An experimental investigation of F, Cl and H₂O mineral-melt partitioning in a reduced, model lunar system

Highlights

• First study of mineral-melt partitioning of F, Cl and H₂O in a model lunar system.

• Oxygen fugacity has limited control on volatile partitioning.

• Data imply a volatile-depleted mantle source region for lunar basalts.

• Lunar F/Cl and F/H₂O partly inherited from lunar magma ocean solidification.

Abstract

It is believed that the Moon formed following collision of a large planetesimal with the early Earth. Over the ∼4 Gyr since this event the Moon has been considerably less processed by geological activity than the Earth, and may provide a better record of processes and conditions in the early Earth-Moon system. There have been many studies of magmatic volatiles such as H, F, Cl, S and C in lunar materials. However, our ability to interpret variable volatile contents in the lunar sample suite is dependent on our understanding of volatile behaviour in lunar systems. This is currently constrained by limited experimental data. Here, we present the first experimental mineral-melt partitioning coefficients for F, Cl and H₂O in a model lunar system under appropriately reduced conditions (log fO₂ to IW-2.1, i.e. oxygen fugacity down to 2.1 log units below the Fe-FeO buffer). Data are consistent with structural incorporation of F, Cl and OH⁻ in silicate melt, olivine and pyroxene under conditions of the lunar mantle. Oxygen fugacity has a limited effect on H₂O speciation, and partitioning of H₂O, F and Cl is instead largely dependent on mineral chemistry and melt structure. Partition coefficients are broadly consistent with a mantle source region for lunar volcanic products that is significantly depleted in F, Cl and H₂O, and depleted in Cl relative to F and H₂O, compared to the terrestrial mantle. Partitioning data are also used to model volatile redistribution during lunar magma ocean (LMO) crystallisation. The volatile content of lunar mantle cumulates is dependent upon proportion of trapped liquid during LMO solidification. However, differences in mineral-melt partitioning during LMO solidification can result in significant enrichment on F relative to Cl, and F relative to H₂O, in cumulate phases relative to original LMO composition. As such, Cl depletion in lunar volcanic products may in part be a result of LMO solidification.

Introduction

The presence of ‘water’ or related H-bearing species, i.e. H₂O, OH⁻, H₂, etc., in the interior of the Earth has a controlling influence on a range of mantle properties from melting behaviour to rheology (Peslier, 2010, and references therein). The additional role of water in sustaining life means that the origin of Earth’s volatiles and the timing of their delivery are key areas of study in Earth and planetary sciences. However, many geological processes have modified the Earth since its initial formation, redistributing these volatiles and masking geochemical signatures of their origin. If it is accepted that the Earth and Moon share some common history, the question is when did their compositions diverge and to what extent (Albarède, 2009). According to the canonical “Giant Impact” hypothesis, the Moon formed following collision of a Mars-sized planetesimal with the early Earth (Canup and Agnor, 2000, Canup, 2004). This catastrophic heating event was previously believed to be consistent with an anhydrous lunar interior and lunar sample suite (Lucey et al., 2006, Taylor et al., 2006). The subsequent detection of water and other volatiles in volcanic lunar glasses (Saal et al., 2008) has heralded a new era of lunar research and reinvestigation of the lunar sample suite. This has revealed the presence of ‘water’ in volcanic lunar glasses (Saal et al., 2008, Saal et al., 2013, Greenwood, 2018), in apatites in mare basalts (Boyce et al., 2010, McCubbin Francis et al., 2010, McCubbin et al., 2010, Greenwood et al., 2011, Barnes et al., 2013, Greenwood, 2018), in olivine-hosted melt inclusions in mare basalts (Chen et al., 2015, Hauri et al., 2011, Hauri et al., 2015, Ni et al., 2017, Ni et al., 2019), in plagioclase from lunar anorthosites (Hui et al., 2013, Hui et al., 2017), as well as halogens in mesostasis in lunar basalts (Greenwood et al., 2020). It is now clear that various lunar mantle source regions contain appreciable H, in addition to Cl, F, C and S. As the Moon shares a common origin with the Earth, the logical conclusion is that a similar process may have been involved in volatile delivery to both components of the early Earth-Moon system. Volatiles were either delivered to the Earth prior to the Moon-forming event or delivered to the Earth-Moon system shortly after the Moon-forming event, whilst the newly-formed Moon was still in a largely molten state. As the Moon has remained less affected by large-scale geological processes since its formation, the lunar volatile budget may provide much needed insight into the early Earth-Moon system, and ultimately, provide constraints on the origin of Earth’s hydrosphere.

The lunar sample suite is consistent with a high temperature origin for the Moon. For example, the lunar anorthositic crust probably represents a plagioclase floatation crust formed during lunar magma ocean (LMO) solidification (Warren, 1985). The later mare basalts most likely formed by partial melting of deeper mantle cumulates, also produced by the same LMO solidification process, following late-stage cumulate overturn (Ryder, 1991). As a result, the volatile content of mare basalts and associated volcanic glasses provides insight into the volatile budget of the lunar mantle melts and the LMO solidification products (e.g. Hauri et al., 2011, Saal et al., 2008). With the detection of ‘water’ in lunar materials and inferred ‘water’ contents in parental lunar magmas, it is tempting to make direct comparison with terrestrial magmas. For example, Hauri et al. (2011) noted that volatile contents of melt inclusions in lunar olivine are similar to those in terrestrial mid-ocean ridge basalts, implying similar volatile budgets in terrestrial and lunar mantle source regions. However, any such comparison can be misleading because of differences in magmatic conditions between evolved terrestrial and lunar systems. Mare basalts and associated volcanic glasses are reduced compared to terrestrial volcanic materials, with mineral assemblages and direct oxygen fugacity (fO₂) measurements indicating fO₂ below that of the iron-wüstite (IW, i.e. Fe-FeO) solid buffer, and either at or near Fe-metal saturation (Longhi, 1992, Shearer et al., 2006). The reduced nature of lunar magmas implies melting of a similarly reduced lunar interior (Longhi, 1992), with the oxidation state of the source region for mare basalts and volcanic glasses generally assumed to be at least 2 log units lower than Earth’s upper mantle (Rutherford and Papale, 2009, Sato, 1976). Oxygen fugacity has a fundamental control on ‘water’ or H incorporation in silicate materials. In terrestrial systems at high mantle temperatures and pressures ‘water’ is incorporated into nominally anhydrous minerals such as olivine and pyroxenes as interstitial H associated with underbonded O atom sites (e.g. Skogby et al., 1990, Smyth et al., 1991, Ingrin and Skogby, 2000, Stalder and Skogby, 2003, Bromiley et al., 2004a, Bromiley and Bromiley, 2006, Smyth et al., 2006), forming various O–H defects (Lemaire et al., 2004, Matveev et al., 2005, Grant et al., 2007). In terrestrial silicate melts, water is incorporated initially as OH⁻, with molecular H₂O becoming more dominant at higher total water contents, at least under more oxidising conditions (e.g. Stolper, 1982, Pandya et al., 1992, Dixon et al., 1995). This has recently been contested by Cody et al. (2020) who suggest that H₂O dominates at low H abundances, at least in highly polymerised silicate melts, although whether this is true in less polymerised, basaltic melts remains unclear. Under reducing conditions H₂O becomes unstable as a species, and other H-related defects such as H₂ and CH₄ are observed in spectroscopic investigations of melts at the expense of OH⁻ (Kadik et al., 2006, Mysen et al., 2011, Hirschmann et al., 2012, Ardia et al., 2013). Magmas are commonly used as probes of planetary interiors as they represent escaped partial melts that transport information about source regions. The ‘water’ content of magmas is controlled in part by mineral-melt partitioning during partial melting of the mantle source. Any change in speciation of ‘water’ as a function of fO₂ could result in significant changes in partitioning. The ‘water’ content of a magma will likely depend on both the H concentration and the fO₂ in its mantle source region, and any comparison between lunar and terrestrial magmas must take this into account. Kadik et al. (2006) noted that at high pressure in silicate melts at fO₂ below IW, H₂ becomes increasingly important at the expense of more oxidised species. Hirschmann et al. (2012) noted that the proportion of H₂ in lunar magmas is low under the low total water contents of the lunar interior. However, in reduced lunar systems over a range of inferred bulk ‘water’ contents (Elkins-Tanton and Grove, 2011) the proportion of H₂/H₂ + H₂O could range up to 20–50%. In solid phases under similarly reducing conditions, Yang et al. (2016) demonstrated that mantle silicates can also incorporate some molecular H₂, in addition to OH defects. The effects of any change in speciation on volatile behaviour remain unclear. Newcombe et al. (2017) demonstrated that degassing mechanisms in lunar melts over a range of highly reducing conditions (below the IW buffer) are strongly controlled by H speciation, although they noted that diffusion is dominated by flux of OH⁻ rather than H₂. Recently, Lin et al. (2019) demonstrated that fO₂ has a strong control on H₂O partitioning between plagioclase and silicate melt, but aside from that study there is an absence of data on the effect of fO₂ on volatile partitioning in lunar systems. Here we present the first experimental data on partitioning of H₂O, F and Cl in a reduced, model lunar mantle lithology system. F and Cl are important tracers in lunar materials (e.g. Hauri et al., 2011, Chen et al., 2015), and are increasingly used to provide insight into volatile redistribution during terrestrial geological processes (e.g. Urann et al., 2017, Klemme and Stalder, 2018). Furthermore, isotopic variations of Cl in lunar materials provide significant additional insight into lunar magmatic processes (McCubbin et al., 2011, Potts et al., 2018, Boyce et al., 2018). Therefore, combined study of F, Cl and H₂O distribution during magmatic processes provides a powerful tool for reinvestigating the importance of measured volatile concentrations in lunar samples, and for drawing inferences on volatile delivery to the early Earth-Moon system.