A Petrological Model for the Origin of Martian Shergottite Magmas Based on Their Major Element, Trace Element, and Isotopic Compositions

Introduction: Defining the mineralogy and composition of martian mantle source regions has proved to be a difficult task because: (1) the martian sample set does not include samples of the mantle, and (2) many of the samples are either cumulates or are highly fractionated and therefore not in equilibrium with the mantle. Despite these complexities, isotopic evidence suggests that Mars had a magma ocean that crystallized at ~4.53 Ga, and that melting of magma ocean cumulates between ~175-575 Ma produced the shergottite parental magmas [e.g., 1-3]. This scenario has permitted the Rb/Sr, Sm/Nd, and Lu/Hf ratios of the martian magma sources to be estimated from their initial Sr, Nd, and Hf isotopic compositions [4]. Below we estimate the relative abundances of additional incompatible elements in the magma source regions using a magma ocean crystallization model. Approach: The models presented below follow the approach used by Snyder et al. [5] to estimate the composition of lunar magma source regions. First, the mineralogy of the martian magma ocean cumulates is estimated from the results of high pressure experiments. Next, a combination of equilibrium and fractional melt models is used to estimate the major and trace element abundances in the various cumulate packages. Finally, the mineralogy of depleted martian mantle (DMM) is estimated by determining which combination of cumulates: (1) yield melts with major element compositions that match those of the most mafic parental magma [Am; ALH77005; 6], and (2) have the Rb/Sr, Sm/Nd, and Lu/Hf ratios of the most incompatible element depleted martian meteorite (QUE94201) source region. Experimental constraints on the source regions: We chose Homestead L5 ordinary chondrite as an analog for early martian mantle immediately following core formation, i.e. as the bulk composition of an early martian magma ocean. This composition was used in the partitioning and phase equilibrium experiments of Draper and coworkers [7-8]. These results, as well as those of Agee et al. [9] and Ohtani et al. [10], were used to infer a magma ocean crystallization sequence. We investigated several crystallization sequences to generate magma ocean cumulate packages. These ranged from those having olivine and pyropic garnet as near-liquidus phases (representing lower-P conditions) to those having majoritic garnet on, or near, the liquidus (representing higher P). In each step of a given sequence, we calculated progressive melt compositions using equilibrium and fractional (at melt fractions <20%) crystallization models and the appropriate mineral assemblage. Coexisting solid bulk compositions were calculated by mass balance. Each sequence yielded four or five separate cumulate packages after 98% crystallization of the starting magma ocean composition. We then calculated 10% partial melts of each of these packages for comparison with Am and other proposed martian parental melts. Fig. 1 compares calculated melts from the various sequences to the estimated parental melts. The crystallization sequence that yields partial melts that are the best match to Am [blue line Fig. 1] consists of these five steps: (1) Crystallization of majorite until 5 percent solid (PCS); (2) olivine + majorite (80:20) to 20 PCS; (3) olivine + orthopyroxene (50:50) to 60 PCS; (4) orthopyroxene + clinopyroxene (60:40) to 90 PCS; (5) clinopyroxene + ilmenite (90:10) to 98 PCS.