Abstract

A mixture of antigorite, forsterite, and enstatite was reacted at 2 GPa pressure, with water, to study kinetics of the reaction Mg48Si34O85(OH)62 = 10 Mg2Si2O6 + 14 Mg2SiO4 + 31 H2O.

Reaction progress, F, which can vary between +1 and −1, was measured by comparing areas under X-ray diffraction peaks for run products with corresponding peaks for the starting material. Rates for dehydration and hydration can be regressed with the equation:

 
\[\mathit{r}\frac{F_{R}}{t}\ =\ K_{r}A{^\circ}_{{\theta}}\left[{-}171\left(1\ {-}\ \frac{\mathit{T}_{eq}}{\mathit{T}}\right)\right]^{\mathit{n}}\ mol/cm^{3}_{rock}/s\]

The function FR accounts for the decrease in A𝛉, specific surface area, from A𝛉° at F = 0 to 0 at F = 1:

 
\[F_{R}\ =\ \frac{1}{1\ {-}\ \mathit{p}}\left[1\ {-}\ (1\ {-}\ F)^{1\ {-}\ \mathit{p}}\right]\]

where p, ~0.50 for elongate grains, characterizes grain shape. Regression of the rate equation for dehydration runs can be combined with A𝛉°, measured on the antigorite starting material, to give reaction rate Kr: −9.2(l.2) × 10−15 mol/s/cm2. With that rate, we calculate that well-defined conventional reversal brackets of 5 °C around the equilibrium temperature would require run lengths of 729(99) h, considerably longer than in this or any previous study.

The rate equations can be applied to the question of overstepping of antigorite dehydration below arcs. One modeled geotherm 40 km below a slab surface crosses the antigorite dehydration reaction at about 2 GPa; the slab takes 3 × 105 years to warm one degree. For grain sizes in a serpentinite in the 0.1 to 10 cm range, complete dehydration would take 104–105 years. During that time, the plate would travel no more than a kilometer past the point of first dehydration. If earthquakes associated with dehydration occurred on timescales of 103–104 years, complete dehydration of a volume of plate would require 10–100 separate dehydration events.

You do not currently have access to this article.