Late stage magmatic evolution of A-type rocks around and to the southeast of Olympic Dam, South Australia

The Mesoproterozoic Gawler Range Volcanics (GRV) of South Australia include several voluminous felsic lavas that are important extrusive analogues to the coeval Hiltaba Suite (HS) intrusive rocks. These rocks are notable for their voluminous and high temperature nature, the large area over which they occur (850 by 500 km), and because they both host and are implicated in the formation of iron oxide-copper-gold deposits in the Olympic Province. A range of conflicting petrogenetic scenarios have been proposed for the GRV and related HS rocks. These models variously propose fractionation from mantle-derived melts or crustal fusion driven by mantle-derived melts to produce the magmas which formed these rocks; however, the development of a consistent model has been hindered by the large size of the volcanic province, the scarcity of unaltered rock, and the relatively small datasets of these studies. These challenges have resulted in a paucity of research examining the geochemistry and mineralogy of the GRV units in detail, particularly within the recently unified stratigraphic framework and the emergence of a consensus that several GRV units are effusive lavas rather than large, welded 'ash-flow' sheets of pyroclastic origin. This thesis addressed this shortfall through evaluation of the textures and compositions of crystals, crystal clusters and clustered distribution of minerals in newly collected, unaltered samples of the Upper GRV lavas and the Roxby Downs Granite (RDG), with the aim of understanding the origins of the crystals and clustered distributions of crystals and appraising their implications for the evolution of their host rocks. The RDG and Upper GRV lavas both contain clusters of touching silicate and oxide minerals. Crystal clusters in the RDG comprise magnetite + apatite ± titanite ± biotite ± zircon that are typically enveloped in a plagioclase crystal-matrix. Consistent Fe-Ti oxide + apatite + zircon + plagioclase assemblages and textures in both the Upper GRV and RDG suggest that for their hot and dry magmas parental magmas, cooling cause pyroxene + Fe-Ti oxide + plagioclase in GRV-type crystal clusters to react with both each other and the melt, forming HS-type magnetite + titanite + biotite + feldspar crystal clusters. Pyroxene and titanomagnetite compositions suggest that the Upper GRV lavas are related through differentiation in the upper crust, and imply that all analysed crystals were formed within, or were at least equilibrated to, the host lavas' precursors. Different intergrowth textures between crystal clusters in the GRV lavas, and the widespread presence of magmatic enclaves, suggest that the majority of GRV crystal clusters were liberated from closely packed crystal masses through magma rejuvenation and associated crystal resorption. Within- and between-crystal pyroxene compositions suggests some phenocrysts did not grow in situ, adjacent to the other crystals found in individual clusters in the solidified rock, but rather aggregated or settled with other crystals following a stage of growth as independent, melt-bound phenocrysts. The last-erupted lava contains the highest proportion of clusters comprising complex and anhedral intergrowths. This is consistent with a normally zoned reservoir where the crystal:melt ratio and crystal concentration are greatest at depth. Loosely packed frameworks and free crystals are interpreted to occur in the middle and upper sections of the reservoir, respectively, represented by the first erupted Upper GRV lava and quartz-phyric portions of the lavas. Plagioclase is the most abundant phenocryst throughout the lavas but is present in only a subset of crystal clusters. The low abundance and absence of feldspar and quartz in clusters is likely due to mineral resorption and subsequent separation and loss of these components from other cluster components prior to and during the eruption the studied lavas. Once solid-solid crystal contacts were dissolved, the separation of the cluster components was driven by the strain placed on them by melt flow. Zircon is principally associated with Fe-Ti oxides and clusters of touching crystals in these rocks. There are few published reports of concentrations of zircon with Fe-Ti oxides, nor explicit evaluation and explanation of the potential origins of zircon-rich crystal clusters found in the ~1.59 Ga Gawler Craton rocks and similar rocks of western North America and elsewhere. The lack of pre-magmatic zircon, consistent intra-grain and inter-grain zircon compositional trends, the predominance of oscillatory zoned zircon with morphologies indicating growth from hot, evolved silicate melts, and the lack of evidence for zircon recrystallisation, indicates that zircon crystallised in the host GRV and RDG magmas. Variable zircon compositions within individual clusters does not support epitaxial nucleation of zircon on Fe-Ti oxides, however it is likely that some zircon grew from seed crystals formed by exsolution of Zr from Fe-Ti oxides. Aggregation of isolated, liquid-bound crystals is energetically favourable, and the grain size discrepancy between larger crystals (Fe-Ti oxides, pyroxenes) and smaller accessory minerals (zircon, apatite) maximises the disparity in particle velocities and hence enhances the opportunities for collisions and adhesion between these crystals. We propose that zircon adheres to Fe-Ti oxides with greater ease and/or with greater bond strengths, than to other phases present in the parental magmas. It is possible that this association is related to interactions between zircon and Fe-Ti oxide surface sites with opposing charges, presuming the distance between phase surfaces is sufficiently small. This thesis establishes a more precise model for the formation of the Upper GRV lavas and related Hiltaba Suite intrusive rocks, and provides insights into the origins of crystals found within these rocks, crystal cluster formation mechanisms and settings, establishment of magma chamber zonation, quartz-feldspar recycling and mafic magmatism in the genesis of the Upper GRV lavas, and the formation of HS-type clusters from GRV-type clusters. These findings contribute to the theory that voluminous and hot magmatic systems that express a dominant evolved component, such as the GRV and HS, likely depend on a stable connection to actively melting asthenospheric mantle. Furthermore, it is likely that a complete spectrum of mafic to felsic magmas being generated during the Hiltaba event (i.e. any compositional gaps are related to sampling bias and igneous developmental paths that are less favourable for the preservation of intermediate magma compositions), and that the lavas do not accurately reflect the long-term state of the magma reservoir.