Evidences and consequences of slow hydrogen diffusion in olivine

In the most abundant upper mantle phase, olivine, the presence of hydrogen significantly modifies the timescale of chemical diffusion, plastic deformation, electrical conductivity and the attenuation of seismic waves. Early experiments showed that hydrogen is the fastest species able to diffuse through the olivine lattice. We have found, however, experimental and natural evidence suggesting that hydrogen diffusion can also be orders of magnitude slower.

In olivine there are four different hydrogen substitution mechanisms, associated with Mg vacancies, Si vacancies, trivalent cations and titanium substitution, hereafter referred to as H[Mg], H[Si], H[triv] and H[Ti] respectively. We experimentally investigated the dehydroxylation of synthetic forsterite with two contrasting hydrous defect populations: (1) dominated by H[Si], and H[Ti] with subsidiary H[Mg] and H[triv]; and (2) H[Si] exclusively. The loss rates of H[Mg] and H[triv] are in agreement with previous measurements of bulk hydrogen diffusion in forsterite, but the decrease in H[Ti] and H[Si] are ~1.5 and ~ 3 orders of magnitude slower, respectively. The activation energy and pre-exponential terms derived in these experiments are in agreement with the empirical correlation recently proposed based on the Meyer-Nedel compensation law (Jones 2014, G3, 15, 2616-2631).

Natural observations attest further to slow hydrogen diffusivity in olivine dominated by H[Ti] and H[Si]. Metamorphic olivines formed after dehydration reactions in the Alpine orogeny preserve their original water contents despite long times of exhumation (2-3 Ma). Closure temperature calculations suggest that using previous fast diffusion rates, these olivines should reequilibrate down to 400°C. Only slow hydrogen diffusion coefficients such as those corresponding to H[Si] are able to explain the observed preservation of water content at the peak metamorphic temperature (700-800°C).

These findings have implications for estimating the ascent rate of xenoliths, which are more consistent with other independent constrains, as recently noted (Hilchie et al. 2014, Lithos, 202-203, 429-441). They are also required to assess the time necessary to equilibrate experimental charges, since for the different hydrous defects this time would vary by orders of magnitudes.