Geotherm with peridotite melting
2.8 Why? Evidence for hydration melting

Water is released from hydrated ocean crust (spilite) and subducted sediment as the downgoing slab is heated and metamorphosed. It is likely that some of that water moves back up the subduction fault zone. But if dehydration reaction water migrates into the mantle overlying the subducted slab, the mantle will react with the water to form basaltic magma. We know this because of the results laboratory experiments on peridotite in the presence of water. Look at again at Figure 2.09. H2O is very soluble in silicate liquids and is a signficant "antifreeze" for rocks. Mouse over Figure 2.10 (above) to see the region (colored yellow) of the overlying mantle that has P-T conditions for peridotite melting due to added water as identified by Grove and others (2006, fig. 8). Grove and others (2006) show that the amount of mantle melting depends on the amount of water that is added. The yellow region ends at a depth of 108 km because Grove and others (2006) argue that the bulk of the available water in the slab will have been released by that depth.

Basaltic liquid is produced by partial melting of peridotite. Some basalt is found in volcanic arcs. However, the most common magma in volcanic arcs over subduction zones is andesitic, both for ocean-ocean and for ocean-continent convergent plate boundaries. Therefore, processes must occur along convergent plate boundaries that start with basalt magma and lead to andesite and even rhyolite. Other parts of this book will consider those processes.

Decompression causes mantle melting at divergent plate boundaries. Adding water causes mantle melting at convergent plate boundaries. Why are there volcanoes within tectonic plates?