Effects of phosphorus on Martian magmas: an experimental study

Introduction: Experimental studies showed that elevated phosphorus concentrations (> 1 wt. %) in terrestrial basaltic magmas have significant effects on the crystallization sequence, the composition and physical properties of melts [1-2]. [1] demonstrated that the polyvalent nature of P⁵⁺ favors the polymerization of silica tetrahedra, promoting the silica liquidus boundary in favor of that of pyroxene and the pyroxene liquidus boundary in favor of that of olivine. The enhancement of polymerized minerals can have significant implications regarding the petrogenesis of igneous rocks and mineralogy of the crust. Mars’ primitive mantle contains >10 times more phosphorus than Earth’s [3], which could result in considerable differences in magma compositions and crustal mineralogy between Earth and Mars.

Preliminary thermodynamical partial melting models based on a martian mantle composition using pMELTS software showed that phosphorus likely have implications in terms of magma compositions [4]. This study aims to explore with piston-cylinder experiments the effects of phosphorus on martian melt compositions and residual mineral assemblages when considering partial melting of a primitive mantle composition [6].

Methods: A set of nominally anhydrous partial melting experiments has been run on a primitive mantle composition [5] with two P₂O₅ contents (0 wt.% and 0.5 wt.%). We followed the exact same procedure as [6], which considered partial melting experiments in the same laboratory of a primitive mantle composition [5] with P₂O₅=0.2 wt.%, the actual mantle content. We will compare our results with those from [6].

Using oxides and carbonates, all reagents were fired to avoid any water contaminations, and mixed together using an agate motar for 2 hours. The mixture was then loaded in a graphite inner capsule, which was placed into a platinum outer capsule. Before welding shut the platinum capsule, the loaded capsule was dried at 300°C during 24 hours to minimize water contamination. The welded shut capsule was then emplaced into a crushable MgO internal spacers, cylindrical graphite furnaces, BaCO₃ pressure sleeve, and the whole assembly was wrapped within a Pb-foil. A half-inch piston cylinder apparatus at Rice University has been used at a pressure of 2GPa and temperatures between 1210 and 1450°C during 24-72 hours. Texture and composition have been analyzed using an electron probe microanalyzer at Rice university, and quantitative elemental maps have been acquired using Wavelength Dispersive Spectrometers (WDS). Phase proportions have been calculated by mass balance calculations.

Equilibrium: Equilibrium conditions are likely met according to the defined delimitations between grain boundaries, exchange coefficients between Fe and Mg K_D mostly varying from 0.30 to 0.40 (accepted value 0.35 [7]), the high quality of mass balance calculations (sum of squared residuals <0.30), and the low loss of Fe (relative FeO loss <2 wt.%).

Phase Proportions: The extent of melting corresponds to the amount of mixture between glass and dendritic orthopyroxene. Although the degree of melting is similar for each temperature regardless the amount of P₂O₅, a mixture of glass and quenched orthopyroxene is observed at low temperature in P-bearing samples compared to P-free samples (≥1260 °C and ≥1300 °C, respectively).

As shown on Figure 1, regardless the extent of melting, the addition of phosphorus promotes residual orthopyroxene in favor of olivine.

Figure 1. Proportion (in wt.%) of (left panel) orthopyroxene and (right panel) olivine as a function of extent of melting (%).Model martian mantle compositions with P₂O₅ = 0 wt.% (blue line), P₂O₅=0.2 wt.% ([6];orange line), and P₂O₅=0.5 wt.% (red line).

WDS maps of the lowest melting degree sample with P₂O₅=0.5 wt. % reveals scarce µm-sized grains depleted in SiO₂, Al₂O₃ and FeO, and enhanced in CaO and P₂O₅ that are likely apatite.

Melt Compositions: Melt compositions correspond to that of the mixture between glass and quenched orthopyroxene. As presented on Figure 2, the addition of phosphorus lowers SiO₂ up to 13 wt. % in P-bearing samples, especially at low melting degree ≤10%. At melting degrees ≤18%, P-free samples are depleted in FeO_T and CaO/Al₂O₃ ratio in comparison with P-bearing samples.

Figure 2. Averaged composition (wt.%) of SiO₂, FeO_T, and CaO/Al₂O₃ of experimental partial melts as a function of the extent of melting (%). Error bars represent the standard deviation of the set of analyses for each experimental sample.

Discussion: The occurrence of melt mixtures at low temperature in P-bearing samples compared to P-free ones supports the stabilization of melt by the polyvalent cation P⁵⁺. The enhancement of orthopyroxene in favor of olivine in P-bearing samples also suggest an expansion of polymerized mineral liquidus boundary as shown for terrestrial basalts [1]. The enhancement of orthopyroxene can explain the decrease of SiO₂ in P-bearing melts.

Polymerized minerals are promoted due to the ability of P⁵⁺ to remove network-modifying cation (M) like Fe²⁺ from silica network as illustrated by [1-2]. The occurrence of twice FeO_T in the primitive martian mantle relative to Earth’s [3] likely enhances the complexation P-Fe(II), promoting polymerization of the Si-network. Stabilization of orthopyroxene may favor the accumulation of orthopyroxene cumulates within the martian crust. Partial melting of a mantle composition with P₂O₅=0.2 wt.% forms P-rich liquids at low melting degrees (P₂O₅>1 wt.%), likely explaining high P₂O₅ contents in martian basalts like those analyzed by Mars Exploration Rover at Columbia Hills [8], and the large amount of apatite detected by the Curiosity rover in Gale crater [9]. Amounts of phosphorus as low as P₂O₅=0.2 wt.% in the martian mantle are sufficient to significantly impact magma compositions and mineral assemblages of the upper mantle and crust.

[1]Kushiro, I. (1996) Earth Process:Reading the Isotopic Code, 109-122. [2]Toplis, M.J., et al (1994) GCA, 58,2, 797-810. [3]Dreibus G. and Wänke H. (1985) Meteoritics, 20, 267-381.[4]Toplis, M.J., et al. (2008) LPSC XXXIX Abst. #1282. [5]Lodders, K., & Fegley Jr,B. (1997) Icarus, 126(2), 373–394. [6]Ding, S. and Dasgupta, R. (2015) LPSC 48, 2079. [7]Filiberto, J. and Dasgupta, R. (2011) EPSL, 304(3), 527-537. [8]McSween, H.Y. et al. (2006) JGR : Planets, 111, E09S91. [9]Forni, O. et al. (2015) GRL, 42.