Demand for scandium is increasing due to the use of scandium-aluminum alloys in the aerospace and sporting goods industries, and in solid-oxide fuel cells. Scandium deposits are associated with mafic intrusions and laterites, carbonatites, and granitic pegmatites and the element is recovered as a byproduct of uranium, titanium, apatite, and rare earth element mining. Understanding the processes that form scandium-rich deposits is vital in order to inform exploration for such deposits. The deposits and mafic intrusions at Kiviniemi, Finland and Nyngan, Australia, as well as the granitic pegmatites of Evje-Iveland, Norway are of particular interest. Experiments and thermodynamic modeling of magmatic-hydrothermal systems have been performed in order to constrain the petrogenesis of these deposits.

Cold-seal pressure vessel experiments have been performed on systems with basaltic to rhyolitic compositions in order to evaluate the behavior of scandium in upper crustal magmas. Partition coefficients for scandium between olivine, pyroxene, plagioclase, biotite, spinel, cordierite, aluminosilicates, ilmenite, rutile, apatite and silicate melts, were determined and found to vary as a function of mineral and melt compositions. These partition coefficients were combined with MELTS modeling (MELTS is a software package that is used for performing thermodynamically constrained phase equilibria calculations) to evaluate the behavior of scandium during fractional crystallization of a mafic melt and formation of a cumulate, the subsequent partial melting of that cumulate, then the isothermal decompression and final cooling of that melt. Fractional crystallization can produce scandium-rich cumulates, such as those found at Kiviniemi and Nyngan. However, felsic melts produced by partial melting of a scandium-rich cumulate have, at most, scandium concentrations consistent with the upper continental crust.

Amphibolite partial melting experiments were performed in a piston-cylinder to constrain the petrogenesis of the Evje-Iveland pegmatites. These experiments are inconsistent with the long-held hypothesis that the pegmatites formed by partial melting of their host amphibolite. Instead, magmatic differentiation is the preferred petrogenic model. This model requires that few ferromagnesian phases occur during crystallization of a felsic melt or the presence of scandium complexes that reduce scandium partition coefficients.