Numerical modelling of dynamic fluid-rock reactions in subduction settings

At subduction zones, geophysical and petrological observations suggest that forearc mantle wedges may be serpentinised by fluids released from the devolatilization of subducting slabs [1]. This pervasive serpentinisation of the wedge may be a substantial source of abiotic hydrogen (H₂) and methane (CH₄): gases with the potential to feed extremophile microorganisms in the deepest parts of the continental lithosphere that overlie the wedge. Characterisation of mantle wedge serpentinisation is therefore paramount to constraining the limits within which this deep biosphere can exist. However, the geochemical and geodynamical controls on wedge serpentinisation remain a subject of immense uncertainty. The magnitude of H₂ and CH₄ concentrations and fluxes generated from wedge serpentinisation are therefore poorly constrained at present.

Owing to the inaccessibility of the mantle wedge, constraints on H₂ and CH₄ generation within the mantle wedge must be predicted through geochemical models. In this contribution we present the preliminary results of an ongoing modelling study into mantle wedge serpentinisation. Our approach utilises the Deep Earth Water model [2] to calculate fluid-rock reactions at relayed pressure-temperature conditions in the wedge, which are dictated by geodynamical models of subduction zone thermal structure [3]. The resultant fluids of prior reactions are used as reactant fluids for subsequent reactions at new pressures and temperatures; a chain of individual reactions therefore simulates the whole-scale serpentinisation of a column of mantle rock by slab fluid as the fluid migrates upwards through the wedge. By recording the composition of the overall mantle column at each pressure-temperature step, the introduction of new fluid to the resultant column provides a time element, which we use to track the evolution of bulk mantle mineralogy as subduction progresses.

Our preliminary results suggest that a heavily serpentinised layer forms rapidly at the slab-wedge interface, thereby strongly shielding the overlying mantle from significant alteration. Over more time steps, while bulk mantle density continues to decrease with time and increasing serpentinisation, our model suggests that new fluid does not significantly alter the mineralogical composition of the bulk mantle as observed within the first few time steps, and H₂ and CH₄ concentrations remain invariant throughout the column. However, the rate at which this fluid equilibration is achieved is strongly dependent on the initial conditions applied to the model. Our approach therefore provides a means to test multiple different parameters on H₂ and CH₄ generation at subduction zones, with scope for investigating the impact of variable fluid-rock ratio, initial mantle wedge and slab fluid compositions, and mantle wedge thermal structure.

[1] Vitale Brovarone et al., 2020. Nature Comms. 11(1), 3880.
[2] Sverjensky et al., 2014. Geochim. Cosmochim. Acta 129, 125-145.
[3] Holt and Condit, 2021. Geochem. Geophys. Geosyst. 22(6), e2020GC009476.