Effect of Titanium on REE and HFSE Partitioning Between Garnet and Melt

Garnet is a strong fractionator of trace elements and plays an important role in the petrogenetic history of planetary interiors at high pressure. In order to model petrogenetic processes that operate within terrestrial planets accurately, it is important to understand how garnet partitions rare earth and high field strength elements. Here we assess the influence of Ti on garnet-melt trace element partitioning with a view both to constrain important crystal-chemical effects and to evaluate possible roles for garnet in lunar petrogenesis. Experiments were performed at ∼5 GPa and 1650-1675° C in a Walker-style multi-anvil high pressure apparatus using an Apollo 14 black picritic glass composition ( ∼17 wt% TiO₂) to assess the effect of Ti on garnet partitioning. These experiments were also designed to examine the possible presence of garnet in mare source regions. Experimental charges were analyzed for major and trace elements by EPMA and SIMS, respectively. D-values measured in this study using the Apollo 14 black Ti-rich composition are consistently higher than those measured by Draper et al. (2004, LPSC XXXV:1297), who used Apollo 15 green C glass (<0.5 wt% TiO₂). D vs. ionic radii are well-described for the trivalent cations by the lattice-strain partitioning model of Blundy and Wood (1994, Nature 372:452), with D₀ = 2.27 ± 0.40, E = 159 ± 58 GPa, and r₀ = 0.879 ± 0.044 Å (r₂ = 0.957). For comparison, this model applied to the low-Ti experiments of Draper et al. (2004) yields D₀ = 2.93 ± 0.25, E = 572 ± 40 GPa, and r₀ = 0.926 ± 0.005 Å (r₂ = 0.996) at ∼3.5 GPa. Both these fits show significant mismatch to the partitioning predicted by the formulations of van Westrenen et al. (2001, CMP 142:219), as previously shown for Fe-rich systems by Draper et al. (2003, PEPI 139:149). Use of our D-values (for rare earth and high field strength elements in batch-melting models) provisionally supports the hypotheses of Neal (2001, JGR 106:27865) and Neal and Shearer (2004, LPSC XXXV:2135), who proposed that garnet was present in the source regions of lunar mare basalts. The results of these calculations are similar to those using data from garnet-melt partitioning in very Ti-poor systems (Draper et al. 2004).