fig. S4c). Figures S3c, S3d, and S4b also show rings of plagioclase laths which radiate outward from chemical heterogeneity in the groundmass, with Mg- the pyroxene cluster (Figs. 4a and 4c). We have versus Al-rich zones or schlieren. Zones with few described examples of this pyroxene-plagioclase clumps are richer in plagioclase and poorer in pyroxene, association as clot-aureole structures (Hewins et al. that is, more Al-rich, than zones with many clumps, 2013a; Humayun et al. 2013). The most obvious which are Mg-rich. ellipsoidal objects are around 100 lm in size, but many Microprobe analysis of minerals in the groundmass smaller ones (down to ~30 lm) are recognized in X-ray of the melt rock is difficult because of the small grain maps (Figs. 4c, S3c, and S3d), especially Si Ka and size. Good analyses of pyroxene, however, are 96 R. H. Hewins et al.

consistently magnesian and show moderate Al2O3 (2–3 wt%) and TiO2 (up to 1 wt%) contents. The results are shown in the pyroxene quadrilateral (Fig. 5a) and Table S2 in supporting information. The analyses for 12 melt bodies in 4 sections are very similar and have a nearly constant Mg number of 69 Æ 2. The continuity in Ca concentrations (Wo1.8–47.4) suggests mixtures of orthopyroxene and augite, which is confirmed by BSE and X-ray images. However, the clumps are dominantly orthopyroxene En68.1 Æ 1.3Fs29.5 Æ 1.3Wo2.4 Æ 0.3,whereas the groundmass pyroxene is on average more Ca-rich (En52.3 Æ 6.1Fs23.3 Æ 4.5Wo24.4 Æ 12.9). The clump orthopyroxene is close in major-element composition to ALH 84001 orthopyroxene, which we show in this (Fig. 5a) and subsequent pyroxene quadrilaterals to emphasize similarities (and differences in the case of the more fractionated rocks). However, it is more Al-rich (2.9 Æ 0.9 wt% Al2O3) than ALH 84001 pyroxene and other orthopyroxenes in the breccia, and also extremely Cr-poor (0.03 wt% Cr2O3). Furthermore, whereas pyroxene in the breccia generally has an FeO/MnO ratio of ~34, with scattered high Mn exceptions, pyroxene in the clast-laden melt rocks is consistently more Mn-rich than this typically Martian value, with an FeO/MnO ratio of ~25 (Fig. 5b). The plagioclase is mainly andesine, An56–32 (An41 Æ 7), and not significantly different in the aureoles surrounding the clumps than in the groundmass (Fig. 5c, Table S3 in supporting information). It is similar in composition to that in the less fractionated lithic clasts.

Shards and Melt Spherules

Orthopyroxene-rich objects probably related to the melt rock clumps occur as clasts in the meteorite matrix, and tend to have an irregular or polygonal form. Most have concave surfaces bounded by plagioclase with the orthopyroxene dendrites in the interior (Figs. 6a and S5a in supporting information). These appear to be shards of fine-grained vesicular melt rock or devitrified glass. Other shard-like clasts have a foliated or spherulitic appearance. Rare small spherules Fig. 5. Mineral compositions for the groundmass and the clump-aureole structures of the clast-laden melt rocks: (a) resembling droplets of the clump material are described pyroxene quadrilateral, (b) FeO-MnO of pyroxene compared below. The mineral compositions of the shards are like to pyroxene elsewhere in the breccia, and (c) feldspar. those of the clumps, and some are included in Tables Orthopyroxene and augite in ALH 84001 are shown for S2, S3, and Fig. 5. comparison. Large melt spherules are prominent on the sawn surfaces of this breccia (e.g., Figs. 2 and S1) and of and feldspar laths in fine spherulitic to variolitic textures paired meteorites (e.g., Agee et al. [2012] fig. 1; (e.g., Figs. 6b–d). The spherule 7533-2-2 in Figs. 6b and Wittmann et al. 2015 fig. 1). The spherules can be 6c is 3 mm in diameter, with a groundmass speckled with grouped into three types: vitrophyric, crystalline, and submicron Fe-rich phase(s), including Cr-bearing altered. The largest spherules (diameters of several mm) magnetite and chromite, especially just under its surface, are vitrophyric, although their groundmass is not glassy and it develops coarsely crystalline patches near its but finely crystalline (grain size 1–5 lm) with pyroxene borders. It has no large dendrites and is traversed by Martian regolith breccia Northwest Africa 7533 97

Fig. 6. BSE images. a) Shard with peripheral plagioclase resembling pyroxene clumps. Concave margins indicate vesicle walls. b) Spherule 7533-2-1 with feldspar veins and magnetite-rich border. c) Close-up of (b) showing igneous texture and hyalophane- bearing veins. d) Detail of olivine-bearing spherule 7533-2-2 in Fig. 2 (bottom right). Olivine dendrite with plagioclase inclusions and overgrowth of orthopyroxene on both sides of dendrites are continuous with orthopyroxene intergrown with plagioclase laths in the groundmass of the spherule.

veins of plagioclase and hyalophane (Figs. 6b and 6c) Olivine occurs in NWA 7533 only in one melt consistent with hydrothermal activity. Another veined spherule 74533-2-2 (Fig. 2) as long chain dendrites feldspar-rich spherule (7533-7-sph1; Fig. S5b) has a Na- (Fig. 6d) along with long pyroxene dendrites. In NWA depleted signature consistent with melting of an altered, 7034, olivine dendrites occur in a large partial spherule, clay-bearing protolith (Humayun et al. 2014c). Oval but also as one grain in a microbasalt, probably in objects with a patchy texture, in some cases with a reaction relationship with pyroxene (Santos et al. vitrophyric interior region, have more coarsely crystalline 2015a). Plagioclase occurs as small insets in the olivine dendritic/spherulitic borders with pigeonite and subcalcic dendrites. The olivine dendrites are bordered by augite (Fig. S5b), in turn coated with probable orthopyroxene that is continuous with pyroxene accretionary rims. Similarly, spherules in NWA 7475 alternating with plagioclase in the spherulitic tend to preserve quenched mesostasis in their interiors groundmass (Fig. 6d). A similar object from NWA (Wittmann et al. 2015). Sampling of nonequatorial 7034, a vitrophyre with olivine dendrites (Agee et al. sections of larger spherules and differences in alteration 2013; Udry et al. 2014), has a high Na/K ratio and may account for variations in the appearance of resembles the more basaltic components of the breccia. spherules, which require detailed study. The derivation of the spherules from both clay-rich and 98 R. H. Hewins et al. basaltic sources suggests an analogy with S-type and I-type granites. Smaller spherules (a few hundred lmin diameter) are more coarsely crystalline. In texture and mineralogy, these show resemblances to aspects of clast- laden melt rocks, for example, clumps and groundmass (Figs. S5c and S5d). Mineral compositions for spherules are shown in Fig. 7, and Tables S4 and S5 in supporting information. For olivine-bearing spherules, olivine (Fo74–65) and pyroxene (En69Wo2-En63Wo5) compositions are shown in Fig. 7a, along with selected data for the similar material in NWA 7034 (Agee et al. 2013; Udry et al. 2014). The pyroxene in the latter is similar (En73Wo2- En49Wo9) but more zoned, and the olivine is more ferroan (Fo65–53; Udry et al. 2014) than that in 7533-2- 2. Pyroxenes from a small finely crystalline spherule, 7533-1-ENS (see Fig. 10c), are also shown: they are orthopyroxene En71.9Fs25.6Wo2.5 and augite En35.1Fs18.0 Wo46.9. As shown in Fig. 7a, the orthopyroxene in the small spherule is similar to that in the vitrophyric spherules in 7533-2 and -7, as is its plagioclase, but augite has also crystallized. The orthopyroxene is in all cases quite close in composition to that of ALH 84001. For the 3 mm extremely fine-grained spherule in section 2, we have only vein feldspar analyses, in part mixtures of plagioclase An41.8Ab53.2Or5.0Cn0.3 and hyalophane An4.5Ab20.4Or63.7Cn11.4. These are shown in Figs. 7b and 7c, along with plagioclase from four other spherules, none of which have Ba-rich feldspar. The vein feldspar in 7533-7-1 is mainly plagioclase similar in composition to all the other plagioclase, for example, the spherule with olivine dendrites (An33.8Ab62.6Or3.6) and the small 7533-1 ENS plagioclase An39.6Ab58.5Or1.9. The latter also contains chromite Sp19.1Chr56.8 Usp1.4Mgt22.4. Analyses of oxide minerals for spherules Fig. 7. Mineral compositions for olivine vitrophyre and and all other clast types are shown in Table S6 in feldspathic spherules. Feldspars mainly in veins for 2-1 and 7- supporting information. 1. a) Pyroxene quadrilateral with olivine below the baseline. b) Feldspars Ba- (Ca+Na)-K-Cn. c) Feldspars Ca-Na-K. Fine-Grained Igneous Clasts this material, but their occasional presence (Fig. S6a in NWA 7533 contains basaltic clasts with grains supporting information) suggests that the microbasalt is ~50 lmto~100 lm (Fig. 8) and subophitic to clast-poor impact melt, and both of them have similar granoblastic textures, which we have called microbasalts basaltic and Ni-Ir-rich bulk compositions (Humayun (Hewins et al. 2013a) and analyzed samples plot in the et al. 2013). The microbasalt is not observed as clasts basalt field of an SiO2ÀNa2O+K2O plot (Humayun within the clast-laden melt rock. The subophitic texture et al. 2013). Similar clasts in NWA 7034 were called is illustrated in Figs. 8a and 8b, where pyroxene grains igneous clasts by Santos et al. (2015a) who classified are molded around euhedral plagioclase laths. them as basaltic, andesitic, etc., based on their bulk Plagioclase laths are irregular in size and distribution, compositions. They contain pyroxenes, plagioclase, with trachytic patches where they are subparallel, in oxide minerals, chlorapatite, and in some cases K which case a pyroxene grain (pigeonite in Fig. 8b) may feldspar. They are coarser grained than the groundmass range from poikilitic to apparently interstitial. There are of the clast-laden melt rocks, and, according to Santos also clasts with metamorphic (granoblastic) textures et al. (2015a, 2015b) have grains typically 200 lmor with anhedral plagioclase and chains of pyroxene less in their longest dimension. Crystal clasts are rare in crystals as shown in Fig. 8c. Martian regolith breccia Northwest Africa 7533 99

Fig. 8. BSE images showing textures of microbasalts: (a) subophitic 7533-2-cent3; (b) poikilitic-intersertal 7533-4-L; (c) granoblastic 7533-1-obj2. d) Mg Ka image of layered sedimentary clast containing pyroxene (light gray) and feldspar (dark gray) with mudstone-size and clay-size layers.

The microbasalts, like the clast-laden melt rocks, may Plagioclase has a composition range of An55–23 in 15 contain magnesian orthopyroxene, but pigeonite is also clasts of microbasalt from 5 sections, with little variation identified with certainty (Table S7 in supporting between clasts or in %Or (Fig. 9b; Table S8 in information). The presence of orthopyroxene versus supporting information). A sixteenth clast, 7533-2-84, the pigeonite does not correspond to metamorphic versus one with zoned pigeonite and augite; however, it has a igneous texture, and the two phases have a similar range similar range of Ca/Na ratios (An50–24) but is enriched in of Fe/Mg ratios. All microbasalt pyroxenes are plotted in K, up to Or22 (Fig. 9b), probably due to small grains or Fig. 9a showing that their most magnesian compositions exsolution blebs of K feldspar in andesine. A seventeenth are close to those of ALH 84001, and of melt rock clast (7533-7-MH4; Fig. S6b) contains discrete K feldspar pyroxenes. Pyroxene compositions for individual clasts in (Or84; Fig. 9b); its pyroxene is also fractionated although six polished sections are shown and discussed in Fig. S7 it is orthopyroxene. Such clasts may have basaltic in supporting information. The rocks tend to have either andesite major-element compositions (Santos et al. orthopyroxene or pigeonite, with only one clast augite- 2015a, 2015b) and they have mineralogical similarities to rich. Orthopyroxene extends to more Mg-rich the medium-grained rocks with zoned pyroxenes compositions than pigeonite but otherwise the discussed below. Fe-rich oxide compositions are shown in composition ranges are similar. One clast, 7533-2-84, has Fig. 9c and Table S6. The spinel is Fe-rich, but contains both zoned pigeonite and zoned augite, like coarser significant Cr and Ti, with a composition range Chr1–13 grained rocks described below and shergottites. and Usp1–29 with an outlier at Usp63. Coexisting 100 R. H. Hewins et al.

Sedimentary Clasts

Fine-grained clasts in NWA 7533 with layers a few hundred microns in thickness are interpreted as sedimentary in origin. They have clastic textures unlike the granoblastic texture of the fine matrix, with the smallest pyroxene grains independent and not molded around anhedral plagioclase grains. Variations in grain size and mineral abundance (pyroxene versus feldspar) are responsible for this lamination or bedding. One such clast is observed in 7533-2 (Fig. 1b) immediately below and to the left of the 3 mm spherule. A more detailed image, also an Mg Ka map, is given in Fig. 8d, which shows layers of clay-size particles (<4 lm) probably transported in suspension alternating with layers of very fine siltstone- or mudstone-size particles (4–64 lm). The latter are very poorly sorted with both clay-size and larger particles, mainly crystal clasts, up to 100 lm. Indeed, the two flat igneous “pebbles” (2-1521 and 2-84) appear to be part of the sedimentary sequence, as shown in the BSE image Fig. S8 in supporting information. Both BSE and Mg Ka show that there are alternating pyroxene- and plagioclase-rich layers. Similar lithic clasts, with ~100 lm layers, have been observed in NWA 7475 (Wittmann et al. 2015; fig. 17c). The finest grains are similar in size to observed atmospheric dust grains (Lemmon et al. 2004) and we concur with Wittmann et al. (2015) that these may have been eolian dust deposits.

MEDIUM-GRAINED LITHIC CLASTS

A major distinction in clast types in the breccia is between fine-grained type described above and those with 1–2 mm grains. We have no barometry for these rocks but they have coarser exsolution features in pyroxene and alkali feldspar than the subvolcanic/ hypabyssal SNC meteorites. This suggests slow cooling and we have chosen to use plutonic rock names: they are mainly noritic to monzonitic fragments (Hewins et al. 2013a; Humayun et al. 2013). Plutonic rocks are normally classified by abundance of minerals but here typical crystals and crystal fragments are ~1 mm long Fig. 9. Mineral compositions of NWA 7533 microbasalts. a) and small clasts may be made up of only a few of the All data plus ALH 84001, to show that different phases possible phases. We have used a combination of phase (orthopyroxene, pigeonite, and augite) can be present. b) compositions and associations to make a classification, Feldspar compositions for microbasalts for 15 clasts, and for and use adjectives rather than names to indicate for the clast 7533-2-84, with pigeonite and augite, and K feldspar-bearing 7533-7-MH4 with orthopyroxene. c) Cr- uncertainty in modal abundances. For example, as we bearing spinel compositions for microbasalts. are not sure of the K feldspar abundance, some monzonitic clasts could be syenites or diorites. Ca-poor magnetite and ilmenite are present and, as for other pyroxene is the key mineral for noritic clasts and K ilmenite-rich clast types (zoned-pyroxene and feldspar for monzonitic clasts, and both may contain monzonitic), indicate fO2 2–3 log units above the FMQ augite and plagioclase. In the absence of the key buffer (Table S9 in supporting information). minerals, plagioclase composition (more calcic in noritic Martian regolith breccia Northwest Africa 7533 101 clasts, more sodic in monzonitic clasts) and spinel composition (Cr-bearing in noritic clasts and Ti-bearing in monzonitic clasts) can be used. Crystal clasts also give clues to lithologic types, and the presence of magnesian orthopyroxene suggests a source from orthopyroxenites (see below). However, in addition to plutonic material, there is a minority of clasts whose strongly zoned pyroxenes suggest a near-surface origin, and we discuss these first.

Clasts with Zoned Pyroxenes

Four clasts contain zoned pyroxenes, pigeonite and augite with no exsolution (Table S10 in supporting information), like those in shergottites and in microbasalt 7533-2-84, plus two feldspar phases and abundant chlorapatite (analyses of phosphates from all clasts types are given in Table S19 in supporting information). NWA 7533-4-K and 7533-1-12 shown in Figs. 10a and 10b, respectively, contain up to 1 mm pyroxene and 500 lm chlorapatite. Accessory phases include Ti-magnetite, ilmenite, a TiO2 phase, plus late hydrothermal pyrite, and terrestrial alteration in the form of Fe oxyhydroxides and veins with calcite. Clast 1–12 is similar to but has less augite than 4-K. The feldspars in the medium-grained clasts (Table S11 in supporting information) are very sodic plagioclase An35–18, and homogeneous orthoclase or sanidine Or69–88 (Fig. 11a). The plagioclase in microbasaltic clast 7533-2-84 is also shown: with An50–24 it is more calcic but overlapping the plagioclase in the clasts with zoned pyroxenes. Discrete orthoclase was not recognized in the microbasalt but spikes in the K content of “plagioclase” (Or22) indicate overlaps of the two phases. Oxide phases in the medium-grained clasts (Tables S6 and S9) are Ti magnetite notably poor in Cr, Sp3–0Chr0.5–0.1Usp23–12Mgt62–86, and ilmenite Fig. 10. BSE images of medium-grained clasts with zoned Ilm80–91Hem18–8. Spinel compositions are plotted in pyroxenes. a) NWA 7533-4-K K (clast 1 of Bellucci et al. Fig. 11b. 2015). b) NWA 7533-1-12. PI = pigeonite, AG = augite, The pyroxene compositions of these clasts are PF = plagioclase, AF = alkali feldspar, AP = chlorapatite, = = = shown all together in Fig. 11c, again with Mt magnetite, Il ilmenite, Pt pyrite. microbasaltic clast 7533-2-84 that has similar mineral compositions despite its texture, and separately in Fig. S9 Noritic-Monzonitic Clasts in supporting information. In clast 4-K, the pyroxenes are pigeonite En60–36Fs31–50Wo8–14 and augite En48–23 Clasts with orthopyroxene or inverted pigeonite, Fs21–39Wo28–38 (Figs. 11c and S7), like those of and plagioclase (An  50–30), plus augite, Cr-Ti- shergottites. The unusual microbasalt clast 7533-2-84, magnetite, and rarely alkali feldspar, are classed as with a range of pigeonite compositions, is seen in Fig. S9 noritic; those with alkali feldspar as well as to have pigeonite and augite compositions more plagioclase (An<  30) and augite, plus rare inverted magnesian on average than, but overlapping with, the pigeonite, abundant chlorapatite, Ti-bearing magnetite, compositions of pyroxenes in the four zoned-pyroxene ilmenite, and zircon, as monzonitic. Orthopyroxene clasts. Some noritic-monzonitic clasts containing both including inverted pigeonite is more abundant in the unzoned pigeonite and K feldspar may be related to the noritic clasts and augite in the monzonitic clasts, but zoned-pyroxene rocks. there is considerable overlap in pyroxene compositions 102 R. H. Hewins et al.

crystallization from impact melts (Humayun et al. 2013). Noritic clasts apparently representing rocks with hypidiomorphic granular textures (Turner and Verhoogen 1951) are illustrated in Figs. 12a, 12b, and S10 in supporting information. The Ca-poor pyroxene may be homogeneous orthopyroxene (Figs. 12b, S10a, and S10b), but inverted pigeonite (Fig. 12a) and rarely pigeonite also occurs (Fig. S10c). Fresher orthopyroxene is more ferroan, and dusty speckled orthopyroxene is more magnesian (OF and OM in Fig. S10a). Augite can be single phase (Fig. S10a), or exsolved (Fig. S10b). Most of the augite grains coexisting with orthopyroxene in other clasts equilibrated at temperatures (Lindsley and Andersen 1983) of 800–900 °C (Fig. S11 in supporting information). Chromite-magnetite is common, and accessories include zircon (Fig. S10a) and chlorapatite (in rare cases overgrown on merrillite or containing lacy monazite, as in Fig. S10c, Table S19). Monzonitic clasts contain either two homogeneous feldspar phases (Fig. 12d) or plagioclase plus intergrowths (perthitic or antiperthitic; Figs. 12c and 12e). The minority phase occurs as irregularly distributed ragged lamellae or flames. As perthitic feldspar is found only in monzonitic clasts, a crystal clast of perthitic feldspar is automatically monzonitic (Fig. 12c). Most clasts contain subhedral to anhedral augite, partly molded around feldspar grains, but there is also peculiar stringy augite interstitial or subophitic to alkali feldspar laths (Fig. 12e). Euhedral chlorapatite is abundant. Zircon, although present in some noritic clasts, is more abundant in the monzonitic ones (Figs. 12c–e), as is baddeleyite (Fig. 22 discussed below). Individual mineral analyses are given in Tables S12, S13, S6, and S9, and Fig. S12 in supporting information. Average mineral compositions for lithic clasts are shown in Fig. 13a (feldspars), 13b (pyroxenes), and 13c (spinels). The separation into two groups is seen well in the feldspar triangle (Fig. 13a), where there is one exception of K feldspar coexisting with calcic plagioclase. Another case of intermediate properties is shown where low calcium pyroxene coexists with sodic plagioclase. Oxide minerals, if present, also serve to distinguish the rock types. Spinel in noritic rocks is Cr-rich or Cr-bearing, whereas those in monzonitic rocks are Ti-bearing (Fig. 13c). Spinels are also more Al-rich in noritic clasts, Fig. 11. Mineral compositions for four zoned-pyroxene clasts up to Sp24. There is a small overlap of the two spinel and for 7533-2-84, a slightly fractionated microbasaltic clast; trends near the magnetite pole. (a) feldspars, (b) Ti-bearing spinels, (c) pigeonite and augite. The monzonitic fragments are enriched in K, REE, P, Ti, Cl, and Zr and contain zircon, and occasionally between the two types, with pyroxene in monzonite baddeleyite (Humayun et al. 2013) that have provided a only slightly more ferroan on average. Concentrations crystallization age for the noritic-monzonitic rocks, as of siderophile elements in these clasts indicate discussed below. Compositions of minerals associated Martian regolith breccia Northwest Africa 7533 103

Fig. 12. BSE images of noritic and monzonitic clasts: (a) noritic clast 7533-5-B with pigeonite and andesine, (b) norite 7533-1-5, (c) antiperthite clast with zircon, Z2, (d) monzonitic 7533-4-Z1, (e) monzonitic clast with zircons Z5,6. Opx, OF, OM = orthopyroxene; aug = A augite; P = plag; PF = plagioclase; KF = K feldspar; Cr = chromite; Mgt = magnetite; IL = ilmenite; AP = chlorapatite; PT = pyrite; Z = zircon.

with zircons are included in Tables S12, S13, and S6. bulk composition of An2.8Ab69.6Or28.6, with host Zircon Z1 (Fig. S10a) is almost completely included by An2.0Ab96.0Or2.0 and lamellae An1.7Ab14.2Or84.1. Similar a 300 lm clast rich in pyroxene that is dominantly a feldspars are found in monzonitic clasts in NWA7533 fractured ferroan orthopyroxene grain (En45.3Fs52.1 and their exsolution suggests formation at depth, Wo2.6). The clast also contains two small grains of although the patchiness of the coarse lamellae may plagioclase (An39.5Ab57.8Or2.6) and Cr-bearing magnetite indicate hydrothermal or deuteric modification, making Sp3.59Cr1.40Usp10.14Mgt79.54. This assemblage is them unsuitable for cooling rate determination. characteristic of the noritic clasts in NWA7533. Zircon Figure 12e shows a third monzonitic clast with two Z2 (Fig. S12c) is a skeletal, hopper-shaped crystal analyzed zircons Z5 and Z6, and a fourth baddelyite- (Humayun et al. 2013; Fig. S2) inside a shard of bearing monzonitic clast (discussed below in Fig. 22) antiperthite. Its shape suggests rapid growth, possibly yields age information from four phases (Humayun due to supercooling of the melt. The antiperthite has a et al. 2013; Bellucci et al. 2015). 104 R. H. Hewins et al.

orthoclase. Tie-lines for exsolved phases in perthite and antiperthite are also shown. Two unusual clasts are shown separately. One (7533-5 C) contains both Ca- and Na-rich plagioclase, suggesting alteration.

Fe-Ti-P-Rich Clasts (FTP)

The second uncommon clast type containing abundant ilmenite and chlorapatite blades, an Fe-, Ti-, and P-rich FTP (Agee et al. 2013; Santos et al. 2015a, 2015b; Nyquist et al. 2016), has only albite-rich feldspar (Fig. 14). However, the analyses of three grains of this anorthoclase feldspar extend between the most sodic plagioclase and bulk antiperthite compositions, resembling a segment of the continuous crystallization trend from plagioclase to alkali feldspar found in some alkali basalts (e.g., Keil et al. 1972). The FTP clasts in our sections are too small and too rare for detailed study.

Crystal Clasts and Rock Types

Crystal clasts are angular anhedral grains distinct from the igneous or metamorphic textured material containing them. The bigger clasts (up to ~2 mm) are dominantly crystal clasts (pyroxenes and feldspars, as well as magnetite-chromite and chlorapatite). Those analyzed are mainly >100 lm, so there are very few derived from the finer grained materials. The compositions of crystal clasts of orthopyroxene, pigeonite, and augite (Table S14) are shown in the pyroxene quadrilateral (Fig. 15). Their average FeO/ MnO ratio is 34.5 Æ 6. Note that most of the compositions are like those of the noritic-monzonitic rocks, but that the orthopyroxene extends to more magnesian compositions. Orthopyroxene with En80–74 has not been seen attached to plagioclase, although it may be associated with augite and/or chromite. We assume such orthopyroxene-dominated clasts represent orthopyroxenites, and treat them below with crystal clasts. ALH 84001 (Mittlefehldt 1994) has orthopyroxene compositions falling close to the range of the breccia pyroxenes not known to coexist with plagioclase. The orthopyroxene minor element concentrations of orthopyroxenite ALH 84001 fall in the center of the clouds for NWA 7533 crystal clast orthopyroxene, while its augite on the other hand is at the extremity of the breccia augite range (Fig. S13 in Fig. 13. Average mineral compositions for noritic and supporting information), probably because augite is a monzonitic lithic clasts: (a) feldspars; (b) pyroxenes, with ALH 84001 for reference; (c) spinels, both Cr-bearing and Ti- minor phase in ALH 84001 and thus contributions of bearing. augite from similar or more magnesian orthopyroxenites to the polymict breccia would be negligible. Individual single-phase feldspar analyses To investigate the question of clast provenance, and (Table S13) are shown in Fig. 14, along with the possibility of pristine clasts, for example, of representative tie-lines for coexisting plagioclase and orthopyroxenite, we identified the carriers of Ni in the Martian regolith breccia Northwest Africa 7533 105

Fig. 14. Coexisting feldspars in noritic and monzonitic clasts.

average compositions of 11 orthopyroxene crystal clasts (plus one augite and two pigeonites), plus pyroxenes in lithic clasts, are shown in Fig. 16c. Excluding two high values (possibly from noritic rocks), the crystal clasts, mostly magnesian orthopyroxene, have Ni concentrations (mean 21 ppm, SD 13) close to the detection limit and to the Ni content of orthopyroxene from ALH 84001, consistent with an origin in pristine crustal rocks. The pyroxenes in plagioclase-bearing lithic clasts have between 100 and 800 ppm Ni, mean 299, SD 184 (excluding 2 clasts with low values), like Fig. 15. Major-element compositions of pyroxene crystal those in the groundmass of clast-laden melt rock. clasts in the breccia. Inverted pigeonite is common as crystal clasts, but augite with exsolved Ca-poor pyroxene is less abundant breccia. The most Ni-rich phase is pyrite with up to 4.5 (Hewins et al. 2013a, 2013b; Humayun et al. 2013). wt% Ni (Lorand et al. 2015), but this occurs as an Lamellae are generally up to about 8 lmwide accessory mineral. Spinel (chromite-magnetite) is more (Figs. S14a and S14b in supporting information). abundant, and is found in the matrix and most clasts. Exceptionally they are ~50 lm wide (Fig. S14c). The more Fe-rich spinels contain ~0.3 wt% NiO Analyses of hosts and lamellae are given in Table S16 in (Hewins et al. 2013a, 2013b), and Cr-rich spinels have supporting information. Composition profiles across the lower (<1000 ppm) Ni contents (Fig. 16a), approaching inverted pigeonite and augite of Figs. S14a and S14b those for chromite in ALH 84001 (Gleason et al. 1997). are shown in Figs. 17a and 17b. These are top-hat Ni analyses for silicates are given in Table S15 in profiles, with no concentration gradients. Compositions supporting information. The Ni contents of silicate of coexisting phases are shown in the pyroxene standards (Jarosewich et al. 1980; Smith et al. 1983; quadrilateral and of augite in the Lindsley-Andersen Barrat et al. 1999; El Goresy et al. 1999; De Hoog et al. geothermometer plot (Figs. S15a and S15b in 2010) were easily measured at 20 KeV with error bars supporting information). Despite the equilibrium (SD of replicates) bracketing literature values for those suggested by the top-hat profiles, the estimated with the lowest concentrations, olivine in Eagle Station temperatures range from 700 to 900 °C, possibly due to and orthopyroxene in Tatahouine (Fig. 16b). The excavation of hot rocks. 106 R. H. Hewins et al.

Fig. 17. Exsolution lamellae in pyroxene crystal clasts, NWA 7533-1-9 and -10. a) Augite in inverted pigeonite. b) Orthopyroxene in augite. See also Fig. S2 and Table S16.

supporting information) and their compositions are plotted in Fig. S15c. It is striking that the sodic plagioclase (An<30) of the monzonitic clasts is not represented in the clasts analyzed. Perhaps there is a sampling bias because such compositions are inconspicuous relative to the brighter (higher Z) more calcic plagioclase and the two-tone perthites. However, the most calcic plagioclases of the noritic clasts are also missing, with the modes of the two distributions similar Fig. 16. EMP analyses of Ni-bearing phases. a) Wt% NiO (Ab – ), so this may just reflect the low number of versus mole% chromite in spinel in matrix and lithic clasts in 50 60 NWA 7533. Magnetite-rich spinels are Ni-rich, whereas analyses of clast feldspars. chromites approach the low Ni contents of chromite in ALH 84001 (Gleason et al. 1997). b) Analyses of Ni in ppm versus ALTERATION AND VEINS mole% Fo (or En) match well known olivine and orthopyroxene (Jarosewich et al. 1980; Smith et al. 1983; Pyroxene Alteration Barrat et al. 1999; El Goresy et al. 1999; De Hoog et al. 2010). c) Many orthopyroxenes have low Ni contents consistent with a source in pristine rocks like ALH 84001. In some pyroxene clasts, in many cases those Other pyroxenes, including orthopyroxene-augite pairs found showing exsolution, magnetite is observed as small in noritic and monzonitic rocks, have high Ni contents grains in conjunction with a silica phase. In some cases, consistent with crystallization from impact melt. magnetite is abundant, but silica cannot be located (e.g., Fig. S14c). This grain shows extremely coarse lamellae, Feldspar clasts (Table S17 in supporting indicating very slow cooling, and has somewhat kinked information) include plagioclase An54–31, anorthoclase, cleavage, fracturing, and very irregular exsolution K feldspar, plus perthite and antiperthite (Table S18 in lamellae, suggestive of annealing. Martian regolith breccia Northwest Africa 7533 107

In lithic clasts, a pyroxene phase may show a striking bimodal composition distribution indicating disequilibrium in the pyroxene assemblage, as illustrated in Fig. S16 in supporting information (Table S16). For example, the orthopyroxene crystals of clast 7533-4 Z1 in Fig. S10a are both high-Z (OF) and low-Z (OM), respectively OF (En45) and OM (En65) in the figure. Augite in this clast is speckled with Fe-rich material, and gives both stoichiometric and nonstoichiometric (near the large concordant zircon) analyses. Similarly 7533-1-obj D with a distribution of pigeonite compositions like those in zoned-pyroxene lithic clasts, shows wispy patches of magnesian pigeonite in ferroan pigeonite (not zoning). Lithic clast 7533-2-2229 contains augite, which is more magnesian than expected for the coexisting orthopyroxene. In 7533-4 clast Z4-Z12 augite En13Wo40 contains a patch of salitic augite En42Wo49 (Fig. S16). The magnesian pyroxenes are often patchy or vein-like and appear secondary, possibly as a result of oxidation causing the Fe-rich speckling. They are dark and tend to follow cleavage (Fig. S4). They show low Fe/Mn consistent with oxidation. Similar alteration of pyroxene is described by Santos et al. (2015a, fig. 17), and Lorand et al. (2015); and Wittmann et al. (2015, fig. 14a) describe a discordant vein of salitic augite En37Fs15Wo48 in orthopyroxene.

Pyrite Veins and Oxyhydroxide

None of the rock types sampled by NWA 7533 contains magmatic sulfides, except for very rare grains of pyrrhotite totally enclosed in plagioclase. Pyrite is present, however, and was fractured and subsequently altered to oxyhydroxide (Fig. 18a). Some crystals of pyrite that contain mineral inclusions must have grown when the breccia was incompletely lithified. Fine- grained matrix, clast-laden impact melt rocks, melt Fig. 18. BSE images showing (a) replacement of pyrite by terrestrial Fe-oxyhydroxide, which has spread from shock spherules, microbasalts, medium grained-lithic clasts, planar fractures, and (b) euhedral pyrite in vug in NWA 7533-3. and crystal clasts are all cut by fractures or veins containing euhedral pyrite (Lorand et al. 2015). Euhedral pyrite also occurs as a late mineral in vugs fO2 > FMQ+2, and near-neutral solutions (Lorand (Fig. 18b.) S-rich hydrothermal fluids were injected et al. 2015). Os-Ir-rich particles (200–700 nm) were during lithification of the breccia and after the breccia found in altered pyrite, converted from the parental became brittle, generating vein systems by hydraulic meteoritic phases into As-S rich phases by hydrothermal fracturing (Lorand et al. 2015). Pyrite veins or fractures sulfurization reactions (Lorand et al. 2015, 2016). are of late origin but predate an impact event that The fracture systems interpreted as a penultimate shocked pyrite in NWA 7533 (see below). event on Mars would also have allowed the passage of Pyrite displays mixed signatures, preserving fluids on Earth, which we argue below caused the refractory metals (PGE, Re) as rare micronuggets partial to complete replacement of pyrite by Fe mostly inherited from repeated meteorite bombardment oxyhydroxides. There are micro- and nano-scale Fe-rich while less resistant refractory metals (Mo and Re) were phases in the breccia, difficult to identify because of redistributed by late alteration (Lorand et al. 2015, grain size, although many have been identified by 2016). It contains up to 4.5% Ni that, in association different techniques. We identified maghemite by XRD, with magnetite, indicates a temperature up to 500 °C, as did Agee et al. (2013) for NWA 7034, which was 108 R. H. Hewins et al. confirmed by FTIR (Muttik et al. 2014b). Goethite was Shock Effects identified in NWA 7034 by Mossbauer€ spectroscopy (Gattacceca et al. 2014). Beck et al. (2015) identified Despite a probable major reheating at 1.35–1.4 Ga, ferrihydrite in NWA 7533 by IR spectroscopy. EMP and the later removal of the breccia from the surface of analyses of the pyrite pseudomorphs show that these Mars, NWA 7533 shows only weak shock effects (after have compositions similar to that of goethite, but the impact melting events). These are primarily intense containing ~4% SiO2 (Table S20 in supporting fracturing of the crystals, particularly of plagioclase. information). The silica is seen in TEM images as NWA 7533 lacks two of the most useful shock indicator nanolayers (Lorand et al. 2015). Electron diffraction phases, olivine, and quartz. Feldspars in some cases shows that the structure of the replacement phase is have a crackled appearance resembling perlitic compatible with hematite, but other iron-oxides or fracturing, and may be deficient in Na cations. oxyhydroxides cannot be excluded (Lorand et al. 2015). Maskelynite has not been observed. A TEM Muttik et al. (2014b) showed the banded structure and investigation of augite showed (100) mechanical twins identified the phase as maghemite from HRTEM, which have accommodated the plastic deformation, and although maghemite normally has low water contents. shear of the exsolution lamellae crossed by the twins, If the main phase is anhydrous oxide, the low EMP which can be activated at relatively low shock intensity totals must be explained by porosity and/or a hydrated (Leroux et al. 2016). Shock markers in pyroxene such as phase, possibly silica as a gel. NanoSIMS imaging of twinning on (001), mosaicism, planar deformation the Fe oxyhydroxides indicates a terrestrial dD value of features were not observed, although Wittmann et al. 10 Æ 85& with respect to SMOW, similar to values for (2015) reported undulatory extinction and mechanical NWA 7034 (Liu et al. 2014), requiring alteration by twinning in some orthopyroxene clasts in NWA 7475. terrestrial water, which explains gradients in pyrite The most strongly shocked clasts they observed had replacement from fusion crust to the interior of the reduced birefringence and shock melt veins. The meteorite (Lorand et al. 2015). Alteration of pyrite in generally low shock intensity in the breccias is in strong the desert would have yielded acidic pore fluid that contrast to that of SNC meteorites (e.g., El Goresy could have caused alteration along grain boundaries in et al. 2013). the fine-grained matrix, possibly producing the The most distinctive sign of shock is found in amorphous material filling them. pyrite. This fragile mineral shows planar elements (fig. 18a in Lorand et al. 2015) resembling planar Feldspar and Calcite Veins deformation features (PDF). Pyrite also shows fracture networks, and disruption into subgrains. Pyrite The large olivine-free melt spherules are cut by occurring in veins or fractures postdates all rock types veins (Figs. 6b and 6c and S5b) but they do not in the breccia, and other late phases appear to be continue into the material outside the spherules, unlike terrestrial. Pyrite appears to be the last phase to those cutting the pyroxene-plagioclase cluster-aureole crystallize in the breccia on Mars, so that the shock structures in clast-laden melt rocks. This alteration creating its planar fractures was a very late event, probably occurred before the consolidation of the possibly related to the excavation event (Lorand et al. breccia. The veins contain two feldspars, intermediate 2015). plagioclase like that in the spherules and Ba-rich K feldspar (hyalophane) in one case (0–14 mole% Cn, DISCUSSION Fig. 7) and Ba-poor K feldspar in another (0–3 mole% Cn), in addition to several pyrite crystals. Terrestrial Classification of the Breccia hyalophane is dominantly a hydrothermal mineral (McSwiggen et al. 1994). There are well-developed classification schemes for Calcite veins cut all components of the breccia, breccias from the Earth and the Moon (Stoffler€ et al. and in some cases calcite was deposited in sulfide- 1980; French 1998; and particularly Stoffler€ and Grieve filled fractures, where it is in contact with the 2007). NWA 7533 contains crystal and lithic clasts oxyhydroxide replacing pyrite (Lorand et al. 2015). derived from pyroxenitic to monzonitic source rocks, We show examples in clump-aureole structures and in making the breccia distinctly polymict. The dark melt an inverted pigeonite clast (Figs. S4 and S14c). bodies up to ~1 cm embedded in its fine matrix are its Calcite in NWA 7034 is inferred to have a terrestrial most prominent feature (Fig. 2) and have the distinctive oxygen isotopic composition (Agee et al. 2013) textures of clast-laden impact melt rocks (Hewins et al. although the fractures in the pyroxene clumps appear 2013a). Similar objects in NWA 7475 were described as to predate the fall to Earth. clast-rich amoeboid vitrophyres (Wittmann et al. 2015). Martian regolith breccia Northwest Africa 7533 109

Such rocks were recognized in relation to both lunar melt sheets can give rise to hydrothermal systems and terrestrial cratering shortly after the Apollo (Newsom 1980; Osinski et al. 2001, 2013). Impact program (Grieve et al. 1974; Simonds 1975; Stoffler€ and structures are associated with 55% of the occurrences of Grieve 1994) and are widespread in both lunar and hydrous minerals prevalent in the phyllosian alteration other meteorites (Warren et al. 2005; Hudgins et al. of the PreNoachian (Carter et al. 2013). Although 2007; Schmieder et al. 2016). Broadly speaking, NWA hydrous minerals are not abundant in the breccia 7533 is a polymict breccia containing melt particles, that (Muttik et al. 2014b), there is petrological evidence of is, compared to terrestrial rocks it is suevite-like: it hydrous precursors for spherules and melt rocks consists of fine matrix with melt rock bodies in it (Humayun et al. 2014a, 2014b, 2014c) and oxygen (Hewins et al. 2013b). However, it contains both clast- isotope evidence of melt-fluid-atmosphere interaction in rich and clast-free impact melt rocks, with the latter the zircons of noritic and monzonitic melt rocks sometimes enclosed in the former (e.g., Fig. 4d), as in (Nemchin et al. 2014). The high D17O of the zircons “cored bombs.” Thus, the monzonitic and noritic indicates mass-independent fractionation in the xenoliths are an older generation, indicating at least two atmosphere, and due to weathering or hydrothermal impact melting events, the former at ~4.43 Ga. This alteration, this signature was incorporated into the evidence is consistent with an origin for NWA 7533 as a protoliths of the melt rocks which crystallized the regolith breccia (Stoffler€ and Grieve 2007). zircons at 4.43 Ga. In regolith breccias, surface fines are important and Mars regolith breccias should differ from lunar and the matrix in NWA 7533 (Fig. 3) is finely crystalline, meteorite regolith breccias, because of a lack of containing submicron grains (Hewins et al. 2013a; gardening by micrometeorite impacts (Stoffler€ and Muttik et al. 2014a; Leroux et al. 2015). Although dust Grieve 2007), and of implanted solar wind (Bogard grains could be part of impact ejecta (Wittmann et al. et al. 1973; Wieler 1998). Indeed, the absence of 2015), Humayun et al. (2013) also proposed on agglutinates in NWA 7533 is explained by the filtering geochemical grounds that this breccia is derived in part effect of the Martian atmosphere on small projectiles. from wind-blown dust. The presence of vesicular shards Similarly, the Martian surface is shielded from the solar (Fig. 6a) is also consistent with a surficial origin. wind by the Martian atmosphere, so that the absence of Several kinds of melt spherules up to 6 mm in apparent solar wind and the presence of Martian atmospheric diameter are present (Figs. 2 and 6), and they are gases in this regolith breccia (Cartwright et al. 2014) is typical of a surface origin as in regolith breccias clear evidence of a Martian origin. There is evidence of (Stoffler€ et al. 1980; French 1998), but can be formed multiple events, spherule deposition, and the by volcanic fire fountains or deposited as ejecta from accumulation of a meteoritic component on the surface, large craters (French 1998). Siderophile element as for the Moon, but also interaction with the concentrations in NWA 7533 components require ~5% hydrosphere and atmosphere (Humayun et al. 2013; chondritic component and confirm an impact origin for Nemchin et al. 2014). Assuming NWA 7533 is the melt rocks, including the older monzonitic-noritic representative of the southern hemisphere, it represents clasts, and for at least some of the spherules (Humayun a new kind of regolith breccia unique to early terrestrial et al. 2013; Udry et al. 2014). The similarity to Gusev planetary surfaces, providing clues to the nature of soils (Humayun et al. 2013) is also consistent with an Hadean surfaces on Earth. origin as a regolith breccia. Rather than a simple regolith breccia, however, NWA 7034, 7475, and 7533 Formation of Clast-Laden Melt Rocks may be a suevitic breccia formed by recycling of regolith and bedrock during a large event depositing the The NWA 7533 breccia was not formed clast-laden melt rock splashes (Wittmann et al. 2015). instantaneously: it is an accumulation of melt rock Despite previous lack of evidence of impact melts in bodies and matrix, somewhat younger than the 4.43 Ga SNC meteorites, except as shock veins, inevitably they monzonitic clasts they both contain. Abundances of Ir must exist on Mars, given the scale of cratering. Impact and other siderophile elements indicate the presence of melt flows can be recognized at very young craters, like ~5% projectile component in most clasts, as in lunar Pangboche on Olympus Mons (Mouginis-Mark 2015). breccias (Humayun et al. 2013). This contamination Many Mars surface features formed since the late includes the monzonitic and microbasaltic fragments Hesperian, such as dark dunes, dark layers in crater with normal igneous textures, except clasts from pristine floor deposits, and dark streaks, can be interpreted in orthopyroxenite (Hewins et al. 2014). The pervasive terms of impact melt breccias and glassy impact debris siderophile enrichment even in the melt rocks with (Schultz and Mustard 2004). For large complex craters, pyroxene exsolution apparently formed at depth the long-lived heat particularly associated with impact suggests a thick sequence of impact-generated rocks 110 R. H. Hewins et al. including mature regolith. Only the fine-grained clast- (Grieve et al. 1974; Simonds 1975; Stoffler€ and Grieve laden rock bodies have a texture characteristic of 1994; Warren et al. 2005; Hudgins et al. 2007; Osinski impact melting and some of them have sculpted forms et al. 2008), with textures characteristic of melts as in suevitic breccias (Hewins et al. 2013a; Humayun formed by impact, their oval somewhat spherulite-like et al. 2013). The presence of schlieren in impact glasses clump-aureole structures (Figs. 4–5 and S2) are shows that melts formed in different locations can be unique. Other phases in the orthopyroxene clusters intimately mixed during excavation (See et al. 1998), may include chromite (early-forming), fine Fe-oxide and heterogeneity in the groundmass of the melt rocks dust, and (late-forming) pyrite. Even in low (Figs. S3 and S4) is consistent with this process. The magnification, BSE images (Figs. S3f and S4a; Santos presence of aligned ellipsoidal structures in a breccia et al. 2015a, fig. 8) these melt rocks may be recognized by clast in NWA 7034 section 3A,3 (Santos et al. 2015a; the presence of the dark plagioclase aureoles. Fractures Figs. 8a and 8b), similar to the clump-aureole structures partially filled with calcite (Figs. 4 and S4c) emanate described above in clast-laden melt rocks, but deformed, from and cross in the clumps, and are at their thickest in suggests flow in melt rock before the end of the centers of the clumps suggesting pull apart by solidification. shrinkage as in septarian nodules. The groundmass to the Several origins have been discussed for clast-laden clump-aureole structures contains plagioclase and more melt rocks. For coarser grained impact melt rocks, augite than orthopyroxene. The dominance of plagioclase superheated melt mixes with fractured target rock at the laths in the texture of the melt rocks and local subophitic contact between the melt and the crater footwall. Melt texture suggests that plagioclase is the equilibrium jetted from the crater overtakes colder fragmental liquidus phase. We considered several possibilities for the debris, the incorporation of which partially quenches it origin of orthopyroxene clumps. (Grieve et al. 1974; Simonds 1975). Such rocks were Inversion of clasts of a high-pressure phase such as recognized in relation to both lunar and terrestrial majorite to pyroxene would explain its concentration, cratering shortly after the Apollo program (Grieve et al. but in this case textural evidence of expansion rather 1974; Simonds 1975; Stoffler€ and Grieve 1994) and are than contraction should be present. widespread in both lunar and other meteorites (Warren Thin reaction coronas of pyroxene grains may be et al. 2005; Hudgins et al. 2007; Schmieder et al. 2016). observed around olivine and quartz clasts in melt rock Recent detailed investigations of terrestrial craters and they may also contain plagioclase laths (Hewins (Grieve et al. 2010; Stoffler€ et al. 2013) have indicated 1984) or glass depleted in Fe and Mg (Grieve 1975). that suevites are not necessarily fallback or ejecta from The clumps in NWA 7533 are unlikely to represent the initial plume but can be generated coronas from totally transformed relicts, as olivine is phreomagmatically when hydrated rocks and/or water rare and quartz has not been observed. However, the fall into impact melt pools. Vaporization at depth can plagioclase aureoles in NWA 7533 probably formed in cause explosive fragmentation of melt and mixing of similar boundary layer liquids depleted by pyroxene melt splashes with solid debris. The presence of shards crystallization. with concave surfaces (Fig. 6a) indicates vesiculation in Glass spherules are often encountered in proximal the melt rocks. There are ample indications of the action and distal impact ejecta deposits (French 1998; of water on the precursors to the melt materials Wittmann et al. 2013b). Devitrified glass spherules are incorporated into NWA 7533. Thus, these early impacts often surrounded by inward-growing fibro-radial were probably on a wet Pre-Noachian terrain with feldspar (Simonson 2003). However, the clumps in phyllosian weathering effects. Formation of clast-laden NWA 7533 melt rocks are of different composition and melt rocks may therefore be due to late contamination the feldspar forms laths, not fibers, which like the and phreatic eruptions of an impact melt pool. The large pyroxene extend outwards into the melt groundmass volume of Onaping breccias above the Sudbury Igneous (Figs. 4b and 4c). Aggregates of pyroxene like those in Complex suggests that seawater reacted with the impact melt rock clumps occur in shards in the breccia, often melt in the crater (Grieve et al. 2010). Evidence of wide with concave surfaces (Fig. 6a), consistent with spread occurrence of breccias like NWA 7533 in the vesiculated melts. The textural similarity suggests a Noachian would have interesting implications for water similar crystallization process in both vesicle-rich and abundance, if the suevite analogy is correct. vesicle-free melt rocks. In some cases, a large chromite grain in the center Clump-Aureole Structures in Clast-Laden Melt Rocks of the clump constitutes a possible nucleation site for orthopyroxene in a highly supercooled liquid, if Although the clast-laden melt rocks are similar in nucleation of plagioclase were inhibited. However, in many ways to those known on other planetary bodies many other cases chromite is absent from the plane Martian regolith breccia Northwest Africa 7533 111 exposed. Nucleation on unknown sites similar to the diagram of silica vs. total alkalis (Fig. 19a). The fine- formation of spherulites remains a possibility. The grained materials plot in the basaltic field, where many similarity to textures in irregular fragments, concave Gusev soils fall and where Agee et al. (2013) showed fragments, and the pyroxene-rich vesicular shards that the bulk composition of NWA 7034 (measured on supports this idea. Schlieren in impact glasses show fine-grained groundmass) plots. mixing of melts formed in different locations (See et al. There is partial overlap in composition between our 1998). The heterogeneity in the groundmass (Fig. 5) and fine-grained samples, but variation in Al/Si shows that the similarity in pyroxene composition to that in ALH microbasalts are closer to SNC meteorites than matrix 84001 suggest the presence of melted orthopyroxenite. and clast-laden melt rocks (fig. S4 in Humayun et al. Rapid crystallization of these magnesian schlieren, 2013). This is probably due to mixing with feldspar supercooled by mixing with cold clasts, would be clasts from the medium-grained noritic-monzonitic natural before the growth of plagioclase in the more rocks, or similar rocks. The REE data (Fig. 19b) aluminous groundmass. This chemical heterogeneity emphasize the overlap between our generally clast-free suggests assimilation by the impact melt with the microbasalt and the clast-laden melt rocks, with the clumps forming on unresorbed orthopyroxene relicts, difference between the two being due to the absence or although overgrowths rather than clusters of grains presence (surviving or digested) of clasts. The would be expected. monzonitic rocks are distinctly higher in alkali Some small aggregates of micron-sized pyroxene are (Fig. 19a) and REE (Fig. 19b) contents. In particular, observed in breccia matrix (Fig. S2d) and some are one monzonitic sample (clast II) falls in the middle of traversed by cracks. On early Mars, alteration produced the mugearite (basaltic trachy-andesite) field, and has a phyllosilicates including Fe-Mg clays (Bibring et al. composition resembling that of sample Jake M. at Gale 2006) and the phreomagmatic model of suevitic impact Crater (Stolper et al. 2013). Such rocks appear to be melt rocks (Grieve et al. 2010; Stoffler€ et al. 2013) absent at Gusev. involves addition of water or hydrated material into Though the Curiosity rover is especially concerned melts. Dehydration of altered mafic dust pellets might with sedimentary rocks at Gale Crater, many rocks have led to shrinkage on heating before or after analyzed by ChemCam are highly feldspathic like lithic incorporation in the melt, and subsequent deposition of clasts in NWA 7533. Of course the rover has calcite in the fractures. Dust balls of pyroxene or encountered a more diverse suite of rocks, including altered pyroxene might survive better in the impact melt olivine-bearing and quartz(?)-bearing rocks not seen in if dehydration occurred before incorporation. Thus, the the breccia, as well as the sedimentary rocks. Here, we clump-aureole structures may have formed by examine the similarities between our monzonitic and spherulitic growth on remnants of pyroxene or hydrated noritic clasts and anhydrous samples analyzed by dust pellets. ChemCam with H lacking in the laser-induced breakdown spectra (LIBS). While individual grains can Clast Bulk Compositions and Mars Surface be imaged and identified by physical properties, laser Measurements data are generally of overlapping grains. The mixing lines in ratio diagrams can be used to extrapolate We have described clast-laden melt rocks, endmember single mineral compositions (Sautter et al. microbasalts generally free of clasts, and medium- 2014, 2016). In Fig. 20a, we compare monzonitic clast grained noritic-monzonitic clasts. The siderophile II with Black Trout sample from Gale Crater (Sautter element concentrations of all three indicate et al. 2014). Ideal augite, orthopyroxene, and olivine incorporation of a projectile component (Humayun plot at 0.5, 1.0, and 2.0 on the (Fe+Mg)/Si axis; et al. 2013). The high Ni in the pyroxenes of the noritic- anorthite CaAl2Si2O8 plots at 1.0 and both albite monzonitic clasts of NWA 7533 is consistent with NaAlSi3O8 and orthoclase KAlSi3O8 plot at 0.33 on the crystallization from impact melts, as proposed by Al/Si axis. Our electron probe analyses of pyroxenes Humayun et al. (2013). Ni from projectiles is not and feldspars plot on or near the axes, and can be fractionated out of the silicate, due to high fO2 and low identified by stoichiometry. The laser spot analyses, by fS2 in the melt, indicated by the presence of magnetite ICP-MS (Humayun et al. 2013) and LIBS/ChemCam and absence of magmatic sulfides (Lorand et al. 2015). (Sautter et al. 2014) are generally mixtures of phases. Santos et al. (2015a) have shown that the bulk For NWA 7533 clast II we also have a laser bulk compositions of “microbasaltic” clasts in NWA 7034 analysis performed by rastering. The LIBS data for are mostly basaltic but extend to mugearitic. Major- Black Trout clearly require the presence of two element ICP-MS data were acquired on NWA 7533 feldspars (similar to the intermediate-sodic plagioclase clasts (Humayun et al. 2013) (Table S1) and we show a and alkali feldspar of the breccia) and also suggest the 112 R. H. Hewins et al.

Fig. 20. Comparison of chemical data for rocks at Gale – Fig. 19. a) SiO2 alkalis diagram to show compositions of Crater (Sautter et al. 2014) with breccia noritic and clast-laden melt rock, microbasalt, and monzonitic clasts in monzonitic clasts. NWA 7533. b) REE concentrations of these clasts. Data from Humayun et al. (2013).

229, the pigeonite having more Ni (648 ppm, SD presence of a silica-rich phase such as quartz. In Fig. 20b, 5 ppm) than the late subcalcic augite. Based on these we compare analyses of noritic clasts in the breccia with data, the zoned-pyroxene rocks are likely to have sample Mara (Sautter et al. 2014) from Gale Crater. incorporated projectile material, and to represent impact Here, laser spots can be explained as mixtures of calcic melts, like the other lithic clast types analyzed by LA- plagioclase, augite, and orthopyroxene, possibly with a ICPMS (Humayun et al. 2013). Bellucci et al. (2015) higher augite/orthopyroxene ratio in Mara than in the found a U–Pb age for chlorapatite in clast 7533-4-K breccia norite. Matrix, melt rock, and microbasalt in (their clast 1) of 1.357 Æ 81 Ga. They also used Pb NWA 7533, not shown here, are similarly noritic, but isotope measurements to show a model age near there the data points are pulled across the plagioclase- 4.43 Ga for one plagioclase and near 1.35 Ga for most orthopyroxene tie-line by the presence of Fe-rich phases, feldspar in the same zoned-pyroxene lithic clast. Ody especially chromite and magnetite. et al. (2015) report four locations with spectral The zoned-pyroxene clasts resemble enriched signatures (strongly influenced by pyroxenes) matching shergottites mineralogically. Although the clasts have basaltic shergottites in Sirenum Planum. Because of the the same composition ranges for pigeonite and augite as Noachian age of this terrain, another possible Shergotty, they differ from shergottites (even Los interpretation of these spectra would be impact melt Angeles) in having abundant K feldspar. The nakhlites rocks rather than basaltic shergottites. and chassignites are even more distinct in mineralogy. Along with the 4.43 Ga noritic-monzonitic impact In addition the pyroxene in 7533-1-12 was analyzed for melt clasts, NWA 7533 contains orthopyroxenes with Ni (Table S15), and is Ni-rich with 486 ppm Ni, SD low Ni contents. Many of these orthopyroxene clasts Martian regolith breccia Northwest Africa 7533 113 are Mg-rich with compositions not associated with rocks, were analyzed by SHRIMP. Two dated zircons plagioclase in lithic clasts. This suggests a pristine Z2 and Z1 are shown in their respective noritic and orthopyroxenite crustal component, which resembles monzonitic host in Figs. 12c and S10a, and described ALH 84001 in major and minor element abundances by Humayun et al. (2013, SOM). The morphology and (Figs. 15 and S13). Orthopyroxene in NWA 7033 also oscillatory zoning of many such zircons (e.g., Z14, a has d17O and d18O very close to those of ALH 84001 matrix grain with a concordant age, Fig. S10d) suggest (Franchi et al. 1999; Ziegler et al. 2013). Pyroxene clasts that such grains have crystallized during solidification of are not yet individually dated, although pyroxene these rocks. separates in NWA 7034 have a crystallization age close The crystallization age of the zircon in noritic- to that of the zircon (Nyquist et al. 2016). ALH 84001 monzonitic clasts in NWA 7533 is 4.428 Æ 0.025 Ga has an age of 4.1 Ga (Bouvier et al. 2009; Lapen et al. (Humayun et al. 2013) and zircons in NWA 7034 have 2010), both indicating early Noachian terrain and similar ages (Tartese et al. 2014; Yin et al. 2014). suggesting an origin in the southern hemisphere. Ody Similar crystallization ages were inferred for pyroxene et al. (2015) have examined orthopyroxene-rich terrains and plagioclase based on Sm-Nd dating for NWA and found that the IR spectra indicate the presence of 7034 (Nyquist et al. 2016) and a common-Pb model plagioclase, that is, noritic rocks, or an association of age for plagioclase (Bellucci et al. 2015). For NWA orthopyroxenite and norite, as in the breccia. Based on 7533, the hosts to zircon and baddeleyite are matrix the clasts in NWA 7533, the noritic rocks might equally and noritic-monzonitic melt rocks. The least radiogenic well be melt rocks or breccia as pristine norite. Pb isotopic compositions measured in plagioclase and K feldspar lie within error of the 4.428 Ga Geochron, The History of the Breccia and hosts include monzonitic melt rock and medium- grained zoned-pyroxene melt rock. We conclude that To understand the earliest history of Mars we must the crystallization of medium-grained rocks was at find the relative ages of major processes, such as the ~4.4 Ga. crystallization of earliest magmatic rocks and the transition from primary to secondary atmospheres. The Relative Ages of Melt Rocks timing of crust formation, the extent of the early heavy The history of the breccia as well as it can currently bombardment, and the onset of aqueous alteration have be deciphered is outlined in Fig. 21. While an early age been uncertain (Carr and Head 2010; Fassett and Head of crystallization is firmly established for the oldest 2011). There are over 40 impact basins with diameters components at ~4.35–4.43 Ga, the timing of assembly, >400 km in the Noachian (Hartmann 2005). Craters and lithification or compaction age of the breccia is not >150 km in diameter have been dated by crater well understood, at present. Lithification may be related counting and about 73% of them formed in the Early to heating by initially hot components, compaction by Noachian, that is, as a result of the early heavy impacts, and cementation by circulating fluids. We bombardment completing accretion (Tanaka et al. consider two possible models here. 2014). Most of the formative events in Mars’ history (1) The lithification age is ~4.4 Ga. One could occurred before the limit where crater counting becomes expect that the impact record of Mars should include a a viable dating tool, and has been consigned to the broad range of ages between 4.4–3.8 Ga, related to the PreNoachian, that is, buried under the crater-saturated late, heavy bombardment, as observed for the lunar surface (Frey et al. 2003; Carr and Head 2010; Fassett breccias (Grange et al. 2013; Joy et al. 2014). The and Head 2011; Tanaka and Hartmann 2012). Clearly number of zircon grains dated is still relatively small, but impact breccias must be important in this story and the no upper intercept age younger than 4.35 Ga has been dating of NWA 7533 and NWA 7034 (Humayun et al. observed (Humayun et al. 2013; Tartese et al. 2014; Yin 2013; Nyquist et al. 2016) give us a unique glimpse of et al. 2014). One interpretation of this absence of post- geological processes in this earliest time period. 4.35 Ga zircons is that the breccia was compacted at ~4.3 Ga, and that all subsequent disturbances recorded Crystallization Absolute Ages by zircons reflect later impacts or other geological events The monzonitic fragments are enriched in K, REE, that affected the assembled breccia. This model implies P, Ti, Cl, and Zr; and contain zircon, and occasionally rapid assembly and closure within the Pre-Noachian baddeleyite, which have provided a crystallization age period. Certain aspects of this model are unsatisfactory, for the noritic-monzonitic rocks (Humayun et al. 2013). including that the abundant Fe-oxides are expected to Zircon and baddeleyite in one noritic and four form in the “siderikian” (Bibring et al. 2006), a time monzonitic fragments, and others occurring as clasts or (<3.5 Ga) overlapping the Amazonian period (<3.0 Ga) xenocrysts in fine-grained matrix and clast-laden melt of Martian history. However, the absence of excesses of 114 R. H. Hewins et al.

(2) The lithification age is ~1.4 Ga. Yin et al. (2014) argued that some of the zircon analyses at the lower intercept were actually concordant grains that crystallized from melts at ~1.4 Ga. This led them to interpret the lithification age of the breccia as ~1.4 Ga. This model of the assembly time cannot explain the absence of zircons between 4.35–1.4 Ga, but is more permissive to sampling a very wide range of Martian history (~3 Ga), even if not recorded by zircon-forming events. Resolving these two models is essential to fully reading the record of ancient Mars from this breccia. The ages of components such as clast-laden melt rocks, microbasaltic melt rocks, and spherules are not strongly constrained like those of the monzonitic melt rocks. However, the zoned-pyroxene rocks, microbasalts, and clast-laden melt rocks differ from each other and noritic-monzonitic rocks mainly in clast abundance, grain size, and extent of fractionation. The microbasalts in part resemble zoned-pyroxene clasts in mineral compositions (Fig. 11) and the clast-laden rocks in containing clasts in a few cases (Fig. S6a). Furthermore, these components appear to have the same zircon and plagioclase crystallization ages as the monzonitic clasts (Nemchin et al. 2014; Yin et al. 2014; Bellucci et al. 2015; Santos et al. 2015a). Specifically, a clast in NWA 7034 with basaltic trachyandesite composition (Santos et al. 2015a), which forms part of the microbasaltic group of clasts, contains concordant and discordant zircons (4333 Æ 38 Ma and 1434 Æ 65 Ma; Yin et al. 2014); zoned-pyroxene clast 7533-4-K has a plagioclase model age near 4.5 Ga with apatite and K feldspar reset (Bellucci et al. 2015). However, the fine-grained melt rocks must have formed in a later impact event than the source rocks of the noritic and monzonitic melt rock clasts they contain (Fig. 4b), and the orthopyroxene clasts they contain, although not necessarily much later. Wittmann et al. (2015) suggested that the melt rock (“amoeboid clast- rich vitrophyres”), spherules, and pulverized target rock were deposited in a single event, possibly at ~1.4 Ga. The zircon and feldspar age data cited above suggest that this event is followed closely after the crystallization of the medium-grained melt rocks. The occurrence of both igneous and annealed textures in microbasaltic clasts shows that some of them were Fig. 21. Schematic development of the breccia, showing two metamorphosed before incorporation in the breccia. impact events slightly separated in time. 1) Regolith (reg) on This suggests burial by hot debris and close formation primitive crust with felsic rocks (FLC) and orthopyroxenites in time in a series of impact deposits, the last of which (pxt). 2) Noritic-monzonitic impact melt rocks (NO-MO). 3) formed the breccia. The new Sm-Nd dating of bulk Microbasaltic and clast-laden melt rocks (MR). breccia chips and mineral separates (Nyquist et al. 2016) yields an age of 4.42 Æ 07 Ga, indistinguishable Cl, S, and Zn in the breccia (Humayun et al. 2013) may from the zircon crystallization ages and showing that all be consistent with a pre-theiikian age (>4.0 Ga) of the rock types making up the breccia formed within a lithification or at least alteration. very short lapse of time. Martian regolith breccia Northwest Africa 7533 115

Late Events Despite the low level of shock in the breccia, there is evidence of late disturbances of the crystallization ages of the melt rocks. Imaging of some zircon grains (Z1 and Z2) reveals variable later degradation of internal structures as a result of damage induced by radioactive decay, consistent with discordant ages and annealing at ~1.7–1.4 Ga (Humayun et al. 2013). Some zircon U-Pb data for monzonitic clasts fall on a discordia line with a lower intercept of 1.712 Æ 85 Myr indicating a major subsequent resetting, presumably by a later impact (Humayun et al. 2013). Re-examination of the lower intercept for NWA 7533 zircons shows that of the four zircons that plot near this intercept, the Pb- Pb ages of at least two of these zircon analyses (Z2, Fig. 21; Z3) yield lower intercepts of ~1.4–1.5 Ga. For NWA 7034, Yin et al. (2014) and Tartese et al. (2014) also report zircon resetting at ~1.4–1.5 Ga. The U-Pb isotopic compositions measured for the chlorapatite in NWA 7533 (Bellucci et al. 2015) yield a 1.357 Æ 81 Ga age and in NWA 7034 (Yin et al. 2014) a 1.345 Æ 47 Ga age. Feldspar Pb-Pb dating by Bellucci et al. (2015) also shows that some Pb-Pb model ages for Fig. 22. Monzonitic clast 7533-4-Z4 (BSE) showing two distinct plagioclase in lithic clasts are within error of 4.4 Ga, U-Pb and Pb-Pb ages recorded in four phases, baddeleyite (B), but K feldspar and plagioclase also have reset ages close plagioclase (P), K feldspar (K), and chlorapatite (C). Data from to the 1.35 Ga Geochron. Humayun et al. (2013) and (as “clast 2”) from Bellucci et al. Some monzonitic clasts and one zoned-pyroxene (2015). melt rock clast (NWA 7533-4 K) show both preserved crystallization ages and reset ages in different phases, their protoliths (Lorand et al. 2015). The Martian origin demonstrated by data in figs. 3 and 4 of Bellucci et al. of the pyrite is demonstrated by its fracturing caused by (2015). As an example of this, in Fig. 22 we show a shock (Lorand et al. 2015), possibly during the launch monzonitic clast with baddeleyite ≥4.298 Æ 0.020 Ga, event. Wittmann et al. (2015) using Fe and Th plagioclase ~4.4 Ga, chlorapatite 1.357 Æ 0.081, and K abundances, and the ~5 Ma exposure age of Cartwright feldspar ~1.35 Ga ages (Humayun et al. 2013; Bellucci et al. (2014), suggest one possible launch site for the et al. 2015). This clast 7533-4-Z4 is referred to as clast 2 breccias, the young rayed crater Gratteri. This crater in Bellucci et al. (2015). Bellucci et al. (2015) proposed has also been suggested as the source of ALH 84001 that this strong disturbance indicated by the isotope (Tornabene et al. 2006). NanoSIMS imaging of pyrite data was a thermal event related to impact in the grains shows that Fe oxyhydroxides replacing them Amazonian toward 1.35. The partially sintered sub-lm along their fractures exhibit a homogeneous terrestrial granoblastic texture (Muttik et al. 2014a; Leroux et al. D/H ratio (Lorand et al. 2015). The average dD value 2015) is consistent with a moderate annealing intensity, (10 Æ 85&) indistinguishable from the terrestrial values, corresponding to brief heating possibly during the and calcite veins with terrestrial oxygen isotopic 1.35 Ga event. The role of other disturbances in the compositions (Agee et al. 2013), record the breccia’s isotopic record, and the ages of specific clast types, may relatively brief residence (typically thousands of years need to be clarified, but the 1.35 Ga event probably for desert meteorites) on Earth. represents the lithification of the breccia. The latest breccia event before excavation, during Implications for Conditions on Mars and after annealing of the matrix, is hydrothermal activity depositing pyrite in vugs in matrix and in Radiometric dating has yielded crystallization ages fractures cutting all facies of the breccia (Lorand et al. of 4.43–4.35 Ga for all rock types in the breccia 2015). Pyrite precipitation in these oxidized rocks (Humayun et al. 2013; Nyquist et al. 2016). Thus, a required T up to 400–500 °C and near-neutral rather crustal reservoir containing debris from pristine igneous than acidic solutions, possible because of the absence of rocks heavily contaminated by projectiles was in place magmatic sulfides in the breccia components and/or on Mars at 4.43 Ga, a little over 100 Ma after its 116 R. H. Hewins et al. accretion. Debaille et al. (2007) showed that combined (Nemchin et al. 2014). The zircons probably crystallized 142Nd-143Nd isotope evidence in shergottites implies from melts that had incorporated surface material with mixing between two distinct source reservoirs formed at the atmosphere-hydrosphere isotopic signature, 4.535 Gyr and 4.457 Gyr ago and argued that the although are also partly altered (the most extreme D17O younger reservoir was residual melt from a magma values being found in metamict portions of the zircons ocean, rather than ancient crust. However, the with an age of ~1.4 Ga). The excess of norite- solidification of any magma ocean is thought to have monzonite zircon D17O values over SNC values occurred within 5 to 10 Ma after Martian accretion (Nemchin et al. 2014) implies that the atmosphere was (Elkins-Tanton 2012), a much shorter time frame than thin enough to create reactive 17O species before impact the 30–110 Ma required by Debaille et al. (2007). The melting and zircon crystallization, and hydrated crust early crust rather than a late magma ocean reservoir is was already present at 4.43 Ga (Humayun et al. 2013; available to contaminate and enrich shergottite magmas Nemchin et al. 2014). emplaced through it. The Martian Nd isotope systematics (Debaille et al. 2007) can be explained by Early Aqueous Alteration formation of this crust and depleted mantle, without the All crystallization ages for clasts in NWA 7533 date need for a magma ocean (Agee et al. 2013; Humayun from the earliest Noachian (PreNoachian) characterized et al. 2013). In that case, H2OandCO2 would have by phyllosilicate alteration in remote sensing. Newsom been outgassed continuously during early crust (1980) argued for the production of clays as well as formation to form the (Pre-) Noachian atmosphere, ferric hydroxides by hydrothermal alteration associated allowing clement surface conditions (Nemchin et al. with Martian impact melt sheets. 2014). In this period of the most intense basin Geochemical evidence that the precursors of the formation, hydrothermal activity would have been monzonites and some spherules were altered and clay- important, and could have promoted an extensive rich (Humayun et al. 2013, 2014a, 2014b, 2014c) is aquifer even if surface material was frozen (Pope et al. consistent with the zircon record of melt-hydrosphere- 2006; Abramov and Mojzsis 2014). atmosphere interaction (Nemchin et al. 2014). Humayun et al. (2014c) have shown that spherules, although Oxygen Isotope Data themselves little altered, have alkali element ratios An orthopyroxene mineral separate (Ziegler et al. (enrichments of K, Rb, Cs, and Tl and depletion in Na) 2013), likely to be Ni-poor orthopyroxene (Hewins characteristic of terrestrial clay minerals and consistent et al. 2014) from pristine orthopyroxenites like ALH with a weathered source. Vesicular shards (Fig. 6a) also 84001 (Mittlefehldt 1994), exhibited D17O of 0.33&, indicate the presence of volatiles. The bulk compositions indistinguishable from SNC compositions, and reflecting (Humayun et al. 2013) also indicate degassing or mantle isotopic composition. There is an overall trend leaching of S from the precursors and the melt rocks of increasing D17O with increasing d18O in NWA 7034 lack magmatic sulfides in consequence (Lorand et al. separates, toward the composition of known Martian 2015). Thus, phyllosian conditions (Bibring et al. 2006) alteration phases. These D17O values were attributed to are indicated before 4.43 Ga. However, there is no clear photochemically induced isotope fractionation and evidence that phyllosilicates formed in the assembled transfer of the atmospheric signal to rocks by surface breccia in the Noachian, for although NWA 7034 does alteration (Farquhar and Johnston 2008). The Amazonian contain clays, they occur as nanogranoblastic crystals in and younger Martian atmosphere has been proposed to the annealed matrix (Muttik et al. 2014a). Their have had a high D17O value of >0.6& (Farquhar and Martian origin is based on the observation that water Thiemens 2000; Ziegler et al. 2013; Shaheen et al. 2015). from NWA 7034 is Martian in oxygen isotope The similar O isotopic compositions of ALH 84001 and composition (Agee et al. 2013). Possibly these are SNC alteration phases suggest that exchange with the annealed Pre-Noachian clay dust grains, but their time atmospheric reservoir occurred through most of Martian of formation is not well defined. history (Farquhar and Thiemens 2000). Fractional crystallization of a Martian basalt The zircon crystals in NWA 7533 dated by the U- magma like those in Fig. 19 does not yield evolved Pb method were subsequently analyzed for O isotopes rocks like the zircon-bearing noritic-monzonitic clasts. (Nemchin et al. 2014). Different zircon grains show a The various clast types in the breccia may be derived by difference in D17O values, many higher than the SNC melting different Pre-Noachian protoliths with different values defining the Mars fractionation line. Some are degrees of alteration or sedimentary geochemical also higher than bulk NWA 7034 and magnetite characteristics, rather than by fractionation of thick separate data (Ziegler et al. 2013), and include the melt sheets. If all the iron is ferrous, olivine will largest D17O reported from a Martian meteorite crystallize instead of orthopyroxene and magnetite. The Martian regolith breccia Northwest Africa 7533 117 magnetite-ilmenite assemblages in these rocks indicate Alteration of spherules and melt rock clasts consists oxygen fugacities at FMQ+2–3 log units (see below and of a fine speckling with high-Z Fe-rich material, which is Table S9) and show that the melt and therefore the difficult to characterize in situ, and different techniques protolith were highly oxidized. This oxidation, have identified different phases. The micro- and nano- accomplished by weathering or alteration of olivine in scale Fe-rich phases indicated include maghemite, precursors to serpentine + magnetite, explains the hematite, goethite, and ferrihydrite, a complication being crystallization of pyroxene and magnetite in its stead in the terrestrial alteration of pyrite. A poorly crystalline the noritic-monzonitic melt rocks (Humayun et al. Fe–OH bearing mineral gives the best match to the IR 2014a, 2014b), although the high alkalis and high K/Na spectrum of the breccia. IR and XANES measurements of the monzonitic rocks remain a problem in modeling. of NWA 7533 show that a strong 3 lm absorption band NWA 7533 could possibly record global climate is due to Fe3+-OH, and corresponds to one observed in changes during the Noachian, Hesperian, and Mars surface spectra particularly of the dark areas (Beck Amazonian periods with different weathering styles et al. 2015). Although there are also terrestrial (phyllosian, theiikian, and siderikian; Bibring et al. 2006). oxyhydroxides in the breccia these are confined to However, the breccia does not contain the weathering specific areas corresponding to former pyrite crystals sequence phyllosilicate/sulfate/anhydrous oxide and this and there must be a widespread occurrence of a phase suggests that local environmental effects might have like ferrihydrite on Mars. Abundant fine-grained countered the global-scale patterns. Murchie et al. (2009) magnetite and maghemite explain the breccia’s magnetic describe both phyllosilicate-rich and silica-, sulfate- properties especially the high saturation remanence bearing layered deposits deposited on ancient cratered (Gattacceca et al. 2014). terrain. These deposits may both have formed from Parts of the Noachian terrain display anomalies in volcanic and impact glasses, but the phyllosilicates would crustal remanent magnetization reported by Acuna~ have formed by abundant neutral fluid alteration, and the et al. (1999). This could be explained if the crust were sulfates from more acid solutions. Thus, the general constructed of certain SNC meteorites, but implausible climatic transition may depend on local environmental thicknesses would be required, whereas only ~1kmof conditions, due to variations in volcanic emissions and breccia could cause the observed remanent magnetism availability of S and/or water (Murchie et al. 2009). (Gattacceca et al. 2014). This is consistent with Either magmatic sulfide was virtually absent in NWA formation of the NWA 7533 breccia in the southern 7533 (Lorand et al. 2015) or sufficient water was present hemisphere while the Martian dynamo was active, that to remove S along with Cl and Zn in solution (Humayun is, in the Noachian (Rochette et al. 2013; Gattacceca et al. 2013). In addition, there is no record of the et al. 2014). Magnetic anomalies are restricted spatially Hesperian (theiikian alteration style) in the dating. Fluid but the 3 lm absorption indicating dust rich in Fe3+ is compositions and alteration phases may vary locally, but widespread on Mars. The source of the hydrated and as similar breccias are probably widespread in the oxidized dust covering may be the ancient breccias in southern hemisphere this may also be a regional effect. the heavily cratered southern highlands eroded and transported by wind (Beck et al. 2015). The similar Oxidation and Magnetics compositions of fine-grained materials in NWA 7533 The magnetite-ilmenite temperature assemblage in and the soils of Gusev Crater (Humayun et al. 2013) is lithic clasts in the breccia indicates a high fO2 of IW+2 in accord with this idea. to IW+3 log units at ~750 °C (Table S9). Pyroxene has been oxidized to magnetite plus silica (Leroux et al. CONCLUSIONS 2016). Alteration of NWA 7533 lithic clasts includes disequilibrium pyroxene assemblages probably related NWA 7533 is a polymict Martian breccia paired to oxidation, and pyroxene decomposition into Mg-rich with NWA 7034, 7475, 7906, 7907, 8114, and 8171. The pyroxene, silica, and Fe-oxide (Leroux et al. 2015). This abundance of fine-grained clast-laden melt rock bodies, reaction requires a high temperature and a high oxygen spherules and shards, and their high siderophile element fugacity, at least 2–3 log units above FMQ, consistent contents is consistent with classification as a regolith with a water-rich environment. Late-forming pyrite breccia. NWA 7533 is essentially a deposit of cogenetic veins also formed at a minimum log fO2 of >FMQ+2at melt particles differing texturally and chemically as a 400–500 °C (Lorand et al. 2016) from near neutral function of clast abundance. In addition, the breccia (6 < pH<10), H2S-HS-rich fluids. Hydrothermal contains a (slightly) earlier generation of medium- alteration under oxidizing conditions appears to have grained noritic and monzonitic melt rocks. There is a given rise to fine-grained Fe-rich phases influencing the paucity of material that can be interpreted as pristine magnetic properties of the breccia. crustal rock, although the low-Ni orthopyroxene clasts 118 R. H. Hewins et al. are interpreted as pristine orthopyroxenite of a deep- Acuna~ M. H., Acuna M. H., Connerney J. E. P., Wasilewski crustal origin similar to ALH 84001, a conclusion P., Kletetschka G., Ness N. F., Reme H., Lin R., and consistent with their low siderophile element Mitchell D. 1999. Global distribution of crustal magnetization discovered by the Mars Global Surveyor abundances. The crystallization of the monzonitic- MAG/ER experiment. Science 284:790–793. noritic, zoned-pyroxene, and microbasaltic melt rocks Agee C. B., Wilson N. V., McCubbin F. M., Ziegler K. M., occurred at ~4.43 Ga. Then orthopyroxenites and Ziegler K. 2012. Basaltic breccia NWA 7034: New crystallized in the Martian crust from before 4.43 Ga at ungrouped planetary achondrite (abstract #2690). 44th least until 4.09 Ga (Lapen et al. 2010). The noritic and Lunar and Planetary Science Conference. CD-ROM. Agee C. B., Wilson N. V., McCubbin F. M., Ziegler K., monzonitic clasts are essentially two-pyroxene and two- Polyak V. J., Ziegler K., Asmerom Y., Nunn M. H., feldspar rocks, respectively, but can also be Shaheen R., Thiemens M. H., Steele A., Fogel M. L., distinguished by plagioclase and spinel compositions. Bowden R., Glamoclija M., Zhang Z., and Elardo S. E. Exsolution in this clast suite pyroxene is coarser than in 2013. Unique meteorite from early Amazonian Mars: SNC meteorites, suggesting a thick sequence of impact Water-rich basaltic breccia Northwest Africa 7034. Science 339:780–785. melt rocks. The nanocrystalline matrix of the breccia Ashley G. M. and Delaney J. S. 1999. If a meteorite of was partly annealed, possibly in the ~1.35 Ga event Martian sandstone hit you on the head would you disturbing K feldspar, chlorapatite, and zircon ages, but recognize it? 30th Lunar and Planetary Science it preserves some porosity. Matrix and clasts are cut by Conference. CD-ROM. pyrite veins, probably caused by this heating event. Bandfield J. L., Hamilton V. E., and Christensen P. R. 2000. A global view of Martian surface compositions from Oxygen isotopic analyses of zircon grains and the MGS-TES. Science 287:1626–1630. geochemistry of the fine-grained rocks show that the Barrat J. A., Gillet P., Lesourd M., Blichert-Toft J., and protoliths interacted with atmosphere-hydrosphere Poupeau G. R. 1999. The Tatahouine diogenite: before melting. These breccias document the formation Mineralogical and chemical effects of sixty-three years of of evolved crust on Mars with a wet surface terrestrial residence. Meteoritics & Planetary Science 34:91–97. environment before 4.43 Ga. Beck P., Pommerol A., Remusat L., Zanda B., Lorand J. P., Gopel€ C., Hewins R., Pont S., Lewin E., Quirico E., Acknowledgments—We are indebted to L. Labenne, J. Schmitt B., Montes-Hernandez G., Garenne A., Bonal L., Gattacceca, and A. Jambon, for sample and insights, Proux O., Hazemann J. L., and Chevrier V. C. F. 2015. A oxygen isotope measurements, and Fe/Mn ratios, Noachian source region for the “Black Beauty” meteorite, and a source lithology for Mars surface hydrated dust? respectively; to A. Wittmann for discussions; to D. W. Earth and Planetary Science Letters 427:104–111. Mittlefehldt, J. G. Spray, and associate editor G. K. Bellucci J. J., Nemchin A. A., Whitehouse M. J., Humayun Benedix for thorough reviews which led to much M., Hewins R., and Zanda B. 2015. Evidence for an early improvement of the manuscript; to M. Fialin for help high l crust on Mars. Earth and Planetary Science Letters – with the electron probe and particularly with Ni 410:34 41. Bibring J.-P., Langevin Y., Mustard J. F., Poulet F., Arvidson analyses; to M. J. Carr for IGPET; to L. Lepage for R., Gendrin A., Gondet B., Mangold N., Pinet P., and ILMAT; and to D. Troadec for the preparation of FIB Forget F. 2006. Global mineralogical and aqueous Mars sections. We are grateful for funding from NASA history derived from OMEGA/Mars Express data. Science Cosmochemistry grant NNX13AI06G (M. H.), CNES- 312:400–404. INSU grant 2013-PNP (B. Z.) and grant 2014-PNP (J.- Bogard D. D., Nyquist L. E., Hirsch W. C., and Moore D. R. 1973. Trapped solar and cosmogenic noble gas abundances P. L.). The National NanoSIMS Facility at the in Apollo 15 and 16 deep drill samples. Earth and Museum National d’Histoire Naturelle was established Planetary Science Letters 21:52–69. by funds from the CNRS, Region ^Ile de France, Borg L. E., Edmunson J. E., and Asmerom Y. 2005. Ministere delegue a l’Enseignement superieur et ala Constraints on the U-Pb isotopic systematics of Mars – – – Recherche, and the Museum itself. inferred from a combined U Pb, Rb Sr, and Sm Nd isotopic study of the Martian meteorite Zagami. Geochimica et Cosmochimica Acta 69:5819–5830. Editorial Handling—Dr. Gretchen Benedix Bouvier A., Blichert-Toft J., Vervoort J. D., Gillet P., and Albarede F. 2008. The case for old basaltic shergottites. REFERENCES Earth and Planetary Science Letters 266:105–124. Bouvier A., Blichert-Toft J., and Albarede F. 2009. Martian Abramov O. and Kring D. A. 2005. Impact-induced meteorite chronology and the evolution of the interior of hydrothermal activity on early Mars. Journal of Mars. Earth and Planetary Science Letters 280:285–295. Geophysical Research 110:E12S09. doi:10.1029/ Carr M. H. and Head J. W. III 2010. Geologic history of 2005JE002453. Mars. Earth and Planetary Science Letters 294:185–203. Abramov O. and Mojzsis S. J. 2014. Thermal effects of impact Carter J. and Poulet F. 2013. Ancient plutonic processes on bombardments on early Mars (abstract #1484). Eighth Mars inferred from the detection of possible anorthositic International Conference on Mars. terrains. Nature Geoscience 6:1008–1012. Martian regolith breccia Northwest Africa 7533 119

Carter J., Poulet F., Murchie S., and Bibring J.-P. 2013. soils in Gusev Crater from the alpha particle X-ray Automated processing of planetary hyperspectral datasets spectrometer. Science 305:829–832. for the extraction of weak mineral signatures and Gellert R., Berger J. A., Boyd N., Brunet C., Campbell J. L., applications to CRISM observations of hydrated silicates Curry M., Elliott B., Fulford P., Grotzinger J., Hipkin V., on Mars. Earth and Planetary Science Letters 76:53–67. Hurowitz J., King P. L., Leshin L., Limonadi D., Pavri Cartwright J. A., Ott U., Herrmann S., and Agee C. B. 2014. B., Pradler I.., Marchand G.., Perrett G.. M., and Scodary Modern atmospheric signatures in 4.4 Ga Martian meteorite A. 2013. Initial MSL APXS activities and observations at NWA 7034. Earth and Planetary Science Letters 400:77–87. Gale Crater, Mars (abstract #1432). 44th Lunar and De Hoog J. C. M., Gall L., and Cornell D. H. 2010. Trace- Planetary Science Conference. CD-ROM. element geochemistry of mantle olivine and application to Gleason J. D., Kring D. A., Hill D. H., and Boynton W. V. mantle petrogenesis and geothermobarometry. Chemical 1997. Petrography and bulk chemistry of Martian Geology 270:196–215. orthopyroxenite ALH 84001: Implications for the origin of Debaille V., Brandon A. D., Yin Q. Z., and Jacobsen B. 2007. secondary carbonates. Geochimica et Cosmochimica Acta Coupled 142Nd-143Nd evidence for a protracted magma 61:3503–3512. ocean in Mars. Nature 450:525–528. Grange M. L., Nemchin A. A., Pidgeon R. T., Merle R. E, El Goresy A., Kong P., and Palme H. 1999. Discovery of Cu-, and Timms N. E. 2013. What lunar zircon ages can tell Ni-, Zn-, Fe-metal alloy impregnations in the SNC (abstract)? 44th Lunar and Planetary Science Conference. Meteorite Chassigny: Fingerprints of a pristine CD-ROM. metallogenic activity on the SNC parent body (abstract Greeley R., Foing B. H., McSween H. Y., Neukum G., Pinet #1078). 30th Lunar and Planetary Science Conference. P., van Kan M., Werner S. C., Williams D. A., and Zegers CD-ROM. T. E. 2005. Fluid lava flows in Gusev crater, Mars. El Goresy A., Gillet P., Miyahara M., Ohtani E., Ozawa S., Journal of Geophysical Research 110:E05008. Beck P., and Montagnac G. 2013. Shock-induced Grieve R. A. F. 1975. Petrology and chemistry of the impact deformation of Shergottites: Shock-pressures and melt at Mistastin Lake crater, Labrador. Geological perturbations of magmatic ages on Mars. Geochimica et Society of America Bulletin 86:1617–1629. Cosmochimica Acta 101:233–262. Grieve R. A. F., Plant A. G., and Dence M. R. 1974. Elkins-Tanton L. T. 2012. Magma oceans in the inner solar Lunar impact melts and terrestrial analogs—Their system. Annual Reviews in Earth and Planetary Science. characteristics, formation and implications for lunar 40:113–139. crustal evolution. Proceedings, 5th Lunar Science Farquhar J. and Johnston D. T. 2008. The oxygen cycle of the Conference. pp. 261–273. terrestrial planets: Insights into the processing and history Grieve R. A. F., Ames D. E., Morgan J. V., and Artemieva of oxygen in surface environments. Reviews in Mineralogy N. A. 2010. The evolution of the onaping formation at the & Geochemistry 68:463–492. Sudbury impact structure. Meteoritics & Planetary Science Farquhar J. and Thiemens M. H. 2000. Oxygen cycle of the 45:759–782. Martian atmosphere_regolith system: 17O of secondary Hartmann W. K. 2005. Martian cratering 8: Isochron phases in Nakhla and Lafayette. Journal of Geophysical refinement and the chronology of Mars. Icarus 174:294– Research 105:1991–1997. 320. Fassett C. I. and Head J. W. 2011. Sequence and timing of Hartmann W. K. and Barlow N. G. 2006. Nature of the conditions on early Mars. Icarus 211:1204–1214. Martian uplands: Effect on Martian meteorite age Franchi I. A., Wright I. P., Sexton A. S., and Pillinger C. T. distribution and secondary cratering. Meteoritics & 1999. The oxygen isotopic composition of Earth and Planetary Science 41:1453–1467. Mars. Meteoritics & Planetary Science 34:657–661. Hewins R. H. 1984. The case for a melt matrix in plagioclase- French B. M. 1998. Traces of catastrophe: A handbook of POIK mesosiderites. Proceedings, 15th Lunar and shock-metamorphic effects in terrestrial meteorite impact Planetary Science Conference. Journal of Geophysical structures. LPI Contribution 954. Houston, Texas: Lunar Research 89:C289–C297. and Planetary Institute. 120 p. Hewins R. H., Zanda B., Humayun M., Pont S., Fieni C., and Frey H. V., Frey E. L., Hartmann W. K., and Tanaka K. L. Deldicque D. 2013a. Northwest Africa 7533, an impact T. 2003. Evidence for buried “pre-Noachian” crust pre- breccia from Mars (abstract #2385). 44th Lunar and dating the oldest observed surface units on Mars (abstract Planetary Science Conference. CD-ROM. #1848). 34th Lunar and Planetary Science Conference. Hewins R. H., Zanda B., Humayun M., Lorand J.-P., CD-ROM. Deldicque D., Pont S., Fieni C., Nemchin A., Grange M., Gaboardi M. and Humayun M. 2009. Elemental fractionation Kennedy A., Gopel€ C., and Lewin E. 2013b. Petrology of during LA-ICP-MS analysis of silicate glasses: NWA 7533: Formation by impacts on ancient Martian Implications for matrix-independent standardization. crust (abstract #5252). Meteoritics & Planetary Science 48. Journal of Analytical Atomic Spectrometry 24:1188–1197. Hewins R. H., Zanda B., Humayun M., Lorand J.-P., and Gattacceca J., Rochette P., Scorzelli R. B., Munayco P., Agee Pont S. 2014. Impact melt rocks and pristine clasts in C., Quesnel Y., Cournede C., and Geissman J. 2014. Northwest Africa 7533 (abstract #1416). 45th Lunar and Martian meteorites and Martian magnetic anomalies: A Planetary Science Conference. CD-ROM. new perspective from NWA 7034. Geophysical Research Hofmann B. A., Greber N. D., and Greenwood R. C. 2014. Letters 41:4859–4864. New data for NWA 7906, 7907 AND 8171, pairings of Gellert R., Rieder R., Andersen R. C., Bruckner€ J., Clark B. Mars breccia NWA 7034 (abstract #5230). Meteoritics & C., Dreibus G., Economou T., Klingelhofer€ G., Lugmair Planetary Science 48(Suppl.). G. W., Ming D. W., Squyres S. W., d’Uston C., Wanke€ Hudgins J. A., Walton E. L., and Spray J. G. 2007. H., Yen A., and Zipfel J. 2004. Chemistry of rocks and Mineralogy, petrology, and shock history of lunar 120 R. H. Hewins et al.

meteorite Sayh al Uhaymir 300: A crystalline impact-melt the NWA 7533 Martian meteorite. Meteoritics & Planetary breccia. Meteoritics & Planetary Science 42:1763–1779. Science 51:932–945. Humayun M., Simon S. B., and Grossman L. 2007. Tungsten Lindsley D. H. and Andersen D. J. 1983. A two-pyroxene and hafnium distribution in calcium-aluminum inclusions thermometer. Proceedings 13th Lunar and Planetary CAIs from Allende and Efremovka. Geochimica et Science Conference. JGR Suppl 88:A887–A906. Cosmochimica Acta 71:4609–4627. Liu Y., Guan Y., McCubbin F. M., Eiler J. M., Agee C. B., Humayun M., Davis F. A., and Hirschmann M. M. 2010. Williford K. H., and Allwood A. C. 2014. The Martian Major element analysis of natural silicates by laser surface water in breccia meteorite NWA 7034 (abstract ablation ICP-MS. Journal of Analytical Atomic #2368). 12th Lunar and Planetary Science Conference. Spectrometry 25:998–1005. CD-ROM. Humayun M., Nemchin A., Zanda B., Hewins R. H., Grange Lorand J.-P., Hewins R. H., Remusat L., Zanda B., Pont S., M., Kennedy A., Lorand J.-P., Gopel€ C., Fieni C., Pont Leroux H., Marinova M., Jacob D., Humayun M., S., and Deldicque D. 2013. Origin and age of the earliest Nemchin A., Grange M., Kennedy A., and Gopel€ Christa Martian crust from meteorite NWA7533. Nature 503:513– 2015. Nickeliferous pyrite tracks pervasive hydrothermal 517. alteration in Martian regolith breccias: A study in NWA Humayun M., Hewins R. H., Lorand J.-P., and Zanda B. 7533. Meteoritics & Planetary Science 50:2099–2120. 2014a. Weathering and impact melting determined the Lorand J.-P., Hewins R. H., Humayun M., Remusat L., mineralogy of the early Martian crust preserved in Zanda B., La C., and Pont S. 2016. Chalcophile Northwest Africa 7533 (abstract #1880). 45th Lunar and siderophile trace element systematics of hydrothermal Planetary Science Conference. CD-ROM. pyrite from Martian regolith breccia NWA 7533 (abstract Humayun M., Hewins R. H., and Zanda B. 2014b. #6082). Meteoritics & Planetary Science 50:932–945. Weathering, transport and impact melting in the Martian MacArthur J. L., Bridges J. C., Hicks L. J., and Gurman S. J. highlands crust viewed from Martian meteorite NWA 7533 2015. The thermal and alteration history of NWA 8114 (abstract #1459). Eighth International Conference on Martian regolith (abstract #2295). 45th Lunar and Mars. Planetary Science Conference. CD-ROM. Humayun M., Hewins R. H., and Zanda B. 2014c. Evidence Mangold N., Forni O., Ollila A., Andersen R., Berger G., for phyllosilicate precursors in an aphanitic spherule from Bridges J. C., Clegg S. M., Cousin A., Dietrich W. E., NWA 7533 (abstract #5413). Meteoritics & Planetary Gupta S., Lewin E., Fabre C., Gasnault O., Herkenhoff Science 48(Suppl.). K., Le Mouelic S., Maurice S., Meslin P.-Y., Sautter V., Jarosewich E., Nelen J. A., and Norberg J. A. 1980. Wiens R. C., and Williams R., and the MSL Science Corrections. Geostandards Newsletter 4:257–258. Team. 2013. Chemcam analysis of conglomerates at Joy K. H., Nemchin A., Grange M., Lapen T. J., Peslier A. Bradbury site, Mars (abstract #1267). 44th Lunar and H., Ross D. K., Zolensky M. E., and Kring D. A. 2014. Planetary Science Conference. CD-ROM Petrography, geochronology and source terrain McSween H. Y. Jr. 2013. Planetary science: A chunk of characteristics of lunar meteorites Dhofar 925, 961 and ancient Mars. Nature 503:473–474. Sayh al Uhaymir 449. Geochimica et Cosmochimica Acta McSween H. Y. Jr., Murchie S. L., Britt D. T., Brukner€ J., 144:299–325. Dreibus G., Economou T., Ghosh A., Golombek M. P., Keil K., Fodor R. V., and Bunch T. E. 1972. Contributions to Greenwood J. P., Johnson J. R., Moore H. J., Morris R. V., the mineral chemistry of Hawaiian rocks II: Feldspars and Parker T. J., Rieder R., Singer R., and Wanke€ H. 1999. interstitial material in rocks from Haleakala and West Chemical, multispectral, and textural constraints on the Maui volcanoes, Maui, Hawaii. Contributions to composition and origin of rocks at the Mars Pathfinder Mineralogy and Petrology 37:253–276. landing site. Journal of Geophysical Research 104:8679–8715. Korotev R. L., Rumble D., and Irving A. J. 2013. Elemental McSween H. Y. Jr., Grove T. L., and Wyatt M. B. 2003. and oxygen isotopic composition of Martian mafic regolith Constraints on the composition and petrogenesis of the breccia NWA 7475 (abstract# 5046). Meteoritics & Martian crust. Journal of Geophysical Research 108:5135. Planetary Science 47(Suppl.). McSween H. Y. Jr., Wyatt M. B., Gellert R., Bell J. F. III, Lapen T. J., Brandon A. D., Debaille V., Beard B. L., Shafer Morris R. V., Herkenhoff K. E., Crumpler L. S., Milam J. T., and Peslier A. H. 2010. A younger age for ALH K. A., Stockstill K. R., Tornabene L. L., Arvidson R. E., 84001 and its geochemical link to shergottite sources in Bartlett P., Blaney D., Cabrol N. A., Christensen P. R., Mars. Science 328:347–351. Clark B. C., Crisp J. A., Des Marais D. J., Economou T., Lemmon M. T., Wolff M., Smith M., Clancy R., Banfield D., Farmer J. D., Farrand W., Ghosh A., Golombek M., Landis G., Ghosh A., Smith P., Spanovich B., Whitney B., Gorevan S., Greeley R., Hamilton V. E., Joliff B. L., Whelley P., Greeley R., Thompson J., Bell J. III, and Johnson J. R., Klingelhofer€ G., Knudson A. T., Squyres S. 2004. Atmospheric imaging results from the McLennan S., Ming D., Moersch J. E., Rieder R., Ruff S. Mars exploration rovers: Spirit and opportunity. Science W., Schroder€ C., de Souza P. A. Jr., Squyres S. W., 30:1753–1756. Wa€̈nke H., Wang A., Yen A., and Zipfel J. 2006. Leroux H., Marinova M., Jacob D., Hewins R. H., Zanda B., Characterization and petrologic interpretation of olivine- Humayun M., and Lorand J. P. 2015. A TEM study of rich basalts at Gusev crater, Mars. Journal of Geophysical the fine-grained material of the NWA 7533 Martian Research 111(E2):1–17. regolith breccia (abstract #1832). 45th Lunar and McSween H. Y. Jr., Jeffrey Taylor G., and Wyatt Michael B. Planetary Science Conference. CD-ROM. 2009. Elemental composition of the Martian Crust. Science Leroux H., Jacob D., Marinova M., Hewins R. H., Zanda B., 324:736–739. Pont S., Lorand J.-P., and Humayun M. 2016. Exsolution McSwiggen P. L., Morey G. B., and Cleland J. M. 1994. and shock microstructures of igneous pyroxene clasts in Occurrence and genetic implications of hyalophane in Martian regolith breccia Northwest Africa 7533 121

manganese-rich iron-formation, Cuyuna Iron Range, Martian meteorites as identified by OMEGA/Mex. Icarus Minnesota, USA. Mineralogical Magazine 58:387–399. 258:366–383. Mittlefehldt D. W. 1994. ALH 84001, a cumulate Osinski G. R., Spray J. G., and Lee P. 2001. Impact-induced orthopyroxenite member of the Martian meteorite clan. hydrothermal activity within the Haughton impact Meteoritics 29:214–221. structure, arctic Canada: Generation of a transient, warm, Moser D. E., Chamberlain K. R., Tait K. T., Schmitt A. K., wet oasis. Meteoritics & Planetary Science 36:731–745. Darling J. R., Barker I. R., and Hyde B. C. 2013. Solving the Osinski G. R., Grieve R. A. F., Collins G. S., Marion C., and Martian meteorite age conundrum using micro-baddeleyite Sylvester P. 2008. The effect of target lithology on the and launch-generated zircon. Nature 499:454–458. products of impact melting. Meteoritics & Planetary Mouginis-Mark P. J. 2015. Cratering on Mars with almost no Science 43:1939–1964. atmosphere or volatiles: Pangboche crater. Meteoritics & Osinski G. R., Tornabene L. L., Banerjee N. R., Cockell C. Planetary Science 50:51–62. S., Flemming R., Izawa M. R. M., McCutcheon J., Parnell Murchie S. L., Mustard J. F., Ehlmann B. L., Milliken R. E., J., Preston L. J., Pickersgill A. E., Pontefract A., Sapers Bishop J. L., McKeown N. K., Noe Dobrea E. Z., Seelos H. M., and Southam G. 2013. Impact-generated F. P., Buczkowski D. L., Wiseman S. M., Arvidson R. E., hydrothermal systems on Earth and Mars. Icarus 224:347– Wray J. J., Swayze G., Clark R. N., Des Marais D. J., 363. McEwen A. S., and Bibring J.-P. 2009. A synthesis of Pope K. O., Kieffer S. W., and Ames D. E. 2006. Impact melt Martian aqueous mineralogy after 1 Mars year of sheet formation on Mars and its implication for observations from the Mars Reconnaissance Orbiter. hydrothermal systems and exobiology. Icarus 186:1–9. Journal of Geophysical Research 114:E00D06. Poulet F., Bibring J.-P., Mustard J. F., Gendrin A., Mangold Mustard J. F., Poulet F., Gendrin A., Bibring J.-P., Langevin N., Langevin Y., Arvidson R. E., Gondet B., and Gomez Y., Gondet B., Mangold N., Bellucci G., and Altieri F. C. 2005. Phyllosilicates on Mars and implications for early 2005. Olivine and pyroxene diversity in the crust of Mars. Martian climate. Nature 438:623–627. Science 307:1594–1597. Poulet F., Gomez C., Bibring J.-P., Langevin Y., Gondet B., Muttik N., Keller L. P., Agee C. B., McCubbin F. M., Santos Pinet P., Belluci G., and Mustard J. 2007. Martian surface A. R., and Rahman Z. 2014a. A TEM investigation of the mineralogy from Observatoire pour la Mineralogie, l’Eau, fine-grained matrix of the Martian basaltic breccia NWA les Glaces et l’Activite on board the Mars Express 7034 (abstract #2763). 46th Lunar and Planetary Science spacecraft (OMEGA/MEx): Global mineral maps. Journal Conference. CD-ROM. of Geophysical Research 112:E08S02. Muttik N., McCubbin F. M., Keller L. P., Santos A. S., Robbins S. J., Di Achille G., and Hynek B. M. 2011. The McCutcheon M. C., Provencio P. P., Rahman Z., Shearer volcanic history of Mars: High-resolution crater-based C. K., Boyce J. W., and Agee C. B. 2014b. Inventory of studies of the calderas of 20 volcanoes. Icarus 211:1179–1203. H2O in the ancient Martian regolith from Northwest Rochette P., Gattacceca J., Scorzelli R. B., Munayco P., Agee Africa 7034: The important role of Fe oxides. Geophysical C. B., Cournede C., and Geissman J. 2013. Searching for Research Letters 41:8235–8244. the lithology responsible for large crustal magnetization on Nemchin A. A., Humayun M., Whitehouse M., Hewins R. H., Mars: A changing perspective from NWA 7034 (abstract Lorand J.-P., Kennedy A., Grange M., Zanda B., Fieni #1880). 44th Lunar and Planetary Science Conference. C., and Deldicque D. 2014. Record of ancient Martian CD-ROM. hydrosphere preserved in zircon from NWA 7533. Nature Santos A. R., Agee C. B., McCubbin F. M., Shearer C. K., Geoscience 7:638–742. Burger P. V., Tartes R., and Anand M. 2015a. Petrology Newsom H. E. 1980. Hydrothermal alteration of impact melt of igneous clasts in Northwest Africa 7034: Implications sheets with implications for Mars. Icarus 44:207–216. for the petrologic diversity of the Martian crust. Niihara T., Kaiden H., Misawa K., Sekine T., and Mikouchi Geochimica et Cosmochimica Acta 157:56–85. T. 2012. U-Pb isotopic systematics of shock-loaded and Santos A. R., Agee C. B., McCubbin F. M., and Shearer C. annealed baddeleyite: Implications for crystallization ages K. 2015b. Petrology of iron, titanium, and phosphorus of Martian meteorite shergottites. Earth and Planetary rich clasts within Martian meteorite Northwest Africa Science Letter 341:195–210. 7034 (abstract #1941). 44th Lunar and Planetary Science Nyquist L. E., Bogard D. D., Shih C., Greshake A., Stoffler€ Conference. CD-ROM. D., and Eugster O. 2001. Ages and geologic histories of Sautter V., Fabre C., Forni O., Toplis M. J., Cousin A., Ollila Martian meteorites. Space Science Reviews 96:105–164. A. M., Meslin P. Y., Maurice S., Wiens R. C., Baratoux Nyquist L. E., Shih C., McCubbin F. M., Santos A. R., D., Mangold N., Le Mouelic S., Gasnault O., Berger G., Shearer C. K. Jr., Peng Z., Burger P. V., and Agee C. B. Lasue J., Andersen R. A., Lewin E., Schmidt M., Dyar 2016. Rb-Sr and Sm-Nd isotopic and REE studies of D., Ehlmann B. L., Bridges J., Clark B., and Pinet P. igneous components in the bulk matrix domain of Martian 2014. Igneous mineralogy at Bradbury Rise: The first breccia Northwest Africa 7034. Meteoritics & Planetary ChemCam campaign at Gale crater. Journal of Geophysical Science 51:483–498. Research Planets 119:1–17. Ody A., Poulet F., Langevin Y., Bibring J.-P., Bellucci G.., Sautter V., Toplis M., Beck P., Mangold N., Wiens R. C., Altieri F., Gondet B., Vincendon M., Carter J., and Pinet P., Cousin A., Maurice S., LeDeit L., Hewins R. H., Manaud N. 2012. Global maps of anhydrous minerals at Gasnault O., Quantin C., Forni O., Newson H., Meslin the surface of Mars from OMEGA/MEx. Journal of P.-Y., Wray J., Bridges N., Payre V., and Lemouelic S. Geophysical Research 117:E00J14. 2016. Magmatic complexity on early Mars: Getting to the Ody A., Poulet F., Quantin C., Bibring J.-P., Bishop J. L., bottom of the Noachian crust through a combination of and Dyar M. D. 2015. Candidate source regions of orbital, in-situ and meteorite data. Lithos 254–255:36–52. 122 R. H. Hewins et al.

Schmidt M. E., Campbell J. L., Gellert R., Perrett G. M., modeling Part I: Observations. Meteoritics & Planetary Treiman A. H., Blaney D. L., Olilla A., Calef III F. J., Science 48:515–589. Edgar L., Elliott B. E., Grotzinger J., Hurowitz J., King Stolper E. M., Baker M. B., Newcombe M. E., Schmidt M. P. L., Minitti M. E., Sautter V., Stack K., Berger J. A., E., Treiman A. H., Cousin A., Dyar M. D., Fisk M. R., Bridges J. C., Ehlmann B. L., Forni O., Leshin L. A., Gellert R., King P. L., Leshin L., Maurice S., McLennan Lewis K. W., McLennan S. M., Ming D. W., Newsom H., S. M., Minitti M. E., Perrett G., Rowland S., Sautter V., Pradler I., Squyres S. W., Stolper E. M., Thompson L., and Wiens R. C., and MSL Science Team. 2013. The Van Bommel S., and Wiens R. C. 2014. Geochemical petrochemistry of Jake Matijevic: A Martian mugearite. diversity in first rocks examined by the Curiosity Rover in Science 341:1239463-1-7. Gale Crater: Evidence for and significance of an alkali and Tanaka K. L. and Hartmann W. K. 2012. The planetary time volatile-rich igneous source. Journal of Geophysical scale. In The geologic time scale 2012, edited by Gradstein Research 119:64–81. F. M., Ogg J. G., Schmitz M., and Ogg G. Boston: Schmieder M., Kring D., Swindle T. D, Bond J. C., and Elsevier. Elsevier, Boston, USA, pp. 275–298. Moore C. B. 2016. The Gao-Guenie impact melt breccia— Tanaka K. L., Skinner J. A. Jr., Fortezzo C. M., Hare T. M., Sampling a rapidly cooled impact melt dike on an H Irwin III R. P., Platz T., Michael G., Dohm J. M., Kolb chondrite asteroid? Meteoritics & Planetary Science E. J., and Robbins S. J. 2014. The “new” geology of 51:1022–1045. Mars: Top ten results of post-Viking global mapping and Schultz P. H. and Mustard J. F. 2004. Impact melts and crater-dating (abstract #1087). Eighth International glasses on Mars. Journal of Geophysical Research 109: Conference on Mars. E01001. Tartese R., Anand M., McCubbin F. M., Santos A. R., and See T. H., Wagstaff J., Yang V., Horz F., and McKay G. A. Delhaye T. 2014. Zircons in Northwest Africa 7034: 1998. Compositional variation and mixing of impact melts Recorders of crustal evolution on Mars. 45th Lunar and on microscopic scales. Meteoritics & Planetary Science Planetary Science Conference #2020. CD-ROM. 33:937–948. Tornabene L. L., Moersch J. E., McSween H. Y. Jr., McEwen Shaheen R., Niles P. B., Chong K., Corrigan C., and Thiemens A. S., Piatek J. L., Milam K. A., and Christensen P. R. M. H. 2015. Carbonate formation events in ALH 84001 2006. Identification of large (2–10 km) rayed craters on trace the evolution of Martian atmosphere. Proceedings of Mars in THEMIS thermal infrared images: Implications the National Academy of Sciences 112:336–341. for possible Martian meteorite source regions. Journal of Simonds C. H. 1975. Thermal regimes in impact melts and the Geophysical Research 111:E10006. doi:10.1029/ petrology of the Apollo 17 Station 6 boulder. Proceedings, 2005JE002600. 6th Lunar Science Conference. pp. 641–672. Turner F. J. and Verhoogen J. 1951. Igneous and metamorphic Simonds C. H., Phinney W. C., Warner J. L., McGee P. E., petrology. New York: McGraw-Hill Inc., New York, 601 Geeslin J., Brown R. W., and Rhodes J. M. 1977. Apollo p. 14 revisited, or breccias aren’t so bad after all. Udry A., Lunning N. G., McSween H. Y. Jr., and Bodnar R. Proceedings, 8th Lunar Science Conference. pp. 1869– J. 2014. Petrogenesis of a vitrophyre in the Martian 1893. meteorite breccia NWA 7034. Geochimica and Simonson B. M. 2003. Petrographic criteria for recognizing Cosmochimica Acta 141:281–293. certain types of impact spherules in well-preserved Warren P. H., Ulff-Møller F., and Kallemeyn G. W. 2005. Precambrian successions. Astrobiology 3:49–65. “New” lunar meteorites: Impact melt and regolith beccias Smith J. V., Steele I. M., and Leitch C. A. 1983. Mineral and large-scale heterogeneities of the upper lunar crust. chemistry of the shergottites, nakhlites, Chassigny, Meteoritics & Planetary Science 40:989–1014. Brachina, pallasites, and ureilites. Proceedings, 14th Lunar Wieler R. 1998. The solar noble gas record in lunar samples and Planetary Science Conference. Journal of Geophysical and meteorites. Space Science Reviews 85:303–314. Research 88:B229–B236. Wittmann A., Korotev R. L., Jolliff B. L., Kuehner S. M., Stoffler€ D. and Grieve R. A. F. 1994. Classification and and Irving A. J. 2013a. Petrology of impact melt rich nomenclature of impact metamorphic rocks: A proposal to Martian regolith breccia Northwest Africa 7475 (abstract the IUGS Subcommission on the Systematics of #5272). Meteoritics & Planetary Science 47. Metamorphic Rocks. 25th Lunar and Planetary Science Wittmann A., Goderis S., Claeys P., Vanhaecke F.k, Deutsch Conference. pp. 1347–1348. A., and Adolph L. 2013b. Petrology of impactites from Stoffler€ D. and Grieve R. A. F. 2007. Impactites. In El’gygytgyn crater: Breccias in ICDP-drill core 1C, glassy Metamorphic rocks: A classification and glossary of terms, impact melt rocks and spherules. Meteoritics & Planetary recommendations of the International Union of Geological Science 48:1199–1235. Sciences, edited by Fettes D. and Desmons J. Cambridge, Wittmann A., Korotev R. L., Jolliff B. L., Irving A. J., Moser UK: Cambridge University Press. pp. 82–92, 111–125, and D. E., Barker I., and Rumble D. 2015. Petrography and 126–242. composition of Martian regolith breccia meteorite Stoffler€ D., Knoll€ H.-D., Marvin U. B., Simonds C. H., and Northwest Africa 7475. Meteoritics & Planetary Science Warren P. H. 1980. Recommended classification and 50:326–352. nomenclature of lunar highland rocks - A committee Wray J. J., Hansen S. T., Dufek J., Swayze G. A., Murchie S. report. In Proceedings, Conference on the Lunar Highlands L., Seelos F. P., Skok J. R., Irwin III R. P., and Ghiorso Crust, edited by Papike J. J. and Merrill R. B.. New York M. S. 2013. Prolonged magmatic activity on Mars inferred and Oxford: Pergamon Press. pp. 51–70. from the detection of felsic rocks. Nature Geoscience Stoffler€ D., Artemieva N. A., Wunnemann€ K., Reimold W. 6:1013–1017. U., Jacob J., Hansen B. K., and Summerson I. A. T. 2013. Yin Q.-Z., McCubbin F. M., Zhou Q., Santos A. R., Tartese Ries crater and suevite revisited—Observations and R., Li X., Li Q., Liu Y., Tang G., Boyce J. W., Lin Y., Martian regolith breccia Northwest Africa 7533 123

Yang W., Zhang J., Hao J., Elardo S. M., Shearer C. K., Ziegler K., Sharp Z. D., and Agee C. B. 2013. The unique Rowland D. J., Lerche M., and Agee C. B. 2014. An NWA 7034 Martian meteorite: Evidence for multiple Earth-like beginning for ancient Mars indicated by alkali- oxygen isotope reservoirs (abstract). 44th Lunar and rich volcanism at 4.4 Ga (abstract #1320). 45th Lunar and Planetary Science Conference. CD-ROM. Planetary Science Conference. CD-ROM.

SUPPORTING INFORMATION Fig. S5: BSE images of shards and spherules. a) Triangular shard with orthopyroxene surrounded by Additional supporting information may be found in plagioclase in breccia matrix. b) Detail of large aphanitic the online version of this article: spherule in NWA 7533-7 with feldspar veins and coarse Fig. S1: Sawed surface of a 2.5 cm wide slab of crystals near the periphery. c, d) Small spherules NWA 7533 with two melt spherules (black) and dark containing pyroxene, plagioclase, chromite, and magnetite. bodies of clast-laden melt rock; feldspathic clasts are Spherules (c) in 7533-LM and (d) in 7533-5 resemble light colored. orthopyroxene clumps. Fig. S2: Breccia matrix. a) Low magnification BSE Fig. S6: a) Microbasaltic clast in 7533-6, with two view, with many crystal clasts (plagioclase dark gray, small lithic clasts near its center. b) Microbasaltic clast pyroxene medium gray). b) High magnification BSE 7533-7-4, containing potash feldspar (K), as well as image, with light orthopyroxene embedded in dark orthopyroxene (O), augite (A), and plagioclase. plagioclase. c) TEM image showing granoblastic texture Fig. S7: a–f) Pyroxene compositions for microbasalt. with porosity on grain boundaries. d) BSE. Micron- a) One clast E in 7533-4 with primary orthopyroxene only, sized medium gray orthopyroxene grains in a cluster and one with pigeonite only, 7533-4-L. Both phases have like a dustball. compositions extending to more Fe-rich values Fig. S3: Clast-laden melt rock BSE images. a) (En69–55Wo~3 and En66–49Wo~10) than for clast-laden melt somewhat rounded 40 lm clasts in subophitic fine- rocks. b) In granoblastic clast 7533-1-obj2 (Fig. 8c) fine grained groundmass; (b) less fine- grained, fewer clasts; mixtures of orthopyroxene and augite are present with little (c, d) Mg and Al Ka X-ray maps complementary to variation in Fe/Mg ratio. Magnesian orthopyroxene Fig. 4c showing pyroxene and plagioclase, light/dark and En63.7Æ0.6Fs32.0Æ0.7Wo4.3Æ0.2 is less abundant than more dark/light, respectively. These images show the clusters ferroan pyroxene whose average composition and aureoles very clearly, because of the enhanced En52.7Æ2.5Fs38.6Æ3.5Wo8.8Æ5.4 indicates that primary contrast between pyroxene and plagioclase relative to pigeonite was originally present. Fe-Mg fractionation is BSE. In Mg Ka, the top half is seen to be an Mg-rich recorded in the orthopyroxene, En67–53Wo~3.c,d)A groundmass with radial orthopyroxene in the clumps. In crystallization sequence of orthopyroxene followed by the Al image, the groundmass laths are seen to be pigeonite, and/or augite. e) The fine-grained clast 2-84 skeletal-dendritic and the aureole plagioclase is radial; the (which sits at the bottom of the sedimentary clast in plagioclase lath abundance defines the schlieren. e) Fig. 8d) has zoned pigeonite and augite En66.4–51.9 Orthopyroxene granules/dendrites in clump and (arrows) Fs28.8–37.8Wo6.3–14.2 and augite En49.3–40.5Fs17.8–22.2 fractures partly filled with calcite (f) elongated and large Wo28.1–32.0, and is discussed below along with similar irregular orthopyroxene clumps in 7533-5, suggesting medium-grained rocks, that also have zoned pyroxenes like flow. The large amoeboid clump has an aureole that is those of shergottites. discontinuous on its borders and partially in its interior, Fig. S8: BSE image of sedimentary clast in 7533-2. suggesting fragmentation. Compare to Santos et al. Border (white) established by locating the characteristic (2015a, 2015b; Figs. 8a and 8b). texture of breccia matrix using a 1.2 Gpix FEG SEM image Fig. S4: a) Detail of NWA 7533-1 showing crescent- of true dimensions ~6 9 9 m. Noritic clast 2-1521 at top shaped clast-laden melt rock body (MR), with crystal and microbasaltic clast 2–84 at bottom are attached to the clasts, several clump-aureole structures (small dark sedimentary layers. ellipses), and also microbasalt (MB) and spherule (S). b) Fig. S9: Individual pyroxene quadrilaterals for four Red-green-blue coded X-ray (Mg, Al, Si, Ka) image of the small zoned-pyroxene lithic clasts and one microbasalt clump-aureole structures (orthopyroxene clusters with (7533-2:84), showing pigeonite with or without augite or plagioclase borders) in melt rock in Fig. 4c: orthopyroxene subcalcic augite (clast size effect). purple, augite dark purple (with orthopyroxene rims), Fig. S10: BSE images. a) Noritic clast 7533-4-C with plagioclase light blue; Fe-oxides and crack-filling calcite magnesian (OM) and ferroan (OF) orthopyroxene, augite are black. Plagioclase-rich and pyroxene-rich schlieren are (A), plagioclase (P), and Cr-bearing magnetite (M). Z is evident. c) Ca Ka map shows white calcite in fractures. zircon Z1, which partially retains a concordant U-Pb age 124 R. H. Hewins et al. despite dark speckled pyroxene alteration in the clast. b) Lindsley and Andersen (1983), and (c) compositions of Noritic clast 7533-4-C containing orthopyroxene (opx), feldspar crystal clasts. augite (aug, with opx lamellae), plagioclase, and Cr-bearing Fig. S16. Examples of disequilibrium pyroxene magnetite. c) Noritic clast 2-nor7479X; Merrillite (Me) assemblages due to oxidative alteration forming magnesian overgrown by chlorapatite (Ap); similar two-phosphate pyroxene. These are the clasts from 7533-4 shown in clasts are found in breccia matrix. Lacy monazite (Mo) in Fig. s10a and Fig. 22, plus 1-objD and 2-2229. chlorapatite. Pigeonite (pig) with augite (exsolved and as Table S1: ATEM analyses (atom%) of pyroxenes alteration, associated with pyrite) and chromite. d) and feldspars in fine-grained breccia matrix. Igneously zoned zircon clast Z14 in breccia matrix (black) Table S2: Analyses of pyroxenes in clumps and has concordant U-Pb age (Humayun et al. 2013). groundmass of clast-laden melt rocks, and shards. Fig. S11: a) The coexisting pyroxenes of several noritic Table S3: Analyses of feldspars in aureoles to clumps clasts (Table S14), including those in Figs. 12b, S10a, and and groundmass of clast-laden melt rocks, and shards. S10b, are plotted. b) The augites after recalculation with Table S4: Analyses of olivine and pyroxene in spherules. IGPET for the Lindsley and Andersen (1983) Table S5: Analyses of feldspars in spherules and veins in geothermometer. We show separately clast NWA 7533-4-C spherules. (Fig. S10a) where the large augite, unlike the small Table S6: Analyses of spinel, ilmenite, and maghemite orthopyroxene granules, is unmixed to more calcic augite in lithic and crystal clasts. and orthopyroxene lamellae. This thermometer is not Table S7: Analyses of pyroxenes in microbasaltic (melt rigorously applicable for moderate amounts of rock) clasts. nonquadrilateral components and there is a risk of Table S8: Analyses of feldpars in microbasaltic (melt underestimating the Ca content of thin augite lamellae. rock) clasts. Nevertheless clast C shows the expected behavior where Table S9: Magnetite-ilmenite geothermobarometry. igneous equilibration is about 300 °Chigherthan Table S10: Analyses of pigeonite and augite in lithic equilibration within the unmixed augite. However, most of clasts with zoned pyroxenes. the augite grains coexisting with orthopyroxene in the other Table S11: Analyses of feldspars in lithic clasts with clasts equilibrated at temperatures of 800–900 °C, little zoned pyroxenes. different from that of the augite lamellae in clast C. Table S12: Analyses of pyroxenes in noritic, Exsolution in pyroxene crystal clasts is discussed below. monzonitic, and intermediate lithic clasts. Fig. S12: Compositions of individual analyses of (a) Table S13: Analyses of feldspars in noritic, monzonitic, feldspars and (b) pyroxenes in noritic and monzonitic and intermediate lithic clasts. clasts. Compositions for clasts of intermediate properties Table S14: Analyses of pyroxene crystal clasts. are not shown, for clarity. Table S15: Pyroxenes and controls analyzed for Ni by Fig. S13: Minor elements in pyroxene crystal clasts in EMPat20kVand300nA. NWA 7533. Table S16: Coexisting pyroxenes in equilibrated two- Fig. S14: Clasts with exsolution. a) Orthopyroxene pyroxene rocks, in exsolved crystals, and in disequilibrium with augite lamellae (BSE), 7533-1 -10. b) Augite with assemblages. orthopyroxene lamellae (BSE), 7533-1-9. c) Coarsely Table S17: Analyses of feldspar crystal clasts. exsolved pyroxene 7533-7-MH3. X-ray map Mg red, Fe Table S18: Analyses of perthitic feldspar crystal clasts. blue, Ca green. Note tiny magnetite granules in host Table S19: Analyses of phosphates in lithic and crystal orthopyroxene (salmon), coarser magnetite (blue), irregular clasts. augite lamellae (greenish), and calcite veins (green). The Table S20: Analyses of oxyhydroxide in lithic and dark lines are relics of LA-ICPMS work. crystal clasts. Fig. S15: Crystal clasts; (a) compositions of host and lamellae in pyroxenes, with (b) temperature estimates after