Effect of water on the magnesite–iron interaction, with implications for the fate of carbonates in the deep mantle

Highlights

• We studied the effect of hydrous fluid on the carbonate-iron reaction at 6–16 GPa

• We observed formation of magnesiowüstite, graphite and carbide

• In peridotite, pyroxenes react with magnesiowüstite resulting in olivine enrichment.

• MgCO₃ in hydrous system will be reduced during subduction to the transition zone.

• MgCO₃ can survive only at the coldest subduction P-T path or in anhydrous systems.

Abstract

The subduction of carbonates beyond 250–300 km, where redox conditions favour the presence of metallic iron (Fe), will result in redox reactions with the Fe dispersed in the silicate rocks. Here, we studied the effect of water on the carbonate–Fe interaction in the hydromagnesite–Fe system at 6, 8 and 16 GPa and the peridotite–CO₂–H₂O–Fe system at 8 GPa, using a multianvil apparatus. In all of the studied samples, we observed the formation of magnesiowüstite, graphite and carbide. Additionally, in the peridotite–CO₂–H₂O–Fe system, magnesiowüstite reacted with pyroxenes, resulting in olivine enrichment.

Kinetic calculations performed at 8 GPa showed that, at the pressure–temperature (P–T) parameters of the ‘hot’, ‘medium’ and ‘cold’ subduction, about 40, 12 and 4 vol% of carbonates, respectively, would be reduced in the hydrous system within 1 Myr, assuming direct contact with Fe. Based on the present results, it is suggested that carbonates will largely be consumed during the characteristic subduction time to the mantle transition zone by reaction with the reduced mantle in the presence of hydrous fluid.

Introduction

Investigations of mantle xenoliths, carbonatite and kimberlite and numerous experimental studies have revealed an important role of carbonates in the formation of the Earth's mantle heterogeneity and source regions for sublithospheric magmatism (Dasgupta and Hirschmann, 2010; Litasov, 2011; Yaxley et al., 1991). Carbonates drastically reduce the melting temperature of mantle rocks, peridotites and eclogites (Dasgupta and Hirschmann, 2010; Litasov, 2011), which leads to the formation of kimberlites and related alkaline rocks (Shatskiy et al., 2017). Due to their low density, viscosity and high reactivity, alkaline magmas affect the geochemistry of the surrounding mantle, and play a significant role in diamond formation (Stachel and Harris, 2008; Yaxley et al., 1991).

The subduction of oceanic slabs delivers a large amount of carbonates, predominantly calcite (Cal), dolomite (Dol) and magnesite (Mgs), into the mantle (Dasgupta and Hirschmann, 2010; Staudigel, 2014). Roughly 20–80% of the initial amount of carbonates may survive decarbonation beneath island arcs, and be further transported, beyond ~150 km depth, to the deep mantle (Kerrick and Connolly, 2001a, Kerrick and Connolly, 2001b). Indeed, findings of syngeneic inclusions of Ca- and Mg-carbonates in super-deep diamonds (Brenker et al., 2007; Bulanova et al., 2010) and carbonate-bearing marbles, gneisses, eclogites and ultramafic rocks from ultrahigh pressure metamorphic complexes (Shatsky et al., 1995), confirm the existence of carbonates in the upper mantle, transition zone and even in the lower mantle.

Cal and Dol are predominant among the carbonates of subduction zones at near-surface conditions (Plank and Langmuir, 1998; Staudigel, 2014). However, at high pressure, Cal reacts with silicates, such as garnet (Grt), enstatite or bridgmanite, forming Dol or Mgs (Brey et al., 1983; Yaxley and Brey, 2004). After the decomposition of Dol at about 4.5 GPa and 1000 °C (Sato and Katsura, 2001), Mgs remains as a major oxidised carbon phase in the peridotites and eclogites. However, although it has been experimentally confirmed that polymorphs of Mg‑carbonate are stable at extremely high temperatures and pressures (Fiquet et al., 2002), their presence in mantle assemblages mostly depends on the redox conditions.

Oxygen thermobarometry measurements on peridotites from the cratonic lithosphere have revealed that mantle rocks become progressively more reduced with depth due to the effect of pressure on the Fe³⁺/Fe²⁺ equilibrium (Frost and McCammon, 2008; Woodland and Koch, 2003). Stabilisation of the Fe³⁺-bearing end-members in the solid solutions of Grt, pyroxene and bridgmanite leads to the disproportionation of Fe²⁺ (Fe²⁺ → Fe³⁺ + Fe⁰), with the release of metallic Fe. Thermodynamic calculations and experiments have suggested that the mantle becomes metal saturated at depths exceeding 250–300 km (Ballhaus, 1995; Frost et al., 2004; Rohrbach et al., 2007). Studies on diamond inclusions have confirmed the presence of metallic Fe in the mantle. Inclusions of native Fe and Fe-Ni alloy have been detected in diamonds, together with carbide, wüstite (Wus), magnetite and sulphides, in equilibrium with typical upper mantle peridotite mineral assemblages, including olivine (Ol), Grt, and chromite (Sobolev et al., 1981; Stachel et al., 1998), and with Ca-perovskite in diamonds with a lower mantle origin (Kaminsky and Wirth, 2011; Smith et al., 2016).

Consequently, significant chemical heterogeneity in the redox conditions exists in the mantle, and the oxygen fugacity varies from approximately C–CO₂ buffer (CCO) in the carbonate–bearing subducting slabs to the Fe–FeO buffer (IW) in the metal-saturated surrounding mantle. As a result of the existing redox gradient, redox reactions occur, and diamonds, or other reduced carbon species, can be formed directly by the reduction of carbonates or carbonatite melts (Palyanov et al., 2013; Rohrbach and Schmidt, 2011; Stachel and Harris, 2008). The mechanism and kinetics of these processes are extremely important for understanding carbon behaviour and its distribution in the deep Earth.

Carbonate–Fe interaction represents a simplified model of the processes that occur between the Fe-saturated mantle and the carbonate-bearing subducting slabs. Most of the previous experimental and theoretical studies have examined simplified carbonate–Fe systems, including MgCO₃–Fe (Martirosyan et al., 2015a, Martirosyan et al., 2018; Zhu et al., 2018), CaCO₃–Fe (Martirosyan et al., 2015b, Martirosyan et al., 2016) and (Ca,Mg)CO₃–Fe (Dorfman et al., 2018; Palyanov et al., 2013). These works revealed the occurrence of redox reactions, with the formation of graphite (Gr)/diamond, Fe₃C/Fe₇C₃ carbide and magnesiowüstite (Mws) or calcium Wus, depending on the initial composition of the carbonate. It is evident that, in order to further understand the carbonate–Fe reaction, an experimental study of the natural-like systems is required. This would include investigations on silicate- and H₂O-containing systems.

Water is the dominant volatile component in the subduction zone (Peacock, 1990). The amount of water that survives dehydration at shallow depths is poorly constrained. Thermodynamic calculations have shown that the slab may retain from 7% to 35% of initial H₂O, concentrated in basaltic crust, gabbro and peridotites, to 200–300 km depth (Schmidt and Poli, 1998; van Keken et al., 2011). Furthermore, the finding of fluid inclusions containing H₂O in diamonds indicates the existence of hydrous fluid in the mantle (e.g.Izraeli et al., 2001 ; Klein-BenDavid et al., 2004 ; Navon et al., 1988). H₂O in such fluids is commonly associated with C-bearing compounds, such as CO₂ and carbonates, revealing that the C and H cycles are intimately related (Izraeli et al., 2001; Klein-BenDavid et al., 2004; Navon et al., 1988).

High-pressure peridotites with hydrous phases, phlogopite and amphibole, together with carbonates, discovered in ultrahigh-pressure terranes, provide another link between the carbon and water cycles in the deep Earth (Naemura et al., 2009; van Roermund et al., 2002; Zhang et al., 2007). The peridotite–COH system has been experimentally studied under oxidised conditions up to 32 GPa, with the main focus being on the melting relations (Litasov, 2011; Tumiati et al., 2012). However, the existing data for the systems with reduced COH fluid under the conditions relevant to the IW buffer are limited (Litasov et al., 2014; Taylor and Green, 1988).

Despite the fact that reactions in the MgCO₃–Fe and CaCO₃–Fe systems have been carefully studied, only a few works have been published that model the process in systems containing H₂O. (Mukhina et al., 2017; Sonin et al., 2014) or in the peridotite–COH system (Rohrbach and Schmidt, 2011). In this study we explored the effect of water on the carbonate–Fe redox interaction in the hydromagnesite (hMgs)–Fe and peridotite–CO₂–H₂O–Fe systems under upper mantle and transition zone pressure–temperature conditions.