- Semi-volatile trace metals (e.g., Cd, Sn, Pb, Zn, Cu, Mo) have been analyzed by using laser ablation ICP-MS in a number of silicate glasses (GSE-1G, GSD-1G, NIST 610, NIST 612, BCR-2G, BHVO-2G). Our work investigates and highlights sources of uncertainty in the analysis of semi-volatile metals using LA-ICP-MS. We identify within- cell transport fractionation as a primary source of uncertainty in these analyses. We found no evidence for significant fractionation of elements on the basis of volatility over typical 45 second ablation intervals. However, systematic fractionation of up to 20% was found for a number of siderophile and chalcophile trace elements as well as volatile lithophile elements when comparing analyses from different locations with a simple single-volume ablation chamber. Some fractionation was also observed for analyses in GSE-1G that intersected fractures or cracks in the glass, although overall GSE-1G appears to be homogeneous with respect to semi-volatile element distributions at the ~10 % level. Calibration using GSE-1G produces more accurate results on basaltic glass compositions than NIST 610. This work thus confirms the suitability of GSE-1G as a calibration standard for analysis of semi-volatile metals in mafic composition glasses.
This thesis also reports major and trace element data for glass, olivine, and olivine-hosted melt inclusions from the 1959 eruption of Kilauea Iki, Hawaii. Major element compositions of glasses match the results of earlier studies and suggest that dominant fractionation of olivine + Cr-spinel and secondary mixing between two magma batches control compositional variations. Melt inclusions trap melts with high sulfur concentrations (~0.1-0.15 wt.%) suggestive of a melt at or near sulfide saturation, while matrix glass is highly degassed. Lithophile trace element variations more clearly illustrate the effects of mixing at a late stage of magmatic development, and require magmas with two distinct mantle source regions and/or different degrees of partial melt. A number of non-traditional volatile trace elements were also analyzed including those with a range of volatility and geochemical affinity. Of these, most (e.g., Sn, Mo, Pb) display typical incompatible behavior while others appear to be compatible in known phases (primarily olivine: Zn, Co). Copper concentrations cannot be explained by removal of major phenocryst phases. Scatter in Cu concentrations could be achieved by either variable volatile mobility at depth or removal with a Cu-sulfide phase (either fractionated or residual). Previous studies of rare sulfides present in the 1959 eruption and more abundant sulfides crystallized in the lava lake suggest high Cu (>40 wt.%) in the sulfides. In this case even removal of a small amount of sulfide (<<1% by volume) could strongly deplete the melt of Cu while having little affect on other metals since their presence was not observed in the sulfides. There is no evidence for loss of volatile or semi-volatile trace metals during subaerial degassing.