Challenges in the determination of highly accurate experimental mineral-melt partition coefficients

Partition coefficients (D) depend on temperature, pressure, redox conditions, and on the chemical composition of solid and liquid. Trace element modeling of evolution of the Lunar Magma Ocean (LMO) rely on modeled and experimentally determined D’s. However, most of these D’s have large uncertainties, due to the in-situ analytical methods commonly used. Our aim is to develop an improved technique to obtain more accurate trace element partition coefficients from experimental samples. Since clinopyroxene (Cpx) is the most important carrier of trace elements in the crystallizing LMO, we currently focus on Cpx/melt partitioning.

We conducted 1-atm experiments with lunar compositions based on previous studies in gas-mixing furnaces. Starting material compositions contain a variety of trace elements including REE, transition metals, large ion lithophile and high field strength elements. Using the Re-wire loop technique, the experiments cover fO2 relevant for the Moon (i.e., IW to IW-2).

We measured mineral separates and glasses using isotope dilution ICP-MS methods. The separation of Cpx and glass from the experimental mounts is a key challenge. We compare mechanical separation by drilling using a Micro-Mill 2.0 (Element Scientific Lasers) with separation by hand-picking and in-situ LA-ICP-MS measurements to assess which method yields the most accurate D’s.

Our new partition coefficients of key trace elements in the LMO will improve models for its evolution.