Aluminum Enrichment in Silicate Melts by Fractional Crystallization: Some Mineralogic and Petrographic Constraints

The degree of aluminum saturation of an igneous rock may bedescribed by its Aluminum Saturation Index (ASI) defined asthe molar ratio Al₂O₃(CaO + K₂O + Na₂O). One suggested originfor mildly peraluminous granites (ASI between 1 and about 1.1)is by fractional crystallization of subaluminous (ASI < 1)magmas; hornblende, having ASI < 0.5, could be a major drivingforce in such a fractionation process. The efficacy of the processdepends not only on precipitation of hornblende and its effectiveremoval from the reacting system, but on the composition andnature of other coprecipitating phases, weighted by their modalabundances in the reactive system. Precipitation of feldspar(ASI = 1), for instance, would retard or even prevent aluminumenrichment in the melt if the ASI of melt is < 1, but wouldenhance such evolution if the ASI of the melt is > 1. Discussionof the efficacy of any mineral must be made in the context ofthe total reacting system. For hornblende to effectively cause a melt to evolve into aperaluminous composition, it must be able to coexist with peraluminousmagmas. Experimental phase equilibrium data show that at pressure> 5 kb hornblende can coexist with strongly peraluminousmelts (ASI {small tilde} 1.5). Scantily phyric volcanic rocksshow that hornblende can coexist with granitic magma havingASI {small tilde} 1.1 –1.2. The aggregate ASI of last-stageminerals of a typical granite is less than this value; therefore,even after hornblende has reacted out, the residual magma maybe expected to continue to evolve toward more aluminous compositions. Potentials and problems of using coarse-grained granitic rocksto probe courses of magmatic evolution are illustrated by asuite of samples from the Grayling Lake pluton of southwesternMontana. Such rocks probably always contain a large cumulatecomponent in their texture and should not be used as primarymeans to test the occurrence or efficacy of a fractionationprocess that might lead to peraluminous melts. The process isunlikely to give rise to peraluminous plutons of batholithicdimensions. A simple differential equation is presented thatallows the direct use of petrographic data (mineral chemistryand modal abundance) to predict the path of incremental evolutionof a given magma.