Preliminary decompression experiments of 1902 and 1929 andesites from Mount Pelée (Martinique): implications for “pelean style” degassing.

Recent evidence from textural, petrological and experimental studies has shown that the 1902 and 1929 dome and nuée ardente products, including those from violently explosive 1902 events, were erupted from once H2O-rich andesite magmas (5-6 wt% H2O) that underwent strong open-system degassing upon ascent, and erupted as low-vesicularity, high-cristallinity, H2O-poor lavas. We are currently carrying out decompression experiments in internally-heated pressure vessels at various, mostly low (´ 0.5 kb) pressures and varying decompression rates, in order to simulate syn-eruptive degassing processes such as vesiculation and microlite crystallization. The experimental data are systematically compared with textural and compositional data (microlite size, abundance, shape and composition, glass composition and water content) from representative natural samples covering the whole range of vesicularity and eruptive styles (e.g. 1929 and 1902 domes, 1929 block-and-ash flows, 8 May and 30 August 1902 surges). Preliminary data have been obtained from experiments in the following conditions: two glass compositions representing the andesite whole-rock (62.5°0.5% SiO2) and the rhyolite phenocryst-devoid composition (77.5°1.5% SiO2), respectively; initial water content corresponding to saturation; starting pressures 2.5, 1 and 0.5 kb; decompression rates of 10, 100 and 1200 bar/h; temperature 900°C. Equilibrium experiments at constant 2.5, 1 and 0.5 kb pressures have also been performed for reference. The main observations to date are as follows. All the final experimental products are vesiculated even for low initial water content (< 1wt%). For high P0 and initial water content (2.5 kb and 6 wt% respectively), the final water contents are close to theoretical values at the final pressure (suggesting degassing close to equilibrium) for a low decompression rate of 10 bar/h, but are significantly higher than theoretical values for a higher decompression rate of 100 bar/h. For comparatively low P0 and initial water content (0.5 kb and <1 wt% respectively), final water contents are significantly lower than the theoretical values, which suggests an additional degassing process to vesiculation at high liquid viscosities. A low decompression rate of 10 bar/h does not induce any microlite crystallisation when the H2O content is low (< 2wt%). At 100 bar/h, the nucleation of plagioclase and orthopyroxene is delayed. However, crystallisation upon decompression is favoured by isobar steps of a few days. High-silica phases similar to those observed in the natural products have been to date experimentally reproduced in one experiment with the andesitic composition, low pressure (from 500 to 200 bar) and high decompression rate (1200bar/h)