Paired nakhlites MIL 090030, 090032, 090136, and 03346: Insights into the Miller Range parent meteorite


  • Arya UDRY,

    Corresponding author
    1. Department of Earth and Planetary Sciences and Planetary Geosciences Institute, University of Tennessee, Knoxville, Tennessee 37996, USA
      *Corresponding author. E-mail:
    Search for more papers by this author
  • Harry Y. McSWEEN Jr,

    1. Department of Earth and Planetary Sciences and Planetary Geosciences Institute, University of Tennessee, Knoxville, Tennessee 37996, USA
    Search for more papers by this author

    1. Department of Geosciences, Virginia Polytechnic Institute and State University, Blacksburg, Virginia 24061, USA
    Search for more papers by this author
  • Robert J. BODNAR

    1. Department of Geosciences, Virginia Polytechnic Institute and State University, Blacksburg, Virginia 24061, USA
    Search for more papers by this author

*Corresponding author. E-mail:


Abstract– Miller Range (MIL) 03346 is the most oxidized and least equilibrated nakhlite known and displays the highest amount of intercumulus phase. The discovery of three new nakhlites, MIL 090030, MIL 090032, and MIL 090136, in the Miller Range, Antarctica, geographically close to the location of MIL 03346, suggests that they may come from the same parent meteorite. In this study, we investigate the mineralogy and texture of cumulus and intercumulus phases, in situ major and trace element compositions for the cumulus phases, as well as pyroxene crystal size distribution patterns and spatial distribution patterns of MIL 090030, 090032, and 090136. Using these combined results, we conclude that the three nakhlites studied here are paired with MIL 03346. However, modal mineral abundances of MIL 090030, 090032, 090136, and 03346 exhibit variations indicating that a single sample of a Miller Range nakhlite is not modally representative of the parent meteorite and that analyses of multiple samples for a single nakhlite may be necessary to obtain representative modal data for placement in the cumulate pile. Our calculated parental melt composition based on all the paired samples confirms a previous study suggesting that the nakhlite parent melt crystallized as a closed system.


Nakhlites are clinopyroxenites that are thought to derive from Mars (e.g., McSween 1985; Treiman et al. 2000). At present, a total of eight nakhlites have been recovered, assuming that the Miller Range (MIL) nakhlites studied here come from the same parent meteorite. Nakhlites share similar cumulate textures, whole-rock (light rare earth elements [LREE]-enriched) compositions, and mineralogies. Furthermore, they all have a similar cosmogenic exposure age of approximately 10–11 Ma, indicating a similar ejection time from Mars (Nyquist et al. 2001), and the same igneous crystallization age of approximately 1.3 Ga, which has been unambiguously defined by four independent dating techniques: 87Rb/87Sr, 147Sm/144Nd, 39Ar/40Ar, and U/Pb (Nyquist et al. 2001).

Nakhlites are cumulates consisting of euhedral to subhedral augite and less abundant olivine with a fine-grained intercumulus phase. This intercumulus phase mainly consists of glass, usually with laths of sodic plagioclase (except in MIL 03346; Day et al. 2006), silica, phosphate, pyroxene, and minor opaque phases such as titanomagnetite and sulfides (e.g., Treiman 2005). The large pyroxene and olivine grains that dominate nakhlites are interpreted as cumulus phases that crystallized at low pressure in a shallow magma body (Day et al. 2006). Moreover, the presence of cumulus and intercumulus materials indicates a two-stage cooling history. Both pyroxene and olivine are zoned. Augite and olivine cores are relatively magnesium-rich, whereas the rims of both minerals are increasingly ferroan with an average thickness of 10–20 μm (Treiman 2005). The thin rims indicate continuous diffusive exchange with a fractionating magma (McSween and Treiman 1998).

Due to the fact that nakhlites all have textural, compositional, and chronologic similarities, it is believed that they all were ejected from a single thick lava flow or sill sequence. Based on their modal abundances and equilibration differences (e.g., thickness of the rims, enrichment in Fe, and depletion in Ca in rims), each nakhlite is considered to have originated from a different location in the nakhlite cumulate pile. The model put forth by Treiman (2005), Day et al. (2006), and Mikouchi et al. (2003) suggests that cumulus phases first crystallized at a shallow level from magma derived from a LREE-depleted mantle source. Treiman (2005) and Day et al. (2006) suggested that the LREE enrichment observed in the whole-rock composition is due to limited partial melting of the mantle source, which probably contains garnet, or to LREE enrichment prior to segregation of the parental magma. The crystallization of cumulus phases was followed by an entrainment of cumulus phases within the melt and eruption at, or emplacement near, the Martian surface. Pyroxene and olivine then settled, and partially equilibrated with the intercumulus phase. Thus, nakhlites that experienced the greatest extent of equilibration are inferred to have been located near the base of the cumulate pile, whereas nakhlites showing less equilibration are assigned to positions nearer the top of the pile. Because stratigraphically lower samples would likely have squeezed out intercumulus melt that would be redistributed upward, the proportion of intercumulus phases, as well as its crystallinity and the phases present in it could also indicate relative position within the pile (e.g., Sautter et al. 2002; Mikouchi et al. 2003, 2012; Day et al. 2006; Jambon et al. 2010).

MIL 03346 was recovered in the Miller Range, Antarctica, during the 2003 season. It has been postulated that MIL 03346 was located at or near the top of the cumulate pile because it displays the least extensive equilibration of cumulus minerals, has a high proportion of intercumulus phase (Day et al. 2006), and is the most highly oxidized of the nakhlites (Dyar et al. 2005). Another nakhlite, NWA 5790, displays a higher amount of intercumulus phase; consequently, it was placed at the topmost section of the cumulate pile by Jambon et al. (2010). However, complete results have yet to be published, so this nakhlite will not be considered here. Nakhlites MIL 090030 (452.63 g), MIL 090032 (532.19 g), and MIL 090136 (170.98 g) were collected during the 2009 season in the Miller Range, geographically close to the location of MIL 03346 (Fig. 1). The fact that multiple samples of such a rare meteorite type were found in such geographically close proximity indicates the possibility that the four meteorites may be paired. Using textural, geochemical, and petrographic evidence, we confirm the pairing of MIL 090030, MIL 090032, MIL 090136, and MIL 03346. We also examine the scale of petrologic variability in the combined, more representative sample, and consider whether modes vary systematically at different levels within the nakhlite cumulate pile.

Figure 1.

 Map of the Miller Range, Antarctica, showing the location of MIL 090030, MIL 090032, MIL 090136, and MIL 03346. The meteorites were found in two pairs approximately 4 km apart, and each specimen in the pair was found about 200 m apart. The lines represent undisclosed latitude and longitude and smaller points are other MIL meteorites. Map courtesy of J. Schutt. The inset represents the Antarctica continent showing the Miller Range and Yamato location sites.


One thin section of each nakhlite was analyzed, with surface areas of 58 mm2 for MIL 090030,21; 101 mm2 for MIL 090032,21; and 29 mm2 for MIL 090136,22.

The modal abundances of cumulus pyroxene and olivine as well as intercumulus phase were estimated using pixel counting with Image J software. Each cumulus phase grain was drawn in Adobe Illustrator and the rest of the thin section was assumed to correspond to the intercumulus phase (Fig. 2). Then, a gray value was assigned for each phase. Using Image J, we calculated the proportion of pixels of each gray tone.

Figure 2.

 Rendering of the distribution of textures and phases from reflected optical microscopic images of MIL 090030 (a), MIL 090032 (b), and MIL 090136 (c) nakhlites, with pyroxene shown in black, olivine in light gray, and intercumulus phase in dark gray. The three images are at the same scale and were made using Adobe Illustrator.

In situ major and minor element compositions were determined using the CAMECA SX-100 electron microprobe (EMP) at the University of Tennessee. Operating conditions for the analyses of pyroxene, olivine, chromite, glass, and sulfides were an accelerating voltage of 15 kV, beam currents of 20 nA (pyroxene, glass, chromite, and metals) to 30 nA (olivine), and beam sizes of 2 μm (pyroxene, olivine, and chromite) to 5 μm (glass) using the wavelength dispersive (WDS) mode and standard PAP corrections. The counting times for all elements were 20 s, except for Ca (30 s), Cr (30 s), and Ni (50 s) in olivine. The detection limits (3σ above background) were 0.03 wt% for SiO2, TiO2, Al2O3, Cr2O3, MgO, CaO, NiO, P2O5, Na2O, and K2O, and 0.05 wt% for FeO and for Co and Ni in metals. Natural and synthetic standards were used for calibration.

Trace element abundances were measured at Virginia Tech using an Agilent 7500ce inductively coupled plasma mass spectrometer (ICP-MS) combined with an Excimer 193 nm GeoLasPro Laser Ablation (LA) system. We used spot sizes between 16 and 32 μm depending on the grain size. Reference glass NIST SRM610 was used as the external standard. Ca abundances measured with the EMP were used as the internal standard to quantify elemental concentrations using the ratios obtained from the LA-ICP-MS analysis. For each of the analyses, an approximately 60 s background signal was collected before the ablation process was initiated. The NIST standard was measured two times for 60 s before and after analysis of each sample. Each point analysis was measured for 40–60 s. The samples were ablated in a helium atmosphere. The helium gas carrying the ablated particles was then mixed with the make-up gas (argon) before entering the ICP-MS system. The analytical precision for the LA-ICPMS system for the elements analyzed for this study (compared to NIST612) is better than ±5% (for V, Co, Cr, Ce, Zn, Ga); between ±5 and 10% (for Sr, Ba, Pr, Eu, Hf); between ±10 and 20% (for Y, Zr, La, Nd, Sm, Gd, Dy, Er, Yb, Lu, Th, Tb, Ho), and poorer than ±20% for Sc. The detection limits are dependent on several factors such as spot size, element analyzed, and the duration of the analysis. Therefore, each analysis and element has a different detection limit. For all of the trace elements analyzed, the detection limit obtained during our calibration runs described above was less than one ppm. Data reduction was done using the Analysis Management System (AMS) software (Mutchler et al. 2008) enabling time-resolved signal analysis.

Petrology and Geochemistry

Petrographic Descriptions

MIL 090030, MIL 090032, and MIL 090136 all display the same mineralogy. They are cumulate clinopyroxenites with minor amounts of olivine set in a vitrophyric intercumulus phase (Fig. 3). Modal abundances for the three nakhlites, listed in the order MIL 090030, MIL 090032, MIL 090136, are as follows: 66.3%, 74.3%, and 64.1% pyroxene; 9.8%, 3.9%, and 9.5% olivine; and 23.9%, 21.9%, and 26.4% intercumulus phase, respectively (also see Table 1).

Figure 3.

 Plane-polarized photomicrographs (a, c, e, g, and h) and backscattered electron images (b, d, and f) of a) and b) MIL 090030, c) and d) MIL 090032, e) and f) MIL 090136, and g) and h) MIL 03346. The top photographs show whole thin sections all at the same scale. The bottom images, at higher magnification, show cumulus pyroxene and minor olivine with the dark intercumulus phase. Images g) and h) reprinted from Day et al. (2006) with permission.

Table 1.   Modal abundances (%) of nakhlites. Different modal analyses are represented for MIL 03346.
SamplePyroxeneOlivineIntercumulus matrix
  1. a Hallis and Taylor (2011); bDay et al. 2006; cImae and Ikeda 2007; dDyar et al. 2005; eMcKay and Schwandt 2005; fWeighted modal abundances using MIL 090030/32/136 areas and modal abundances from this study and MIL 03346 values from b and c; gSautter et al. 2002; hImae et al. 2003; iLentz et al. 1999; jTreiman 2005.

MIL 090030,2166.39.823.9
MIL 090032,2174.33.821.9
MIL 090136,2264.19.526.4
MIL 090030,23a66.48.322.6
MIL 090032,24a75.03.021.2
MIL 090136,21a75.25.618.1
MIL 03346b78.41.319.8
MIL 03346c67.70.831.5
MIL 03346d70.83.026.2
MIL 03346e79.84.615.7
Weighted modal abundancef72.33.923.7
NWA 817g69.012.518.5
Gov. Vala.i81.28.99.8
NWA 998j70.010.019.0

The pyroxenes are euhedral to subhedral augite prisms, which display Fe-rich rims averaging 10 μm in thickness (Figs. 3 and 4). Their average sizes are 0.38 × 0.23 mm, 0.40 × 0.24 mm, and 0.41 × 0.24 mm for MIL 090030, MIL 090032, and MIL 090136, respectively. Pyroxene lengths can reach 1.8 cm. No Fe-rich rims are present at pyroxene grain boundaries, indicating that the rims formed after pyroxene grains were in contact with each other. They sometimes show single twin planes, a feature probably resulting from impact shock. Some rare Fe-rich pyroxenes are present in the intercumulus phase. Large patchy melt inclusions (∼100 μm), as observed by Sautter et al. (2006) in MIL 03346, were also found in the pyroxenes of MIL 090030/32/136. However, the sizes of the different phases within these melt inclusions were too small to obtain accurate major element compositions with the electron microprobe. No Cl-amphiboles, which were observed in MIL 03346 (Sautter et al. 2006), were detected in this study.

Figure 4.

 Backscattered electron images of intercumulus phase. a) Skeletal Ti-magnetite in MIL 090030, b) pyrrhotite-fayalite-Ti-magnetite “filaments” in MIL 090032, and c) fayalite and P-rich needles in MIL 090136. The Fe-rich pyroxene rims appear lighter than the cores.

Two types of olivine are distinguished in MIL 090030/32/136: rare, large euhedral to subhedral olivines, which display Fe-rich rims (Fig. 4a), and skeletal fayalites found in the intercumulus phase (Fig. 4c). Unlike the pyroxenes, the large olivines display thick Fe-rich rims that can reach approximately 100 μm. Olivines enclose rounded pyroxene grains, showing that pyroxene crystallized before olivine. Olivine also displays rounded crystallized melt inclusions ranging up to 100 μm.

The intercumulus phase is mainly composed of incompletely crystallized feldspathic glass containing fine-grained fayalite laths (∼5 μm, see Fig. 4c), pyrrhotite, pyroxene, small acicular P-rich phases, and “filaments” (Fig. 4b) composed of aligned sulfide, fayalite, and Fe-Ti-oxides (e.g., Day et al. 2006). P-rich phases are more abundant in MIL 090136 than in the other two nakhlites. The intercumulus phase also displays cruciform (Haggerty 1976) titanomagnetite with skeletal growth morphologies (Fig. 4a), which has only previously been found in MIL 03346 (Day et al. 2006) and one other nakhlite, NWA 817 (Sautter et al. 2002). These textures are indicative of rapid cooling. No plagioclase was found in the intercumulus phase, unlike in nakhlites farther down in the cumulus pile. The main alteration phase is orange-brown iddingsite, found mainly in veins and cracks within olivine. Iddingsite in Miller Range nakhlites, described by Hallis and Taylor (2011), consists of a mixture of Fe-phyllosilicates and fine-grained Fe-oxides with possible presence of an amorphous silica gel component. Iddingsite forms through low-temperature aqueous alteration of olivine and basalt glass (Treiman 2005). We did not find other alteration minerals observed by Day et al. (2006) in MIL 03346 and by Hallis and Taylor (2011), who thoroughly described the textures and compositions of secondary minerals in MIL 090030/32/136.

Crystal Size Distribution (CSD) and Spatial Distribution Patterns (SDP) Analyses

Crystal size distribution (CSD) is a useful tool that allows quantification of crystal sizes and an assessment of the growth histories of crystal populations. Magma residence time, which can be useful for understanding nakhlite formation, can be estimated from CSD analyses (Marsh 1988). CSD is based on steady-state open-system crystallization, during which continuous nucleation and growth result in a negative log-linear plot of population density versus crystal size. The population density n is represented by n = n0 exp(−L/), where n0 is the final nucleation density, L is the crystal size, G is the growth rate, and τ is the magma residence time (Marsh 1998).

Crystal size distribution analyses were applied to pyroxene grains in MIL 090030, MIL 090032, and MIL 090136 (Table 2), using the CSDslice software (Morgan and Jerram 2006). Pyroxene is the only mineral used in this analysis because it is the most abundant phase crystallizing in nakhlites (Lentz and McSween 2003). CSDslice converts two-dimensional crystal sizes of random slices through grains by converting them into a true three-dimensional crystal size distribution using a database of 703 discrete crystal shapes. The best matching habit is obtained, representing the short-, intermediate-, and long-axis ratios. The calculated grain shape ratios are 1.00:1.25:1.30, 1.00:1.25:1.40, and 1.00:1.20:1.50 for pyroxenes in MIL 090030/32/136, respectively. Morgan and Jerram (2006) argue that CSD calculations have to be conducted on a minimum of 250 grains per sample to establish a statistically significant distribution. Three hundred and eighty pyroxene grains were measured in MIL 090030, 712 grains in MIL 090032, and 153 grains (all that were available) in MIL 090136. Thus, the data for MIL 090136 should be treated with caution. These measured grain size ratios were subsequently entered in CSDcorrection software (Higgins 2000), which calculates pyroxene population density versus crystal size (Fig. 5; Table 2). The three nakhlites display the same CSD pattern with an overall negative slope and similar intercept. A decrease in population density is observed for the smallest grain sizes. Lentz et al. (1999), who applied CSD measurements to other nakhlites (although without the newer CSDslice software), reported that the turnover at small grain sizes observed for MIL 090030/32/136 reflected the absence of new grains nucleating. MIL 090136 exhibits a discrepancy for small crystal sizes, which can likely be explained by the low number of measured pyroxene crystals in this small thin section.

Table 2.   Crystal size distribution results for MIL 090030/32/136 and MIL 03346.
SampleOrientationArea (mm2)No. grainsAverage dimensions (mm)Slope (mm−1)Intercept R valueResidence time (10−10)b (Earth years)Residence time (5 × 10−10)c(Earth years)
  1. aFrom Day et al. (2006).

  2. bUsing growth rate from Jerram et al. (2003).

  3. cUsing growth rate from Leu (2010).

MIL 090030,21Length583800.38−
 Width  0.23−9.07.71 357
MIL 090032,21Length1017120.40−4.95.691.426513
 Width  0.24−7.27.12 449
MIL 090136,22Length291530.41−4.35.341.347415
 Width  0.24−6.26.58 5210
MIL 03346,111aLength1105960.43−
 Width  0.26−5.86.08 5511
MIL 03346,118aLength1195450.4−3.74.581.358617
 Width  0.25−6.15.88 5210
Figure 5.

 Crystal size distribution (CSD) analysis plots (population density versus crystal size) for a) MIL 090030/32/136 augites and b) MIL 090030/32/136 and MIL 03346 augites (the latter from Day et al. 2006). Crystal sizes were calculated for both pyroxene crystal lengths and widths. The four meteorites display similar CSD patterns.

Residence times of pyroxenes were calculated using the slopes of the CSD plots with the equation: m = −1/() (Marsh 1998). The residence times of all three nakhlites are 54.5 ± 19.5 Earth years, using the growth rate of 1 × 10−10 mm/s used by Day et al. (2006) (Table 2).

Spatial Distribution Pattern is a useful tool to determine arrangement of crystals and constrain processes of rock formation. This technique is based on the R-value and the porosity of the thin section (Jerram et al. 1996, 2003). The R-value represents the ratio of the observed versus predicted mean distribution of nearest neighbor distances using the density of the observed distribution. R is defined as = (2√ρΣr)/N where ρ is the density of the observed distribution, r the nearest neighbor distance, and N the total number of measured grains (Jerram et al. 1996). Similar to MIL 03346 and other nakhlites (Nakhla, Lafayette, and Governador Valadares), the MIL 090030/32/136 pyroxene population consists of a clustered and touching framework, comparable to terrestrial phenocrysts in igneous cumulate rocks (Jerram et al. 2003) (Fig. 6; Table 2). As mentioned by Day et al. (2006), these phenocrysts form by poor sorting during accumulation in the cumulate pile.

Figure 6.

 SDP cluster analysis diagram (Jerram et al. 2003) with new data for MIL 090030/32/136 augites, and MIL 03346 (Day et al. 2006), Nakhla, Lafayette, and Governador Valadares (G.V. Lentz et al. 1999) data.

Mineral Major/Minor Element Compositions

Cumulus pyroxene cores are homogeneous and display consistent augite compositions in all three meteorites (Table 3; Fig. 7). The core compositions are magnesian (Wo39–41En35–38Fs22–25) and are similar to pyroxene cores in other nakhlites (e.g., Treiman 2005). They display Fe-rich hedenbergite rims (Fig. 7) with a composition of Wo34–42En6–24Fs37–49. As shown in Fig. 7, pyroxene rim wollastonite contents decrease slightly as they become more hedenbergitic in composition. Rims are also enriched in minor incompatible element oxides such as Al2O3 and TiO2, compared with the cores. Fe-rims are only present where they are in contact with the intercumulus phase and do not occur where pyroxenes are in contact with other pyroxenes or with olivine. The pyroxene compositions are indistinguishable among the three meteorites.

Table 3.   Representative major and minor element compositions of cumulus pyroxenes in MIL 090030/32/136.
wt%MIL 090030 - PxMIL 090032 - PxMIL 090136 - Px
Figure 7.

 Pyroxene core (triangles) and rim (circles) compositions as well as olivine (squares) compositions in MIL 090030/32/136. All three nakhlites from this study have similar cumulus phase compositions. Envelopes are shown for pyroxenes in MIL 03346 (Mikouchi et al. 2003; Day et al. 2006; Imae and Ikeda 2007; Treiman and Irving 2008), NWA 817 (Sautter et al. 2002), Y000593, Nakhla, Governador Valadares, Lafayette, and NWA 998 (Mikouchi et al. 2003; Treiman and Irving 2008). Olivine envelopes from Sautter et al. (2002) and Day et al. (2006) for MIL 03346 and from Mikouchi et al. (2003) for NWA 817, Nakhla, Governador Valadares, and Lafayette. As all the nakhlites have similar pyroxene core compositions, the MIL 03346 pyroxene core envelope corresponds to the other nakhlite pyroxene core composition.

The large olivines are zoned, ranging from Fo43 in the cores to Fo13 in the rims (Table 4; Fig. 7). The olivine cores are similar in composition to those in MIL 03346 and NWA 817 but are slightly more magnesian than in other nakhlites (Day et al. 2006). The intercumulus olivines have fayalitic compositions of Fo2–15. As for pyroxenes, the olivine compositions are similar for all three meteorites. Cumulus phase major element compositions reported here are consistent with the results of Hallis and Taylor (2011).

Table 4.   Representative major and minor element compositions of cumulus and intercumulus olivines in MIL 090030/32/136.
wt%Cumulus olivineIntercumulus fayalite
 MIL 090030 - OlMIL 090032 - OlMIL 090136 - OlMIL 090030MIL 090032MIL 090136

The intercumulus phase mainly consists of feldpathic glass, with a variable composition from trachy-basalt to trachyte (Fig. 8; Table 5). A pure glass analysis has been difficult to obtain because of the many tiny grains present in the intercumulus phase. Although the range of compositions found in these glasses is similar to those reported by Hallis and Taylor (2011), we do not observe a compositional cluster for MIL 090032 at the extreme right of the MIL 03346 envelope (i.e., MIL 090032 glass is not systematically silica rich). “Filaments,” found throughout the intercumulus phase, consist of fayalite, titanomagnetite, and sulfides. Larger intercumulus sulfides than those present in the filaments are pyrrhotites having a composition of Fe0.89–0.92S (Table 5). The compositions of skeletal titanomagnetite are shown in Table 6 and illustrated in Fig. 9. Compared with MIL 03346 (Day et al. 2006), the magnetites display a smaller range of composition from Usp33-Mt67 to Usp65–Mt35. Pure magnetite, reported in MIL 03346 (Day et al. 2006), was not found in any of our samples. P-rich needles and Si-rich phases of unresolved mineralogy also occur in the intercumulus phase.

Figure 8.

 Total alkalis versus silica diagram showing the intercumulus phase compositions in MIL 090030, MIL 090032, and MIL 090136, with an envelope of MIL 03346 intercumulus phase compositions from Day et al. (2006). Black dots represent intercumulus glass compositions of Nakhla + Governador Valadares, Lafayette (Berkley et al. 1980), and Yamato 000749 (Imae et al. 2005). Most of the glass compositions of MIL 090030/32/136 fall within the MIL 03346 envelope.

Table 5.   Representative major and minor element compositions of intercumulus glass and pyrrhotites in MIL 090030/32/136.
wt%MIL 090030MIL 090032MIL 090136
Intercumulus Al-rich glass
Table 6.   Representative major and minor element compositions of skeletal titanomagnetites in MIL 090030/32/136.
MIL 090030MIL 090032MIL 090136
Figure 9.

 FeO-Fe2O3-TiO2 diagram for MIL 090030/32/136 titanomagnetite, with envelope of MIL 03346 spinels from Day et al. (2006). A cluster of our data compared with the MIL 03346 envelope is observed, probably due to sampling bias.

Trace Element Compositions

Pyroxene rims are enriched in incompatible elements such as Y and Ti, and depleted in compatible elements such as Cr (Fig. 10). Pyroxene cores and Fe-rich rims display similar REE patterns, but the rims are enriched in REEs compared with the cores (Fig. 11; Table 7). The LREEs show a positive slope, whereas a slight negative slope is observed for the heavy REEs (HREEs). The pyroxenes are not enriched in HREEs, contrary to typical terrestrial augites. Small negative Eu anomalies are observed in most pyroxenes. MIL 090030/32/136 REE compositions are similar to the MIL 03346 compositions (Day et al. 2006) (Fig. 11). Furthermore, MIL 090030/32/136 REE patterns are parallel to those of other nakhlites (e.g., Wadhwa et al. 2004).

Figure 10.

 Abundances of Y, Ti, and Cr versus Fe in pyroxene cores and rims in MIL 090030/32/136. The rims are enriched in incompatible elements (Y and Ti) and depleted in compatible elements (Cr).

Figure 11.

 REE concentrations normalized to CI chondrite (Anders and Grevesse 1989) in pyroxene rims (circles) and cores (triangles) of a) MIL 090030, b) MIL 090032, and c) MIL 090136. Envelopes of REEs in MIL 03346 pyroxene rims (light gray) and cores (dark gray) from Day et al. (2006).

Table 7.   Representative trace element compositions of cumulus pyroxenes in MIL 090030/32/136.
ppmMIL 090030 - PxMIL 090032 - PxMIL 090136 - PxPrecision (%)
  1. n.d. = not detected. In addition, Cu, Rb, Nb, Cs, Tm, Ta, Pb, and U were below detection limit in all pyroxene analyses.



Evidence of Pairing

As noted earlier, the geographic proximity of the Antarctic locations where MIL 090030, MIL 090032, MIL 090136, and MIL 03346 were found, was the first indication of pairing. We compared textures, CSD patterns and SDP analyses, in situ major and trace element compositions for cumulus phases and mineralogy and major element compositions for intercumulus phases.

Figure 3 compares plane-polarized light images of the four meteorites at the same scale. All show similar overall textures with equivalent size cumulus pyroxene grains (see also Table 2). Skeletal titanomagnetites have been found in the intercumulus phases of all four meteorites (Figs. 12a–d). This phase was also observed in NWA 817 (Sautter et al. 2002), but not in other nakhlites. The skeletal texture of the titanomagnetite indicates fast quenching, presumably at or near the top of the cumulate pile.

Figure 12.

 Backscattered electron images of a) titanomagnetite in MIL 090030 and b) titanomagnetite in MIL 03346 (Righter et al. 2008; reprinted with permission), c) P-rich phases and fayalite in MIL 090136 and d) P-rich phases and fayalite in MIL 03346 (McKay and Schwandt 2005; reprinted with permission). Images a) and b) and images c) and d) are at the same scales.

Pyroxene habits of MIL 090030/32/136 calculated with CSDslice are very similar to that of MIL 03346 calculated by Day et al. (2006). CSD patterns and intercepts for MIL 090030/32/136 and MIL 03346 are similar (Fig. 5). Moreover, the calculated magma average residence time of 54.5 ± 19.5 Earth years (Table 2) is consistent with the residence times of Day et al. (2006) of 20–100 yr using the minimum growth rate (1 × 10−10 mm. s−1) from Jerram et al. (2003). The residence time calculated using the maximum growth rate (5 × 10−10 mm. s−1) from Jerram et al. (2003), which also corresponds to the growth rate from Leu (2010), is 10.8 ± 5 Earth years, consistent with residence times estimated by Day et al. (2006) (Table 2). The growth rate from Jerram et al. (2003) corresponds to silicate phases in basaltic melt, whereas the growth rate from Leu (2010) is for clinopyroxene in mafic melt. Comparable data obtained using the same technique for different nakhlites are not available in the literature, so comparisons of residence times with nakhlites from deeper levels in the cumulate pile are not possible. However, we should not expect significant differences in pyroxene residence times between the different nakhlites. This is due to the fact that the pyroxenes crystallized at deeper levels in a common magma chamber, before being emplaced at or near the Martian surface to form the cumulate pile. Assuming nakhlites all come from the same lava flow sequence (e.g., Day et al. 2006), we would expect to observe the same crystal population, with similar residence times. Thus, pyroxene grain sizes should not vary with depth, except if overgrowths occurred. Furthermore, SDP analyses, which are similar for the Miller Range nakhlites, show that the cumulus phases underwent poor sorting during accumulation in the cumulate pile.

One of the most obvious similarities between MIL 090030/32/136 and MIL 03346 is the pyroxene major element composition. Major and trace element compositions of pyroxene cores and rims in MIL 090030/32/136 plot within the field of MIL 03346 (Figs. 7 and 11). Although cumulus pyroxene cores display the same augitic composition in all nakhlites, pyroxene rim compositions are different (Fig. 7): NWA 998, which is inferred to have come from the base of the pile (Treiman and Irving 2008), exhibits pyroxene rims with the highest Mg/Fe, whereas MIL 03346, inferred to be near the top of the pile (Day et al. 2006), is enriched in the wollastonite component. Furthermore, the olivine major element compositions for MIL 090030/32/136 plot within the MIL 03346 field (Fig. 7), as do the glass compositions (Fig. 8) and the Ti-magnetite compositions (Fig. 9). Figure 9 shows the MIL 090030/32/136 data clustered in a narrower compositional range compared with the MIL 03346 envelope, probably due to sampling bias. Trace element compositions of MIL 090030/32/136 pyroxenes are indistinguishable from other nakhlite compositions, except for some REE-enriched rims (e.g., Wadhwa et al. 2004).

Taken together, similar textures and grain sizes, the presence of skeletal titanomagnetite, pyroxene CSD patterns and SDP analyses, cumulus mineral major and trace element compositions, as well as intercumulus phase major element compositions, all support pairing of the four meteorites, as also suggested by Hallis and Taylor (2011) and Corrigan et al. (2011). This confirmation of pairing significantly increases the amount of material available for study, so that collectively the four meteorites provide a more representative sampling of this nakhlite.

Nakhlite Modal Abundances

The degree of crystallinity and modal abundance of intercumulus phases, the inferred cooling rate, and the degree of equilibration of cumulus material, are characteristics that could assist with placement of nakhlites within the cumulate pile (e.g., Sautter et al. 2002; Mikouchi et al. 2003, 2012; Day et al. 2006; Jambon et al. 2010). Cumulus phases are expected to be more abundant at the base of the pile whereas the intercumulus phase should be more abundant near the top. This is because of packing of cumulus phases at the base (Mikouchi et al. 2003) as well as quenching near the top, which prevents crystallization of cumulus phases (Sautter et al. 2002). Modal abundances are significantly different for MIL 090030/32/136 and MIL 03346. MIL 090030,21 and 090136,22 contain significantly more olivine than MIL 03346 (see Table 1; Figs. 2 and 13). Moreover, pyroxene content is higher and intercumulus phase is lower in MIL 0990136,21 (Hallis and Taylor 2011) than in MIL 090136,22 (our thin section). As observed in Fig. 13 and Table 1, modal abundances for pyroxene, olivine, and intercumulus phase among the four paired nakhlites in this study vary, as does the intercumulus/cumulus area ratio, indicating that individual Miller Range samples are not modally representative of the entire parent meteorite. Such variations are consistent with studies of terrestrial cumulate piles, in which modal abundances can vary significantly (Lentz et al. 1999). The surface areas of our three nakhlite sections are comparable to nakhlite thin sections reported in the literature (e.g., Lentz et al. 1999; Sautter et al. 2002; Imae et al. 2003; Day et al. 2006; Imae and Ikeda 2007) but the modal abundance variations of the nakhlites and intercumulus/cumulus ratio do not show clear trends throughout the cumulate pile (Fig. 13). Thus, we can conclude that modal analyses of multiple samples for a single nakhlite may be necessary to obtain representative modal data. Modal abundances of pyroxene, olivine, and intercumulus phase for various Miller Range nakhlite samples assumed to come from the same parent sample were used to calculate a weighted mode, taking into account the corresponding areas for each analyzed thin section, to provide a more representative sampling of the parent meteorite (Table 1). A weighted modal abundance was calculated using data from MIL 090030/32/136 from this study, and MIL 03346 data from Day et al. (2006) and Imae and Ikeda (2007). The weighted modal abundance does not include other Miller Range thin sections because not all papers published the areas of the analyzed thin sections. The modal abundance of the parent meteorite is 72.3% pyroxene, 3.9% olivine, and 23.7% intercumulus phase (Table 1). Modal analyses of multiple samples of other nakhlites are required for valid comparisons with MIL nakhlites.

Figure 13.

 Histograms showing modal abundances of pyroxene, olivine, and intercumulus phase from nakhlites. The sources of data are given in Table 1. The different MIL 03346 samples and MIL 090030/32/136 samples are not presented in any particular order.

Nakhlite Parental Melt Compositions for MIL 090030/32/136

Nakhlites are cumulate rocks and therefore their compositions are not representative of the parental melt composition. However, it is possible to calculate the parental magma melt composition by inverting REE compositions of the most primitive pyroxene cores using suitable partition coefficients (D-values). The experimental partition coefficients used in this study are the same as those previously employed by Day et al. (2006), Wadhwa et al. (2004), and Oe et al. (2001) and were calculated from nakhlite compositions. The fact that MIL 090030/32/136 REE patterns of the calculated parental magma compositions and the measured whole-rock MIL 03346 pattern are parallel (Fig. 14a) (Day et al. 2006) supports the conclusion that nakhlites are formed by closed-system crystallization of the parental magma (Wadhwa and Crozaz 1995; Day et al. 2006). Moreover, the average compositions for the intercumulus phases of MIL 090030/32/136 correspond to those of MIL 03346 (Day et al. 2006), confirming closed-system crystallization. These compositions fall on the 30% fractional crystallization line according to the model of Day et al. (2006) (Fig. 14b), which is the amount of fractional crystallization necessary to explain the intercumulus phase REE-enriched compositions. The identical intercumulus phase REE patterns for the four MIL nakhlites and the 30% late-stage fractional crystallization line support the REE calculations and conclusions of Day et al. (2006). Furthermore, because the REE pyroxene rim composition range is the same for the four Miller Range nakhlites (Day et al. 2006) (Fig. 11), 30–100% fractional crystallization is necessary to explain the Fe-rich pyroxene rim compositions.

Figure 14.

 CI chondrite-normalized (Anders and Grevesse 1989) REE patterns for: a) Calculated pyroxene parental magma of MIL 090030/32/136 and MIL 03346 (Day et al. 2006) and measured whole-rock of MIL 03346 (Day et al. 2006). The parental magma patterns of MIL 090030/32/136 are parallel to the MIL 03346 whole-rock (see text for details). b) Fractional crystallization model calculated by Day et al. (2006) with amount of fractional crystallization and average intercumulus phase REE compositions of MIL 090030/32/136 and MIL 03346 (Day et al. 2006). The average intercumulus phase compositions correspond to 30% fractional crystallization, supporting the Day et al. (2006) model.


The MIL 0900030, MIL 090032, and MIL 090136 nakhlites appear to be paired with MIL 03346, as demonstrated by the following:

  • 1 similar mineralogy and texture for the cumulus and intercumulus phases
  • 2 similar in situ major and minor element compositions for the cumulus minerals, glass and Ti-magnetites, and trace element compositions for pyroxenes.
  • 3 similar CSD patterns and spatial distribution pattern analyses.

However, modal mineral abundances of MIL 090030/32/136 and MIL 03346 show variations indicating that individual Miller Range samples alone are not modally representative of the parent meteorite. Modal abundances show variations throughout the nakhlite cumulate pile, indicating that analyses of multiple samples of a single nakhlite may be necessary to obtain representative modal data for placement in the nakhlite cumulate pile. Modal analyses of multiple samples of other nakhlites are required for valid comparisons with MIL nakhlites.

Calculated parental melt compositions confirm closed-system fractionation and that a late-stage fractional crystallization of 30% is required to explain REE enrichments of the intercumulus phase.

Acknowledgments—  We thank the NASA/Johnson Space Center for providing the Miller Range nakhlite thin sections; A. Patchen for assistance with microprobe analysis; and J. Day, K. Righter, J. Schutt, B. Balta, K. Thaisen, P. Barry, and S. Singerling for beneficial discussions and comments. L. Hallis, V. Sautter, J. Day, and A. Ruzicka provided extremely useful reviews. This study was supported by Cosmochemistry Program grant NNX10AH48G to HYM.

Editorial Handling—  Dr. Alex Ruzicka