Phase Relations of Basalts in their Melting Ranges at PH2O = 5 kb. Part II. Melt Compositions

Abstract

This paper describes the melting relations of three basalts, a Picture Gorge tholeiite, the 1921 Kilauea olivine tholeiite, and the 1801 Hualalai alkali basalt, at 5 kb water pressure, 680–1045 °C, at the oxygen fugacities of the quartz-fayalite-magnetite (QFM) and hematite-magnetite (HM) buffer. All melts produced within the hornblende stability field are strongly quartzo-feldspathic. All are quartz-normative, including those from the alkali basalt, and all but two of the melts are corundum-normative. Melt composition shows very little dependence on oxygen fugacity within the hornblende stability field, as MgO and FeO contents are very low. When hornblende begins to melt extensively (1000°–1045 °C), the TiO2, FeO, and MgO contents of the melt increase abruptly. In this range, melts formed on the HM buffer have much higher Mg/Fe ratios and lower TiO2 than melts formed on the QFM buffer. Melt composition is also quite insensitive to changes in basalt composition, within the hornblende stability field. The chief exception is the Na/Ca ratio, which varies directly with Na/Ca in the starting basalt. When projected into the Ab-An-Or-Qz quaternary system, all melts produced follow a rather narrow spiral path through the tetrahedron; they descend from the An corner, moving toward Qz at constant Ab/Or, moving toward Or only when plagioclase± quartz begin to precipitate. The melting behavior of hornblende, plagioclase, and augite in these experiments has been examined closely, with the following results: successive partial melts may differ from each other by compositional increments which are very different in composition from the minerals contributing to the melt in the temperature interval under consideration. These increments can almost never be expressed solely in terms of members of the one or two mineral solid solutions from which they are actually derived. In a few cases the increments cannot be expressed in terms of any reasonable combination of minerals. This pattern contrasts markedly with that observed in fractional crystallization, in which the difference between successive melts must always correspond to present or possible phenocryst minerals. The contrast implies that magma series generated by any kind of melting process, equilibrium or fractional, should be recognizably different from series generated by fractional crystallization, if minerals like hornblende or pyroxene are involved.