Deep water gives up another secret.

Just as water is essential for life as we know it, so too is it critical to our planet’s inner workings. In PNAS, Pan et al. (1) report results that constrain one of water’s most important properties at deep-earth conditions: its dielectric constant. A host of geologic processes signal water’s important role at depth. The volcanoes of the Pacific Ocean’s “ring of fire” erupt explosively because the magmas carry water from the deep interior. Water influences the still-mysterious, sudden breaking and sliding of rock we sense as earthquakes. Furthermore, many concentrations of valuable metals on which our society depends are created through complex natural reaction processes mediated largely by H2O. Indeed, without deep water, we might not have continents (2), and the plate tectonic engine controlling the outer part of the planet would probably operate quite differently (3). Aqueous fluids in the subsurface are confined in pores and fractures in sediment and rock; as temperature rises with depth, porosity and permeability decline (4), but the solubility of rock components increases dramatically (5). Thus, the rock hosts exert strong control on the available solutes, whereas the properties of H2O govern their concentrations and structure in solution. At depths more than several tens of kilometers (i.e., the lower continental crust and the upper mantle), the chemistry of aqueous fluids remains enigmatic because direct sampling from deep drill holes is limited to less than about 10 km, small deep-fluid inclusions trapped in minerals often have complex histories that can hinder their study, and experimental approaches have been challenging. An alternative to direct sampling or experimental study is thermodynamic modeling of water–rock reactions, for which a key ingredient is the dielectric constant (static relative permittivity) of H2O (eH2O). Seminal work by Helgeson and coworkers provided geoscientists with the tools necessary to model such systems, most significantly via the Helgeson–Kirkham–Flowers (HKF) equation of state for ions (7, 8). In this approach, the stabilizing energy of transferring an ion into H2O (the Gibbs energy of ion solvation) and its pressure and temperature derivatives depend chiefly on eH2O. The dielectric constant quantifies a solvent’s ability to hold ions in solution; solvents with higher e better shield ions such as Na and Cl from interacting to bond with another ion or precipitate. Polar condensed phase water possesses a high dielectric constant of ∼80 times vacuum at ambient conditions. This explains water’s remarkable solvent properties relative to other geologically relevant compounds such as CO2. Previous determinations of eH2O extend to modest pressures (1.2 GPa; 600 °C) and indicate that it declines with increasing depth and rising temperature (9). The limited pressure–temperature range has prevented extending the HKF model to mantle conditions, where H2O participates in important geologic processes. To date, it has not proven possible to measure eH2O in situ at the requisite conditions. However, Pan et al. (1) show that ab initio molecular dynamics offers a solution. They used density functional theory (DFT) to derive eH2O from fluctuations in dipole moment at high density and temperature. Good agreement between simulation pressures and those from accepted equations of state gives confidence in the approach. Achieving a high level of accuracy depends on tracking water molecules over sufficiently long trajectories (times) in the simulation cells. Pan et al. (1) take advantage of the fact that at high density and temperature, diffusion and rotation of water molecules are fast compared with ambient conditions. As a result, the authors could derive eH2O from simulation times of 20 ps, or about 300 times shorter than is required at ambient conditions (10). Pan et al. (1) calculate eH2O to 11 GPa and 2,000 K. They find that it increases with density along isotherms. At low pressure, their results agree with existing compilations (9). This is in contrast to the significantly lower eH2O reported in a previous molecular dynamics study using the simple point charge extended (SPC/E) potential with a constant dipole moment (11). At 1,000 K, an increase in pressure from 1 to 11 GPa causes eH2O to rise from ∼15 to ∼40, which is only about half that at ambient conditions. This indicates that hot, dense fluids of the upper mantle should be better solvents for ions than their shallower crustal equivalents, at equal temperature. At constant pressure, increasing temperature leads to a decline in eH2O, so along such a trajectory increased ion pairing should be expected. The values for eH2O calculated by Pan et al. (1) are used to revise the HKF model and compute equilibrium constants for the dissolution of carbonate minerals (MCO3, where M is Ca, Mg, or Fe) in H2O at 1,000 K to 10 GPa (Fig. 1). The predicted concentrations of Mg and CO3 2− in equilibrium with magnesite (MgCO3) increase from effectively zero (submicromolal levels) to several millimolal in otherwise pure H2O, illustrating the role of increasing eH2O with depth. The results add to a growing body of information about the properties of dense, hot water. It appears that there are two types of local structure in liquid H2O, the low-density water familiar at ambient conditions, and high-density water at high pressure (e.g., 12). DFT calculations reveal that increasing Fig. 1. Carbonate anion in H2O. Ab initio molecular dynamics results on H2O improve estimates of the static dielectric constant (1), which allows extension of thermodynamic models of aqueous fluids to the high pressures and temperatures of the earth’s mantle. The predicted solubilities of metal and carbonate ions are higher than previously recognized. Image courtesy of Q. Wan (University of California, Davis, CA).