Chemical and isotopic fractionation of wet andesite in a temperature gradient: Experiments and models suggesting a new mechanism of magma differentiation

Piston-cylinder experiments were conducted to investigate the behavior of partially molten wet andesite held within an imposed temperature gradient at 0.5 GPa. In one experiment, homogenous andesite powder (USGS rock standard AGV-1) with 4 wt.% H₂O was sealed in a double capsule assembly for 66 days. The temperature at one end of this charge was held at 950 °C, and the temperature at the other end was kept at 350 °C. During the experiment, thermal migration (i.e., diffusion in a thermal gradient) took place, and the andesite underwent compositional and mineralogical differentiation. The run product can be broadly divided into three portions: (1) the top third, at the hot end, contained 100% melt; (2) the middle-third contained crystalline phases plus progressively less melt; and (3) the bottom third, at the cold end, consisted of a fine-grained, almost entirely crystalline solid of granitic composition. Bulk major- and trace-element compositions change down temperature gradient, reflecting the systematic change in modal mineralogy. These changes mimic differentiation trends produced by fractional crystallization. The change in composition throughout the run product indicates that a fully connected hydrous silicate melt existed throughout the charge, even in the crystalline, cold bottom region. Electron Backscatter Diffraction analysis of the run product indicates that no preferred crystallographic orientation of minerals developed in the run product. However, a significant anisotropy of magnetic susceptibility was observed, suggesting that new crystals of magnetite were elongated in the direction of the thermal gradient. Further, petrographic observation reveals alignment of hornblende parallel to the thermal gradient. Finally, the upper half of the run product shows large systematic variations in Fe-Mg isotopic composition reflecting thermal diffusion, with the hot end systematically enriched in light isotopes. The overall δ⁵⁶Fe_IRMM-14 and δ²⁶Mg_DSM-3 offsets are 2.8‰ and 9.9‰, respectively, much greater than the range of Fe-Mg isotope variation in high-temperature terrestrial samples. In contrast, no obvious chemical differentiation was observed in a similar experiment (of 33 days duration) where the temperature ranged from 550 to 350 °C, indicating the critical role of the melt in causing the differentiation observed in the 950-350 °C experiment. If temperature gradients can be sustained for the multi-million-year time scales implied by geochronology in some plutonic systems, thermal migration could play a heretofore unrecognized role in the development of differentiated plutons. Elemental distributions, dominated by phase equilibria, cannot be used to discriminate thermal migration from conventional magma differentiation processes such as fractional crystallization. However, the observation of Fe-Mg isotopic variations in partially molten portions of the experiment indicates that these isotopic systems could provide a unique fingerprint to this process. This result could also provide a possible explanation for the Fe-Mg isotope variations observed in high-temperature silicate rocks and minerals.