PII S0016-7037(01)00892-4

Crystallization conditions of Los Angeles, a basaltic Martian meteorite

DIMITRIOS XIROUCHAKIS,1,* DAVID S. DRAPER,1 CRAIG S. SCHWANDT,2 and ANTONIO LANZIROTTI3

1Office of Astromaterials Research, NASA JSC, Mail Code SA13, Houston, TX 77058, and Department of Chemistry, Texas SouthernUniversity, 3100 Cleburne Avenue, Houston, TX 77004, USA

2Lockheed Martin, 2400 NASA Road One, C23, Houston, TX 77058, USA3Consortium for Advanced Radiation Sources/The University of Chicago, at Brookhaven National Lab, Upton, New York 11973, USA

(Received March 24, 2001; accepted in revised form December 7, 2001)

Abstract—The texture of Los Angeles (stone 1) is dominated by relatively large (0.5�2.0 mm) anhedral tosubhedral grains of pyroxene, and generally subhedral to euhedral shocked plagioclase feldspar (maskelynite).Minor phases include subhedral titanomagnetite and ilmenite, Fe-rich olivine, olivine�augite-dominatedsymplectites [some of which include a Si-rich phase and some which do not], pyrrhotite, phosphate(s), and animpact shock-related alkali- and silica-rich glass closely associated with anhedral to euhedral silica grains.Observations and model calculations indicate that the initial crystallization of Mg-rich pigeonitic pyroxenesat �1150 °C, probably concomitantly with plagioclase, was followed by pigeonitic and augitic compositionsbetween 1100 and 1050 °C whereas between 1050 and 920 to 905 °C pyroxene of single compositioncrystallized. Below 920 to 905 °C, single composition Fe-rich clinopyroxene exsolved to augite and pigeonite.Initial appearance of titanomagnetite probably occurred near 990 °C and FMQ-1.5 whereas at and below 990°C and �FMQ-1.5 titanomagnetite and single composition Fe-rich clinopyroxene may have started to react,producing ilmenite and olivine. However, judging from the most common titanomagnetite compositions, weinfer that most of this reaction likely occurred between 950 and 900 °C at FMQ-1.0�0.2 and nearlysimultaneously with pyroxene exsolution, thus producing assemblages of pigeonite, titanomagnetite, olivine,ilmenite, and augite. We deem this reaction as the most plausible explanation for the formation of theolivine�augite-dominated symplectites in Los Angeles. But we cannot preclude possible contributions to thesymplectites from the shock-related alkali- and silica-rich glass or shocked plagioclase, and the breakdown ofFe-rich pigeonite compositions to olivine�augite�silica below 900 °C. Reactions between Fe-Ti oxides andsilicate minerals in Los Angeles and other similar basaltic Martian meteorites can control the T-fO2

equilibration path during cooling, which may better explain the relative differences in fO2 among the basalticMartian meteorites. Copyright © 2002 Elsevier Science Ltd

1. INTRODUCTION AND PREVIOUS WORK

The basaltic Martian meteorite Los Angeles, a microgabbro,was first described by Rubin et al. (2000a, 2000b), Warren et al.(2000a, 2000b), and Greenwood et al. (2000). This meteoritehas several compositional features suggesting it is the mostdifferentiated of the shergottite-nakhlite-chassignite (SNC)group (Rubin et al., 2000b), but it also exhibits compositionaland textural similarities to other basaltic Martian meteorites(e.g., Shergotty and QUE94201), and the lunar mare meteoriteAsuka 881757 (Mikouchi, 2000).

Los Angeles 5103 (hereafter LA 5103) is a 1.14 g sample ofstone 1 allocated to us from the American Museum of NaturalHistory for study. Here, we describe the mineralogy and textureto establish the similarities with the samples previously studiedby Rubin et al. (2000a, 2000b), Warren et al. (2000a, 2000b),and Greenwood et al. (2000), and we present our interpretationof the phase relations and estimates of the T and fO2 conditionsduring crystallization. Of particular interest are late-stage tex-tural features and phases, and low-variance mineral assem-blages that are common in Los Angeles, Shergotty, andQUE94201 (Aramovich et al. 2001, Herd et al. 2001a, Stolperand McSween 1979), and which may constrain the intensiveparameters during crystallization. These features are oxide-

silicate clusters, symplectites that consist primarily of Fe-richolivine and augite, and presumably crystalline silica grainsclosely associated with alkali- and silica-rich glasses.

Rubin et al. (2000a, 2000b) originally argued that the Fe-richolivine and titanomagnetite intergrowths next to silica grainsmay suggest fO2 conditions near the Fayalite-Magnetite-Quartzbuffer curve (FMQ) at a calculated equilibration temperature of420 °C. These estimates do not represent the original igneousconditions, as Rubin at al., (2000b) also noted, and the assem-blage Fe-rich olivine�titanomagnetite�silica is not equivalentto the FMQ buffer curve. Rubin et al. (2000a, 2000b), Warrenet al. (2000a, 2000b), and Greenwood et al. (2000) also arguedthat the symplectites consist of the assemblagehedenbergite�fayalite�silica, which they inferred to resultfrom the breakdown of single-phase pyroxferroite, a pyrox-enoid of (Ca0.13–0.15Mg0.02–0.00Mn0.02–0.00Fe0.83–0.85)SiO3 com-position (Chao et al., 1970; Lindsley and Burnham, 1970;Burnham, 1971). Implicit in the single-phase pyroxferroitebreakdown model is the notion that the bulk of crystallizationhad to take place at �1.0 GPa and 950 °C (Lindsley, 1967;Lindsley and Burnham, 1970) with subsequent emplacementand slow cooling on or near the surface of the planet wheresingle-phase pyroxferroite reacted to hedenbergite�fayalite�silica, e.g., 14Ca2/7Fe12/7Si2O6�4CaFeSi2O6�10Fe2SiO4�10SiO2.

An alternative reaction, that takes place throughout a rangeof pressure conditions, could have been the transformation of a

* Author to whom correspondence should be addressed(dxirouch@ems.jsc.nasa.gov).

Pergamon

Geochimica et Cosmochimica Acta, Vol. 66, No. 10, pp. 1867–1880, 2002Copyright © 2002 Elsevier Science LtdPrinted in the USA. All rights reserved

0016-7037/02 $22.00 � .00

1867

pyroxenoid to pyroxene in the presence of already existingFe-rich olivine and silica (Lindsley, 1967; Lindsley and Burn-ham, 1970). But in the case of Los Angeles this requirement isnot likely met. Nonetheless, a model that involves a pyroxenoidhas to take into account the interplay between the magnesiumcontent and temperature, and to a lesser extent pressure. Forexample, at 1 atm the pyroxenoid to pyroxene phase transitionalong the Fe2Si2O6-CaFeSi2O6 join takes place between 940and 960 °C (Lindsley, 1967; Lindsley and Munoz, 1969; Lind-sley and Burnham, 1970), and it is displaced to higher temper-atures with increasing magnesium content. And, along theCaMgSi2O6-CaFeSi2O6 join the displacement is approximately�6 °C per mol of CaMgSi2O6 added (Turnock, 1962; Turnockand Lindsley, 1981). Aramovich et al. (2001) considered thesymplectites to be the breakdown product of a Ca-poor andFe-rich pyroxene or pyroxenoid to Fe-rich olivine�augite�silica whereas Herd et al. (2001a) appear to favor thepyroxferroite breakdown model. The breakdown reaction of alow-Ca pyroxene is perhaps a more viable mechanism providedthat, (i) the symplectites have a pyroxene-like stoichiometry,(ii) the augite and olivine in the symplectites have the appro-priate compositions and a silica phase is present, and (iii) thesymplectites were not modified by the shock or by interactionwith shock-related phases. Mass balance considerations alsosuggest that the breakdown of a low-Ca pyroxene or pyrox-enoid produces significantly greater amounts of silica and Fe-rich olivine than augite. Thus, Fe-olivine and silica shoulddominate X-ray compositional maps or diffraction spectra ofthe symplectites. Finally, we believe that any interpretation oftheir formation must also account for their close association tothe late-stage Fe-Ti oxides, Fe-rich olivine, and exsolved py-roxenes (Fig. 1).

2. RESULTS

2.1. Analytical Conditions

Two 10�12 mm thin sections (LA 5103,1 and LA 5103,2)were analyzed with the JSC CAMECA SX-100 electron mi-croprobe. Analytical conditions were 20 kV, 5 nA, a beamdiameter of �1 �m for pyroxenes, olivine, and oxides, and 20�m for the symplectites. Peak counting times were 40 s for Naand K, and 60 s for Ca, Fe, Mg, Mn, Al, Cr, Ti, Si, P, and S.Shocked plagioclase feldspar, alkali-rich feldspathic glass, andfusion crust analyses were performed with 15 kV, 20 nA, peakcounting times of 20 s, and a beam diameter of 1 or 5 �m.Background counting times were 10 s for all elements. Thestandards were natural and synthetic oxide and silicate miner-als, and metals for Ni and Cr. To ensure reproducibility weanalyzed the standards at the beginning and end of each ses-sion.

Digital X-ray intensity maps of Ca, Mg, Al, Fe, Si, Ti, K, andP elements for the entire thin sections as well as the symplec-tites were also collected with the JSC electron microprobe.Mapping the full 10�12 mm extent of the two thin sectionsrequired X-ray collection as a composite of 16 maps of 5elements each. Each segment map was 500�600 �m, with apixel step size of 5 �m. The 4�4 map grid with these config-urations was then compiled into 5 element maps each with2000�2400 pixels, for a total point count of 4.8�106 points.Subsequently, a map of each thin section was produced and

analyzed numerically with the Noesys�IDL software. In aneffort to identify �1 �m pure silica grains in the symplectites,the X-ray maps of the symplectite areas were done with stepsizes of 3 (Fig. 2a and b) or 1 �m (Fig. 2c and d).

We also used the X26A beamline at the National Synchro-tron Light Source, Brookhaven National Laboratory, to collectsynchrotron X-ray diffraction and fluorescence spectra (SXRDand SXRF) for phase identification (e.g., Fig. 3) and analyses ofminor elements in selected single phases or the entire thinsections. A full discussion of the SXRF analyses will be pre-sented separately. The incident beam was tuned to the U LIIIbinding energy using a Si(111) channel-cut monochromator.This 350 �m collimated monochromatic beam is then focusedto 10 �m in diameter using the beamline’s system of Rh coatedKirckpatrick-Baez mirrors (Eng et al., 1998). Microbeam(10�14 �m) SXRD data were collected using a BruckerSMART CCD system in reflection mode geometry. The wave-length of the incident monochromatic beam during the diffrac-tion analyses was kept at 0.7976 Å. The identification of theminerals is primarily based on the presence or absence ofclassic Debye-Sherrer rings on the CCD. However, if thecrystallinity is on the order of the spot size, it is possible toobserve discrete spots that make up the entire Debye-Sherrerring. Furthermore, these spots are strongly controlled by crystalorientation relative to the beam and the camera position. So forcoarsely crystalline materials there is the possibility that onemay see no diffraction. The other implication of this is that thepeak intensities may be off (and typically are). Nonetheless, formost of the phases we did get good Debye-Sherrer rings and it

Fig. 1. Back Scattered Electron (BSE) image of an oxide-silicatecluster. Olivine (Fayalite) forms rims around titanomagnetite (TiMt)and is in contact with ilmenite (Ilm), pyrrhotite (Py), and augite (Aug).Ol continues into the symplectites where is in contact only with Aug.Pyroxenes show exsolution features. The symplectites were analyzedhere and appear to have excess Si relative to pyroxene stoichiometry(IVSi�2 p.f.u). Note, the expansion cracks emanating from Msk, thebreaking of the pyroxene grain by Msk, and the modification of thesymplectites and the TiMt�Ilm�Ol cluster by Msk. Here, the modi-fication by Msk also resulted in Ol�Msk areas within either thesymplectites or maskelynite. Crosses mark pyroxene analyses that areshown in Figure 5.

1868 D. Xirouchakis, D. S. Draper, C. S. Schwandt, and A. Lanzirotti

1869Crystallization conditions of Los Angeles meteorite

is unlikely that any potential SiO2 polymorph in the samplewould be any different.

2.2. Mineralogy and Petrology

The texture of LA 5103 is that of a microgabbro, and isdominated by relatively large (0.5�2.0 mm) anhedral to sub-hedral grains of pyroxene and generally subhedral to euhedralshocked plagioclase. Minor phases include, (i) subhedral tit-anomagnetite and ilmenite grains, (ii) olivine, (iii)olivine�augite-dominated symplectites, (iv) pyrrhotite, (v)phosphate(s), and (vi) alkali-rich feldspathic glass closely as-sociated with subhedral to euhedral silica grains (e.g., Fig. 4).Modal proportions (vol.%) and 2� uncertainties as determinedby digital X-ray element maps are, shocked plagioclase53.9�3.2%, pyroxenes 40.9�2.8%, fusion crust 1.5�0.0%,Fe-Ti oxides 1.0�0.1% (Ilm �0.25%, TiMt �0.75%), alkali-and silica-rich glass 1.7�0.2%, and phosphate(s) 1.0�0.1%.We have likely overestimated the proportions of pyroxenegrains, albeit slightly, because we have not accounted for thesymplectites.

The shocked plagioclase feldspar grains in LA 5103 exhibitmany of the features described by Chen and El Goresy (2000)for maskelynite. We find, however, that the grains are stoichi-ometric with a compositional range of An50–62Ab37–48Or1–4.Also, the rims can be relatively enriched in sodium, especiallynext to alkali- and silica-rich glasses. In addition and despitethe shock-induced isotropy, the crystalline structure has par-tially survived; weak plagioclase feldspar peaks are present inat least one grain analyzed by SXRD (d spacings at 0.205,0.242, 0.265, and 0.42 nm).

Pyroxene compositions expressed as Wo (CaSiO3), En (Mg-SiO3), and Fs (FeSiO3) components range from Mg- to Fe-rich(Wo12–41En50–6Fs81–33). The interior of the pyroxene crystalsis Mg-rich grading to Fe-rich rims (En10–6) with a range ofcalcium content (Wo13–41) because of exsolution to iron-richaugite and pigeonite (Fig. 1, 5). The concentration of non-quadrilateral components does not exceed 5 mol.%. Pyroxene

iron enrichment (mg# � molar Mgo

MgO�FeOfrom 0.58 to 0.16) is

accompanied by an increase in TiO2 (see also Wadhwa et al.,2001) and a small decrease in Al2O3 (wt.%). In contrast, Cr2O3

(wt.%) remains constant with increasing iron enrichment. How-ever, for pyroxenes having mg# between 0.16 and 0.04, bothTiO2 and Al2O3 range from as low as 0.2 to as high as 0.8 to

Fig. 3. SXRD spectra of three 10 � 14 �m symplectite spots. Point1 and 2 are marked by the crosses in Figure 2b. The peaks in the XRDspectra can be assigned to iron-rich olivine (e.g., d spacings at 0.18,0.25, 0.35 nm) and augite (e.g., d spacings at 0.15, 0.21, 0.25, 0.26,0.30, and 0.34 nm). Major quartz (e.g., d spacings at 0.33 and 0.43 nm),tridymite (d spacings at 0.39, 0.41, and 0.44 nm), coesite (e.g., 0.27,0.31, 0.34, 0.44 nm), stishovite (e.g., 0.15, 0.20, 0.29 nm), and post-stishovite SiO2 phase (e.g., 0.28 nm) peaks are either absent or cannotbe uniquely identified.

Fig. 4. BSE image of interstitial alkali-rich glass pockets associatedwith SiO2 (dark gray) grains which are marked by the crosses, and theyare surrounded by shocked plagioclase (msk), pyroxene(s), and oxide-silicate and olivine-pyroxene clusters (bright). The shape of the euhed-ral silica grain in the center of the figure is reminiscent of tridymite.Also note the expansion cracks emanating from the silica grains.

Fig. 2. (a) Ca and (b) Si X-ray maps (pixel step size of 3 �m) of asymplectite area. Scale bar is 100 �m. Point 1 and 2 mark spots fromwhich we collected SXRD spectra, and which are shown in Figure 3.The Ca X-ray map suggests that this symplectite area consists of aCa-poor and Ca-rich phase. On the Si X-ray map, if a pure SiO2 phasewas present it would have clearly appeared as bright red spots, but onlya few spots appear to have a relative higher concentration of silica. Thewhite dashed lines in Figure 2b mark optically different symplectiteareas. (c) BSE image (pixel step size of 1 �m) of a symplectite area.Rectangles mark symplectites that optically appear different than therest. Specifically, they lack the darker spots present in the rest but thecompositional differences between the two are within 1 to 2 wt.% forthe major oxides. Arrow points to an alkali- and silica-enriched feld-spathic inclusion inside the symplectites. Other phases are pyroxenes(px), phosphate (ph) which is likely apatite, and an alkali- and silica-rich glass area (white dashed lines). (d) Si-X-ray map (pixel step sizeof 1 �m) of the area shown in Figure 2c. Black lines demarcate analkali-enriched (Na�K) rim in maskelynite which is followed byalkali- and silica-rich glass (bright red area). Red spots inside symplec-tites may be either alkali-and silica-rich glass or possibly SiO2. (e)Blow-up of the area around the smaller rectangle in Figure 2c. (f) BSEimage of a symplectite and an alkali- and silica-rich glass area. Sym-plectites engulfed by the glass primarily consist of Ol�gl.

1870 D. Xirouchakis, D. S. Draper, C. S. Schwandt, and A. Lanzirotti

1.2 wt.%. This likely results from the fact that in this mg# rangesingle pyroxene exsolved to pigeonite and augite that havedifferent preference for non-quadrilateral minor elements.

Olivine forms 5 to 15 �m thick rims around titano-magnetite�ilmenite grains which produces the two commonassemblages titanomagnetite�ilmenite�olivine and augite�olivine�pigeonite, and the less common titanomagnetite�ilmenite�olivine�augite assemblage. In places where olivinecontacts the olivine�augite-dominated symplectites, olivine inthe symplectites has the same composition as olivine in therims. These textural features are best seen in Figure 1 byfollowing augite (small dark gray areas) from within the pigeo-nite host through the symplectites and up to the Ol�TiMt�Ilmcluster. The olivine composition, in terms of the Ca2SiO4 (La),Mg2SiO4 (Fo), and Fe2SiO4 (Fa) components, is within therange: La0.2–1.0Fo5–6.5Fa93–95.

The olivine�augite-dominated symplectites were originallynamed the pyroxferroite breakdown material (PBM) by Rubinet al. (2000a, 2000b), Warren et al. (2000a, 2000b), and Green-

wood et al. (2000). In the two thin sections we examined, thesymplectites typically but not always constitute the neighboringphases to the olivine�titanomagnetite�ilmenite clusters. Oc-casionally, tiny bits of Fe-Ti oxides or pyrrhotite are seen in thesymplectites. Although other authors (Rubin et al., 2000a,2000b; Warren et al., 2000a, 2000b; Greenwood et al., 2000;Herd et al., 2001a) have argued in favor of SiO2 being presentin the symplectites, we observed symplectites both with andwithout SiO2 or a SiO2-enriched phase in them (Fig. 2 and 3).Specifically, many of the detailed (i.e., 1 to 3 �m step size) SiX-ray maps of different symplectite areas in the sample suggestcomplete absence of silica or a silica-rich phase (Fig. 2a, b).Moreover, the XRD patterns of several symplectites do notcontain peaks unique to either a low or high pressure SiO2

polymorph. The peaks can be easily assigned to Fe-rich olivineand augite (Fig. 3). Note also that peaks characteristic ofhedenbergite (CaFeSi2O6) are absent, e.g., peak at 0.47 nm. Incontrast, others show that a silica-rich phase or perhaps silicaitself is probably present (Fig. 2c). Conceivably, the symplec-

Fig. 5. Quadrilateral projection of observed pyroxene (E) and olivine (�) compositions and calculated Aug�Pigisotherms. Px adjacent to Ol�Il�TiMt clusters have 6 to 10 mol.% En and range from �15 to �40 mol.% in Wo becauseof exsolution to Aug and Pig. Crosses (�) represent marked Pig and Aug grains in Figure 1. Filled circles (●) and squares(■ ) denote calculated compositions of Px and Ol in equilibrium with Ilm and TiMt from 990 °C and FMQ-1.6 to 870 °Cand FMQ-0.6 (path ABC in Fig. 7). Exsolution and thus generation of a range of calcic and subcalcic iron-rich pyroxenecompositions is favored below 920 °C. None of the most calcium- and iron-rich pyroxenes in LA approach magnesium-freecompositions. Tie-lines at 905 °C denote the expected olivine (O), pigeonite (P) and augite (A) compositions in equilibriumwith Qz (Q); assemblages QOAP, OAQ, and OQ. Thick dark gray rectangle marks the best Px-like symplectite compositionof Wo14�3En5�1.

1871Crystallization conditions of Los Angeles meteorite

tites may have been affected by the alkali- and silica-rich glassproduced by the impact event (see below) and/or the shockedplagioclase, which we infer from the observed vermicular in-tergrowths of olivine�maskelynite�glass and olivine�augite�maskelynite�glass (e.g., Fig. 1 and 2f). Nonetheless,we do find symplectites that unequivocally consist solely ofolivine�augite next to the late-stage oxide�pyroxene clustersin LA 5103. The average of eighteen analyses from fairly largesymplectite areas has excess tetrahedral Si (2.03�0.04 Si pfu)relative to a pyroxene. However, eleven out of the eighteenanalyses have a Ca-poor and Fe-rich pyroxene-like stoichiom-etry with an average composition of Wo14�3En5�1.

The spinel phase present in LA 5103 is essentially titano-magnetite with less than 2 wt.% Cr2O3 and Al2O3 (Table 1). Interms of the components Fe2TiO4 (Usp), MgFe2O4 (Mg-Ferrite), MnFe2O4 (Mn-Ferrite), and Fe3O4 (Mt) the composi-tions of single titanomagnetite grains covers the range 61 to79 mol.%, 0 to 1 mol.%, 1 to 2.5 mol.%, and 38 to 18 mol.%,respectively. Peripheral to the titanomagnetite grains we

observed individual ilmenite grains having compositions ofHm3–0Gk1–0Py1.5–1Il94–98 in terms of Fe2O3 (Hm), MgTiO3

(Gk), MnTiO3 (Py), and FeTiO3 (Ilm). The occurrence ofilmenite appears to be crystallographically controlled, and thetitanomagnetite- ilmenite textures resemble those resultingfrom (granule) oxy-exsolution. Most of the titanomagnetite andilmenite analyses fall within the ranges Usp67–72Mt22–32 andHm1–2Ilm96–97.

The alkali-and silica-rich phase in Los Angeles is inferred tohave been liquid/glass because it does not appear to have amineral stoichiometry. Although Synchrotron-XRD of a10�14 �m spot within a 200 �m area suggests a mostlyamorphous material with weak feldspar and no silica poly-morph peaks, the electron microprobe and SXRF analyses(Lanzirotti, written communication) suggest that these glassesare a mixture of an alkali- (Na, K, Rb) and a silica-rich phasewith contributions from the adjacent shocked plagioclase grainsand oxide�silicate (TiMt-Ilm-Ol-Pig-Aug) clusters as well(Table 1; Fig. 2f and 4). This conclusion is consistent with the

Table 1. Range (min-max) of Oxides or Element(s) Concentration in Titanomagnetite, Ilmenite, Olivine, Augite, Pigeonite, Maskelynite,Alkali-Rich and Silica-Rich Feldspathic Glass, and Fusion Crust.

wt% Titanomagnetite Ilmenite Olivine Augite Pigeonite Maskelynite Glass F. C.‡

{106, 9} {70, 10} {25, 5} {57, 13} {70, 7} {97, 5} {30, 4} 24Na2O 0.00–0.14 0.00–0.07 0.00–0.08 0.03–0.18 0.00–0.11 3.76–5.34 0.84–3.37 1.29–1.99K2O 0.00–0.03 0.00–0.03 0.00–0.09 0.00–0.10 0.00–0.02 0.08–0.69 0.93–5.15 0.10–0.17CaO 0.00–0.40 0.00–0.64 0.13–0.53 7.48–18.31 5.04–8.88 10.29–12.03 0.25–3.01 9.61–10.93MgO 0.00–0.17 0.11–0.25 1.76–2.54 1.74–14.88 1.82–17.36 0.02–0.15 0.00–0.01 4.55–5.76FeO 65.17–71.37 45.35–47.64 63.44–65.97 19.87–39.69 22.07–43.73 0.43–0.85 0.20–1.04 21.51–23.84MnO 0.45–0.76 0.53–0.71 1.13–1.30 0.61–1.01 0.69–1.31 0.00–0.05 0.00–0.04 0.54–0.71NiO – – – – – – – 0.00–0.06Al2O3 1.13–2.18 0.00–0.11 0.01–0.11 0.51–1.26 0.23–0.82 27.62–29.48 2.76–17.71 5.94–9.68Cr2O3 0.00–0.10 0.00–0.03 0.00–0.03 0.00–0.07 0.00–0.14 0.00–0.02 0.00–0.01 0.00–0.07TiO2 20.93–26.90 49.58–51.90 0.11–1.30 0.24–0.87 0.14–0.81 0.00–0.08 0.06–0.44 0.96–1.77SiO2 0.04–0.82 0.01–0.17 29.70–30.59 45.71–50.23 45.94–51.97 53.45–57.58 70.53–96.66 45.28–49.48P2O5 0.00–0.05 0.00–0.05 0.00–0.08 0.00–0.21 0.00–0.12 0.00–0.09 0.00–0.33 0.68–1.99S 0.00–0.05 0.00–0.02 0.00–0.59 0.00–0.05 0.00–0.02 - 0.00–0.01 0.00–0.04SUM 94.17–97.46 96.29–99.63 98.28–100.74 98.01–100.71 98.04–101.02 98.01–101.99 96.74–101.93 98.79–101.79

Notes: Pair of numbers denotes number of points and grains analyzed, e.g., 106 points in a total of 9 titanomagnetite grains. For presentationpurposes only, we distinguished pigeonite from augite on the basis of their Wo content: 12–18 mol% and 19–40 mol%, respectively. Therefore someoverlap in CaO (wt%) is expected. (‡) F.C. stands for fusion crust and its composition is the average of 24 discrete points.

Table 1 continued: Average Compositions.

wt% TiMt Ilm Ol Msk Apat Sympl.§ Sympl.† SiO2 Glass F. C.

Na2O 0.01 (2) 0.01 (2) 0.03 (3) 4.79 (43) 0.17 (4) 0.05 (3) 0.03 (2) 0.14 (1) 2.11 (64) 1.76 (20)K2O 0.01 (1) 0.01 (1) 0.02 (2) 0.15 (8) 0.03 (5) 0.00 (1) 0.00 (0) 0.35 (5) 3.44 (1.21) 0.13 (2)CaO 0.03 (6) 0.04 (9) 0.27 (10) 11.44 (31) 53.11 (78) 4.83 (1.17) 5.00 (1.34) 0.23 (3) 1.22 (63) 10.38 (27)MgO 0.06 (3) 0.17 (3) 2.20 (23) 0.11 (2) 0.00 (2) 1.59 (32) 1.57 (22) 0.00 (0) 0.00 (0) 5.25 (32)FeO 69.36 (1.01) 46.49 (50) 64.97 (55) 0.55 (7) 1.17 (50) 43.58 (1.77) 44.48 (1.34) 0.13 (3) 0.47 (27) 23.84 (1.99)MnO 0.53 (5) 0.61 (4) 1.20 (5) 0.01 (1) 0.11 (2) 1.02 (9) 1.10 (6) 0.01 (1) 0.01 (01) 0.61 (6)NiO – – – – – – – – – 0.01 (2)Al2O3 1.58 (15) 0.00 (2) 0.03 (2) 28.73 (33) 0.14 (26) 0.26 (7) 0.27 (8) 2.21 (12) 11.24 (3.25) 8.68 (1.04)Cr2O3 0.04 (2) 0.01 (1) 0.00 (1) 0.00 (0) 0.00 (0) 0.00 (1) 0.00 (0) 0.01 (1) 0.00 (0) 0.03 (2)TiO2 24.59 (83) 51.05 (53) 0.60 (38) 0.04 (2) 0.07 (11) 0.24 (9) 0.25 (9) 0.19 (1) 0.20 (9) 1.21 (18)SiO2 0.1 (1) 0.06 (3) 30.00 (21) 55.11 (85) 1.45 (1.16) 46.78 (1.38) 45.80 (0.78) 95.91 (19) 81.53 (5.60) 47.47 (1.12)P2O5 0.01 (1) 0.01 (1) 0.03 (13) 0.02 (2) 37.09 (1.04) 0.05 (3) 0.05 (3) 0.00 (0) 0.06 (8) 1.08 (30)S 0.01 (1) 0.01 (1) 0.05 (13) – 0.08 (8) – – 0.00 (1) 0.00 (0) 0.01 (1)SUM 96.32 (91) 98.45 (79) 99.40 (64) 101.00 (1.34) 93.44 (69) 98.44 (76) 98.57 (77) 99.21 (3) 100.29 (1.18) 100.46 (81)

Notes: Numbers in parentheses represent one standard deviation and refer to the last digit(s). (§) Average of all symplectite analyses (n � 18). (†)Average of symplectite analyses (n�11) with pyroxene-like stoichiometry. SiO2 average is based on two discrete grains.

1872 D. Xirouchakis, D. S. Draper, C. S. Schwandt, and A. Lanzirotti

positive Eu anomaly observed in these glasses by Wadhwa etal. (2001). These glasses also have a Br content of 19.5�2.5ppm (Lanzirotti, written communication) that is identical to theestimated average Br content of �20 ppm of the Martiansurface (Rao et al., 2002). We rule out Br contamination of thesamples during preparation; moreover, Br analyses of an adja-cent shocked feldspar grain are clearly different (0.8 ppm).Therefore, we conclude that these glasses may be an impactshock-produced feature. Nevertheless, it is unclear whether thealkali-rich phase was originally, an alkali-rich silicate glass, analkali-rich feldspar, or alkali-rich plagioclase feldspar rims.Stolper and McSween (1979) observed similar features in theirstudy of Shergotty and Zagami, and they concluded that theymay represent shocked mixtures of alkali-feldspars and silica.

LA 5103 contains 50 to 100 �m anhedral to subhedral SiO2

grains, either next to or within the alkali- and silica rich glassesand outside neighboring olivine�pyroxene�oxides clusters. Itis noteworthy that the crystal shape of the most euhedral SiO2

grains resembles that of tridymite (Fig. 4), and that the SiO2

grains appear to be the loci of expansion cracks that propagatethrough the neighboring grains. The composition of the SiO2

grains is also suggestive of a low-pressure polymorph withsufficiently open structure to accommodate aluminum, alkalies,and other elements in trace quantities. Conceivably, shockpressure may have transformed the original SiO2 phase into ahigh-pressure polymorph that expanded upon relaxation. Thesimilarity between the SiO2 grains in LA and Shergotty (ElGoresy et al., 2000; Sharp et al., 1999) is intriguing andwarrants a more detailed investigation.

Other accessory minerals include pyrrhotite (Fe0.96�0.01S)typically within or close to titanomagnetite grains and occa-sionally in the symplectites, and apatite (sum of oxides �93wt.%) that is always associated with the late-stage titanomag-netite-ilmenite-olivine-augite clusters. A few analyses indi-cated the possible presence of a nearly anhydrous (sum ofoxides �98 wt.%), iron-bearing phosphate besides apatite.Several apatite grains contain inclusions of augite and moreoften either shocked plagioclase or alkali-and silica-rich glass.

We infer from the Ti, Fe, Ca and P X-ray maps of the entiresections, and the occurrence of Fe-Ti oxides, apatite, and silicagrains in the samples that 95 to 97% of Los Angeles hadcrystallized before Fe-Ti oxides, apatite, and silica saturation.Our inference broadly agrees with the findings of Minitti andRutherford (2000) for dry crystallization of sulfur-free Martiancompositions under reducing conditions (� log fO2

FMQ 0) (seebelow).

3. DISCUSSION

3.1. Crystallization Conditions and Phase Relations

In the following section we examine the crystallization con-ditions and phase relations. To do so, we use the programQUILF (Andersen et al., 1993) to compute the T and fO2

conditions during crystallization and to model the oxide-silicatephase relations, and complement these calculations withMELTS (Ghiorso and Sack 1995).

The comparison of the observed pyroxene compositions tothe calculated Aug�Pig solvus using QUILF allows the fol-lowing broad conclusions: First, initial crystallization of Mg-rich pigeonitic pyroxenes at �1150 °C was followed by pigeo-

nitic and augitic compositions between 1100 and 1050 °C.Second, between 1050 to 920 to 905 °C the pyroxene compo-sitions are subcalcic and cluster around the crest of the solvus.We interpret this to mean crystallization of a single-composi-tion pyroxene in this temperature interval. Below 920 to 905 °Csingle pyroxene compositions became unstable and exsolved toaugite and pigeonite. Finally, third, iron-enrichment was con-comitant with decrease in temperature (Fig. 5). At all timespyroxenes were likely in equilibrium with the residual silicatemelt and other crystallizing phases (e.g., plagioclase) but onlythe very iron-rich pyroxene compositions could have been inequilibrium with the clearly late iron-rich olivine and Fe-Tioxides. Even though MELTS predicts plagioclase crystalliza-tion before pyroxene(s), it is unclear from the textures whetherpyroxene preceded, crystallized simultaneously with, or fol-lowed plagioclase. Considering, however, that the plagioclasehas been transformed to maskelynite and the original textureshave been partially disturbed by the shock, our inability to timethe crystallization of plagioclase is not surprising.

Equilibria between ilmenite and titanomagnetite indicatetemperatures between 990 and 870 °C at FMQ-1.5 to FMQ-2.1,respectively (Fig. 6). These estimates are based upon the mostTi-rich titanomagnetite composition (Usp79Mt18), converselyMt-poor, which at these conditions would have been in equi-librium with the most Hm-rich ilmenite composition (Il95Hm4),taking into account variations in the Mg- and Mn-bearing oxidecomponents. However, considering the results from the mod-eling of the oxide-silicate phase relations, which we discuss indetail in the following paragraphs, we conclude that initialsingle-phase titanomagnetite crystallization probably occurredat near 990 °C and FMQ-1.5.

The textures and relations among the late-stage oxides andsilicates as well as the variability in the phase compositions canbe explained by the reactions:

single clinopyroxene � Ti-magnetite � olivine � ilmenite

(1)

single clinopyroxene � pigeonite � augite (2)

and

pigeonite � Ti-magnetite � olivine � augite � ilmenite.

(3)

Reactions 1 through 3 involve several net transfer and ex-change equilibria between oxides, oxides and silicates, py-roxenes, and olivine and pyroxene (Rumble 1970; Lindsley andFrost 1992). Some of the most helpful equilibria in betterunderstanding the model reactions are,

Fe2Si2O6 � 2Fe2TiO4 � 2Fe2SiO4 � 2FeTiO3 (4)

2Fe3O4 � Fe2Si2O6 � 6Fe2SiO4 � O2 (5)

Fe2TiO4 � Fe2O3 � Fe3O4 � FeTiO3 (6)

4Fe3O4 � O2 � 6Fe2O3 (7)

CaFeSiO4 � Fe2Si2O6 � CaFeSi2O6 � Fe2SiO4 (8)

CaFeSi2O6 (aug) � CaFeSi2O6 (pig) (9)

1873Crystallization conditions of Los Angeles meteorite

Fe2Si2O6 (aug) � Fe2Si2O6 (pig). (10)

Equivalent equilibria that involve Mg-bearing components areeasy to write as well. We modeled reactions 1 to 3 in theCa-Mg-Fe-Ti-Si-O system, using QUILF, as follows. Pressurewas kept constant at 0.1 GPa while the temperature waschanged incrementally from 1000 to 800 °C. As these are notpressure-dependent reactions the choice of a lower or higher

pressure value does not affect the results. We allowed theprogram to calculate, at different but fixed temperatures, theoxygen fugacity and compositions of titanomagnetite, olivine,pyroxenes, and the mole fraction of Fe2O3 in ilmenite. Themole fractions of the Mn- and Mg-bearing components first inilmenite, and second in titanomagnetite, were held constant ator near their respective maximum values. These values areconsidered as more representative of the original high-temper-

Fig. 6. T-fO2 conditions relative to FMQ at 0.1 Gpa for Los Angeles. Heavy-lined boxes represent oxygen fugacityconditions as inferred from oxides (gray), and oxides and silicates (black), (see text for discussion). Line ABC representsthe maximum range in T-fO2 conditions for the observed late-stage reactions betweens oxides and silicates. Duringsubsequent subsolidus re-equilibration along a TiMt-dominated path, TiMt and Ilm pairs inside the oxide-silicate clustersmay record temperature and oxygen fugacity conditions shown by the large arrow, i.e., towards more reducing conditionsrelative to the original. Correspondingly, the ilmenite grains may approach Fe3�-free compositions. Light-lined areasrepresent maximum T and the corresponding fO2 conditions for Ilm-TiMt or TiMt crystallization in other Martian basalticmeteorites.

1874 D. Xirouchakis, D. S. Draper, C. S. Schwandt, and A. Lanzirotti

ature concentrations. In both cases we obtained virtually iden-tical results for all the calculated parameters.

We obtained close matches between observed and calculatedmineral compositions, except ilmenite, along a cooling pathfrom 990 °C and FMQ-1.6 to 870 °C and FMQ-0.6 (linelabeled ABC in Fig. 6). Because of pyroxene exsolution below920 to 905 °C, the stable solid assemblage isCpx�Ol�TiMt�Ilm at �920 to 905 °C and it changes toPig�Aug�Ol�TiMt�Ilm at 920 to 905 °C. The majority ofthe observed titanomagnetite compositions falls within therange Usp67 to 72Mt22 to 32. Therefore, it is permissible that mostof the interaction between oxides and silicates occurred be-tween 950 and 900 °C at FMQ-1.0�0.2 (heavy-lined box inFig. 7), nearly simultaneously with pyroxene exsolution, whichprobably proceeded below 905 °C. In this case the product

pigeonitic compositions are inherently unstable with respect toaugite�olivine�silica. At �905 °C, the expected Ca content inaugite and olivine in equilibrium with silica (Qz) from thebreakdown of pigeonite is between 38 to 41 and 1 to 2 mol.%,respectively. These values coincide with a few of the observedaugite compositions and are marginally close to the olivine Cacontent (Fig. 5).

The assemblage olivine�augite�silica (Qz) indicates pres-sure conditions of 0.1 to 0.35 GPa in the range 850 to 900 °C,and the most pyroxene-like bulk symplectite composition of isnot stable relative to augite�olivine�quartz below 0.6 to 0.7GPa and 900 °C. Furthermore, if the late-stage silica phase wasoriginally tridymite, then at 900 °C (see below) the tridymiteto quartz inversion constrains the last stages of crystallizationto 0.06 GPa.

Fig. 7. Model crystallization sequence for Los Angeles (stone 1) calculated with MELTS. The sequence is plagioclase,two pyroxenes, whitlockite, olivine, titanomagnetite, ilmenite and apatite. There is an unexpected and perhaps, but notnecessarily, model-dependent hiatus in olivine crystallization at FMQ-2.5 and �1100°C. Titanomagnetite generally formsafter olivine except between FMQ-1 and FMQ-1.7. At 0.1 Gpa the stable silica phase to crystallize below 890°C is quartz.

1875Crystallization conditions of Los Angeles meteorite

We believe that reaction 3 with possible contributions fromthe breakdown of unstable Ca-poor and Fe-rich pyroxene com-positions at 900 °C and/or shock-related phases offers a morereasonable explanation for the formation of the symplectitesthan does the breakdown of a pyroxenoid. The fact that we didnot observe much ilmenite inside the symplectites does notmake the model inaccurate as ilmenite clearly formed at thesource of Ti (titanomagnetite) while in contact with olivine andalways close to if not locally in contact with augite (Fig. 1, 3).Furthermore, at the inferred most probable temperatures ofsymplectite formation (�900 to 950 °C), both the most Ca-poor and Fe-rich pyroxenes as well as the bulk composition ofpyroxene-like symplectites contain enough magnesium to havecrystallized as pyroxenes and not as pyroxenoids (e.g., Turnockand Lindsley, 1981). Therefore, we argue against the formerpresence of a pyroxenoid in Los Angeles and its replacementby augite�olivine�silica or via a pyroxenoid to pyroxenetransformation.

The close match between calculated and observed composi-tions can be seen for olivine and pyroxenes in Figure 5, and fortitanomagnetite in Figure 6. However, the difference betweenobserved and calculated ilmenite is lost in Figure 6 because theisopleths (shown as dashed lines) are for coexisting Mg-andMn-free Fe-Ti oxides only, and in this case ilmenite is affectedmore than titanomagnetite when these components are notaccounted for. Note that this is not the case for the calculationsdescribed above where Mg and Mn components were included.Specifically, the observed ilmenite compositional range isIl94 –98Hm0 –3, in contrast, the calculated ilmenite composi-tions range from Il94Hm4 at point A to Il91Hm6 at point C inFigure 6. We believe that the discrepancy between calculatedand observed ilmenite compositions results from the nearly-complete mantling of the titanomagnetite and ilmenite grainsby iron-rich olivine, which facilitates interoxide re-equilibra-tion with decreasing temperature. Because titanomagnetite isclearly more abundant than ilmenite within each of the olivine-titanomagnetite-ilmenite clusters, the interoxide re-equilibra-tion would have to follow paths, collectively shown as a gray-colored arrow in Figure 6, either parallel or subparallel tocorresponding ulvospinel isopleths. As a result of the Fe-Tiexchange between the oxides which may continue after thecomposition of silicates has locked in, ilmenite gains Ti andloses Fe3� while the Ti content of titanomagnetite may changelittle along such a path (Frost et al., 1988). Thus, individualpairs of titanomagnetite and ilmenite within the oxide-silicateclusters may also record a secondary T-fO2 path towards morereducing conditions. Interoxide re-equilibration may be respon-sible for the very Fe2O3-poor ilmenite compositions inQUE94201 (McSween et al., 1996; Mikouchi et al., 1998).Although we concur with other authors (Herd et al., 2000,2001a, 2001b; Wadhwa 2001) that QUE94201 crystallizedunder reducing conditions (Fig. 6), the low Fe3� concentrationsin QUE94201 ilmenite grains is not a full proof indicator ofsuch conditions (McSween et al., 1996). In addition, estimatesof fO2 from R3�/R2� ratios in plagioclase and/or pyroxene mayhave significant uncertainties associated with them. The parti-tioning of R3� and R2� elements between minerals and melt isa rather complex process, and it depends on mineral and meltcomposition as well as intensive parameters, e.g., T, aSiO2

, fO2.In the MELTS calculations of the crystallization sequence at

0.1 GPa we used the LA stone 1 bulk composition (Rubin et al.,2000a, 2000b) and we considered the following phases: pla-gioclase, clinopyroxene, olivine, spinel phase, ilmenite, quartz,whitlockite, and apatite. To include apatite we had to add atleast 0.02 to 0.05 wt.% H2O, to the original bulk composition.In the calculations we fixed the fO2 at different values withinthe range FMQ-3 to FMQ-0.2 while temperature was allowedto decrease from the liquidus down to 850 °C at steps of 2 or5 °C. The results are shown in Figure 7. Although the MELTScalculations broadly reproduce the observed mineralogical fea-tures, there are discrepancies between predicted and observedphase relations. For instance, MELTS predicts that two discretepyroxenes crystallize early from the liquid and continue to doso during the whole crystallization sequence. In addition, oli-vine is predicted to precipitate directly from the liquid, exceptbetween 1080 and 1120 °C at FMQ-2.5 where there is perhaps,but not necessarily, a model-dependent hiatus in olivine crys-tallization. In contrast, the textures clearly indicate that olivinecrystallized after and at the expense of titanomagnetite. How-ever, the MELTS calculations predict crystallization of titano-magnetite well before olivine near FMQ-1.5 and 1075 °C, infairly good agreement with our estimates for titanomagnetitecrystallization based on QUILF. Titanomagnetite crystalliza-tion before olivine may have suppressed the precipitation ofolivine from the melt in LA, as it would have depleted theresidual liquid in iron but also enriched it in silica. Silicaenrichment of the residual liquid and increases in silica activitymay have favored crystallization of pigeonite over olivine (e.g.,Toplis and Carroll, 1995). Although ilmenite is predicted byMELTS to crystallize from the melt, ilmenite crystallization issignificantly delayed relative to that of titanomagnetite. None-theless, both Fe-Ti oxides are predicted to coexist at �975 °C,in good agreement with the QUILF-based estimates (Fig. 6, 7).

3.3. Los Angeles and Other Martian Meteorites

To date Los Angeles is the most iron- and REE-rich Martianmeteorite (Rubin et al., 2000a, 2000b; Warren et al., 2000a),and it has similar ejection (�3 Ma) and crystallization (�175Ma) ages, respectively, with the basaltic meteorites Zagami andShergotty, and the peridotitic meteorites LEW88516, Y793605,and ALHA7705 (Nyquist et al., 2001). Though the basalticmeteorite QUE94201has a similar ejection age, its crystalliza-tion age is �330 Ma (Nyquist et al., 2001), and it is richer inTiO2 and P2O5 and depleted in LREE (Meyer 1998 and refer-ences therein). But QUE94201 and Los Angeles also exhibitsimilarities in pyroxene chemistry and perhaps oxide-silicateinteraction.

The similarities between the �175 Ma-old subset of Martianmeteorites can also be seen using Pearce Element Ratio (PER)analysis (Pearce, 1968; Russell and Nicholls, 1988; Nicholls,1988). The key feature of PER analysis is that the elementratios used have as their denominator a constituent, termed aconserved element, that does not take part in the process(es)that resulted in the variance exhibited by a given data set.Highly incompatible elements are excellent candidates for thispurpose, because they are not removed by crystallizing miner-als. Thus, ratios of conserved elements in a suite of comag-matic rocks remain essentially constant (within analytical un-

1876 D. Xirouchakis, D. S. Draper, C. S. Schwandt, and A. Lanzirotti

certainty) provided they are not perturbed by other processes,such as crustal contamination.

Ratios of conserved elements provide tests for whether agiven set of magmatic rocks could be cogenetic. For threeconserved elements X, Y, and Z, a plot of the ratios X/Z versusY/Z for a set of cogenetic rocks would, ideally, show all thepoints in exactly the same location. However, chemical analy-sis is not ideal, so the criterion would instead be whether thepoints fall within the propagated analytical uncertainties for thetwo ratios plotted. PER plots cannot prove a hypothesis iscorrect, but are used rather to reject hypotheses that are notconsistent with the data. Thus, if the data fall far outside theuncertainty envelope on a PER diagram plotting two ratios ofconserved elements, the hypothesis that the samples are coge-netic must be rejected.

To apply a test of the hypothesis that the �175-Ma-oldsubset of SNCs are cogenetic, we must identify at least threeconserved elements. There are surprisingly few elements forwhich most of the SNCs have been analyzed that are highly

incompatible in the minerals potentially important in SNCmagma genesis (olivine, pyroxenes, garnet, plagioclase), im-mobile during terrestrial weathering, and not concentrated in aputative martian crust (many authors have argued that crustalcontamination could be an important process for SNCs). Tak-ing these factors into account, the best choices for conservedelements are the light to middle REE, and high field-strengthelements such as Zr and Ta. LREE may be more susceptible tocrustal contamination than the middle REE.

Figure 8 is a set of PER cogenetic-test diagrams that plotratio pairs of the elements La, Sm, Tb, and Ta for all the SNCmeteorites except ALH84001, whose ancient age and otherfeatures make it unique, and the clinopyroxenite Lafayette, forwhich there are no published data for these elements. Error barson this diagram are arbitrarily chosen by assuming 10% relativeuncertainties on all the data plotted, which is admittedly sim-plistic but probably not unreasonable given that the data wereproduced from many different laboratories over the past threedecades; a rigorous assessment of the uncertainties on all these

Fig. 8. Pearce Element Ratio diagrams testing possible cogenetic relationships among SNC meteorite subsets (ALH84001not plotted). Symbol types show general age ranges. Error bars plotted under arbitrary assumption of 10% relativeuncertainty on each datum. Abbreviations used: 77005, ALH77005; 79001/A,B, EETA79001 lithology A and B; LEW,LEW88516; Dhofar, Dhofar 019; Dar al Gani, Dar al Gani 476; Yamato, Y793605; LA-1,2, Los Angeles stone 1 and 2;QUE, QUE94201; GV, Governador Valadares; Sherg, Shergotty. Data are averaged values fro sources referenced in Meyer(1998) except for Los Angeles (Rubin et al. 2000b), Dhofar 019 (Neal et al. 2001), and Dar al Gani 476 (Zipfel et al. 2000).On each PER diagram, the same three groupings emerge. One: the nakhlites Governador Valadares and Nahkla (possiblywith Chassigny). Two: Shergotty, Zagami, LEW88516, LA-1 (and possibly LA-2). Three: Dar al Gani 476, EETA79001,ALH77005, and Dhofar 019.

1877Crystallization conditions of Los Angeles meteorite

data is beyond the scope of this study. Nevertheless, it can beseen on all these plots that three groupings appear fairly con-sistent, which suggests either that all four elements are con-served to a similar extent, or that they are all perturbed bysimilar processes; we prefer the former interpretation.

In the first grouping in Figure 8, the basaltic meteorites LosAngeles, Shergotty, and Zagami cluster near the lherzoliticmeteorites LEW88516. In the second, the other �175 Ma oldrocks EETA79001 (lithologies A and B) and ALH77005 and,particularly on the plot of Tb/Sm vs. Ta/Sm, the �500 Ma oldDar al Gani 476 and Dhofar 019, tend to group together. Thethird grouping includes the �1.2 Ga-old clinopyroxenites Na-khla and Governador Valadares, as well as the dunite Chas-signy. QUE94201 plots significantly away from all the otherSNCs on these plots, largely because its Ta content is verysmall. On the basis of these relationships, and assuming thatcrustal contamination does not perturb all of the elements used,the hypothesis that all of the �175 Ma-old samples are coge-netic must be rejected. However, Figure 8 permits the interpre-tation that the meteorites in each of the first two SNC groupingslisted above could be cogenetic, subject to further, completeassessment of analytical uncertainties on the data plotted in thatfigure as well as additional geochemical information. We em-phasize that cogenetic in this context is not equivalent tocomagmatic; for the second grouping, for example, the differ-ence in crystallization ages obviously negates that possibility.

This �175 Ma subset of Martian meteorites also follow atholeiitic trend on an AFM diagram, originally described byMcSween et al. (1999) for all meteorite finds, Mars Pathfinderrocks, and estimates of intercumulus melts of Martian meteor-ites. For Los Angeles, Zagami, and Shergotty, our oxide oxy-gen-barometry and thermometry estimates indicate oxygen fu-gacity conditions between 1 and 2 logfO2 units below FMQ inthe range 700 to 1000 °C, in good agreement with tholeiiticrocks on Earth (e.g., Frost and Lindsley 1992). For instance,using mineral chemistry data from Stolper and McSween(1979), Smith and Hervig (1979), and Ghosal et al. (1998) at acommon pressure of 1 bar, we estimate 1 to 2 logfO2 unitsbelow FMQ at 700 to 810 °C for Shergotty, and 1 to 1.5 logfO2

units below FMQ at 750 to 810 °C for Zagami (see also Herdand Papike 2000; Herd et al., 2001a, 2001b). Using the samedata base, we also estimated the T-fO2 conditions (Fig. 6) forcoexisting titanomagnetite-ilmenite in EET79001A (860 to 940°C and FMQ-2.7 to FMQ-1.5) and EET79001A (875 to 990 °Cand FMQ-2.7 to FMQ-1.8), and titanomagnetite crystallizationin QUE94201 (1020 to 1090 °C and FMQ-3.6 to FMQ-1.9).The latter estimates are in good agreement with the 1 atmexperimental data of Koizumi et al. (2001) who observedappearance of titanomagnetite, but not ilmenite, at 1050 °C andFMQ-2.5 in controlled crystallization experiments of a syn-thetic QUE94201 bulk composition. Also noteworthy is theapparent continuity in the mg# of the pyroxenes in the peridot-itic and basaltic meteorites regardless of calcium content, gen-erally and particularly in this subset.

The textures and phase relations, bulk composition and min-eral chemistry trends, and oxygen fugacity estimates suggestthat the basaltic Martian meteorites Shergotty and Los Angeles,and to a lesser extent Zagami, appear to share the followingcharacteristics: (i) Crystallization at relatively reducing condi-tions but not near iron metal saturation, i.e., 1 to 2 logfO2 units

below FMQ. (ii) Iron enrichment of the mafic silicate phases.(iii) Late crystallization of the Fe-Ti oxides. (iv) Oxide-silicateinteraction probably in the presence of an iron-depleted, silica-rich liquid. And (v) crystallization of a silica polymorph only atthe very late stages of crystallization.

The oxide-silicate interaction in Los Angeles and other ba-saltic meteorites has the following implications which shouldbe considered in interpretations of geophysical and geochemi-cal data of Martian meteorites and/or the surface of the planet.First, it controls the T-fO2 equilibration path of a given rock.Hence, the relative differences in fO2 among the basaltic me-teorites may conceivably reflect the interplay between fO2, bulkcomposition, and phase relations (e.g., Toplis et al., 1994;Toplis and Carroll, 1995; Frost and Lindsley, 1992). Second,oxide-silicate interaction may affect the stability and composi-tion of magnetite, one of the principal carriers of rock magne-tism. Therefore reactions that destabilize, consume, or changethe composition of magnetite or other magnetic minerals (e.g.,iron carbonates, and sulfides) have to be considered in theinterpretations of Martian magnetic fields.

4. SUMMARY

We conclude that in Los Angeles, an iron-and REE-richMartian microgabbro, pyroxene and plagioclase crystallizationwas nearly simultaneous. Initial crystallization of Mg-rich pi-geonitic pyroxenes at �1150 °C was followed by pigeoniticand augitic compositions between 1100 and 1050 °C. Subse-quently and from 1050 to 950 °C only a single pyroxenecrystallized, and below 950 °C it became unstable and exsolvedto augite and pigeonite. Fe-Ti oxides, apatite, and silica satu-ration in Los Angeles occurred late after 95 to 97% solidifica-tion. Although some phosphate crystallization may have oc-curred early, the majority of it appears to have taken placenearly simultaneously with titanomagnetite. In addition, eventhough pyroxene and titanomagnetite may have started to reactafter the latter crystallized at �990 °C and �FMQ-1.5, most ofthe reaction between titanomagnetite and pyroxene probablyoccurred between 950 and 900 °C at FMQ-1.0�0.2 and atabout the time single pyroxene started to exsolve, producing thefinal Pig�Aug�Ol�Ilm�TiMt assemblage(s). The combina-tion of the proposed reaction between titanomagnetite andpyroxene, and pyroxene exsolution is likely responsible for thesymplectites in Los Angeles. Whether this mechanism mayexplain similar symplectites in other basaltic Martian meteor-ites remains to be evaluated; however, it seems plausible fromthe textural descriptions. Possible contributions to the symplec-tites from the shock-related alkali- and silica-rich glass and/orshocked plagioclase, and finally, the breakdown of a low-Capyroxene during cooling below 900 °C cannot be precluded.

We contend that oxide-silicate interaction in Martian mete-orites has to be carefully evaluated as it bears on the T-fO2

crystallization conditions, rock magnetism, and the composi-tion of the residual silicate melts. In addition, the nature of thesilica polymorph and its relation to the late iron-rich olivine andpyroxene rim compositions remains unresolved, although thesefeatures may help constrain the depth of crystallization of theMartian meteorites that contain them. Finally, the inferredpresence of pyroxferroite in other basaltic Martian meteoriteshas to be carefully re-evaluated.

1878 D. Xirouchakis, D. S. Draper, C. S. Schwandt, and A. Lanzirotti

Acknowledgments—We thank AMNH and the late M. Prinz for the LAsample, two anonymous reviewers, H. Palme, D. Mittlefehldt, J.H.Jones, and J.B. Parise for comments, and D. H. Lindsley and B.R. Frostfor discussions. We bear, however, sole responsibility for the conclu-sions. We gratefully acknowledge financial support by NASA grantNCC 980 to B. Wilson and 344-31-20-25 to C. Agee. The micro-XRFand micro-XRD studies at X26A were supported by DOE grantDOEFG0292ER14244 to S. Sutton (CARS/U. Chicago) and NSF grantEAR-9724501 to J.B. Parise (SUNY Stony Brook, Geosciences), re-spectively. We thank them all.

Associate editor: H. Palme

REFERENCES

Andersen D. J., Lindsley D. H. and Davidson P. M. (1993) QUILF: APascal program to assess equilibria among Fe-Mg-Mn-Ti oxides,pyroxenes, olivine, and quartz. Comp. Geosc. 19, 9., 1333–1350.

Aramovich C. J., Herd C. D. K. and Papike J. J. (2001). Possible causesfor late-stage reaction textures associated with pyroxferroite andmetastable pyroxenes in the basaltic Martian meteorites. LunarPlanet. Sci. Conf., XXXII, Lunar Planet. Inst., Houston, #1003 (ab-str.).

Burnham, C. W. (1971) The crystal structure of pyroxferroite fromMare Tranquillitatis. Proc. 2nd Lunar Sci. Conf. 1, 47–57.

Chao E. C., Minkin J. A., Frondel C., Klein C. Jr., Drake J. C., FuchsL., Tani B., Smith J. V., Anderson A. T., Moore P. B., ZechmanG. R. Jr., Trail R. J., Plant A. G., Douglas J. A. V. and Dence M. R.(1970) Pyroxferroite, a new calcium-bearing iron silicate from Tran-quillity Base. Proc. Apollo 11 Lunar Sci. Conf., Geochim. Cosmo-chim. Acta, Suppl. 1, 65–79.

Chen M. and El Goresy A. (2000). The nature of maskelynite inshocked meteorites: not diaplectic glass but a glass quenched fromshocked-induced dense melt at high pressures. Earth Planet. Sci.Lett. 179, 489–502.

Eng P. J., Newville M., Rivers M. L., and Sutton, S.R. (1998) Dynam-ically figured Kirkpatrick Baez Synchrotron-XRD focusing optics. InX-ray Microfocusing: Applications and Technique (ed. I. McNulty)Proc. SPIE. 3449, 145.

El Goresy A., Dubrovinsky L., Sharp T. G., Saxena S., and Chen M.(2000). A monoclinic post-stishovite polymorph of silica in theShergotty meteorite. Science. 288, 1631–1634.

Frost B. R., Lindsley D. H., and Andersen D. J. (1988) Fe-Ti oxide-silicate equilibria: Assemblages with fayalitic olivine. Am. Mineral.73, 727–740.

Frost R. B. and Lindsley D. H. (1992) Equilibria among Fe-Ti oxides,pyroxenes, olivine, and quartz: Part II. Appl. Am. Mineral. 77,1004–1020.

Greenwood J. P., Warren P. H., and Rubin A. E. (2000). Late-stagecrystallization of Los Angeles, a new basaltic shergotttite. LunarPlanet. Sci. Conf., XXXI, Lunar Planet. Inst., Houston, #2074 (ab-str.).

Ghiorso M. S. and Sack R. O. (1995) Chemical mass transfer inmagmatic processes IV. A revised and internally consistent thermo-dynamic model for the interpolation and extrapolation of liquid-solidequilibria in magmatic systems at elevated temperatures and pres-sures. Contrib. Mineral. Petrol. 119, 197–212.

Ghosal S., Sack R. O., Ghiorso M. S., and Lipschutz M. E. (1998)Evidence for a reduced, Fe-depleted martian mantle source region ofshergottites. Contrib. Mineral. Petrol. 130, 346–357.

Herd C. D. K, Papike J. J. and Brearley A. J. (2001a). Oxygen fugacityof the Martian basalts from electron microprobe oxygen and TEM-EELS analyses of Fe-Ti oxides. Am. Min. 86, 1015–10024.

Herd C. D. K., Borg L. E., and Papike J. J. (2001b). Controls on oxygenfugacity during Martian basaltic petrogenesis: Clues from geochemi-cal correlations. Lunar Planet. Sci. Conf., XXXII, Lunar Planet. Inst.,Houston, #1150 (abstr.).

Herd C. D. K. and Papike J. J. (2000). Oxygen fugacity of the Martianbasalts from analysis of iron-titanium oxides: Implications for man-tle-crust interaction on Mars. 63rd Ann. Meteor. Soc. Meet. #5085(abstr.).

Koizumi E., McKay G. A., Schwandt C. S., and Le L. (2001) Crystal-lization of Martian basalt QUE94201. 17th Ann. Summer Intern Conf.13–15. Lunar Planet. Inst., Houston, TX.

Lindsley D. H. (1967) The join hedenbergite-ferrosilite at high pres-sures and temperatures. Carnegie Inst. Wash., Year B. 65, 230.

Lindsley D. H. and Munoz J. L. (1969) Subsolidus relations along thejoin hedenbergite—ferrosilite. Am. J. Sci., Schairer volume. 267-A,295–324.

Lindsley D. H. and Burnham C. W. (1970) Pyroxferroite: Stability andX-ray Crystallography of Synthetic Ca0.15Fe0.85SiO3 Pyroxenoid.Science 168, 364–367.

Lindsley D. H. and Frost B. R. (1992) Equilibria among Fe-Ti oxides,pyroxenes, olivine, and quartz: Part I. Theory. Am. Mineral. 77,987–1003.

McSween H. Y. Jr., Murchie S. L., Crisp J. A., Bridges N. T., AndersonR. C., Bell J. F. III, Britt D. T., Bruckner J., Dreibus G., EconomouT., Ghosh A., Golombek M. P., Greenwood J. P., Johnson J. R.,Moore H. J., Morris R. V., Parker T. J., Rieder R., Singer R., andWanke H. (1999) Chemical, multispectral, and textural constraintson the composition and origin of rocks at the Mars Pathfinder site. J.Geophys. Res. 104. 4, 8679–8715.

McSween H. Y. Jr., Eisenhour D. D., Taylor L. A., Wadhwa M., andCrozaz G. (1996) QUE94201 shergottite: Crystallization of a Mar-tian basaltic magma. Geochim. Cosmochim. Acta. 60, 22., 4563–4569.

Meyer C. (1998) Mars Meteorite Compendium-1998, Houston, TX,NASA Johnson Space Center, 237 p.

Mikouchi T. (2000). Pyroxene and plagioclase in the Los AngelesMartian meteorite: Comparison with the Queen Alexandra Range94201 Martian meteorite and Asuka 881757 Lunar meteorite. 63rd

Ann. Meteor. Soc. Meet., #5193 (abstr.).Mikouchi T., Miyamoto M., and McKay G.A. (1998) Mineralogy of

Antarctic basaltic shergottite Queen Alexandra Range 94201: Sim-ilarities to Elephant Moraine A79001 (lithology B) martian meteor-ite. Meteor. Planet. Sci. 33, 181–189.

Minitti M. E. and Rutherford M. J. (2000). Genesis of the MarsPathfinder “sulfur-free” rock from SNC parental liquids. Geochim.Cosmochim. Acta. 64, 14., 2535–2547.

Neal C. R., Taylor L. A., Ely J. C., Jain J. C., and Nazarov M. A.(2001). Detailed geochemistry of new shergottite, Dhofar 019. LunarPlanet. Sci. Conf., XXXII, Lunar Planet. Inst., Houston, #1671 (ab-str.).

Nicholls J. (1988), The statistics of Pearce element diagrams and theChayes closure problem. Contrib. Mineral. Petrol. 99, 11–24.

Nyquist L. E., Reese Y., Wiesmann H., Shih C.-Y. (2001) Age ofEET79001B and implications for shergottite origins. Lunar Planet.Sci. Conf., XXXII, Lunar Planet. Inst., Houston, #1407 (abstr.).

Pearce T. H. (1968) A contribution to the theory of variation diagrams.Contrib. Mineral. Petrol. 1, 142–157.

Rao M. N., Bogard D. D., Nyquist L. E., McKay D. S., and Masarik J.(2002) Neutron Capture. Isotopes in the Martian Regolith: Implica-tions for Martian Atmospheric Noble Gases. Icarus (in press).

Rubin A. E., Warren P. H., Greenwood J. P., Verish R. S., Leshin L. A.,Hervig R. L. (2000a). Los Angeles: The most differentiated basalticmartian meteorite. Geology 28, 11, 1011–1014.

Rubin A. E., Warren P. H., Greenwood J. P., Verish R. S., Leshin L. A.,Hervig R. L. (2000b) Petrology of Los Angeles: A new basalticshergottite find. Lunar Planet. Sci. Conf., XXXI, Lunar Planet. Inst.,Houston, #1963 (abstr.).

Rumble D. III (1970) Thermodynamic analysis of phase equilibria inthe system Fe2TiO4-Fe3O4-TiO2. Carnegie Inst. Wash., Year B. 69,198–207.

Russell J. K., and Nicholls J. (1988) Analysis of petrologic hypotheseswith Pearce element ratios. Contrib. Mineral. Petrol. 99, 25–35.

Sharp T. G., El Goresy A., Wopenka B., and Chen M. (1999) APost-Stishovite SiO2 Polymorph in the Meteorite Shergotty: Impli-cations for Impact Events. Science. 284, 1511–1513.

Smith J. V. and Hervig R. L. (1979) Shergotty meteorite: mineralogy,petrography and minor elements. Meteor. 14, 121–142.

Stolper E. and McSween H. Y. Jr. (1979) Petrology and origin of theshergottite meteorites. Geochim. Cosmochim. Acta. 43, 1475–1498.

Turnock A. C. (1962) Preliminary results on melting relations of

1879Crystallization conditions of Los Angeles meteorite

synthetic pyroxenes on the diopside-hedenbergite join. CarnegieInst. Wash., Year B. 61, 81–82.

Turnock A. C. and Lindsley D. H. (1981) Experimental determinationof pyroxene solvi for �1 kbar at 900 and 1000 °C. Can. Min. 19,255–267.

Toplis M. J., Libourel G., Carroll M. R. (1994) The role of phosphorusin crystallization processes of basalt: An experimental study.Geochim. Cosmochim. Acta. 58, 2, 797–810.

Toplis M. J. and Carroll M. R. (1995) An experimental study of theinfluence of oxygen fugacity on Fe-Ti oxides stability, phase rela-tions, and mineral-melt equilibria in ferro-basaltic systems. J. Petrol.36, 5, 1137–1170.

Wadhwa M., Crozaz G., Lentz R. C. F., and McSween H. Y. (2001).Trace element microdistributions in Los Angeles: A new basaltic

shergottite similar to, yet distinct from, the others. Lunar Planet. Sci.Conf., XXXII, Lunar Planet. Inst., Houston, #1106 (abstr.).

Wadhwa M. (2001) Redox State of Mars’ Upper Mantle and Crust fromEu Anomalies in Shergottite Pyroxenes. Science 291, 5508, 1527–1530.

Warren P. H., Greenwood J. P., Richardson J. W., and Rubin A. E.(2000a) Geochemistry of Los Angeles, a ferroan, La- and. Th-richbasalt from Mars. Lunar Planet. Sci. Conf., XXXI, Lunar Planet.Inst., Houston, #2001 (abstr.).

Warren P. H., Greenwood J. P. and Rubin A. E. (2000b) Los Angelesat Chicago. 63rd Ann. Meteor. Soc. Meet., #5323 (abstr.).

Zipfel J., Scherer P., Spettel B., Dreibus G., and Schultz L. (2000).Petrology and chemistry of the new shergottite Dar al Gani 476.Meteor. Planet. Sci. 35, 95–106.

1880 D. Xirouchakis, D. S. Draper, C. S. Schwandt, and A. Lanzirotti