Evidence for metasomatism of the lunar highlands and the origin of whitlockite

Abstract

The presence of rare earth element-rich (REE-rich) whitlockite in the relatively primitive, REE-poor lunar highlands presents a paragenetic problem to the petrologist. Comparison with occurrences in mare basalt mesostasis assemblages demonstrates that whitlockite in the lunar highlands has not resulted from the crystallization of the residual interstitial melt. Furthermore, by the nature of REE partition coefficients for whitlockite, it is demonstrated that they cannot have crystallized from KREEPy melts. Rather, we suggest that the only lunar melt which contains high enough REE abundances is the basic immiscible melt derived from a KREEPy liquid. We envisage that whitlockite found in the lunar highlands is a post-cumulus metasomatic phase precipitated from a melt formed when urKREEP underwent silicate liquid immiscibility. The basic immiscible melt thus produced was able to percolate through the lunar highlands due to its low viscosity and act as a metasomatizing agent. Apatite is occasionally found as either epitactic overgrowths on the whitlockites, mantled by whitlockite, or in composite grains with whitlockite. In all cases, these apatites contain much lower REE contents than the corresponding whitlockite. Experimental results suggest either 1. (1) apatite and whitlockite have essentially the same REE Kd's 2. (2) apatite has lower REE Kd's than whitlockite. Both scenarios can be used to explain certain apatite-whitlockite associations. For example, apatite and whitlockite having the same REE Kd's can explain the situation where apatite appears to have crystallized after whitlockite. This is because apatite crystallization requires a certain level of halogens in the melt. These levels are only realized after whitlockite crystallization, which enriches the residual melt in F and Cl, but depletes it in the REEs. Therefore, when apatite crystallizes it will contain lower REE abundances than whitlockite. However, relationships where apatite and whitlockite appear to have co-crystallized, or apatite crystallized before whitlockite, require apatite to have lower REE Kd's. Even if both experimental results are correct, the whitlockite parental melt still requires REE abundances greater than in KREEP/urKREEP. The parental melt has similar REE abundances to a calculated REEPy immiscible melt from KREEP/urKREEP.