Statistical Perspective on the Petrological Utility of Polyphase Groundmass Compositions Inferred via Defocused Beam Electron Probe Microanalysis

Polyphase groundmasses (micro-scale minerals with or without glass) are generated from silicate liquids during the cooling of natural lavas often alongside larger minerals formed long before eruption. Many researchers have posited that compositions gleaned from the analysis of groundmasses closely approximate the compositions of the melts they were derived from, and these have been used frequently to model pre-eruptive magma conditions. However, it is difficult to confidently identify and sample these groundmasses once they are formed. Using a sample of lava that exhibits a wide degree of textural variation (ranging from holocrystalline to hypohyaline) we show that compositions of groundmasses sampled using defocused electron beams are significantly different from glass compositions in terms of mean composition and covariance. Despite this, several groundmass compositions qualify as 'in equilibrium' with matrix/rim olivine. When processed using available thermometers, however, modelled equilibrium temperatures are significantly higher than those produced using glass data, on average. Because of this, we prescribe caution in using polyphase groundmass data generated using defocused beam analysis even as a rudimentary approach. Glass and polyphase groundmass compositions may both be routinely sampled via defocused beam analysis on the electron microprobe. However, polyphase groundmass compositions are fundamentally different from glass compositions in multivariate space. Use of polyphase groundmass compositions in petrological modelling produces erroneous results as a consequence. image