IGNEOUS PHASE DIAGRAMS: PAST, PRESENT, AND FUTURE DIRECTIONS

Over the past century, there has been a tremendous investment in experiments to establish phase equilibria first in simple systems (where phase relations can be represented on ternary diagrams) and then in complex natural systems (which more accurately represent nature but are more difficult to interpret or to represent graphically). The more recent development of internally consistent thermodynamic databases and algorithms for their manipulation allows for complex phase relations to be plotted not just in composition space, but in other useful formats such as entropy diagrams along prescribed P-T paths (e.g. Asimow et al. 1995 GCA), with the possibility of including fractional crystallization or assimilation. So is there anything left to do?

While phase diagrams tell us the equilibrium assemblage, they do not tell us about textures or what happens under disequilibrium conditions, for example when crystal growth occurs at large degrees of undercooling. The spectacular range of textures achievable in silica-rich melts includes obsidian, rhyolite, granite and pegmatite. Their formation does not correlate directly with cooling rate, because granites cool the slowest, but also results from the suppression of crystal nucleation under hydrous conditions, so texture really depends on the full P-T-X(H₂O)-t history (e.g. Nabelek et al. 2010, CMP). A more extreme example of disequilibrium phenomena is the observation of amorphous solids that represent metastable eutectic compositions, produced in condensation experiments that simulate conditions in the early solar nebula (e.g. Rietmeijer and Nuth 2000, EOS).

The future of igneous phase diagrams may lie in mapping out P-T-X-t histories, in a similar way to how metamorphic rocks are interrogated. The chief difference lies in the potential for igneous processes to occur on much faster timescales, producing amorphous solids (volcanic glass, astronomical smokes) in addition to chemically well-behaved crystalline phases.