Figure 5: Simplified model of mantle melting.

(a) Variation of melting temperatures with depth as a function of mantle CO₂ content (blue curves)^2,50, and electrical conductivity versus depth (red curves) for the upper mantle beneath the Philippine Sea⁵¹ (thin red curve) and north Pacific⁵² (broad red curve). Shaded boxes at the base of the plot represent the range of laboratory electrical conductivity measurements for dry olivine (i), olivine containing 150 p.p.m. H₂O (ii) and melt (iii) (refs 30, 50). (b) Schematic cross-section of the permeable melting regime beneath a mid-ocean ridge based on a, using the same vertical scale, a mantle potential temperature of 1,345 °C and a permeability threshold of 0.05%. At depths below the nominally anhydrous solidus, melting is dominated by small fractions of carbonate and carbonated silicate melt. Shaded grey zones indicate possible, depths for carbonated silicate melt initiation depending on the oxygen fugacity (fO₂) of the mantle^53,54,55,56. Onset at the carbonated silicate solidus¹ assumes that mantle fO₂ is high enough to stabilize carbonate over diamond at 275 km. If the mantle is more reducing, melting will instead initiate at the fO₂-dependent diamond to carbonate transition. Melt initiation between 140 and 180 km encompasses whole rock Fe³⁺/ΣFe ratios from 0.035 to 0.05 based on the continental xenolith record³⁸—depths that are also consistent with a range of MORB Fe³⁺/ΣFe ratio, Fe³⁺ bulk partition coefficients, and primitive mantle Fe₂O₃ contents¹⁹ and the assumption that 37p.p.m. C has the power to reduce Fe³⁺/ΣFe by >1% as it oxidizes to carbonate. Because the subsolidus Fe³⁺/ΣFe ratio of the mantle and the depth of metal saturation are uncertain, the depth of melt initiation is also uncertain.