Earth differs from other terrestrial planets in the Solar System by two striking features:
oceans and active plate tectonics. Although recent evidence points to surface water on Mars (Ojha et al., 2015; Orosei et al., 2018), and the surface of Mars probably hosted liq-uid water in the past (Baker et al., 1991; Grotzinger, 2009), oceans are unique to Earth.
The present-day distribution of water, or more general H2O, on Earth between the surface reservoirs of atmosphere and hydrosphere and reservoirs in Earth’s crust and mantle reflects the time-integrated results of planetary accretion and differentiation. Today, Earth’s mantle differentiates into basaltic crust and depleted peridotite at mid-ocean ridges. Plate tectonics transport oceanic crust away from spreading centers towards subduction zones where cold and dense oceanic lithosphere sinks into the mantle. On the way, basaltic oceanic crust and the underlying peridotite react with ocean water through hydrothermal activity and take up H2O in the form of hydrous minerals such as amphibole, chlorite, and serpentine (e. g.
Humphris and Thompson, 1978; Mével, 2003; Staudigel, 2003; Bach et al., 2004). Most of these hydrous minerals, however, break down during subduction and release H2O in dehy-dration reactions. The released aqueous fluids migrate upward into the mantle wedge above the subducting plate where they can trigger melting and give rise to arc volcanism (Pawley and Holloway, 1993; Schmidt and Poli, 1998; Poli and Schmidt, 2002; Eiler, 2003). Aque-ous fluids released by dehydration reactions may also react with the overlying peridotite and form serpentine minerals (e. g. Evans, 1977; Hyndman and Peacock, 2003; Hirth and Guillot, 2013). In cold subduction zones, hydrous minerals hosted by different lithologies of subducted slabs may carry H2O down to depths in excess of 200 km (Poli and Schmidt,
Figure 1.1:Conceptual drawing of deep H2O cycling and dispersal of basaltic crust in Earth’s man-tle illustrating potential processes (1–9) and seismic observables (A–D); see text for references. 1) Hydrated oceanic lithosphere (basaltic crust in red, lithospheric mantle in dark blue) sinks into the upper mantle at subduction zones. 2) Release of H2O from the slab hydrates the transition zone.
3) Melting atop the 410-km seismic discontinuity gives rise to deep seated magmatism. 4) Dia-monds form in the transition zone and enclose transition zone minerals and fluids. 5) Deformation aligns anisotropic minerals. 6) Mantle convection disperses subducted oceanic crust in the lower mantle.7) Subducted oceanic crust accumulates to form geochemical heterogeneities. 8) Plumes probe geochemical heterogeneities. 9) Silica-rich material exsolves from the outer core and gets dispersed in the lower mantle.A) Seismic tomography images three-dimensional variations in seis-mic velocities. B) Deformed mantle rocks generate seismic anisotropy. C) Seismic discontinuities reflect seismic waves. D) Heterogeneities scatter seismic waves in the lower mantle.
1995; Ono, 1998; Schmidt and Poli, 1998). A chain linking the stability fields of hydrous minerals may allow to transport H2O into deeper parts of the mantle, at least along cold subduction paths (Poli and Schmidt, 2002; Ohtani et al., 2004; Komabayashi, 2006; Nishi et al., 2014).
The transport of H2O from Earth’s surface into Earth’s deep interior via subduction pro-cesses highlights the connection between the evolution of terrestrial H2O reservoirs and plate tectonics. The operation of plate tectonics in a similar way as today, i. e. including cold geotherms, can be traced back to the Neoproterozoic (Stern, 2008) implying the po-tential to transport H2O into the mantle for about 1 Ga. The tectonic regime during the Archean remains debated (Condie and Pease, 2008; van Hunen and Moyen, 2012). Petro-logical evidence can be found in support of both subduction (Foley et al., 2002; Foley, 2008) and delamination-dominated processes (Foley et al., 2003; Johnson et al., 2014; Johnson et al., 2017) that led to the formation of early continental crust. The production of typical
1.1 The Link between Water and Tectonics on Earth
Archean tonalite-trondhjemite-granodiorite (TTG) rocks, however, requires partial melting of hydrated basaltic rocks (Rapp and Watson, 1995; Foley et al., 2002). Similar to mod-ern times, hydrated basaltic rocks might have been the starting point for early cycling of H2O into Earth’s mantle since the Archean. Indeed, the low present-day volume of basaltic crust on Earth indicates substantial recycling of earlier mafic crust into the mantle in the geological past (Anderson and Bass, 1986; Herzberg and Rudnick, 2012), a hypothesis that could also explain geochemical observations (Hofmann and White, 1982; Christensen and Hofmann, 1994; Hofmann, 1997) and reconcile the Mg/Si ratio of the silicate fraction of Earth with those of potential meteoritic building blocks (Anderson and Bass, 1986; Hart and Zindler, 1986; McDonough and Sun, 1995). Two questions emerge from this brief out-line linking tectonics and deep H2O cycling through Earth’s history: What happened to H2O that might have been transported into the deep mantle in the geological past, and what happened to the crustal parts of slabs that potentially carried water into the mantle?
The structure and composition of the mantle today should hold answers to these ques-tions as they reflect the dynamic interacques-tions and feedbacks that arise from the injection of hydrated oceanic lithosphere into the mantle. H2O, for example, affects many critical geodynamic parameters of mantle rocks such as the viscosity of olivine (Karato et al., 1986;
Hirth and Kohlstedt, 1996; Mei and Kohlstedt, 2000) as well as temperatures and degrees of partial melting (Hirth and Kohlstedt, 1996; Asimow and Langmuir, 2003; Asimow et al., 2004; Hirschmann et al., 2006). Subducted oceanic crust is predicted to be denser than am-bient pyrolitic mantle throughout most of the mantle (Irifune and Ringwood, 1987; Kesson et al., 1994; Hirose et al., 1999; Ricolleau et al., 2010) and may therefore sink down to the core-mantle boundary and drive convective motions. Figure 1.1 depicts schematically the overall structure of Earth’s mantle together with processes related to deep cycling and storage of H2O and the fate of oceanic crust in the mantle.
1.1.1 The Transition Zone in Earth’s Mantle
The transition zone, confined between major seismic discontinuities at 410 km and 660 km depths (Figs. 1.1 and 1.2a), appears to play a pivotal role in deep recycling of H2O in Earth’s mantle (Bercovici and Karato, 2003; Ohtani et al., 2004). The seismic discontinuity at 410 km depth has been attributed to the phase transformation of olivine,α-(Mg,Fe)2SiO4, to wadsleyite, β-(Mg,Fe)2SiO4, (Bina and Wood, 1987; Agee, 1998; Frost, 2008), which transforms to ringwoodite, γ-(Mg,Fe)2SiO4, at higher pressures (Akaogi et al., 1989; Kat-sura and Ito, 1989; Frost, 2008). The dissociation of ringwoodite into ferropericlase and bridgmanite, in turn, gives rise to the seismic discontinuity at 660 km (Ito and Takahashi, 1989; Shim et al., 2001; Frost, 2008; Ishii et al., 2018). Alternative explanations for the seismic properties of the transition zone, including the mentioned seismic discontinuities, involve the enrichment of basaltic or eclogitic material within the transition zone (Ander-son, 1979; Bass and Ander(Ander-son, 1984; Anderson and Bass, 1986). Observations by seismic tomography suggest that subducting slabs interact with the transition zone in many ways in-cluding the stagnation of slabs within or just below the transition zone (Zhao, 2004; Fukao et al., 2009; Fukao and Obayashi, 2013).
A transition zone rock of pyrolitic composition (Ringwood, 1991; Ita and Stixrude, 1992) would be composed of high-pressure polymorphs of olivine by up to 60 vol-% (Fig. 1.2b;
Ringwood, 1991; Ita and Stixrude, 1992; Stixrude and Lithgow-Bertelloni, 2011). Both wadsleyite and ringwoodite are nominally anhydrous minerals. High-pressure experiments have shown, however, that both phases can incorporate several percent H2O by weight into
Figure 1.2: a) Variation of density ρ, S wave velocity vS, and P wave velocity vP with depth according to PREM (Dziewonski and Anderson, 1981). Phase assemblages in rocks of pyrolitic (b) and basaltic (MORB) (c) compositions as a function of depth; modified after Irifune and Isshiki (1998), Frost (2008), and Irifune et al. (2010) (b) and Perrillat et al. (2006) and Ricolleau et al.
(2010) (c). Note the high volume fraction of wadsleyite in the shallow transition zone (b) and the phase transition from rutile-structured stishovite to CaCl2-type SiO2in the lower mantle (c).
their crystal structures (Inoue et al., 1995; Kohlstedt et al., 1996; Kudoh et al., 2000). The combination of a pyrolitic, or peridotitic, mineral assemblage with the high solubility of H2O in wadsleyite and ringwoodite turns the transition zone into a potential reservoir for H2O in Earth’s mantle (Smyth and Jacobsen, 2006). A ringwoodite inclusion in diamond with an estimated H2O content of∼1.5 wt-% H2O dissolved in the ringwoodite grain provides direct evidence for the viability of this hypothesis (Pearson et al., 2014). Similarly, the presence of aqueous fluids in the transition zone has been inferred from ice-VII inclusions in diamond (Tschauner et al., 2018).
Diamond inclusions provide localized samples that cannot constrain the extent of hy-dration in the transition zone on a global scale. A hydrous transition zone, however, would have far-reaching consequences for the evolution of the mantle. When rising out of the wadsleyite stability field, a hydrated transition zone rock might expel some of the stored H2O due to the drop in H2O solubility from wadsleyite to olivine (Bolfan-Casanova, 2005;
Inoue et al., 2010; Litasov et al., 2011). The resulting dehydration melting would affect the distribution of geochemical key elements in Earth’s mantle (Bercovici and Karato, 2003;
Karato et al., 2006; Karato, 2011) and may give rise to an extended melt layer above the 410-km seismic discontinuity (Revenaugh and Sipkin, 1994; Tauzin et al., 2013; Freitas et al., 2017). Detecting and quantifying the extent of hydration in the shallow transition zone and in the vicinity of the 410-km seismic discontinuity therefore stands out as a key challenge to understand the evolution of global H2O cycles and reservoirs.