Origin of Miocene andesite and dacite in the Goldfield–Superstition volcanic province, central Arizona: Hybrids of mafic and silicic magma mixing

Abstract

The Miocene Goldfield–Superstition volcanic province (G-SVP), ∼8000km2 in central Arizona, is composed largely of silicic pyroclastic rocks and lavas, and smaller volumes of alkalic basalt and intermediate-composition lavas. Volcanism began ∼20.5Ma as sparse rhyolitic and mainly basaltic lavas followed by intermediate lavas, lasting until ∼19Ma. At that time, ∼1 m.y. of silicic eruptions began, creating most of the G-SVP. Petrologic studies are available for basalts and some for silicic rocks, but petrologic/geochemical information is sparse for intermediate-composition lavas. These latter, andesites and dacites, are the focus of this study, in which we present the processes and sources responsible for their origins.Goldfield–Superstition andesites and dacites have SiO2 ∼56–70wt.% and Na2O+K2O that qualifies some as trachy-andesite and -dacite. A prominent petrographic feature is plagioclase-phyric texture (∼11–30vol% plagioclase), where oligoclase–andesine phenocrysts have cores surrounded by corroded, or reacted, zones, mantled by higher An% plagioclase. Where corroded zones are absent, margins are etched, curved, or embayed. Groundmass plagioclase is labradorite, also more calcic than the phenocrysts. Other minerals are quartz (subrounded; embayed), clinopyroxene, amphibole, biotite, and rare titanite and zircon.A salient compositional characteristic that provides insight to andesite–dacite origins with respect to other G-SVP rocks is revealed when using SiO2 as an index. Namely, abundances of many incompatible elements, mainly HFSE and REE, decrease over the low to high SiO2 range (i.e., abundances are lower in dacites than in co-eruptive andesites and underlying alkalic basalts). As examples: G-SVP basalts have ∼50–70ppm La, and andesites–dacites have ∼59–22ppm La; for Zr, basalts have ∼225–170ppm, but most andesites–dacites have ∼180–50; for Y, basalts >20ppm, andesites–dacites ∼18–9ppm. To understand these trends of lower HFSE and REE with increasing SiO2, we modeled fractional crystallization of G-SVP alkalic basalt (∼50wt.% SiO2; ∼9wt.% MgO), dehydration melting (10–25%) of granodiorite and high-K amphibolite, and basalt–rhyolite magma mixing.Fractionation and melting each require specific modal percentages of titanite, zircon, and allanite (e.g., ⩽1%), the high ends of ranges for accessory-mineral/liquid partitioning coefficients, continual crystallization of accessory minerals from basalt to dacite (for fractionation), and specific source-melting percentages and low titanite, zircon, and allanite melting proportions (∼0.02; dehydration melting). These requirements are too stringent to be realistic. Moreover, accessory minerals are rare in these lavas, and neither fractionation nor melting accounts for the plagioclase textures observed.On the other hand, low-HFSE, -REE rhyolites (e.g., La 9–31ppm; Zr 31–93; Nb 9–17; Y 4–10) containing Na-plagioclase are in the G-SVP and were temporally and spatially available to have mixed with G-SVP basalts. Mixing proportions of 20:80 to 90:10 for different rhyolite:basalt combinations yield hybrid compositions that overlap the G-SVP andesite–dacite compositional fields. Also, basalt invading rhyolite reservoirs containing mush zones can account for Na-plagioclase concentrations of ∼11–30vol% formed after mush disruption and dispersal, plagioclase corroded-cores and higher-An% mantles and groundmass, and subrounded-embayed quartz. The straightforward explanation for G-SVP intermediate lavas, then, is repeated hybridization of basaltic and low-HFSE, -REE rhyolitic magmas during the early stages of G-SVP magmatism.