Quantum versus classical protons in pure and salty ice under pressure

Yael Bronstein, Philippe Depondt, Livia E. Bove, Richard Gaal, Antonino Marco Saitta, Fabio Finocchi
2016 Physical review B  
It is generally accepted that nuclear quantum effects (NQEs) trigger the transition to the nonmolecular form of ice under increasing pressure. This picture is challenged in salty ice, where Raman scattering measurements up to 130 GPa of molecular ice VII containing NaCl or LiCl impurities show that the transition pressure to the symmetric phase ice X is shifted up by about 30 GPa, even at small salt concentrations. We address the question of how the inclusion of salt induces the drastic
more » ... n of NQEs by selectively including NQEs in ab initio calculations of ice in the presence of distinct ionic impurities. We quantitatively show that this is mainly a consequence of the electric field generated by the ions. We propose a simple model that is able to capture the essence of this phenomenon, generalizing this picture to other charged defects and for any concentration. This result is potentially generalizable to most "dirty" ices in which the electric field due to the doping is much more significant than local lattice distortions. Ice under extreme conditions is present in many planets, both within our solar system [1] and beyond [2] . These ices are usually "dirty": Planetary real ices unavoidably contain impurities such as salt. In pure ice, nuclear quantum effects (NQEs) play a crucial role by drastically decreasing the VII-X phase transition pressure [3, 4] . However, recent experiments question the role of NQEs in LiCl-doped ice [5] . Whether this is specific to LiCl ice and the mechanism by which the quantum behavior of the protons can be hindered is still an issue. At high pressure, the two molecular phases of ice, proton disordered ice VII and proton ordered ice VIII [6,7], transform to ice X, the only known atomic phase of ice. Below the transition, the oxygens form a body-centered-cubic structure, and the hydrogen bonds are characterized by a prototypical double-well proton transfer potential. As the atoms get closer under the effect of increasing pressure, the intrinsic quantum nature of the protons produces more evident effects and favors the onset of quantum tunneling. Upon further reducing the distance between oxygens, the proton potential degenerates into a single-well potential [8] , giving rise to a symmetric hydrogen bond, where the hydrogen is located midway between two neighboring O atoms. By including NQEs, the transition pressure P t is drastically reduced from about 90 GPa as classically predicted [9] to approximately 60 GPa [10] [11] [12] [13] [14] [15] [16] , that is, the pressure for which the zero-point energy equals the barrier height [3, 4] . However, the effect of a perturbation on a prototype of structural quantum effects in crystals is not a priori trivial from a fundamental point of view. Recent studies [5] on LiCl-water solutions at different salt to water ratios pointed out that the properties of ice change drastically when LiCl salt is homogeneously included into ice VII and that the transition pressure to the phase X strongly depends on the presence of ionic impurities. Here, we report high resolution Raman scattering experiments on NaCl ice up to Mbar pressures. These challenging conditions have been achieved by the use of a diamond anvil cell equipped with extra low-fluorescence synthetic diamonds and by fast quenching of a 5 μm droplet of 2% NaCl water solution in order to produce ice VII-doped ice following the procedure described in the Supplemental Material [5, 17, 18] . The upper panel of Fig. 1 shows the 200-1000 cm −1 region of the Raman spectra recorded at different pressures up to 130 GPa. The oxygen-oxygen T 2g vibrational mode, which is indicative of the cupritelike structure of phase X, clearly appears at 87 ± 5 GPa, similarly to what is observed in LiCl ice [5] . Thus, the presence of small quantities of salt impurities (here, 1 NaCl for 53 H 2 O), which is likely in natural ices, shifts P t by about 30 GPa, roughly the same pressure shift as observed when quantum effects are neglected in pure ice [3, 4] . This observation confirms that the measured shift of P t in LiCl ice is actually a general effect of ionic impurities on the ice lattice. We address this issue by comparing simulations that do or do not include NQEs and quantify their impact on the transition. The first kinds of calculations are done via quantum thermal bath ab initio molecular dynamics (QTB-AIMD) [19, 20] ; this semiclassical approximation quantitatively describes the VII-X transition in pure ice [4] . The second kinds of calculations are standard ab initio molecular dynamics (AIMD). In both cases, the atomic forces are computed within the density functional theory (DFT), via the generalized gradient approximation [21] , as implemented in the QUANTUM ESPRESSO package [22] . Simulations including NaCl or LiCl impurities were run over a time length of about 25 ps with a 0.484 fs integration time step. Our simulation cell contains 53 water molecules and one LiCl or NaCl pair, corresponding to a concentration of 2% mol. The Li + ion is small enough to occupy an interstitial site within the oxygen lattice, while Cl − is substituted for a water molecule [17] . Two configurations are considered in our calculations for the larger Na + cation: an interstitial site (I), as for Li + , and a substitutional site (S), where Na + replaces a water molecule, as does Cl − (the method and initial configurations are described in the Supplemental Material). The incorporation of salt has non-negligible effects on the hydrogen bonds, as shown by the OH pair correlation function (PCF) (see the figure and related description in the Supplemental Material). In particular, the first peak of the 2469-9950/2016/93(2)/024104 (5) 024104-1
doi:10.1103/physrevb.93.024104 fatcat:x4dsg56gmnhabixrh3tvc5dhmm