Melting of garnet peridotite to 13 GPa and the early history of the upper mantle

By analogy with the evolution of the Moon,it has been suggested that the Earth may have had a `magma ocean' or liquid outer layer before 3,800 Myr ago.¹. Its presence would have had profound implications for the primary stratification of the mantle into upper mantle, transition zone and lower mantle^2,3.An essential constraint on this hypothesis is a knowledge of the high-pressure melting characteristics of mantle material, generally assumed to be peridotite in the upper mantle. We recently described⁴ the first melting experiments on mantle peridotite to 14 GPa. Using spinel Iherzolite KLB-1 from Kilbourne Hole, New Mexico, we showed that partial melts close to the solidus at 5-7 GPa are komatiitic in composition and that the solidus and liquidus converge at high pressures. Here we report further melting experiments on garnet Iherzolite PHN1611 from Thaba Putsoa, Lesotho. We confirm the convergence of the solidus and liquidus and we show a negative dT/dP slope for the liquidus above 7 GPa. These observations are briefly discussed in terms of their importance to the petrological evolution of the upper mantle and the presence of a magma ocean in the early Archaean.