Abstract
For magmas cooling under conditions such that their oxygen contents are held fixed, or are subject to only minor loss or gain, there are several generalizations that are evident. If the early crystallizing phase has a lower redox state than that of the liquid, for example an olivine phenocryst in a tholeiitic liquid in an oceanic environment, then as temperature decreases, the progressive increase of ferric iron and sulphur in the liquid will induce the formation of a spinel phase, and an immiscible sulfide liquid. The latter Fe-S-O liquid may contain more O than S at FMQ (Doyle and Naldrett, 1987), providing an adequate reservoir for the oxygen needed to sustain a tholeiitic liquid line of descent on the FMQ buffer (Ghiorso and Carmichael, 1985). In combination, the continued formation of olivine, spinel and sulfide liquid, will regulate the iron redox state of the residual liquid, so that it may resemble a buffered path. Addition or subtraction of oxygen to the system (liquid plus solids) at constant temperature will initially change the proportions of the solids. But should the oxygen fugacity change, then the iron redox ratio in the crystalline phases (and the liquid) will also change. In those highly oxidized natural liquids whose iron redox ratio is greater than that of the precipitating spinel and rhombohedral oxide assemblage, then ultimately a representative of the pseudobrookite series will form, and thereafter the liquid line of descent involving a progressive increase in Fe 2 O 3 may cease. The assemblage of a rhombohedral (FeTiO 3 -Fe 2 O 3 ) solid solution and a pseudobrookite solid solution will have the effect of constraining the iron redox ratio in the liquid, and at constant temperature will act as a buffer to small changes in the oxygen content of the system. Again if the oxygen fugacity changes, at constant temperature, then the solid phases must either change in composition by changing their individual redox ratios, or one will disappear. Thus one problem of modelling a general liquid line of descent essentially hinges on the relationship between oxygen fugacity and the iron redox ratio of the phases that enter above the solidus. For natural silicate liquids this relationship is well known, but it has yet to be established for any ferromagnesian silicate solid solution series. Today the modelled crystallization path has to constrain the composition of the ferromagnesian minerals to their idealized composition, and for tholeiitic magmas this seems reasonably successful. For more alkaline-rich magmas, such as olivine-basalt or more silica-poor varieties, augite can show a large range in composition but experimental calibration of this is sadly lacking. Of additional concern is the fact that in these magmas a sulfide phase can even occur in combination with a sulphate-bearing solid phase (Luhr et al., 1984). The "bottom line" is that a vast experimental effort is required to establish the connection between composition, iron redox state, temperature and oxygen fugacity for the ferromagnesian minerals, particularly augite, biotite, the igneous amphiboles, in equilibrium with liquids.