Magma differentiation in a T gradient: Thermal migration and Soret effects are not dead!

Most magma differentiation probably occurs within a body having significant temperature gradients. Compositional zoning tied to temperature differences in silicic ignimbrites were attributed to thermogravitational diffusion [1], catalyzing temperature gradient experiments by Walker and co-workers [2]. Experiments showed, however, that compositional zoning in silicic rocks was opposite to Soret effects in a silicate melt [3]. Our experiments/models suggest reassessment of temperature gradient effects. We report on thermal migration experiments using wet andesite (AGV-1) or wet rhyolite (RGM-1). AGV-1 was run in 2 cm capsule with top at 950°C and bottom at 350°C in a piston cylinder apparatus for 66 days at 0.5 GPa. Bulk majortrace element compositions vary with position reflecting changing mineralogy with temperature and form a linear “calc-alkaline trend” on an AFM diagram. Surprisingly, static diffusion-reaction in a temperature gradient produces a granitic bulk composition at the cold end of the experiment. Notably, the all-melt upper third of the experiment that lies in a T gradient does not show the Soret effect of higher SiO2 at higher temperature. Instead SiO2 in the melt slightly decreases down temperature. Using the model Iridium [4] which now includes Soret diffusion, we find that this lack of SiO2 enrichment at higher T reflects the tendency of chemical diffusion driven by the coexisting mineral mush at lower T to overwhelm Soret effects. However, this does not mean that Soret diffusion is not occurring—the melt rich portion of the AGV-1 experiment shows large correlated changes in δFeIRMM and δMgDSM with total offsets of 2.8‰ and 9.9‰. In both systems, the hot end is enriched in the lighter isotope, as expected for Soret diffusion. Using Iridium and changing the Soret or chemical diffusion coefficients by 1% to simulate mass dependent diffusion, the isotopic shift observed in the melt having little concentration gradient can be reproduced. Thus, these isotope systems can be used to identify Soret diffusion effects that are not apparent based on concentration. Measurement of these systems on zoned silicic samples could provide evidence for temperature gradient driven differentiation processes occurring in magmas. We will also report compositional, isotopic and modelling results for a currently running cold seal experiment using RGM-1 in order to better compare to zoned silicic ignimbrite compositions and conditions.