Experimental constraints on the origin of Martian meteorites and the composition of the Martian mantle

Abstract

The major element compositions of partial melts of a chondritic (Homestead L5) model Mars mantle have been determined at 4.7–5.0 GPa in a multi-anvil device over a temperature interval that ranged from near solidus to near liquidus conditions. The purpose of these experiments is to determine if the major element abundances of Martian basalts (shergottite meteorites) or their parent magmas can be derived from a model Martian mantle composition at high pressure. Partial melting of our model composition at 4.7–5.0 GPa produces liquids with super-chondritic CaO/Al₂O₃ similar to those of proposed Martian basalt parent magmas. The concentrations of CaO and Al₂O₃ in the high-pressure experimental liquids are, however, lower than in Martian basalt parent magmas. We conclude that matches for both ratios and concentrations involving CaO and Al₂O₃ between current model Mars compositions and proposed Martian basalt parent magmas would require at least two stages of magmatic differentiation. For example, partial melting at 5 GPa (425 km depth in Mars) produces a magma having super-chondritic CaO/Al₂O₃, but subsequent, lower pressure differentiation of olivine (±low-Ca pyroxene) is needed to increase the CaO and Al₂O₃ concentrations to those of calculated Martian basalt parent magmas. This two-stage polybaric differentiation would be consistent with either a magma ocean or mantle plume-melting scenario. On the other hand, these multi-stage differentiation scenarios cannot reconcile the significant mismatch between the FeO content of our experimental liquids and Martian basalt parent magmas. A remedy for this apparent inconsistency might require a bulk Martian mantle or shergottite parent source region with a composition and Mg# closer to that of H-chondrites, but still much less magnesian than are terrestrial upper-mantle basalt source regions.

Introduction

One of the most distinctive geochemical characteristics of Martian basalts and their estimated parent magmas is a marked aluminum depletion, which is often expressed as superchondritic CaO/Al₂O₃ [1], [2]. According to Treiman [3], the shergottite–nahklite–chassignite (SNC) source region appears to have lost 60% of its primordial aluminum. Fig. 1 shows the super-chondritic CaO/Al₂O₃ of several proposed parent magmas for Martian basalts and peridotites. In contrast, average chondrite values cluster, defining a very tight range of compositions that have lower Al₂O₃ and CaO/Al₂O₃ than do Martian parent magmas. Dreibus and Wänke [4] estimated a bulk composition for the Martian mantle (hereafter DW-Mars) that coincides with the chondritic cluster and is very close to the composition of L and LL ordinary chondrites.

Earlier melting experiments on FeO-rich bulk Mars compositions [5], [6], [7], [8], [9], such as DW-Mars and others similar to the composition in our current study, have demonstrated that the super-chondritic CaO/Al₂O₃ of Martian parent magmas cannot be produced by partial melting in the pressure range 1.5–3.0 GPa (∼150–300 km depth in Mars). Longhi et al. [10] concluded that the SNC parent magmas were produced by partial melting of a garnet-absent, aluminum-depleted source. They proposed that the aluminum could have been removed from the mantle by successive melting events that partitioned it into the Martian crust, leaving the SNC source region Al-depleted. Unfortunately, we have yet no evidence supporting the existence of such an aluminum-enriched Martian crust. Furthermore, fractionation processes such as those envisaged by Longhi et al. probably cannot account for the large deviation from chondritic CaO/Al₂O₃ required by mass balance for the source region of Martian parent magmas. For example, low-pressure partial melting of peridotite depletes the source region and enriches the crust in both Al₂O₃ and CaO, since both components are “incompatible”—hence, large perturbations in CaO/Al₂O₃ are unlikely in this process.

Longhi [11] noted that partial melting with garnet present at 5–6 GPa might produce Al-depleted magmas and explain the missing Martian aluminum. However, he cited the melt density measurements of Agee and Walker [12] on komatiite as evidence that magmas produced at 5–6 GPa in Mars might be too dense to rise, making a high-pressure genesis for SNC parent magmas unlikely. In the meantime, Ohtani et al. [13] showed that Martian mantle melt is in fact buoyant relative to olivine up to 7 GPa (∼600 km depth in Mars), and that the melt would be buoyant relative to a garnet-bearing source region at much higher pressures than 7 GPa. Thus, magma buoyancy constraints no longer seem to preclude the existence of Martian magma source regions at pressures of at least 7 GPa, and probably much higher.

It is now well established from several phase equilibrium studies at P>3 GPa on chondritic compositions and on Earth mantle peridotites [14], [15], [16] that increasing pressure up to ∼15 GPa continuously expands the stability field of garnet at the expense of all other crystalline phases, and decreases garnet solubility in silicate melt. Hence, it can be expected that partial melts at P>3 GPa will be characterized by increasing CaO/Al₂O₃, approaching values observed in shergottites. Therefore, we were motivated to determine whether partial melting of a model Martian mantle composition at higher pressures could give rise to shergottites or their parent magmas. To this end, we present here new data from a series of super-solidus experiments, done at 4.7–5.0 GPa, where it was found that garnet coexists with olivine, low-Ca pyroxene, and silicate liquid. Interestingly, as we describe below, the FeO content and Mg# (molar Mg/Fe+Mg) of these high-pressure liquids, not their CaO/Al₂O₃, pose the most formidable challenge in explaining the origin of Martian basalt parent magmas from model Martian mantle at high pressure.