In the Ol-H2O experiments, the product changed from serpentine + magnetite to serpentine + brucite + magnetite, accompanied by a Si-drop in the solutions. Serpentinization proceeded uniformly throughout the reaction tube, indicating that the supply of water was not the rate-determining process. In the Opx-H2O experiments, orthopyroxenes were dissolved along the cleavages, and the amount of newly formed serpentine was very small. The silica activity of the solutions in the Opx-H2O experiments was 1-3 orders higher than in the Ol-H2O experiments. In the Ol-Opx-H2O experiments, serpentinization proceeded in both the Ol and Opx zones. In the Opx zone, the extent of serpentinization was constant, whereas in the Ol zone, serpentinization was most extensive along the boundary between the Ol and Opx zones, and it decreased gradually away from the boundary.
Serpentinization in the Ol-Opx-H2O experiments was modeled simply by coupled processes involving silica diffusion and two serpentinization reactions: a silica-consuming reaction after olivine and a silica-releasing reaction after orthopyroxene. The spatial pattern of the extent of serpentinization was controlled by the diffusion coefficient of silica in aqueous solution, DSiO2,aq, and the apparent reaction rate constants k¡äOl in the olivine zone, and k¡äOpx in the orthopyroxene zone. Assuming DSiO2,aq = 2.0 ¡Á 10?4 cm2/s, the observed variation in the extent of serpentinization after a run of 1512 h is best fitted by simulation with k¡äOl = 4.4 ¡Á 10?4 and k¡äOpx = 3.2 ¡Á 10?5 s?1, indicating a reaction rate constant ¡«14 times higher in the Ol zone than in the Opx zone. Our experimental results represent an analogue of serpentinization in natural hydrothermal systems with a high porosity, and we suggest that the spatial variation of serpentine as a function of the distance from a source of silica could be a useful indicator of the relative magnitudes of reaction, mass transport, fluid flow as well as temperature during hydrothermal alteration of oceanic lithosphere.