Influence of water on global mantle dynamics

Even though surface water is essential for Earth's habitability, the estimates of total amount of water (at the surface and in the deep interior) throughout Earth's evolution vary from 5-15 ocean masses (OM) based on magma ocean solidification models [Hamano et al., 2013] to 1.2-3.3 OM based on petrological studies [Hirschmann, 2006]. Previous numerical models of coupled surface-mantle system have estimated a lower bound of 9-12 OM [Nakagawa et al., 2018]. Experiments have shown that water lowers the melting temperature, density and viscosity of rocks and it is also required for the generation of felsic magmas. In this work, we use global convection models [Tackley, 2008] spanning the age of the Earth to elucidate the effect of water on mantle dynamics in terms of planetary cooling, surface mobility and production of continental crust.

Our models self-consistently generate oceanic and continental (Archean TTGs) crust while considering both plutonic and volcanic magmatism and incorporate a composite rheology for the upper mantle. Pressure-, temperature-, and composition-dependent water solubility maps calculated with Perple_X [Connolly, 2009] control the ingassing and outgassing of water between the mantle and the surface [Jain et al., 2022]. Irrespective of the initial water content used, our models exhibit mobile-lid regime (high surface mobility with subduction) throughout the 4.5 Gyr with episodes of short-lived plutonic-squishy-lid regime (low surface mobility with delamination or dripping) in the Hadean. These models are also consistent with the cooling history of the Earth inferred from petrological observations [Herzberg et al., 2010]. A strong positive correlation is observed between continental crust production and the total amount of water available, with the former's cumulative mass increasing by roughly three times when water in the planetary system is raised from 1 OM to 10 OM.

Models that consider a reduction in the density of crustal and mantle materials in the presence of water exhibit mobile-lid regime for the initial 200 Myr. Afterwards, the mobility stays low as the hydrated oceanic crust is less dense and does not subduct. It thickens over time and eventually collapses as global resurfacing events. Mantle stays comparatively warm and a much lower amount of continental crust is produced. This motivated us to make the following improvements to achieve more realistic models. First, mantle minerals only in the top 5 km of the computational domain (as opposed to 10 km considered previously) are ingassed with water. Second, instead of fully saturating the rocks based on their solubility maps, they are partially saturated to control the input of surface water into the lithosphere. Third, different partition coefficients for water are considered: 0.01 for pyrolite to basalt melting and 0.25 for basalt to TTG melting. These changes help in increasing the surface mobility, cooling down the planet and producing more continental crust. These trends are further amplified in models that additionally consider a viscosity reduction of mantle materials in the presence of water.