“…The similar H 2 O-dD negative correlation was also found in clinopyroxene megacryst and ascribed to diffusive H loss during magma ascent to the surface (e.g., Newman et al, 1988). While in principle the H-rich fluid diffuses much faster than the Drich fluid, it is more possible that molecular H 2 O diffuses faster than the structural OH (e.g., Zhang et al, 1991;Doremus, 2004). Thus, the observed H isotope variations in the two CCSD-MH core segments may be principally caused by the preferential loss of molecular H 2 O from garnet and omphacite during the exhumation of deeply subducted continental crust.…”