Development of Melt Interconnectivity During Ductile Deformation of an Amphibolite

Partial melting of hydrous phases such as amphibole, biotite, and muscovite occurs in orogens where distributed ductile thinning is causing exhumation of mid- to lower-crustal rocks. The partial melting of these hydrous phases contributes significantly to the physical and chemical evolution of the crust, as well as affecting the crust’s strength. The Si-rich melts generated from partial melting reactions of mid- to lower-crustal assemblages migrate toward the upper crust leaving a more mafic restite. Previous laboratory experiments conducted on amphibole-, biotite-, or muscovite-bearing rocks performed at rapid strain rates (10-4/s to 10-5/s) result in brittle deformation due to high local pore pressures. These rapid experiments suggest this brittle behavior is the likely mechanism causing melt segregation in the crust. However, field evidence and slower strain rate experiments (10-6/s to 10-7/s) suggest that crystal plastic processes may be dominant during syndeformational partial melting. To investigate grain-scale melt segregation mechanisms in a common lower crustal protolith, I performed a suite of axial compression and general shear experiments on an amphibole-bearing source rock during syndeformational partial melting at T = 800-975°C, Pc = 1.5 GPa, at a strain rate (ε̇) of 1.6 x 10-6/s. I also performed axial compression experiments on a biotite-bearing gneiss and a muscovite-bearing quartzite at T = 950°C, Pc = 1.5 GPa, at a strain rate (ε̇) of 1.6 x 10-6/s to compare the differences in melt development depending on which hydrous phase is partially melting. The Nemo Amphibolite (d = 140 ± 85 μm) is composed of 62 vol% amphibole (Fe-hornblende), 27 vol% plagioclase (andesine; An30Ab69Or1), 8 vol% quartz, and 3 vol% titanite. The biotite-bearing gneiss (d = 80 +/- 40 microns) consists of quartz (43 vol%), plagioclase (andesine (An22Ab77Or1); 40 vol%), biotite (16 vol%), and ~1 vol% muscovite/Fe-Ti oxides. The muscovite-bearing quartzite is composed of 90 vol% quartz and 10 vol% muscovite. All experiments were performed using a Griggs apparatus using solid salt assemblies. The strengths of the Nemo Amphibolite cylinders decreased linearly as a function of increasing temperature (T = 800-975°C) and as melt contents increased from trace amounts at T = 800°C to ~20 vol% at T = 975°C. The Nemo Amphibolite cylinders deformed homogeneously; no evidence of intragranular brittle deformation was observed. Deformation microstructures observed at low strains (e.g. undulatory extinction) and a calculated stress exponent of n = 3.2 ± 0.2 suggests that the dominant deformation mechanism in the amphibolite is dislocation creep. Microstructures observed at high strains (e.g. flattening of grains, undulatory extinction, grain recrystallization, and deformation lamellae) suggest a transition from dislocation creep processes to diffusion-dominated processes operating at high melt contents. At T < 950°C and melt contents ≤ 15 vol%, the amphibole partial melting reaction produces tonalitic-granodioritic melt in isolated lenses at plagioclase-amphibole and amphibole-amphibole grain boundaries oriented roughly parallel to the compression direction, and the melt does not cross-cut grains. At T = 975°C, there is a dramatic decrease in strength that corresponds to a change from a stress-supporting network of amphibole, plagioclase, quartz, and titanite at melt fractions ≤ 15 vol% to a fully interconnected network of melt at melt fractions ≥ 20 vol%. The interconnected melt has a uniform granodioritic composition. The amphibole partial melting reaction results in the production of Si-rich melt and new amphiboles, plagioclase, pyroxenes, and garnets. These results indicate that as buoyant melts form in amphibolites, they will not be able to migrate from the protolith until high (>15 vol%) melt fractions are formed. Once buoyant melt becomes interconnected, it will likely segregate from the source, leaving a denser, eclogitic restite. This sudden migration of melt and densification of the slab during flat slab subduction may be the trigger for initiating delamination of the slab.