Northwest Africa 4797: A strongly shocked ultramafic poikilitic shergottite related to compositionally intermediate Martian meteorites


  • E. L. WALTON,

    Corresponding author
    1. Department of Earth and Atmospheric Sciences, 1-26 Earth Sciences Building, University of Alberta, Edmonton, Alberta T6G 2E3, Canada
    2. Department of Physical Sciences, Grant MacEwan University, Edmonton, Alberta T5J 4S2, Canada
    Search for more papers by this author
  • A. J. IRVING,

    1. Department of Earth and Space Sciences, University of Washington, Seattle, Washington 98195, USA
    Search for more papers by this author
  • T. E. BUNCH,

    1. Department of Geology, Northern Arizona University, Flagstaff, Arizona 86011, USA
    Search for more papers by this author
  • C. D. K. HERD

    1. Department of Earth and Atmospheric Sciences, 1-26 Earth Sciences Building, University of Alberta, Edmonton, Alberta T6G 2E3, Canada
    Search for more papers by this author

Corresponding author. E-mail:


Abstract– Northwest Africa (NWA) 4797 is an ultramafic Martian meteorite composed of olivine (40.3 vol%), pigeonite (22.2%), augite (11.9%), plagioclase (9.1%), vesicles (1.6%), and a shock vein (10.3%). Minor phases include chromite (3.4%), merrillite (0.8%), and magmatic inclusions (0.4%). Olivine and pyroxene compositions range from Fo66–72,En58–74Fs19–28Wo6–15, and En46–60Fs14–22Wo34–40, respectively. The rock is texturally similar to “lherzolitic” shergottites. The oxygen fugacity was QFM−2.9 near the liquidus, increasing to QFM−1.7 as crystallization proceeded. Shock effects in olivine and pyroxene include strong mosaicism, grain boundary melting, local recrystallization, and pervasive fracturing. Shock heating has completely melted and vesiculated igneous plagioclase, which upon cooling has quench-crystallized plagioclase microlites in glass. A mm-size shock melt vein transects the rock, containing phosphoran olivine (Fo69–79), pyroxene (En44–51Fs14–18Wo30–42), and chromite in a groundmass of alkali-rich glass containing iron sulfide spheres. Trace element analysis reveals that (1) REE in plagioclase and the shock melt vein mimics the whole rock pattern; and (2) the reconstructed NWA 4797 whole rock is slightly enriched in LREE relative to other intermediate ultramafic shergottites, attributable to local mobilization of melt by shock. The shock melt vein represents bulk melting of NWA 4797 injected during pressure release. Calculated oxygen fugacity for NWA 4797 indicates that oxygen fugacity is decoupled from incompatible element concentrations. This is attributed to subsolidus re-equilibration. We propose an alternative nomenclature for “lherzolitic” shergottites that removes genetic connotations. NWA 4797 is classified as an ultramafic poikilitic shergottite with intermediate trace element characteristics.


Intensive search for meteorites in the North African and Arabian deserts and in the Antarctic icesheet has more than quadrupled the number of known Martian samples in the past decade. These meteorites are all mafic (basaltic to diabasic) and ultramafic igneous rocks, which share mineralogical, geochemical, and isotopic characteristics that indicate a common origin on a large planetary body (McSween 1994). Isotopic compositions and abundances, most notably of C, N, O, and noble gases, closely match that of the Martian atmosphere measured in situ by Viking landers, confirming Mars as the parent body (Bogard and Johnson 1983). The most abundant group of Martian meteorites––the shergottites––has expanded beyond the classic falls of Shergotty and Zagami to encompass petrologically and texturally diverse types including basaltic, olivine-phyric, olivine-orthopyroxene-phyric, and so called “lherzolitic” shergottites (McSween 1994; Goodrich 2002).

All Martian meteorites have been ejected from the near surface of Mars by impact events capable of accelerating material to velocities greater than 5 km s−1 (Melosh 1985). The extreme physical conditions of impact, or “shock” have modified the texture, mineralogy, magnetization, and isotopic composition of these rocks (Bogard and Johnson 1983; Weiss et al. 2000; Fritz et al. 2005). The general convention has been to ascribe shock features to the impact ejection event, with the exception of ancient Martian meteorite Allan Hills (ALH) 84001 (see reviews in Nyquist et al. 2001; Walton et al. 2008). However, a recent study on shergottite Dhofar 378 (Ikeda et al. 2006) has shown that an Ar-Ar age of approximately 141 Ma most likely dates a major impact on Mars, with later ejection at 3 Ma in a smaller impact event (Park et al. 2008). This raises the question whether the shock features observed in other shergottites were formed at the time of Mars ejection or by earlier impacts into the Martian surface. Martian meteorites that have been strongly heated by shock, resulting in complete melting and recrystallization of precursor igneous plagioclase, are crucial in determining the timing of shock metamorphism.

Mafic shergottites display a consistent correlation between incompatible element concentrations, most simply quantified by REE patterns (e.g., La/Yb), and oxygen fugacity, with more incompatible element-enriched lithologies (higher La/Yb) having crystallized from more oxidized magmas (Wadhwa 2001; Herd et al. 2002). This is attributable to variations in mantle source characteristics (Herd 2003), and is observed in basaltic/diabasic and olivine-phyric shergottites. However, although most poikilitic shergottites (along with diabasic specimens NWA 480/1460) fall into a compositionally intermediate group in terms of REE and radiogenic isotopic characteristics (i.e., between those groups termed depleted and “enriched”), the oxygen fugacity conditions that they experienced are poorly constrained. This is partially due to subsolidus re-equilibration, which obscures primary igneous compositions.

Martian meteorite Northwest Africa (NWA) 4797, a relatively fresh specimen with partially intact fusion crust, was found in 2001 near Missour, Morocco as a single stone weighing 15.0 g (Connelly et al. 2008). An ejection age of 3.7 Ma for NWA 4797 was reported by Huber et al. (2012) based on cosmogenic 21Ne abundances. Here, we report on the petrography, mineralogy, and chemical composition of this unpaired Martian meteorite from studies on a portion of the 3.3 g type specimen. The study focuses on shock metamorphism, REE distribution in mineral phases, and oxygen fugacity. We compare NWA 4797 with other Martian meteorites, and propose an alternative nomenclature for the “lherzolitic” shergottite subgroup that removes genetic connotations.


The nomenclature of the Martian meteorites has been a problematic issue raised by a number of workers in the field. The traditional subgroups of shergottites, nakhlites, and chassignites, derived from the names of their archetypal meteorites––Shergotty, Nakhla, and Chassigny––and so named due to the absence of an accepted parent body, continue to be used in spite of the now commonly accepted Martian origin. Whereas most Martian meteorites may be related to members of these subgroups, ALHA 84001 is a notable exception, prompting a proposal to discard the subgroup names in favor of accepted igneous rock names with “Martian” as a descriptor (e.g., Mittlefehldt 1994). In such a scheme, shergottites would be termed Martian basalts, for example, yet they do not have the same compositional characteristics of typical terrestrial basalts (which have much higher bulk Ca and Al).

The term “lherzolitic shergottite” was applied to the first discovered members of this subdivision of the shergottites, ALHA77005 and LEW 88516 (McSween 1994) as well as Yamato 793605, and has been commonly used to describe meteorites with similar textural characteristics. However, the term lherzolite is defined as an “ultramafic plutonic rock composed of olivine with subordinate orthopyroxene and clinopyroxene”; specifically, containing 40–90% olivine, >5% of each of orthopyroxene and clinopyroxene, and <10% plagioclase (Le Maitre et al. 2002). As such, some so-called “lherzolitic” shergottites have insufficient olivine and too much plagioclase to be termed lherzolites (e.g., GRV 99027; Table 1; NWA 2646, NWA 4468, and Y793605; Mikouchi 2009), and most do not contain any orthopyroxene (but instead pigeonite as the low-Ca clinopyroxene). Furthermore, they are not plutonic rocks, as their grain size is too small, and the presence of poikilitic and nonpoikilitic areas demonstrates that they did not crystallize at depth (Mikouchi 2009; see also discussion in Nyquist et al. 2009). Neither do they qualify as true basalts, having too little plagioclase (Lindstrom et al. 1994), with the exception of those with a greater proportion of nonpoikilitic lithology, and therefore more plagioclase, e.g., RBT 04262 and NWA 4468 (Mikouchi 2009). A term denoting a textural characteristic and not as saddled with genetic connotations, may be more appropriate (e.g., “pyroxene-oikocrystic shergottite”; Mikouchi 2009).

Table 1.  . Modal abundance of major and minor phases in NWA 4797 compared to lherzolitic shergottites.
MeteoriteLEW 88516ALHA77005GRV 99027NWA 1950NWA 4797NWA 4797bNWA 1950
  1. aPlagioclase is present as maskelynite in LEW 88516 and NWA 1950 and as a vesicluated plagioclase glass in ALHA77005 and recrystallized plagioclase microlites + glass in NWA 4797.

  2. bModal abundance of minerals and glass in the shock melt vein transecting the NWA 4797 thin section (see Fig. 1).

  3. A: Treiman et al. (1994); B: Walton and Herd (2007a, 2007b); C: Lin et al. (2005); D: This study; E: Gillet et al. (2005).

  4. Opaques in the shock vein are chromite + pyrrhotite. n.a. = not analyzed.

Plagioclasea79.57.45.8119.112 8
Opaques1.645. 5 
Melt inclusionsn.a.n.a.1.6n.a.1.70.4  
Shock melt7.713.712.9n.a.1.810.8  

Irving et al. (2010) suggested a classification scheme for shergottites that uses bulk major element parameters in combination with textural and trace element characteristics. Whether the meteorite is mafic, permafic, or ultramafic is determined from a plot of bulk Mg/(Mg + Fe) versus CaO; although boundaries are arbitrary, mafic rocks have lower Mg/(Mg + Fe) and higher CaO, permafic rocks have intermediate values, and ultramafic have high Mg/(Mg + Fe) and low CaO. In this scheme, the so-called “lherzolitic” shergottites range between permafic and ultramafic; those with higher plagioclase in their modes (e.g., NWA 1950) are permafic, whereas those with a greater proportion of poikilitic lithology are ultramafic. Textural qualifiers such as subophitic, olivine-phyric, and poikilitic are also applied in this scheme to designate the most diagnostic overall textural character of a specimen. Trace element characteristics, especially relative enrichment in HREE over LREE (e.g., as quantified by chondrite-normalized La/Yb ratio), provide a further descriptor: depleted, intermediate, or enriched. Under this scheme, NWA 4797 would be an ultramafic poikilitic shergottite with intermediate trace element characteristics.


One thin section and both sides of a polished thick section (approximately 11 × 9.7 × 3 mm) of NWA 4797 were studied by transmitted and reflected light microscopy. Microtextures were characterized using backscattered electron (BSE) and secondary electron (SE) images at the University of Alberta (UAb) using a JEOL 6301F Field Emission (FE) SEM using an 8 mm working distance and an accelerating voltage of 20 kV. Mineral modes for the thin section were determined by manual point counting on BSE images. Pyroxene modes were further refined by point counts made on elemental X-ray maps. These maps were obtained on a JEOL JXA-8900 electron microprobe (EM) at the UAb using an accelerating voltage of 20 kV and a beam current of 20 nA. Analyzed areas and grain dimensions were measured using the image analysis software ImageJ.

Comparisons between NWA 4797 and Allan Hills (ALHA) 77005, Lewis Cliff (LEW) 88516, and NWA 1950 are made on the basis of observations and data relating to the latter from previous studies by the authors (Herd 2001; Walton and Herd 2007b) and others (e.g., Wadhwa et al. 1999; Mikouchi 2005). Compositional data for chromite in ALHA77005 and LEW 88516 were selected from among data collected by Herd (2001) and Walton and Herd (2007b). Data for NWA 1950 are from Walton and Herd (2007b).

Terrestrial Weathering

The exterior of NWA 4797 is partially covered by fusion crust with minor iron oxide staining. The thin section and polished tile are from the stone interior; no fusion crust is present. Large, open fractures that cut across igneous minerals and shock features are attributed to terrestrial weathering. These fractures are largely devoid of alteration products such as calcium carbonate, which have been reported from other altered hot desert shergottites (Crozaz and Wadhwa 2001). As pointed out by previous authors, the products of shock melting are particularly susceptible to terrestrial alteration because of their highly vesiculated, fine-grained, and glassy nature (Zipfel et al. 2000; Wadhwa et al. 2001). However, in NWA 4797, we find that the vesicles in shock melt products are not filled by alteration minerals, nor do their associated oxide and sulfide minerals display iron oxide staining. Reddish brown rims and haloes around igneous oxide and sulfide grains are only observed where they are cut across by open fractures.

Igneous Lithologies

A cut surface on the hand specimen reveals a texture that is heterogeneous on a cm-scale, composed of interpenetrating light and dark regions with vesiculated clear to brown glass in the latter (Figs. 1 and 2). These represent the two igneous lithologies: poikilitic and nonpoikilitic.

Figure 1.

 NWA 4797 thin section in plane light (top left) and crossed polars (top right). (bottom left) False-color back-scattered electron image of a portion of the same thin section showing compositional variation in both the coarse nonpoikilitic lithology and the cross-cutting shock melt vein. Low-Ca pyroxene is mostly blue to green, high-Ca pyroxene is green to red (note the patterns of irregular zoning), olivine is light yellow, and chromite is white (pig = pigeonite, ol = olivine, chr = chromite, aug = augite). Black areas with irregular margins in the vein are primary void space. Note also the presence of stoped angular fragments of pyroxene, olivine, and chromite within the vein (white arrows). The black box shows the image location on the thin section. A sketch of the thin section showing the location and distribution of the two igneous lithologies and shock melt vein is provided (bottom right).

Figure 2.

 Transmitted light images of plagioclase occurring in interstitial regions of the nonpoikilitic lithology in NWA 4797. a) Plane light, the interstitial region has a brown rim and clear core. b) Under crossed polarized light, plagioclase microlites largely define the interior and exhibit first order interference colors. See online version for color image.

The poikilitic lithology is defined by equant, rounded olivine and chromite chadacrysts contained within, and completely surrounded by mm-size pigeonite oikocrysts with thick, irregular rims of augite (up to 9 mm; Fig. 1). The olivines (150–600 μm across) are dispersed within the oikocrysts and are not in contact with each other. Chromite chadacrysts range in size from 50 to 200 μm, and occur as individual grains or clusters. Those larger (100–200 μm) chromites contain small (1–5 μm) round magmatic inclusions of silica-rich glass.

In nonpoikilitic areas, olivine grains are larger, less rounded, and generally in contact with each other compared with those in poikilitic regions. Assemblages of pigeonite, chromite, merrillite, and glassy enclaves are interstitial to the large olivine grains. The glassy areas comprise plagioclase microlites (approximately 2–10 μm wide × 18–50 μm long) embedded in vesiculated glass. In plane light, the glassy enclaves exhibit a pale brown rim (Fig. 2a). Internal textures fall into two categories; the first (type 1) are clear colored in plane light and BSE images do not show contrast in grayscale between glass and microlites. The second (type 2) appear as clear microlites with brown glass filling the interstices. In BSE images, there is a high contrast in grayscale between microlites and glass indicating a compositional difference (see the Mineral and Glass Compositions section). In both types, plagioclase microlites display first-order interference colors (i.e., they are not isotropic; Fig. 2b). Merrillite occurs as elongate prisms up to 350 μm in length with roughly hexagonal basal sections (50–200 μm). Minor ilmenite and rare baddeleyite are encountered. K-rich mesostasis reported in other poikilitic shergottites (e.g., NWA 1950; Gillet et al. 2005), associated with merrillite in the nonpoikilitic lithology, have not been observed in NWA 4797. Igneous pyrrhotite is also not present in the sections examined; however, small grains (≤1 μm) rich in Fe and S (from EDS spectra) are found as stringers and trails in olivine and pyroxene in both lithologies.

Olivine in both lithologies contains multiphase magmatic inclusions, and round chromite (10–70 μm). The magmatic inclusions are composed of silica, pyroxene, and merrillite grains, with minor sulfide and ilmenite embedded in an alkali-rich glass, are round to ellipsoidal in cross-section, and range in apparent diameter from 30 to 300 μm. Augite and phosphate grains are typically skeletal, whereas silica exhibits blocky or rounded bleb shapes. Chromite grains within olivine lack the glassy, silicate magmatic inclusions observed in larger chromites in the poikilitic portion of the rock.

The mineral mode (in vol%) determined by manual point counting (n = 12,000) on the thin section is olivine 40.3%, pigeonite 22.2%, augite 11.9%, shock melt vein 10.3%, glassy enclaves (plagioclase microlites + glass) 9.1%, chromite 3.4%, vesicles 1.6%, merrillite 0.8%, and melt inclusions 0.4% (Table 1).

Shock Metamorphism

Shock metamorphism in NWA 4797 has been investigated by EM, FE SEM, and optical microscopy. NWA 4797 shows evidence for extensive modification of the igneous texture by shock. Shock effects are listed in Table 2 and described in the following paragraphs.

Table 2.  . Shock effects in major minerals of NWA 4797.
DescriptionOpen, irregularOpen, irregularOpen, irregularCompletely devoid of
 Heavily fractured in areas Parallel glassy filmsfractures
Polysynthetic twinning   
Description  Observed in pigeonite 
   oikocrysts only 
DescriptionLocal recrystallization along grainGrain boundary melting, mostGrain boundary melting between olivine/pigeoniteComplete recrystallization
 boundaries; most pronounced inpronounced in contact with Mobilization of plagioclase melt throughout host rock
 contact with plagioclaseplagioclase  
DescriptionStrongStrongStrongStraight extinction
Other featuresShock blackening alongShock blackeningReduced birefringenceMaskelynite not observed
  bands that follow trails  (partial isotropism) 
  and stringers of iron sulfides  shock blackening 

Shock Metamorphism Observed in Host Rock Minerals

Within both lithologies, olivine exhibits strong mosaicism and pervasive fracturing (Fig. 3a). Fractures are irregular and open; no planar elements have been observed despite a systematic search of over 30 grains. In plane light, olivine is pale brown and weakly pleochroic. In poikilitic regions, olivine grain boundaries show evidence for partial melting and local recrystallization (Fig. 3b). Open fractures radiate from the partial melt at olivine/pyroxene grain boundaries (Fig. 3b). Recrystallized olivine is colorless in plane light. In nonpoikilitic regions, olivine grains exhibit cuspate and curved margins, a texture that is most pronounced in direct contact with the quenched plagioclase melt. Those olivine margins in direct contact with quenched melt exhibit compositional zoning (Figs. 3f and 4b). Trails and stringers of iron sulfides (<1 μm) within olivine give the grains a locally blackened appearance (Fig. 3a). Some magmatic inclusions within olivine have a recrystallized and vesiculated groundmass that is consistent with shock melting.

Figure 3.

 BSE (b, d, e, f) and transmitted light images (a, c) of shock features in NWA 4797. a) Extensive fracturing observed in an olivine chadacryst. Trails and stringers of iron sulfides give the grain a black, banded appearance (labeled “FeS”). Shock blackening by iron sulfide mobilization is also noted in the surrounding pyroxene oikocrysts. b) The margin between pyroxene/olivine chadacrysts is curved and cuspate, and largely free of fractures that are abundant in the interior of both grains. At one margin (arrow), fractures are observed to nucleate from a mixed melt (flow textures) at the olivine/pyroxene contact. The adjacent chadacryst contains a recrystallized magmatic inclusion (mi – arrow). c) The relationship between pigeonite core, augite rims, and olivine–pigeonite progresses from a zone of strong mosaicism (right side of image) to polysynthetic twinning near the contact with augite. Olivine (left) also exhibits strong mosaicism. Pigeonite exhibits low, first order interference colors (see online version for color image). d) Open fractures nucleate from multiple sets of nonplanar glassy films in a pigeonite oikocryst. e) High magnification image of plagioclase microlites and interstitial glass in contact with pigeonite. f) Heavily fractured merrillite.

Figure 4.

 BSE images of plagioclase in NWA 4797. a) Overview of the nonpoikilitic lithology. In this area, the melted and vesiculated plagioclase (Gl-plag) has not migrated from the interstitial region. b) Higher magnification of an interstitial assemblage with cuspate margins between pyroxene/olivine, and compositional (grayscale) zoning along olivine and pyroxene margins in direct contact with quenched plagioclase melt (Gl-plag). Trails of iron sulfides are observed in pyroxene. c) Offshoots of plagioclase melt have invaded neighboring pyroxene and olivine. d) As offshoots of mobilized plagioclase melt migrate through olivine, the vein interiors evolve from a quenched plagioclase glass to olivine crystallites in alkali-rich glass.

Pyroxene shows deformed, recrystallized, and partially melted margins, a feature that is most pronounced in nonpoikilitic regions, as also observed for olivine. The large pigeonite oikocrysts exhibit synthetic polysynthetic twinning, intergrown on a fine scale (Fig. 3c). Reduced birefringence is observed in pigeonite oikocrysts (δ = 0.001–0.008), whereas augite exhibits typical second order interference colors (Figs. 1 and 3c). Pyroxene grains in both lithologies exhibit strong mosaicism. The oikocrysts (pigeonite cores + augite rims) are heavily fractured. One oikocryst contains a set of roughly parallel (but not straight) glass films from which open fractures nucleate (Fig. 4d). Trails and stringers of iron sulfide within pyroxene (Fig. 4b) form opaque bands in plane light.

Minor phases in NWA 4797 also show shock metamorphic effects. Chromite in the poikilitic portion exhibits irregular fractures and sharp contacts with oikocrysts; however, chromite grains in the nonpoikilitic portion exhibit curved, cuspate, and undulating margins, and are also fractured. Tiny (approximately 1 μm) vesicles have been observed along grain boundaries. Merrillite is heavily fractured and exhibits strong mosaicism (Fig. 3f). Those merrillite grains completely surrounded by quenched plagioclase melt have curved, cuspate margins.

Igneous plagioclase, originally present in the nonpoikilitic lithology as approximately 150–800 μm size grains interstitial to olivine and pyroxene, has been completely shock melted to form highly vesiculated, quench-crystallized enclaves (Fig. 4). The plagioclase component in NWA 4797 is now present as minute fibrous needles, often radiating and typically 20–100 μm long and <10 μm wide (hereafter referred to as plagioclases microlites; Figs. 3e and 4). Glassy rims are observed around the margins of the quench-crystallized plagioclase. Offshoots of this melt extend approximately 1000 μm from interstitial regions along grain boundaries and as veins (2–20 μm) that cut across neighboring minerals (Figs. 4c and 4d). With increased volumes of melting, textures progress from thin (1–20 μm) stringers of plagioclase glass (Figs. 4c and 4d) to larger veins and pools of melt (25–200 μm) containing olivine and plagioclase crystals, and finally to unmelted angular olivine and pyroxene clasts invaded and surrounded by quench-crystallized plagioclase melt (Fig. 5).

Figure 5.

 BSE image showing igneous olivine surrounded by pools of shock-melted and quenched plagioclase glass.

Shock Melt Vein

A single, almost straight vein (0.9–1.8 mm apparent width) of shock-melted material cuts across the stone (Fig. 1). In thin section, the vein has an opaque brown-gray color. Volume % abundances of phases within the vein are as follows; olivine (53%), pyroxene (29%), interstitial alkali glass or plagioclase microlites (12%), chromite (5%), and pyrrhotite (1–2%). This vein is distinct from those melt offshoots described in the previous section (Fig. 5). The former can be traced to quenched plagioclase melt in the nonpoikilitic lithology.

Vein microtextures are shown in Fig. 6. In some areas, a granular texture is observed, with crystals embedded in a matrix of fine-grained plagioclase (as opposed to glass; Fig. 6a). A gradation in pyroxene and olivine grain size, shape, and distribution is observed within the shock vein; skeletal grain shapes (5–15 μm × 30–110 μm) characterize the interior, whereas areas near the margin exhibit equant, euhedral olivine crystals (10–25 μm) and pyroxene laths (5–20 μm ×60–90 μm) (Figs. 6b and 6c). An increase in nucleation density, associated with the shock vein margin, is observed as an increase in the number of crystals/area based on visual estimates (Fig. 6b).

Figure 6.

 BSE images of the shock melt vein. a) The vein interior characterized by olivine and pyroxene in a groundmass of plagioclase microlites. b) Olivine shape as a function of vein width. Skeletal shape observed in the vein interior transition to equant, euhedral grains near the margin. c) Pyroxene laths nucleating from the host rock minerals into the vein. d) At the host rock/vein margin, plagioclase melt (now glass + microlites) intermingles with the vein, indicating that they were either molten, or at least plastic, at the same time. e) Quench crystallized plagioclase melt (Gl-plag) from the nonpoikilitic lithology intermingles with the groundmass of the shock melt vein, further supporting textural relationships in (d) that these formed in the same thermal event. f) Fragments of host rock (hr) minerals are entrained within the vein.

Chromite dendrites range from 1 to 4 μm and iron sulfide spheres are ≤1 μm. Olivine and pyroxene grains in the vein interior are large enough for investigation by optical microscopy. Both minerals are colorless in plane light, are devoid of fractures, and exhibit straight extinction under crossed polars. Interference colors are slightly lowered (δ = 0.08–0.014, olivine; δ = 0.08–0.011, pyroxene), most likely a result of the small size of the crystals compared with the thickness of the section.

The opposing walls of the vein are congruent; removing the vein filling and piecing the igneous host rock together forms a coherent unit (Fig. 1). The vein/host rock contact is gradational. Pyroxene laths nucleate from the margin of host rock minerals into the vein (Fig. 6c). Quenched plagioclase melt in the nonpoikilitic lithology and plagioclase microlites from the shock melt vein are intergrown where these two phases are in contact (Figs. 6d and 6e). Fragments of host rock and mineral fragments were stoped and included into the vein, with attendant partial melting or marginal reaction (see host rock fragments, Fig 1; X-ray elemental map inset; Fig. 6f).

Bulk Rock, Mineral, and Glass Compositions and Chemistry

Samples and Analytical Methods

Major and minor elemental abundances were measured using a JEOL JXA-8900 EM at UAb and a JEOL 733 EM at the University of Washington. These studies incorporated BSE imagery, elemental X-ray mapping, and quantitative analysis of major and minor elements using wavelength dispersive spectrometry (WDS). An accelerating voltage of 15 or 20 kV and a beam current of 15 or 20 nA were used for WDS and X-ray mapping. Well-known natural and synthetic minerals were used as standards. A 1 μm (focused) or 5–10 μm (defocused) beam size was used for mineral and glass analyses, respectively. Representative mineral and glass analyses are listed in Tables 3 and 4. Detection limits (ppm) for the EM are as follows; Cr 152, Na 107, Ca 67, Si 94, Mn 63, Ti 103, Mg 74, K 56, Al 64, Fe 95, P 145, S 185, and Ni 81. Ferric iron in chromite and ilmenite was calculated on the basis of stoichiometry (Carmichael 1967). The bulk composition of crystallized plagioclase melts was estimated using preprogrammed grid overlays with 10 μm spacing and a defocused beam (10 μm spot size) (Table 5). Only those oxide wt% totals falling within 100 ± 5 were accepted.

Table 3.  . Representative analyses of major and minor mineral phases in NWA 4797.
wt% oxideOlivinePyroxeneOxidesPlagioclaseMerrShock melt veinPlag Offshoot Glass
  1. Poik = poikilitic lithology, N-Poik = nonpoikilitic lithology, Chr = chromite enclosed in oikocryst, Pig = pigeonite, Aug = augite, Usp = Ti-rich chromite in N-Poik, Ilm = ilmenite, S.V. = shock vein, Merr = Merrilite, Glass types 1 & 2 (see text for details), Ol = olivine, Plag Offshoot = pools of plagioclase melt mobilized from N-Poik (see text for details), b.d. = below detection limits.

Fe2O3 __ __ __ __ __ __ 2.595.592.54 __ __ __ __ __ __ __ __ __
V2O3 __ __ __ __ __ __ 0.590.660.26 __ __ __ __ __ __ __ __ __
K2Ob.d.b.d.b.d. __ b.d.b.d.b.d.0.150.27
P2O5b.d.0.06 __ __ __ __ __ __ __ 0.03b.d. __ __ 46.180.400.601.290.02
SO30.040.03 __ __ __ __ __ __ __ b.d. __ __ __ b.d.b.d.b.d.b.d. __
Fo/En716868516649        7245  
Wo  934934         40  
An          51       
Or          2       
Table 4.  . Representative analyses of minerals in magmatic inclusions.
wt% oxideMelt inclusions
  1. Aug = augite, Phos = phosphate, b.d. = below detection limits.

P2O50.050.460.59 __ b.d.46.00
SO3 __ 0.13 __ __ __  
Fo/En 684349  
Wo  4234  
Table 5.  . Bulk composition of five individual glass + plagioclase assemblages in NWA 4797 estimated from EM defocused beam analyses.
wt% oxideEnclave 1 (type 2)Enclave 2 (type 2)Enclave 3 (type 1)Enclave 4 (type 1)Enclave 5 (type 2)
  1. “Enclave” refers to plagioclase + glass + rim assemblages found in the nonpoikilitic lithology. Five enclaves were analyzed for their bulk composition by averaging individual defocused beam measurements for wt% oxides. See text for details.

  2. Type 1 and type 2 refer to the two distinct glass types that are interstitial to the plagiclase crystals within each enclave.

  3. Type 1 glasses appear homogeneous under optical investigation (clear/colorless), whereas type 2 glasses are brown colored.

  4. b.d. = below detection limits, n = number of analyses averaged to estimate the bulk composition.

Cr2O3 __ b.d.–0.11 __ b.d.–0.04 __ b.d.–0.08 __ b.d.––0.13
MnO0.110.07––0.09 __ b.d.–––0.50
P2O5 __ b.d.–0.11 __ b.d.–0.02 __ b.d.–0.03 __ b.d.–0.021.320.12–8.21
NiO __ b.d.b.d.b.d.–0.01 __ b.d.–0.02 __ b.d. __ b.d.–0.02
SO3 __ b.d.b.d.b.d.–0.01 __ b.d. __ b.d. __ b.d.–0.50
n39 40 85 50 36 
Cations 8ox5.06–5.11 5.02–5.09 5.01–5.05 5.00–5.09 4.56–5.15

Trace element concentrations of minerals, glasses, and the shock melt vein were obtained using Laser Ablation Inductively Coupled Plasma Mass Spectrometry (LA-ICP-MS) at UAb. Laser ablation was carried out using a UP213 (213 nm) Nd:YAG laser with an aperture imaging system (New Wave Research), with an energy density of 15 J cm−2, a 5 Hz repetition rate and with 75 or 150 μm spot sizes focused on the sample surface. Trace element detection was done using a Perkin Elmer Elan 6000 ICP-MS at 1300 W power with He carrying gas at 0.5 L min−1. The NIST 612 basaltic glass standard was analyzed under the same conditions and used to determine concentrations and detection limits for the NWA 4797 analyses.

Mineral and Glass Compositions


The range of olivine compositions spans Fo66–79. Chadacrysts (Fo68–72; avg Fo70, n = 111, FeO/MnO 48.1–53.9) are slightly more magnesian than olivine in the nonpoikilitic portion of the rock (Fo66–70; mean Fo68, n = 121, FeO/MnO 48.6–56.0). Core-to-rim transects across olivine in both lithologies show relatively flat zoning profiles. Olivine in direct contact with quenched plagioclase in the nonpoikilitic lithology exhibits higher Mg-contents (rim Fo73 versus Fo68 in olivine away from plagioclase glass contact). Skeletal olivine in the shock melt vein are zoned from cores of Fo73–79 to more ferroan rims (Fo69–72), contain significant P2O5 (0.11–0.55 wt% oxide) and CaO (0.27–0.85 wt% oxide), and display a wider range in FeO/MnO ratios (=39.5–52.3) compared with igneous olivine (FeO/MnO = 48.1–56.0).


Pyroxenes in NWA 4797 include augite and pigeonite (Fig. 7). The mm-size oikocrysts show chemical zoning from cores of En65–74Fs19–27Wo6–13 (FeO/MnO = 23.9–29.5) with thick, irregular rims of En46–60Fs14–22Wo23–40 (FeO/MnO = 20.7–25.9). These augite rims are in direct contact with olivine from the nonpoikilitic lithology or Mg-rich pigeonite (En60–66Fs24–28Wo7–15; FeO/MnO = 24.3–29.0), intergrown with nonpoikilitic olivine. Pyroxene in the interstitial assemblages exhibits patchy zoning from pigeonite/subcalcic augite (En58–68Fs22–28Wo6–19, FeO/MnO = 28.1–30.4), to augite (En49–51Fs15–17Wo34–35; FeO/MnO = 23.9–29.5). There is no evidence for exsolution lamellae in pyroxenes from either lithology. Augite has crystallized in the shock melt vein as En44–51Fs14–18Wo30–42 (FeO/MnO = 25.1–29.8).

Figure 7.

 Pyroxene quadrilateral, showing the distributions of pyroxene spot analyses for NWA 4797. In general, pyroxene compositions from the nonpoikilitic lithology are more ferroan compared with those of the oikocrysts. The range of pyroxene compositions for ALHA77005 and LEW 88516 (Harvey et al. 1993) are shown for comparison and these compositions overlap with those for NWA 4797.


The plagioclase component in NWA 4797 is the product of a quench-crystallized plagioclase melt. Abundant vesicles indicate volatile release. The glassy enclaves possess thin rims that have nonstoichiometric plagioclase compositions with significant FeO and CaO contents. Plagioclase microlites are stoichiometric with compositions (An43–52Ab46–57Or0.5–0.8) indistinguishable from those of plagioclase in the shock melt vein. The glasses interstitial to plagioclase microlites are of two varieties; type 1 are K-, Na-rich (1.38–3.32 wt% K2O; 5.14–6.99 wt% Na2O) and FeO-, MgO-poor (3.69–7.89 wt% FeO; 0.49–2.60 wt% MgO) glasses, and type 2 are relatively alkali-poor glasses with high FeO and MgO contents (29.41–32.92 wt% FeO; 27.69–30.31 wt% MgO). Cr-, Fe-rich oxides (≤1 μm) are intergrown with the alkali-poor glasses (EDS only).

The bulk compositions of five glassy enclaves occurring in interstitial regions of the nonpoikilitic lithology (plagioclase microlites + interstitial glass + rims) are given in Table 5. The individual enclaves contain either type 1 or type 2 glasses described in the previous paragraph. The bulk composition of one glassy enclave exhibits elevated phosphorus (average P2O5 1.32 wt% oxide; see “Enclave 5” in Table 5).

The grains in the now-crystallized offshoots extending from plagioclase in the nonpoikilitic lithology were too small for quantitative analysis. Imprecise analyses yielded Si-, Fe, and Mg-rich compositions with a few wt% CaO and Al2O3, which could be analytical contamination from the surrounding alkali-rich glass, or indicate that the mineral is pyroxene rather than olivine.


The range of chromite compositions in NWA 4797 is shown in Fig. 8, and is similar to the range observed in poikilitic shergottites such as ALHA77005, LEW 88516, and NWA 1950 (Goodrich et al. 2003; Walton and Herd 2007b). Chromites in the poikilitic areas in NWA 4797 are low in Ti and vary primarily in Cr and Al, with Cr# (=molar Cr/(Cr + Al)) ≤0.88. Chromites in nonpoikilitic areas vary continuously in ulvöspinel content, up to Usp 59%, where Usp = molar 2Ti/(2Ti + Cr +Al + Fe3+). These most evolved chromites nevertheless contain at minimum 27% of the chromite component (molar Cr/(2Ti + Cr + Al + Fe3+). Magnetite contents (molar Fe3+/(2Ti + Cr + Al + Fe3+) are approximately 3–7% in both poikilitic and nonpoikilitic areas, remaining constant with increasing Usp content. The lack of increase in magnetite content suggests that there was no significant change in redox conditions during crystallization between the poikilitic and nonpoikilitic assemblages. Chromites within the shock melt vein are polycrystalline or dendritic; the latter are interpreted as having crystallized from the melt. Both dendritic and polycrystalline chromites have compositions that overlap with the range of compositions observed in the host rock. Representative chromite compositions are listed in Table 3. The overlap in chromite compositions for NWA 4797 and other poikilitic ultramafic shergottites is evident in Fig. 8; chromite from these samples share common compositions at high Cr (Cr# = 0.86–0.89) and high Ti content (Usp = 56–59%), with very little difference in compositional trends.

Figure 8.

 Compositions of chromites in NWA 4797 (gray squares) in the system chromite (molar Cr/[Cr + 2Ti + Al])-ulvöspinel (molar 2Ti/[Cr + 2Ti + Al])-spinel (molar Al/[Cr + 2Ti + Al]). Also shown is the range of compositions of chromite in ALHA77005 (gray region), LEW 88516 (open triangles) and NWA 1950 (open squares).


Compositions of ilmenite grains show little variation, containing 67–75% FeTiO3 (ilmenite, Il), 0.8–4.3% Fe2O3 (hematite, Hm), 22–28% MgTiO3 (geikielite, Gk), and 1.2–1.5% MnTiO3 (Pyrophanite, Py) (Table 3). The ilmenite in NWA 4797 is the most magnesian found in the shergottites. By comparison, the composition of ilmenite in other similar shergottites is ALHA77005: 73–77% Il, 1.1–3.1% Hm, 20–23% Gk, 1.5–1.7% Py (McSween et al. 1979a, 1979b; Treiman et al. 1994; Herd 2001); LEW 88516: 77–81% Il, 0.4–4% Hm, 17–18% Gk, 1.2–1.7% Py (Harvey et al. 1993; Treiman et al. 1994; Gleason et al. 1997); NWA 1950: 75–76% Il, 2.3–5.1% Hm, 18–20% Gk, 1.7–1.9% Py (Walton and Herd 2007a, 2007b); GRV 99027: 74–75% Il, 0% Hm, 23–24% Gk, 1.3% Py (Lin et al. 2005); Yamato 793605: 79–83% Il, <1% Hm, 15–20% Gk, 1.2–1.7% Py (Ikeda 1997; Mikouchi and Miyamoto 1997; Herd 2001).

Trapped Igneous Magmatic Inclusions in Olivine and Chromite

Magmatic inclusions in chromite are CaO-, MgO-, FeO-rich glasses. These inclusions are typically only a few microns in diameter, which limits quantitative analysis. One analysis from a larger chromite inclusion was obtained, as an average of the two glasses visible in BSE images (Table 4). Melt inclusions in olivine are polyphase containing augite ± silica-rich phase ± pyrrhotite ± phosphate ± augite rim, embedded in a Si- or a Si-, Al-rich glass. Tiny oxide and sulfide minerals have also been observed in the inclusions as pyrrhotite, magnetite, and ilmenite, although their small size precluded reliable quantitative analysis. Augite is enriched in Al2O3 (4.12–6.72 wt% oxide) compared with interstitial or oikocrystic pyroxenes and is slightly less calcic than the pyroxene mantling the magmatic inclusion. The silica-rich phase contains appreciable K2O (0.73–1.38 wt%) and Al2O3 (1.83–3.21 wt%).

Other Mineral Phases

Representative analyses of merrillite are also shown in Table 3. Merrillite contains 2.21–2.95 wt% FeO and 3.52–4.31 wt% MgO. Na2O wt% oxide contents vary from 0.45 to 0.70 wt%. Apatite has not been encountered. Igneous pyrrhotite has not been observed in our sections of NWA 4797. Instead, bright grains (in BSE imaging mode) are present exclusively as minute (≤1 μm) grains and stringers within olivine and pyroxene. The small size of this phase precludes quantitative analyses; however, EDS spectra show that they are rich in Fe and S (most likely pyrrhotite).

Modal Recombination

As a bulk composition cannot be determined on a representative powder of such a small (15.0 g), relatively coarse grained and heterogeneous specimen, we have calculated one using the mineral mode (n = 12,000) combined with EM analyses of constituent phases. The distribution of pigeonite and augite was then determined by manual point counts performed on elemental X-ray maps. This technique has been successfully applied to other Martian meteorites (e.g., SaU 094, Gnos et al. 2002; Yamato 980459, Greshake et al. 2004). The minerals and glasses in the shock melt vein were not included in the modal recombination count because of the heterogeneity in glass composition and mineral distribution in this phase. Vesicles were also excluded from the bulk calculation. The procedure was then to remove these two phases (vein + vesicles) from the mode, normalizing olivine, pigeonite, augite, quenched plagioclase glass, chromite, and merrillite to 100. Our estimate of plagioclase composition was obtained for the type 1 glasses, interpreted to more closely approximate the composition of the original igneous plagioclase. The results of the calculation are given in Table 6, with the bulk composition of ALHA77005 and LEW 88516 shown for comparison.

Table 6.  . Bulk composition of NWA 4797 compared with ALHA77005 and LEW 88516.
wt% oxideNWA 4797  
CalcaINAAbALHA 77005cLEW 88516c
  1. aBulk composition of NWA 4797 calculated by modal recombination.

  2. b Irving et al. (2008).

  3. c Treiman et al. (1994).

SiO241.6 40.844.5
TiO20.4 0.610.28
Al2O33.7 3.82.9
Cr2O32.0 1.090.65
MnO0.5 0.480.49
MgO25.6 2825.8
CaO4.7 2.94.5
K2O0.1 0.040.03
P2O50.4 0.340.49
Total99.6 100.33100.76

Centimeter-scale heterogeneity has been noted for poikilitic shergottites (McSween et al. 1979a, 1979b); therefore, the bulk composition calculated in this study is dependent upon the proportions of poikilitic versus nonpoikilitic lithology; because of our small sample size, it is expected that the bulk composition of NWA 4797 might vary. However, we note that the major element abundances from our calculation are similar to those of other similar shergottites (Table 6). In addition, our calculated bulk composition yielded identical FeO results to those values measured by INAA (19.9 wt%; Irving et al. 2008).

Mineral and Glass REE Contents

Average concentrations of REE in the mineral and glass obtained by LA-ICP-MS are given in Table 7, and chondrite-normalized plots are shown in Fig. 9. Only low-Ca (pigeonite) pyroxene was analyzed, in oikocrysts or in interstitial (nonpoikilitic) areas. The concentrations of REE in a mineral or phase reflect the average of two or three analyses. Uncertainties on concentrations reflect the greater of 1-sigma uncertainties from counting statistics (as calculated by the Glitter software) or the 1-sigma standard deviation of the mean of the averaged analyses. The exceptions are the patterns obtained from analysis of the larger areas of recrystallized plagioclase (Fig. 5), where results from the two analyses are both plotted to show the range observed.

Table 7.  . REE Concentrations in NWA 4797 mineral and glass phases
 PyroxenePlagioclaseMerrilliteShock melt veinPools of plagioclase melt
Oikocryst (pigeonite)Interstitial (pigeonite)ppmCI1ppmCI1ppmCI1ppmCI1ppmCI1
Mode (%) 33.9   9.1 0.8 10.8    
La/Yb 0.30 0.65 3.74 0.70 0.67 3.38 0.39
 Parental melt (poik px)Parental melt (mask)Parental melt (merrillilte)Bulk comp (mode)
  1. CI1 values from Palme and Jones (2004).

  2. D values from Lundberg et al. (1990) for low-Ca pyroxene are shown in italics, plag (lunar values), and merrillite.

  3. D values from Blinova and Herd (2009) for opx; D(Eu) is 0.010 at <IW + 1.8.

  4. “Plagioclase” refers to quench-crystallized plagioclase melts in the nonpoikilitic lithology, which are now plagioclase microlites + glass.

Mode (%)       
Ce 0.002 142.240.03670.01264.593.23
Pr 0.005 69.160.03171.28264.483.04
Tb 0.029 21.060.01380.002512.047.14
Dy 0.037 24.510.0177.762214.667.52
Ho 0.047 18.510.00995.041913.526.24
Er 0.059 16.680.009118.471516.176.58
Tm 0.075 20.570.006247.401117.935.71
Lu 0.095 21.55    5.08
La/Yb 17.93 0.46 0.220.73
Figure 9.

 Chondrite-normalized REE concentrations of (a) mineral phases and (b) shock melt vein and pools of quenched plagioclase melt in NWA 4797. Error bars represent 1σ uncertainty, which is the greater of the 1σ error reported by the Glitter software or the average 1σ standard deviation of multiple analyses. Error bars for Nd, Sm, Eu, and Gd in poikilitic pyroxene (a) represent maximum values, as these analyses were at or below detection limits.

Concentrations of REE in poikilitic low-Ca pyroxene have high uncertainties. Many analyses are close to or coincident with detection limits. In the case of Gd, all three analyses were below detection limits. A concentration consistent with that expected for a regular REE pattern was selected for plotting in Fig. 9a, but error bars reflect the maximum value for this element. In spite of the uncertainties, it is evident that poikilitic pyroxene has higher LREE than in other similar shergottites. For example, La concentrations in ALHA77005, LEW 88516, and Yamato 793605 poikilitic pyroxene are on the order of <0.01 to 0.1 X chondritic (Lundberg et al. 1990; Harvey et al. 1993; Wadhwa et al. 1994), compared to 0.1 to 1 X chondritic in NWA 4797. Poikilitic and nonpoikilitic pyroxene in ALHA77005, LEW 88516, and Yamato 793605 have linear REE patterns that slope down to lowest concentrations for LREE. In contrast, poikilitic pyroxene in NWA 4797 has an REE pattern that curves upward from Nd to La (Fig. 9b). Nonpoikilitic low-Ca pyroxene in NWA 4797 has an approximately flat REE pattern with concentrations 1–2 X chondritic. Some component of the LREE enrichment in NWA 4797 poikilitic pyroxene may be attributable to terrestrial weathering (Floss and Crozaz 1991; Harvey et al. 1993; Wadhwa et al. 1994); however, we observe no Ce anomalies that are typical of terrestrial REE remobilization.

Concentrations of REE in quench-crystallized plagioclase show an LREE-enriched pattern with a positive Eu anomaly (Fig. 9b; Eu/Eu* approximately 13, where Eu is the measured chondrite-normalized value and Eu* is the value interpolated between Sm and Gd). REE concentrations are higher than in plagioclase/maskelynite from other similar shergottites. For example, plagioclase glass in ALHA77005 contains REE concentrations between 0.1 and 1 X chondritic, with a slight LREE enrichment relative to HREE (La/Tm approximately 2; Lundberg et al. 1990; their Fig. 7); in contrast, recrystallized plagioclase in NWA 4797 contains between 1 and 10 X chondritic concentrations of REE, with a larger enrichment of LREE relative to the HREE (La/Tm approximately 3; Fig. 9b).

Ca-phosphate, especially merrillite, is the primary carrier of REE in Martian meteorites (Lundberg et al. 1990; Wadhwa et al. 1999). In NWA 4797, as in other similar shergottites, the chondrite-normalized REE pattern of merrillite is flat to decreasing from La to Pr, increasing to Gd with a slight negative Eu anomaly, and then decreasing from Gd to Lu (Fig. 9a). Other workers have noted that merrillite in some weathered Antarctic meteorites shows obvious signs of alteration, such as partial decomposition (Mittlefehldt et al. 1997; Wadhwa et al. 1999). In NWA 4797, merrillite is heavily fractured and exhibits strong mosaicism consistent with shock rather than terrestrial weathering. The glassy shock melt products and oxide minerals associated with merrillite are particularly susceptible to terrestrial alteration; however, these glasses show no obvious signs of terrestrial alteration such as calcium carbonate formation in vesicles. Oxides are largely devoid of iron oxide staining. Therefore, merrillite in this meteorite may reliably preserve its igneous REE abundances.

The shock melt vein in NWA 4797 has an REE pattern that closely mimics that for ALHA77005 and Yamato 793605 (and of NWA 4797 itself), wherein the REE pattern is essentially flat from La to Pr, then smoothly increases to Gd, and decreases by a factor of approximately 2 from Gd to Lu (Fig. 9b). REE concentrations are bracketed between 1 and 10 X chondritic. Larger areas of quench-crystallized plagioclase (Fig. 5) are variable, as shown by the two analyses in Fig. 9b. One analysis is slightly depleted in LREE, similar to the shock melt vein, whereas the other is enriched in LREE, consistent with a greater plagioclase component. This variation is attributable to varying proportions of minerals and plagioclase melt, as discussed below.

Mineral modes were used to calculate the whole rock REE composition. The proportions of pyroxene (34.1%), plagioclase (9.1%), shock melt vein (10.3%), and merrillite (0.8%) were used in the calculation, normalized to 100 under the assumption that the REE concentrations in the other phases, such as olivine and Fe-Ti oxides, are negligible. The resulting pattern is shown in Fig. 10.

Figure 10.

 Chondrite-normalized REE concentrations in melts in equilibrium with mineral phases in NWA 4797, calculated using the D values shown in Table 7. Also shown are the concentrations in the shock melt vein and the calculated whole rock REE pattern based on reconstruction using the modal mineralogy.

The REE compositions of melts in equilibrium with mineral phases in NWA 4797 were calculated for comparison with results from other similar shergottites (Fig. 10), using the partition coefficient (D) values from Lundberg et al. (1990) for plagioclase and merrillite. D values for low-Ca pyroxene were taken from the experimental partitioning study of Blinova and Herd (2009) on shergottite Yamato 980459 for orthopyroxene at 1.6 GPa and 1400 c and oxygen fugacity ≤IW + 1.8 for La, Nd, Sm, Eu, Gd, and Yb. This DEu value is appropriate given the suspected oxygen fugacity conditions of crystallization of NWA 4797 (=IW + 0.6). The remaining D values for pyroxene were taken from Lundberg et al. (1990). The melt in equilibrium with merrillite most closely approaches the bulk rock REE pattern calculated from the mode (as well as the shock melt vein composition; Fig. 10). However, the melt in equilibrium with plagioclase has a broadly V-shaped REE pattern and is much higher in REE than the whole rock. Similarly, the melt in equilibrium with poikilitic low-Ca pyroxene is highly enriched in LREE relative to HREE, with a C1-normalized La/Yb ratio of 18. That these melts do not parallel the whole rock or each other indicates that the pyroxene and plagioclase have been affected by remobilization of REE. Processes responsible for the observed REE patterns are considered in the Discussion section.

Oxygen Fugacity

The compositions of relevant phases were used to estimate the redox conditions of crystallization of NWA 4797, through calculation of oxygen fugacity (fO2) using applicable oxythermobarometers. Olivine, low-Ca pyroxene, and chromite found in the poikilitic areas were used to estimate fO2 at the initial stages of crystallization with the olivine-pyroxene-spinel (Ol-Px-Sp) oxybarometer. The same oxybarometer has been used previously with a similar assemblage (e.g., Herd 2006), with low-Ca pyroxene compositions having Wo < 10%. The most magnesian low-Ca pyroxene (Wo6En74Fs20) and olivine (Fo72) in NWA 4797 are not in equilibrium, having a calculated olivine/pyroxene Fe-Mg KD of 1.4, in contrast to the olivine/pyroxene Fe-Mg KD of 1.2 of Longhi and Pan (1989) for shergottitic melts. The olivine is too ferroan relative to the pyroxene, attributed to subsolidus Fe-Mg exchange. Olivine chadacrysts have most likely lost Mg, as they contain more Mg (approximately 38 wt% MgO) than the host oikocrysts (approximately 30 wt% MgO). If we assume that the olivine has lost Mg, with pyroxene largely unchanged, the olivine composition in equilibrium with the pyroxene is Fo75.5. This assumption is justified by the complete enclosure of the smaller olivine chadacrysts within pyroxene oikocrysts. The observed pyroxene and modified olivine compositions were selected for oxybarometry, along with chromite having the highest Cr#, and lowest Fe#, following the criteria of Goodrich et al. (2003). Similar to their for ALHA77005, it was found that the chromite grains with the highest Cr# and lowest Fe# were found enclosed in the low-Ca pyroxene oikocrysts, consistent with lesser degrees of Fe-Mg re-equilibration relative to those chromites enclosed in olivine. Calculations were carried out using a range of compositions from chromite grains enclosed in olivine and pyroxene. Oxygen fugacity was calculated using the thermodynamic models developed by Sack and Ghiorso (1989, 1991b, 1991c, 1994a, 1994b, 1994c) that are integrated into an online calculator for Ol-Px-Sp oxybarometry found on the CT Server ( Peslier et al. (2010) show that the results of the application of this version of the Ol-Px-Sp oxybarometer are lower than the method of Herd (2003) by about 0.5 log units, a difference that is within the uncertainty of the Ol-Px-Sp method (Wood 1991). Furthermore, Peslier et al. (2010) demonstrate that this model better accounts for olivine-spinel Fe-Mg exchange effects on oxygen fugacity, relative to the procedure outlined by Herd et al. (2002). Oxygen fugacity estimates are summarized in Table 8. Average oxygen fugacity is 2.9 (±0.2) log units below the Quartz-Fayalite-Magnetite (QFM) buffer (formulation of Herd 2008). Temperatures, derived from olivine-spinel equilibria (Sack and Ghiorso 1991a), range from 1016 to 1265 °C, with an average of 1151 °C (±100; 1σ deviation of the mean, n = 8).

Table 8.  . Summary of oxygen fugacity estimates obtained for NWA 4797.
Spinel T (°C)Log10fO2Log10fO2 (QFMa)Log10fO2 (NNOb)Log10fO2 (IWc)Cr#Fe#
  1. Calculations for poikilitic assemblage involved Fo75.5 olivine and Wo6En74Fs20 poikilitic pigeonite.

  2. Calculations for nonpoikilitic assemblage involved Fo68 olivine and Wo7En67Fs26 pigeonite and the most Ti-rich chromite (composition given in Table 3).

  3. aRelative to the Quartz-Fayalite-Magnetite buffer as defined by Wones and Gilbert (1969).

  4. bRelative to the Nickel-Nickel Oxide (NNO) buffer as defined by Herd (2008) using the data of O’Neill and Pownceby (1993).

  5. cRelative to the Iron-Wüstite (IW) buffer as defined by Herd (2008) using the data of O’Neill and Pownceby (1993).

Poikilitic assemblage
Average1151 −2.9−3.70.6  
Nonpoikilitic assemblage

The oxygen fugacity of the latter stages of crystallization was estimated using compositions of more evolved minerals. We chose low-Ca pyroxene of composition Wo7En67Fs26, olivine of composition Fo68, which is in equilibrium with the pyroxene on the basis of KD = 1.2 (Longhi and Pan 1989), and the most Usp-rich chromite. The calculation yields fO2 = QFM − 1.7 at an Ol-Sp temperature of 1025 °C.

The compositions of the Ti-rich chromite and ilmenite were utilized to investigate the potential for Fe-Ti oxybarometry to constrain the latter stages of crystallization. The two oxides are not in Mn-Mg exchange equilibrium, falling outside two standard deviations of the Bacon-Hirschmann line (Bacon and Hirschmann 1988; Ghiorso and Evans 2008). Applications of two different oxythermobarometers further support this: the most Usp-rich chromite and a series of 17 ilmenite analyses yields an average Fe-Ti exchange temperature of 1237 ± 52 °C and fO2 near QFM using the oxythermobarometer of Ghiorso and Evans (2008). The high Fe-Ti temperature suggests that the two oxides are not in equilibrium (Ghiorso and Evans 2008), due to differences in the timing of crystallization or subsolidus exchange. Application of the Ca-QUIlF oxythermobarometer to the same dataset yields an average temperature of 586 ± 20 °C (based on Fe-Ti exchange) and fO2 about 0.5 log units below QFM. The low temperature is consistent with disequilibrium. Therefore we conclude that the Fe-Ti oxides cannot be reliably used to estimate late-stage fO2 for NWA 4797. Most likely, ilmenite crystallization began after chromite crystallization had ceased.

A comparison of the early and late Ol-Px-Sp fO2 estimates shows that the amount of oxygen fugacity increase between the earliest and later formed assemblages is relatively small, at most 1 log unit. This increase is not reflected in the ferric iron contents of Ti-rich chromite grains or rims, which have Fe3+/(Fe3 + 2Ti + Cr + Al) contents of no more than 7%. It is likely that a greater contrast in fO2 between cores and rims is required to cause a significant jump in Fe3+. For example, in the NWA 1068/1110 enriched olivine-phyric shergottite, chromite cores formed under fO2 = QFM − 2.5 and have Fe3+/(Fe3+ + 2Ti + Cr + Al) <10%, whereas chromite rims formed under fO2 = QFM + 0.5 and have Fe3+/(Fe3 + 2Ti + Cr + Al) contents of over 40% (Herd 2006).

The same criteria for selection of the earliest co-crystallizing olivine, pyroxene, and chromite were applied to ALHA77005, LEW 88516, and NWA 1950 to estimate their magmatic oxygen fugacity conditions. The fO2 of ALHA77005 was estimated using the olivine and pyroxene compositions selected by Goodrich et al. (2003) as representative of the most magnesian olivine (Fo76) and orthopyroxene (En77Fs20Wo3) in equilibrium with one another, along with the compositions of four early crystallized chromite grains. The estimate of Goodrich et al. (2003) was also recalculated using an updated version of the Ghiorso-Sack Ol-Sp geothermometer. Ol-Sp temperatures range from 1151 to 1259 °C. The average oxygen fugacity from all estimates is QFM − 3.7 (±0.5; 1σ deviation of the mean, n = 5), with an average temperature of 1204 ± 42 °C.

Oxygen fugacity of LEW 88516 was estimated using the representative orthopyroxene oikocryst composition (En77Fs20Wo3) reported by Treiman et al. (1994), a fictive olivine composition (Fo76) in equilibrium with the orthopyroxene on the basis of ol/px Fe-Mg KD = 1.2 (Longhi and Pan 1989), and the compositions of three early crystallized chromite grains. Olivine of composition Fo76 has not been reported in LEW 88516 (maximum is Fo70; Treiman et al. 1994). We assume that olivine has undergone subsolidus Fe-Mg exchange that has made the olivine more ferroan. Ol-Sp temperatures range from 929 to 1202 °C. The average oxygen fugacity is QFM − 2.4 (±0.1; 1σ deviation of the mean, n = 3), with an average temperature of 1032 ± 148 °C. The lower Ol-Sp temperatures and larger temperature range suggest that the selected olivine and spinel compositions may not be in equilibrium.

Oxygen fugacity of NWA 1950 was estimated using the representative orthopyroxene oikocryst composition (En77Fs21Wo2) reported by Mikouchi (2005), and olivine (Fo75) analyzed in the rock that is in equilibrium with the orthopyroxene, as above, along with the four lowest-Fe#, highest-Cr# chromites, three of which are enclosed in pyroxene. Results yield an average oxygen fugacity of QFM − 3.0 (±0.2; 1σ deviation of the mean, n = 5), with an average temperature of 1203 ± 80 °C (range of 1102 to 1268 °C).

The oxygen fugacity of NWA 4797 is most similar to that estimated for NWA 1950, and in agreement within the uncertainty of the Ol-Px-Sp method (±0.5 log units; Wood 1991). The calculated value for ALHA77005 is lower, but overlaps with those for NWA 4797 and NWA 1950; an estimate for LEW 88516 also overlaps within uncertainties, but is offset to higher values. A comparison is given in Fig. 11.

Figure 11.

 Summary of oxygen fugacity and temperature estimates for four poikilitic ultramafic shergottites. The average values are shown in solid symbols; bars represent the range of temperatures obtained for each dataset. Oxygen fugacity relative to the QFM buffer (formulation of Wones and Gilbert 1969).


Shock Metamorphism

Texture Development

Precursor igneous plagioclase in NWA 4797, inferred to have been present originally in interstitial regions of the nonpoikilitic lithology based on its occurrence in other similar shergottites, has been completely melted and quench-crystallized to form plagioclase microlites in a highly vesiculated glass. We attribute this melting to shock rather than a low-pressure thermal metamorphic event based on the contact between the shock vein and recrystallized plagioclase, which is not discordant. Plagioclase microlites that crystallized within the shock vein and those in the nonpoikilitic lithology intermingle with one another (Figs. 6d and 6e). The range of recrystallized plagioclase composition is slightly more Ab-rich than plagioclase (maskelynite) compositions from other poikilitic shergottites. This is attributed to Na loss during melting and vesiculation of precursor igneous plagioclase during shock.

Plagioclase enclaves have a pale brown rim that is nonstoichiometric with small amounts of FeO and MgO (Fig. 2a; Table 2), interpreted as a mixed melt composition between plagioclase/olivine and plagioclase/pyroxene. Two varieties of glass interstitial to the plagioclase microlites are observed: the first in BSE images has a low grayscale contrast with microlites and is K-, Na-rich (wt% oxides type 1; Table 5); the second glass type exhibits a sharp contrast in grayscale relative to the plagioclase microlites in BSE images and is enriched in Fe and Mg components (wt% oxides type 2; Table 5). Based on bulk compositions (Table 5) and textures (Fig. 4), we interpret the enclaves with type 1 glasses as the quenched products of a melt whose original bulk composition was almost pure plagioclase with contributions from K-rich mesostasis, whereas enclaves with type 2 glasses quenched from a mixed pyroxene-olivine melt with large contributions from plagioclase. This accounts for the absence of K-rich mesostasis in NWA 4797 despite its presence in other less strongly shock heated poikilitic shergottites.

Offshoots of plagioclase melt (types 1 and 2) have migrated through the host rock, exploiting grain boundaries and fractures in olivine and pyroxene. This melt was not in equilibrium with earlier formed igneous minerals. Cooling of these offshoots was rapid enough to reach the glass transition temperature before equilibrium could be established, thereby freezing in the complex reaction textures. Veins cutting across olivine exemplify this; the vein interior progresses from an almost pure plagioclase glass to a glass reflecting increasing contributions from olivine and pyroxene grains. Areas consisting of igneous minerals (olivine/pyroxene/chromite) as “rafts” completely enclosed and isolated by pools of recrystallized plagioclase are an extension of these local offshoots (Fig. 5).

Olivine and pyroxene contain trails of minute iron sulfide grains, indicating that this phase has also been mobilized through the host rock (Figs. 3a and 3b). Preferential melting of iron sulfides has been well documented from naturally shocked rocks and experiments (Ahrens et al. 1969; Jeanloz and Ahrens 1976; Schaal and Hörz 1977).

Constraints on Shock Conditions

Olivine and pyroxene in the shock melt vein are colorless, exhibit straight extinction, and are birefringent. No purple, fawn, or yellow color isotropic minerals were encountered, as would be expected for their high-pressure compositional equivalents (majorite/ringwoodite/wadsleyite; Coleman 1977). The size and shape of olivine (skeletal) and pyroxene (lath) in the shock melt vein argue for crystallization from the melt, rather than back-reaction of ringwoodite or majorite. We constrain temperatures within the vein using the results of previous experiments, performed on synthetic glass that matched the bulk composition of a shock melt pocket in ALHA77005 (Walton and Herd 2007a). In these experiments, the starting conditions were varied from superliquidus, liquidus, and subliquidus temperatures. Each run was then cooled at rates of 10, 500, or 1000 °C h−1. These results are applicable based on similarities between the shock melt pocket in ALHA77005 and the shock melt vein in NWA 4797 in terms of (1) modal % mineral abundance, (2) texture (crystal size/variation in shape from shock melt interior to rim), (3) nucleation density of crystals, and (4) olivine compositions (ALHA77005 cores Fo71–89, rims Fo35–57; NWA 4797 cores Fo73–79, rims Fo69–72). Our observations of skeletal olivine in the NWA 4797 vein interior, progressing to equant, euhedral shapes approaching the host rock margin, coupled with an increase in nucleation density, indicate that the mm-size vein in NWA 4797 cooled at a rate of 500−1000 °C h−1 and that a temperature gradient existed within the vein from 1510−1520 °C in the interior to 1460−1500 °C at the shock melt vein/host rock margin. We note that at this cooling rate, any minerals stable at high pressures and high temperatures would have back-reacted with the melt, consistent with our observations of a lower pressure mineral assemblage in the vein (olivine + pyroxene + chromite).

The suite of shock effects in NWA 4797 (Table 2) indicates that this meteorite has been strongly shock metamorphosed. However, the absence of high-pressure minerals precludes quantitative determination of shock pressure and temperature.

“Anomalous” REE Patterns in NWA 4797

Several explanations can be offered for the REE patterns of melt in equilibrium with poikilitic pyroxene and plagioclase (Fig. 10). It is unlikely that pyroxene and plagioclase crystallized from a LREE enriched melt, because the expectation would be that melt in equilibrium with merrillite would have an REE pattern parallel to that of melt in equilibrium with plagioclase. This is because both minerals crystallized in the nonpoikilitic areas of the rock. We would therefore expect their REE patterns to be related by fractional crystallization. Similarly, we would expect melt in equilibrium with poikilitic pyroxene to have an REE pattern parallel to that of the bulk rock, by comparison with other ultramafic shergottites (Lundberg et al. 1990; Harvey et al. 1993). That no other shergottite shows such enrichment in LREE in melt in equilibrium with pyroxene or plagioclase, and that the poikilitic pyroxene and plagioclase melt patterns are not parallel to one another, suggests that a secondary process has affected REE in poikilitic pyroxene and plagioclase. We consider two possibilities: terrestrial alteration and shock metamorphism. Terrestrial alteration tends to preferentially affect pyroxene (Floss and Crozaz 1991; Harvey et al. 1993; Wadhwa et al. 1994); however, we do not observe Ce anomalies in NWA 4797 poikilitic pyroxene, and this scenario does not provide a ready explanation for plagioclase REE concentrations. The most likely reason for the REE patterns observed for melt in equilibrium with poikilitic pyroxene and plagioclase is mobilization of REE, especially LREE by shock melting. In NWA 4797, this has occurred through melting and mobilization of plagioclase along grain boundaries and fractures in other minerals (Figs. 4c, 4d, and 5). As merrillite is the dominant carrier of the REE, the tendency would be for transfer of REE from merrillite to plagioclase melts through diffusion or partial resorption. This is supported by textural observations of curved and cuspate margins bounding merrillite grains in contact with plagioclase and also elevated phosphorus in the bulk compositions of some glassy enclaves (1.32 wt% P2O5 in “Glass 5”; Table 6). A mixture of <1% merrillite would be required to explain the difference between the REE content of NWA 4797 plagioclase and that of plagioclase in other similar shergottites in which shock-induced REE mobilization is not a factor (e.g., LEW 88516). This mechanism can contaminate pyroxene and plagioclase such that they have REE patterns for equilibrium melts inconsistent with igneous fractionation.

Note that, in spite of cross-contamination between minerals (e.g., merrillite and plagioclase/pyroxene, or Martian/terrestrial alteration), the shock melt vein itself and the bulk rock, as calculated from mineral modes (but highly leveraged by merrillite), have REE patterns consistent with other poikilitic shergottites. Thus, we are reasonably confident that our calculated bulk rock composition is reflective of the true bulk rock. Furthermore, it seems that this has remained a closed system, with shock melt remobilization of REE occurring only on a local scale.

Origin of the Shock Melt Vein

The mm-size shock melt vein in NWA 4797 is distinct from thin (1–100 μm), interlocking networks of black, glassy shock veins documented in many other Martian meteorites (Langenhorst and Greshake 1999; Langenhorst and Poirier 2000a, 2000b). These thinner shock veins contain mineral assemblages stable at high pressures and temperatures (20–25 GPa; 2000–2500 °C) including, but not limited to, omphacite, hollandite, stishovite, akimotoite, and amorphous grains of silicate perovskite. The shock melt vein in NWA 4797 is also distinct from rounded to subrounded pockets of shock melt found in other shergottites (Treiman et al. 1994; Beck et al. 2005; Walton and Herd 2007b), in terms of morphology and the nature of the melt/host rock contact. It is possible that the shock melt pockets represent cross-sections through larger shock veins like the one we observe in NWA 4797; however, the contact between the host rock and shock melt pockets is characterized by increased degrees of shock damage (Walton and Herd 2007a) manifest as a progressive change in intensity of mosaicism, fracture density, birefringence, etc. This increase in shock damage reflects the mode of formation of melt pockets by focusing of shock energy by shock impedance contrasts or closure of pore space in the host rock (Treiman et al. 1994; Beck et al. 2005, 2007; Walton et al. 2010). None of these textures characterize the shock melt vein/host rock margin in NWA 4797. A similar mm-size vein of shock melt has been reported from poikilitic ultramafic shergottite Yamato 000027 (Mikouchi and Kurihara 2008); however, the sharp contact between the vein and surrounding host rock in this meteorite is distinct from textures observed in NWA 4797.

The mineral assemblage in the NWA 4797 shock melt vein implies lower crystallization pressures than thin shock veins in other shergottites. The most likely explanation for the origin of this vein is injection of shock melt in a fracture opened during pressure release. The injected shock melt represents bulk melting of NWA 4797, and does not require an exogenous origin for the melt such as injected Martian regolith proposed for the origin of shock melts in other Martian meteorites (see discussions in Walton et al. 2010). This is based on the following lines of evidence. (1) the modal abundance of the shock melt vein is nearly identical to that of the host rock (Table 1), (2) the crystallization sequence of the vein (chromite → olivine → pyroxene → alkali-rich phases) is similar to that inferred for ultramafic shergottites (McSween 1994), (3) the REE abundances measured by LA-ICP-MS on the shock melt vein and host rock are nearly identical (see the Mineral and Glass REE Contents section), and (4) the opposing walls of the vein are congruent (Fig. 1). The shock melt vein could represent liquids tapped from nearby larger pockets of shock melt that remained liquid after pressure release. This is consistent with the observation that larger shock melt pockets in other shergottites have bulk compositions that overlap that of the bulk igneous rock (Walton et al. 2005, 2010).

Shock History

The contact relationship confirms that plagioclase was melted in the same shock event that also formed the shock melt vein (Figs. 6d and 6e). A two-stage shock history is therefore not required to explain these features; however, we note that the shock effects observed in this study are associated with the strongest shock experienced by the host rock. Whether or not these effects are associated with near surface ejection from Mars in a single impact event or impact relocation on Mars and later ejection in a second impact event cannot be determined by petrography alone. Enclaves of plagioclase microlites + glass assemblages are interpreted to have formed by quench-crystallization of shock melt. Their highly vesiculated nature indicates volatile release and thus this specimen has significant potential for assessing shock ages of shergottites. Park et al. (2008) have successfully used Ar-Ar chronology on a plagioclase separate from mafic shergottite Dhofar 378 to assess the timing of shock. This meteorite shows similar plagioclase textures to those documented in NWA 4797.

Igneous Petrogenesis of NWA 4797

The igneous texture and modal mineralogy of NWA 4797 are very similar to that of the other poikilitic ultramafic shergottites (Table 1). Petrographic and compositional relationships show that the crystallization sequence of NWA 4797 is as follows: (1) initial crystallization of Mg-rich olivine and chromite; (2) accumulation of minerals in (1) to form a fairly open framework from which large low-Ca pyroxene grew, enclosing olivine and chromite grains as chadacrysts; (3) growth of augite rims on low-Ca pyroxene oikocrysts; (4) continued accumulation and formation of interstitial melts from which pigeonite and plagioclase cocrystallize to form the nonpoikilitic textures with cumulus olivine and chromite; and (5) late-stage liquids crystallize and react with earlier formed minerals (olivine becomes more Fe-rich, small amounts of augite crystallize, Ti-rich spinel overgrew cumulus chromite, merrillite, ilmenite, and sulphides crystallize). This crystallization history is similar to those reported from other poikilitic shergottites (e.g., Harvey et al. 1993; Mikouchi and Kurihara 2008; Riches et al. 2011). Although the oxygen isotopic composition of this meteorite has not been determined, the similarity in texture, composition, chemistry, CRE age, and shock metamorphic features confirm its Mars origin. Subtle differences exist in terms of fO2 and REE patterns. Previous studies of mafic shergottites have shown that more incompatible element enriched shergottites formed at higher fO2 (Wadhwa 2001; Herd et al. 2002). Here we examine whether a similar relationship exists within the ultramafic shergottites.

There are two significant challenges to interpretation of results from such shergottites: the degree of accumulation and subsolidus re-equilibration, and sampling volume for bulk rock (REE) analysis. The high degree of accumulation and the relatively coarse-grained nature of these rocks, along with the relatively small amount of material available (<20 g each of LEW 88516, GRV 99027, Yamato 793605, and NWA 4797), make sampling a representative volume for whole rock analysis difficult. As an example, two contrasting whole rock REE analyses were obtained by Ebihara et al. (1997) and Warren and Kallemeyn (1997) for ultramafic shergottite Yamato 793605 (Fig. 12). The REE concentrations obtained by Ebihara et al. (1997) are significantly lower than those of the other ultramafic shergottites, which would suggest that the parent melt of Yamato 793605 is more primitive. However, the REE concentrations obtained by Warren and Kallemeyn (1997) are much higher and coincident with the REE patterns of ALHA77005 and LEW 88516. Wadhwa et al. (1999) demonstrate that the proportions of components, especially merrillite as the major carrier of the REE, can account for the differences between these analyses.

Figure 12.

 Chondrite-normalized REE concentrations for intermediate shergottites, including NWA 4797 shock melt vein and calculated whole rock from this study. Data sources include ALHA77005 and LEW 88516 (Lodders 1998); NWA 1950 (Gillet et al. 2005); Y793605 Ebihara (Ebihara et al. 1997); Y793605 W&K (Warren and Kallemeyn 1997); GRV 99027 (Lin et al. 2008); NWA 480 (Barrat et al. 2002).

The whole rock REE pattern for NWA 4797 (Fig. 12) is among the highest in concentration, and also shows a relative enrichment in LREE, with La/Yb ratio of approximately 0.7. The enrichment in LREE, especially La, may be explained by the relative enrichment in the LREE in the minerals used in the reconstruction; as noted above, even the poikilitic pyroxene is more LREE-enriched than its counterpart in otherwise similar shergottites. The LREE enrichment in the pyroxene may be due to mobilization of plagioclase melt and/or merrillite––both LREE carriers––as a result of shock melting. Notably, Wadhwa et al. (1999) obtain an REE pattern similar to our pattern for NWA 4797––including enrichment in La––by mixing 1.25% merrillite with 70% of an LREE-depleted component (essentially poikilitic pyroxene) and 28.75% of an REE-poor component (olivine and oxides). The implication for NWA 4797 is that the whole rock contains a higher proportion of an LREE-bearing phase, which we attribute to mobilization during shock metamorphism (see previous discussions).

The accumulation of olivine- and chromite-bearing oikocrysts, crystallization of intercumulus melt, and subsolidus re-equilibration pose difficulties for oxygen fugacity determination and interpretation. Subsolidus re-equilibration of the poikilitic assemblages, especially in terms of Fe-Mg exchange, causes deviations in fO2 and temperature from the original magmatic conditions. Ferric iron acts as an incompatible element in the major liquidus phases; consequently, cumulate olivine and pyroxene are low in ferric iron, and subsolidus Fe-Mg exchange can modify chromite compositions and obscure the record of magmatic fO2. Intercumulus melt is relatively enriched in ferric iron and will have accordingly higher fO2 (as shown in this study for NWA 4797), and probably records conditions that are not representative of the parental melt.

In contrast to the findings for mafic shergottites, oxygen fugacity in the ultramafic shergottites does not correlate with their REE patterns. A plot of La/Yb versus fO2 for the shergottites (Fig. 13) shows that ultramafic shergottites are intermediate in their REE characteristics. Yet, the spread of fO2 for the ultramafic shergottites is nearly as large as for mafic shergottites. We interpret this range as reflecting the effects of accumulation and subsolidus re-equilibration.

Figure 13.

 Chrondite-normalized whole rock La/Yb versus oxygen fugacity (relative to the QFM buffer) for shergottites. Solid curve is the polynomial line of best fit for the mafic and olivine-phyric shergottites (Y98 = Yamato 980459; 5789 = NWA 5789; SaU = Sayh al Uhaymir 005; Q = QUE 94201; EA = Elephant Moraine [EETA] 79001 Lithology A; EB = EETA79001 Lithology B; LA = Los Angeles; S = Shergotty, Z = Zagami; Larkman Nunatak [LAR] 06319). Key for ultramafic shergottites: 4797 = NWA 4797; 1950 = NWA 1950; ALH = ALHA77005; LEW = LEW 88516; GRV = GRV 99027; Y79 = Yamato 793605. NWA 4797 includes oxygen fugacity estimates from poikilitic assemblage (solid square), nonpoikilitic assemblage (open square), and weighted average (asterisk; see text). Dashed line represents the approximate trend of increasing oxygen fugacity with increasing La/Yb for the mantle sources. Data sources for oxygen fugacity include NWA 4797, ALHA77005, LEW 88516, NWA 1950 (this study); GRV 99027 (Lin et al. 2005); Y793605 (McCanta et al. 2009); LAR 06319 (Peslier et al. 2010); Y98 (Herd 2004); QUE 94201, Shergotty, Zagami, Los Angeles, EETA79001 (Herd et al. 2001). Data sources for La and Yb include ALHA77005, LEW 88516, NWA 1950, GRV 99027 as in Fig. 12; Y793605 (Warren and Kallemeyn 1997); LAR 06319 (sources in Peslier et al. 2010); NWA 5789 (Irving et al. 2010); Y98 (Shirai and Ebihara 2004); all other shergottites (sources in Herd 2003).

It has been proposed that all intermediate ultramafic shergottites may be derived from the same igneous intrusive body, and that the relative degree of subsolidus re-equilibration reflects their relative depths within the same cumulate pile (Warren and Kallemeyn 1997; Wadhwa et al. 1999; Mikouchi 2005). Mikouchi (2005) used the range of poikilitic and nonpoikilitic olivine compositions as an indicator of relative depth, under the assumption that a greater range of olivine compositions and greater preservation of compositional zoning in poikilitic olivines reflect a lesser degree of equilibration and a shallower depth. Wadhwa et al. (1999) suggested that Y793605 was from a shallower depth than either LEW 88516 or ALHA77005, and Walton and Herd (2007b) suggested that NWA 1950 was shallower than ALHA77005 based on the intensity of shock metamorphism. NWA 4797 has a relatively small range of poikilitic olivine compositions, similar to ALHA77005. Therefore, from deepest (most equilibrated) to shallowest (least equilibrated), the sequence is: GRV 99027, NWA 4797, ALHA77005, NWA 1950, LEW 88516, Y793605. The equilibrated nature of NWA 4797 is further supported by our observations of disequilibrium between the most magnesian olivine and pyroxene compositions, and the large range of temperatures obtained by geothermometry (Fig. 11). Accordingly, the range of temperatures is less for ALHA77005 and NWA 1950. In contrast, the range of temperatures for LEW 88516 is high, and the average temperature is the lowest among the samples studied. This apparently contradictory observation may be explained by subsolidus olivine-spinel Fe-Mg exchange. Chromite in LEW 88516 has the highest Fe# of the shergottites in this study, indicative of such exchange, most likely with olivine hosts. Olivine-spinel exchange could occur without significantly affecting the range of preserved olivine compositions. This effect may also explain the low temperatures, if subsolidus Fe-Mg exchange resulted in disequilibrium between the olivine and spinel compositions used in geothermometry.

Even considering the effects of subsolidus re-equilibration, it is challenging to select a member of this ultramafic shergottite group that records fO2 conditions of crystallization that closely reflect the fO2 conditions of the parent magma (or magmas). If we assume that they are all derived from the same parent magma with the same initial fO2, then the average of all, fO2 = QFM − 2.8, La/Yb = 0.45, may reflect the parent magma. Alternatively, if we assume that a representative member of this group of shergottites crystallized as a closed system, then a recombination of early (poikilitic) and late (nonpoikilitic) fO2, weighted according to mode, may represent the parent magma conditions. If we take NWA 4797 as a representative example, and assume that all olivine and pigeonite represent the poikilitic portion (mode = 62.5%), with fO2 = QFM − 2.9, and that the remainder represents the nonpoikilitic portion (37.5) with fO2 = QFM − 1.7, the weighted average fO2 = QFM − 2.4. Given the uncertainties on the fO2 estimates, the parent magma for these particular shergottites had fO2 = QFM − 2.5 ± 0.5. This overlaps with the upper end of the reduced shergottite endmember, as well as the recently re-evaluated lower end of the oxidized endmember (Peslier et al. 2010), suggesting that the mantle sources of the shergottites are limited to a reduced, depleted endmember with source fO2 ∼ QFM − 3.5 (near the IW buffer), and two more oxidized endmembers with source fO2 ∼ QFM − 2.5 and variable incompatible element enrichment. LAR 06319 is representative of the more enriched endmember (La/Yb ∼ 1), and NWA 4797 and its cohorts are representative of an intermediate endmember, with La/Yb ∼ 0.45. In both cases, oxygen fugacity is partially decoupled from incompatible element enrichment due to the superposition of oxidation or reduction during ascent, eruption, emplacement, and re-equilibration (Peslier et al. 2010; this study).


NWA 4797 is a relatively fresh meteorite specimen recovered from a hot desert environment. Investigation of polished sections from the interior shows that it is a strongly shocked crystalline igneous rock. Shock effects in olivine and pyroxene include strong mosaicism, grain boundary melting and recrystallization, lack of planar elements, coloration, shock blackening, and reduced birefringence. Precursor igneous plagioclase has been completely melted by shock and quenched-crystallized to an assemblage of radiating plagioclase microlites in glass. This melt has migrated along cracks and grain boundaries in the host rock and reacted with igneous mineral, to give rise to complex textures. A mm-size vein of shock melt cuts across both igneous lithologies, from which skeletal phosphoran olivine, pyroxene laths, dendritic chromite, and iron sulfide spherules have crystallized. High-pressure mineral polymorphs have not been observed in this shock vein; the relatively low pressure assemblage that crystallized in the vein is the result of longer cooling times due to the vein thickness and its formation during decompression. The modal abundance, crystallization sequence, and REE profile of shock melt vein and the host rock overlap, indicating that the shock melt vein represents bulk melting of the host rock. NWA 4797 has great potential to yield information on the shock ages of shergottites because igneous plagioclase has been completely melted and recrystallized by shock. Implementing our proposed classification scheme for Martian specimens, NWA 4797 is an ultramafic poikilitic shergottite with intermediate trace element characteristics.

Acknowledgments— This work has been funded by the Canadian Space Agency through the Space Science Fellowships program (E. L. W.) and NSERC Discovery Grant 261740-03 to C. D. K. H. We are very grateful to Scott Kuehner and James Wittke for their assistance with EM analyses and imaging of this specimen. The meteorite described here was brought to the attention of scientists by Ali and Mohammed Hmani, with the liason of Norbert Classen. Thanks to Sergei Matveev, George Braybrook, and Guangcheng Chen (UAb) for assistance with EM, SEM, and LA-ICPMS work, respectively. The authors gratefully acknowledge comments by Takashi Mikouchi, Yangting Lin, and an anonymous reviewer made on an earlier version of this manuscript, which improved the quality of our work.

Editorial Handling— Dr. Ian Franchi