Can Fractional Crystallization of a Lunar Magma Ocean Produce the Lunar Crust

New techniques enable the study of Apollo samples and lunar meteorites in unprecedented detail, and recent orbital spectral data reveal more about the lunar farside than ever before, raising new questions about the supposed simplicity of lunar geolo-gy. Nevertheless, crystallization of a global-scale magma ocean remains the best model to account for known lunar lithologies. Crystallization of a lunar magma ocean (LMO) is modeled to proceed by two end-member processes – fractional crystallization from (mostly) the bottom up, or initial equilibrium crystallization as the magma is vigorously convecting and crystals remain entrained, followed by crystal settling and a final period of frac-tional crystallization [1]. Physical models of magma viscosity and convection at this scale suggest that both processes are possible. We have been carrying out high-fidelity experimental simulations of LMO crystal-lization using two bulk compositions that can be re-garded as end-members in the likely relevant range: Taylor Whole Moon (TWM) [2] and Lunar Primitive Upper Mantle (LPUM) [3]. TWM is enriched in refrac-tory elements by 1.5 times relative to Earth, whereas LPUM is similar to the terrestrial primitive upper man-tle, with adjustments made for the depletion of volatile alkalis observed on the Moon. Here we extend our earlier equilibrium-crystallization experiments [4] with runs simulating full fractional crystallization. Fractional crystallization should yield a more pro-nounced change in residual liquid composition than equilibrium crystallization, and therefore potentially the earlier and more extensive crystallization of plagio-clase which is required to produce the plagioclase flo-tation crust represented by the anorthositic highlands. Both processes have been computationally simulated [5,6], and appear to produce broadly similar assem-blages despite differences in bulk lunar composition and model parameters. However, an important limita-tion of theoretical models is the inability to constrain the last stages of LMO crystallization, in which both the plagioclase crust and KREEP component will form. We present the results of fractional crystallization experiments on TWM and LPUM, and their implica-tions thus far for the crystallization of a LMO and thickness of the anorthosite crust.