Abstract
A series of dehydration experiments in the piston-cylinder apparatus was carried out at 2 GPa and 550-850 degrees C on a natural antigorite sample mixed with 5 wt.% of magnetite. Chemical analyses of experimental products show a progressive decrease of the Mg# in antigorite and dinopyroxene between 550 and 675 degrees C, whereas the Mg# of olivine increases. The observed behavior of Mg# signifies Fe-Mg exchange between coexisting minerals. At higher temperatures, between 700 and 850 degrees C, compositions remain stable for all minerals in experimental assemblages.Thermodynamic parameters of the ferrous antigorite end-member were refined with the use of Holland and Powell (1998) data set and added to the antigorite solid solution. Good agreement between theoretical calculations performed for the studied bulk composition and experimental results confirms extrapolated thermodynamic data for Fe-antigorite. Constrained parameters allowed to calculate phase relationships for various serpentinite compositions. First, we assessed the effect of bulk iron content, from 0 to 10 wt.% FeO, on the stability field of antigorite. The results show significant decrease of the antigorite thermal stability with increasing bulk Fe content. Second, we demonstrated the influence of bulk iron content on dehydration reactions in subduction zones along typical thermal gradients. Dehydration observed in pure MSH (MgO-SiO2-H2O) systems comprised of antigorite appears as a univariant reaction, which happens at 710 degrees C/3.7 GPa and 640 degrees C/6 GPa in "hot" and "cold" subduction, respectively. In contrast, more complex in composition Fe-bearing serpentinites show spread dehydration profiles through divariant reactions from similar to 300 degrees C/0.8 GPa to 700 degrees C/3.6 GPa and from 450 degrees C/4 GPa to 650 degrees C/7.4 GPa for "hot" and "cold" thermal gradients respectively.A comparison between depths of "water-release events" and "earthquake occurrence" in the South Chile slab ("hot" subduction) highlights a clear correlation between each other, suggesting a possible contribution of the dehydration of Fe-bearing serpentinites to the occurrence of seismic events. In contrast, and based on the same comparison applied to the Kermadec slab, our results demonstrate that there is no correlation in depth between serpentinite dehydration and seismicity in "cold" subduction zones.