The effects of variable bulk composition on the melting systematics of fertile peridotitic assemblages

Experiments were conducted at 1 GPa on four starting materials to investiga te the effects of variable mineral proportions on the melting systematics o f compositionally fertile peridotitic assemblages. Starting materials were constructed by recombining Kilbourne Hole xenolith mineral separates by wei ght into four mixtures with mineral proportions olivine (01): orthopyroxene (Opx): clinopyroxene (Cpx): spinel (Sp) of 0.50:0.07:0.40:0.03 (FER-B), 0. 50:0.46:0.01:0.03 (FERC), 0.50:0.30:0.10:0.10 (FER-D), and 0.50:0.235:0.235 : 0.03 (FER-E). Experiments were performed on a 1.27cm (0.5 in.) piston-cyl inder apparatus over the temperature interval 1270-1390 degreesC, using a v ariation of the diamond aggregate melt extraction technique employing vitre ous carbon spheres in place of diamonds as the melt extraction layer. The s olidus temperatures are similar for all the starting materials, with an ave rage value of 1250 degreesC. In FER-D and -E, the near-solidus melting reac tion for a Iherzolite assemblage was determined to be of the form Cpx + Opx + Sp -> melt + Ol. A subsequent reaction of the form Opx + Sp ->,melt + Ol was determined for FER-D after the exhaustion of Cpx. Over the entire temp erature interval investigated for FER-B and -C, reactions were determined t o be of the form Cpx + Sp -> melt + Ol and Opx + Sp -> melt + Ol, respectiv ely. Melt percent (F) vs temperature (T) curves are concave up for all star ting materials, demonstrating that isobaric melt productivity increases wit h progressive batch melting. At any given melt fraction, (dF/dT)(p) increas es with increasing amount of Cpx in the starting material, indicating that the modal proportion of Cpx is one of the primary controls on isobaric melt productivity of upwelling peridotite. The concave up melt productivity fun ctions for peridotitic assemblages determined in this study suggest that as suming linear or concave down melt productivity functions for modeling mant le melting may not be appropriate.