Fate Of Continental Crust Subducted to the Mantle Transition Zone: Experimental Investigations

Although some debate about the depth to which continental crust may be subducted continues among petrologists, geophysical and experimental modeling suggests that continental crust constituents may be carried to great depth into the Earth's upper mantle. Ultra-high pressure metamorphic rocks of granitoid bulk chemistry were only recently recognized due to the discovery of coesite, diamond, garnet and titanite containing six-fold coordinated Si and TiO₂ with the α -PbO₂ structure. Despite considerable scientific conservatism in accepting inferred subduction depths for continental materials, a wealth of new finds over the last decade have confirmed the subduction depths inferred from earlier discoveries of coesite and diamond in those terranes. These observations have pushed the accepted `limit' of continental metamorphism ever deeper and have given rise to new questions about the exhumation of UHP granitoid rocks. We have designed a new experimental program to explore the following questions: What fraction of the continental crust may survive transportation from shallow depth to great depth within the mantle? What is the proportion of solid versus melt? What high-pressure phases are stable in certain lithologies at different depths? We report here our preliminary results on experimental studies of mineralogical assemblies of granitoid rocks subjected to recrystallization at P=10 to 12 GPa and T=1173 to 1474K in a Walker-style multianvil apparatus. The run products consist of K-(Na)-hollandite, stishovite, jadeite-rich clinopyroxene and poikilitic intergrowths of majoritic garnet (Si=3.11 c.p.f.u.) with phengite. Tiny crystals of an Al-silicate phase were also observed. The stoichiometry of the Al-silicate phase greatly differs from all currently known high-pressure Al-silicates such as kyanite, topaz-OH, but it is close to the hydrous aluminosilicate phase, AlSiO<sub>3</sub>(OH), where ratio of Al to Si is 1:1. Although we have not yet verified the structure of this mineral due to its small crystal size (0.5-1 μ m), we assume that this is a phase similar to that observed by Ono (1998) in his 15 GPa `wet' experiments on a similar bulk chemistry. Presence of `garnet-phengite' poikilitic domains, an unusual texture for such high-pressure experiments, may reflect local partial melting initiated by an excess of H<sub>2</sub>O-fluid migrating at the grain boundaries due to the breakdown of biotite. Our results show that `continental crust' mineralogical systems need to be explored in detail by experimental determination of phase relations, phase equilibrium, reproduction of mineralogical diversity, calculations of the density of newly synthesized aggregates and evaluation of the proportion between melt and solids.