Fig. 3

Cu/Ag and S/Se systematics during crustal differentiation of OPB and MORB. a Cu/Ag and b S/Se versus [MgO] of OPB and MORB. Because the D^sulf/sil of Cu and Ag are similar (e.g., ref. ¹), both OPB and MORB have constant Cu/Ag regardless of whether the melts were sulfide-saturated or under-saturated. Sulfide-saturated MORB and the Ori Massif samples show an increase in S/Se with decreasing [MgO], consistent with sulfide fractionation. The Tamu Massif samples have considerably lower S/Se than other OPB samples and MORB, which is consistent with S-loss during degassing. Hence, the S/Se of the Ori Massif samples were used to estimate [S] of the Tamu Massif samples prior to degassing ([S*]). Some of the OPB are offset to lower S/Se than MORB and have comparable S/Se to primitive mantle estimates (ref. ¹³), which might suggest that sulfide was exhausted from the OPB mantle source during melting. However, the Kroenke samples also have comparable S/Se to primitive MORB. Additionally, the lower S/Se are expected because the OPB magmas fractionated a smaller proportion of sulfide compared to MORB (i.e., MORB have lower [Cu] at a given [MgO] compared to OPB samples). Data for MORB are taken from Jenner and O’Neill²³, except Se data (see Methods). Error bars are given as the propagated 1SD Se error derived by repeat analysis of the NWLSC standards (see Methods)