- This dissertation is concerned with the behavior of sulfur in intermediate-silicic arc magmas associated with subduction at convergent margins. In particular it focusses on oxidized, sulfur-rich magmas, the conditions at which they might reach sulfate saturation, and implications of sulfate saturation. It is divided into an investigation of natural samples (Chapters II & III) and experimental work (Chapters IV, V and VI). It is motivated by the simple observation that magmatic-hydrothermal ore deposits, particularly porphyry Cu ± Au ± Mo deposits, contain large quantities of sulfur, and therefore require sulfur-rich magmas. Since the magmas most commonly associated with porphyry deposits have very little ability to carry dissolved sulfur species, it has been suggested that they must instead store sulfur in the form of the mineral anhydrite. Furthermore it has been suggested that anhydrite is replaced by an immiscible sulfate-melt phase at high temperature, although this has been the subject of little previous study.
In Chapter II I have used automated energy dispersive spectroscopy (EDS) mapping to identify rare inclusions of anhydrite trapped in resistant mineral phases including apatite, amphibole, plagioclase, quartz and titanite. Anhydrite inclusions are identified in samples from five different magmatic-hydrothermal ore deposits and one sulfur-rich volcanic center, suggesting that sulfate saturation may be more common in arcs than is generally thought. Electron probe microanalyzer (EPMA) measurements show that primary magmatic anhydrite has elevated Sr (>2500 ppm) and P (>500 ppm) compared to hydrothermal anhydrite, although in some cases post entrapment exchange between anhydrite and host mineral has affected the composition. Cathodoluminescence imaging of host mineral zonation, particularly in apatite, can also be used to demonstrate that anhydrite inclusions are of igneous origin.
Analysis of apatite from the same samples (chapter III) shows that in all cases the anhydrite-bearing magmas also produced high-S apatites (>0.3 wt% S). Phenocrystic apatite and apatite inclusions were identified and analyzed by electron probe micro-analyzer (EPMA) for volatile elements (Cl, F, S) and trace elements (Sr, Ce, Na, Si). Host minerals were also analyzed for a range of major and trace elements. Despite a large range in apatite S contents (~0.05 – 0.5 wt% S in most samples) there is little correlation between apatite S contents and various proxies for magmatic evolution (including apatite Sr content, apatite volatile content and host mineral crystallization temperature). The observed range of apatite S contents necessitates that magmas underwent either a large (>200°C) change in temperature or the loss of sulfur to a fluid phase. That these processes were not reflected in the data suggests trends may have been obscured by reequilibration and exchange between apatite and host mineral.
The experimental work detailed in chapters IV and V investigates the relative stabilities of immiscible sulfate melt and anhydrite in arc magmas, and characterizes trace element partitioning between silicate melt and both sulfate melt and anhydrite. Experiments were conducted using piston-cylinder and gas-pressurized cold seal apparatus at conditions of 800-1200°C, 0.2-1GPa and ƒO2>NNO+2. Synthetic starting materials were based on trachydacite and trachyandesite composition, with 4-7 wt% H2O and ~8-10 wt% SO3. Experimental results show that in most cases sulfate melt is stable and in equilibrium with anhydrite and silicate melt at temperatures ≥1000°C, and that it may be present at lower temperatures in alkalic or water rich melts. Furthermore, at temperatures above ~1150-1200°C sulfate melt entirely replaces anhydrite as the stable sulfate phase. EPMA measurements and mass balance calculations provide evidence that sulfate melts are dominated by CaO and SO3, but also contain, in order of abundance, Na2O, K2O, MgO, FeO, Cl and P2O5.
The trace element content of sulfate melts and coexisting silicate glass, measured using laser ablation-inductively coupled plasma-mass spectrometry (LA-ICP-MS) and secondary ion mass spectrometry (SIMS), is also compared to mass balance calculations. Calculated partition coefficients suggest that large 2+ cations (Ba, Sr, Ca) are particularly compatible in sulfate melts, along with F, Cl, light REE and smaller, highly charged cations Mo and W. When elements are arranged in terms of ionic potential (the ratio of nominal charge to ionic radius, Z / r) partition coefficients have peaks at Z/r close to that of Ca and S, suggesting that the incorporation of trace elements into the sulfate melt structure is dictated by the ease with which these elements can replace the major cations. Partition coefficients are lower in experiments with more mafic silicate composition (<10 for all elements in one experiment with trachyandesite at ~1200°C), and this likely related to depolymerization of the silicate melt.
Anhydrite was also analyzed for trace element composition by SIMS, and 7 sets of anhydrite (Anh) -silicate melt (Sil) partition coefficients are reported covering 900-1100°C, and 0.2 - 1GPa. Nernst-type partition coefficients D_i^(Anh-Sil) for +2 and +3 cations are a function of both the partition coefficient for Ca (D_Ca^(Anh-Sil)) and the temperature (T, K), consistent with exchange reactions involving the anhydrite Ca site. For practical purposes, the partition coefficients for +2 and +3 cations can be described by semi-empirical equations of the form
D_(i(+2))=e^((C_1/T+C_2 ) ).(〖CaO〗_Sil )^(-1) (wt%)
D_(i(+3))=e^((C_1/T+C_2 ) ).(〖CaO〗_Sil )^(-2) (wt%)
where T is in units of K and CaOSil is the CaO content of the silicate melt in wt%.
Partition coefficients for +2 and +3 cations also vary systematically with effective ionic radii, and can be described in terms of lattice-strain models. These relate partition coefficients to the elastic strain associated with incorporating a cation of less than optimal ionic radius into a crystal site, in this case, Ca. The Young’s modulus calculated for the Ca site by simple “one-site” fits to the partitioning data is 240 ± 25 kbar. However, the calculated partition coefficients are better fit by a “two site” model, with optimum radii for sites at ~1.1Å and 1.2-1.3Å, suggesting a change to the anhydrite crystal structure at high temperature and pressure.
Using the temperature and compositionally dependent partition coefficient calculated for Sr, it is possible to model the Sr content of anhydrite precipitating at various stages of magmatic evolution. The Sr content of silicate-hosted anhydrite inclusions from the Luhr Hill granite, Yerington, presented in chapter II (4000-5500 ppm) are consistent with anhydrite crystallizing from a melt with ~1000 ppm Sr at 900-1000°C. This suggests that anhydrite was present as a liquidus phase in the Luhr Hill granite and implies magmatic sulfur contents greater than 1200 ppm S.
Finally, experiments described in chapter VI synthesize anhydrite using a molten CaCl2 flux at temperatures <950°C. Anhydrite crystals were grown up to 2 mm using a cooling rate of ~2°C/hr and were generally free of inclusions of salt flux. Powder XRD results showed that they were also free of any secondary crystalline phases. Anhydrite crystals were doped with Sr (up to 3800 ppm) and P (up to 500 ppm) and were homogenous at the precision available by EPMA. With further work it should be possible to use this technique to synthesize anhydrite for use as standards for in-situ trace element and S-isotope measurements.