ANOMALOUS ELECTRICAL SIGNALS IN SUBDUCTION ZONES: INSIGHTS FROM ELECTRICAL CONDUCTIVITY OF SILICATE MELTS

Magmatism in subduction zones fuels arc volcanism. Partial melting may begin deeper than 80 km depth in a subduction zone. The source rocks for melting, reactive transport during ascent, chemical evolution by fractional crystallization, and/or mixing of distinct magmas result in a chemical spectrum of melt compositions prior to eruption. At depth, the magmas may be dominantly mafic. Yet such processes can produce andesitic to rhyolitic magmas which ultimately create new continental crust. Direct study of these processes at depth by geophysical remote sensing helps us to better understand arc volcanism and the growth of continental crust. Magnetotelluric (MT) surveys have been useful in detecting such magmatic systems due to the high electrical conductivity (EC) of magma compared to its host rock. Laboratory-based EC measurements are often crucial to interpreting complex MT data. However, existing EC measurements are often on specific melt compositions and do not capture the full compositional gradient relevant for subduction zones. We ask, how does EC of silicate melt vary due to silica, alumina, alkali, and alkaline contents across the diversity of melt compositions as melts originate and ascend in subduction zone settings? In this study, we build on previous work and explore the relationship between melt composition and EC in magmas at pressures and temperatures relevant to subduction zones. We performed EC experiments using basaltic to andesitic endmember melt compositions. For realistic melts, we also explored how volatiles, such as H₂O, influence the melt EC. We find that high (Na,K)₂O + (Mg,Ca)O versus SiO₂ + Al₂O₃ may overall decrease the activation volume for EC. Our results indicate that mafic magmas are overall more conductive than felsic magmas at high pressure. The EC of felsic magma is also more sensitive to compression than in mafic magma. Dissolution of H₂O into silicate melt further lowers the activation energy and thereby enhances the EC for any melt composition. Our data allow us to constrain compositions and fractions of melt to explain high EC signals in subduction zone conditions.