Metastability of Liquid Water Freezing into Ice VII under Dynamic Compression

The ubiquitous nature and unusual properties of water have motivated many studies on its metastability under temperature- or pressure-induced phase transformations. Here, nanosecond compression by a high-power laser is used to create the nonequilibrium conditions where liquid water persists well into the stable region of ice VII. Through our experiments, as well as a complementary theoretical-computational analysis based on classical nucleation theory, we report that the metastability limit of liquid water under nearly isentropic compression from ambient conditions is at least 8 GPa, higher than the 7 GPa previously reported for lower loading rates.

Figure 1

Phase diagram of water relevant to our experiments. Ordinary water ice, ice Ih (hexagonal, $P{6}_3/\mathrm{mm}c$), lies off the lower left corner, and the ice VII-ice X (bcc, Pn-$3m$) transition lies to the right at $\sim 40$ to 80 GPa [12]. The equilibrium phase boundaries are from Ref. [10], and the principal Hugoniot, isentrope, and isotherm are from Ref. [13].

Figure 2

(a) Target schematic and (b) VISAR image from experiment 29419 showing motion of the upper baseplate-witness interface and lower water-window interface.

Figure 3

Results from experiment 29419 and its simulations. Initial motion of the baseplate front surface (deduced from simulations of the witness side of the target) defines time 0. (a) Interface velocities and corresponding pressures (applicable to all curves). Lagrangian position versus time diagram of the pressure (colorbar) inside the water layer from the simulation using (b) the multiphase $\mathrm{liquid}+\mathrm{ice}$ VII equation of state (EOS) and CNT-based kinetics model and (c) only the liquid EOS (no phase transitions). Lagrangian position 0 and 14 correspond to the baseplate and window interface positions, respectively. The vertical dashed lines in (b) mark the onset and completion of freezing (same as Fig. 5).

Figure 4

(a) Liquid–ice-VII freezing pressure versus compression rate (defined in the text). (b) Water-window interface pressure histories from our experiments, ordered by decreasing compression rate and shifted in time for clarity. In the legend of (a), $S$ and $Q$ denote sapphire and $\alpha$-quartz windows, respectively, and $\mathrm{\Delta }{T}_0$ is the initial temperature increase above the principal isentrope, either known by controlled preheating of the sample (Nissen et al. [26]) or estimated for a possible initial shock that is indistinguishable from the steep ramp (this Letter). In (b), asterisks mark the pressure relaxation interpreted as freezing. The plateau several nanoseconds before freezing in the three right-most profiles using $\alpha$-quartz windows is confirmed, using hydrocode simulations, to result from wave interactions arising from the $\sim 150\text{−}\mu \mathrm{m}$-thick baseplate. The plotted values from Nissen et al. [26] were taken directly from their publication (rates calculated over 5 to 11 GPa).

Figure 5

Calculated observables from the simulation of experiment 29419 including the multiphase water EOS and CNT-based kinetics model [same as Figs. 3 and 3]. (a) Water-window interface pressure [same red curve as Fig. 3], (b) dimensionless driving force and nucleation rate, and (c) critical-cluster size and ice-VII phase fraction (dotted curves) at various locations from the baseplate (front) side to the window (back) side of the water layer.