Collective proton transfer in ordinary ice: local environments, temperature dependence and deuteration effects

The transfer of multiple protons in hydrogen-bonded networks usually occurs one proton at a time. At sufficiently high temperatures, each proton transfers via thermally activated hopping along its hydrogen bond, thereby moving a charge defect through the network. At low temperatures, quantum-mechanical tunnelling might set in instead, thus avoiding hopping over the energy barriers. In the case of several transferring protons, independent thermal hopping or quantum tunnelling of the individual protons becomes less favourable because of a significant creation of charge defects. In individual molecules or hydrogen-bonded molecular complexes, for instance, double proton transfer is often found to be concerted. Multiple proton transfer that avoids charge defects can occur in cyclic topologies built from several hydrogen bonds that allow for directional chains of proton transfer. This requires perfect proton order within these rings, which imposes handedness and thus chirality, and changes parity upon transfer of all protons. Ordinary ice, which is hexagonal ice I_h, is the most stable form of crystalline ice obtained upon freezing liquid water at ambient pressure and consists of interconnected six rings of oxygen atoms that host six protons each. These hexagonal rings remain proton disordered even down to low temperatures, as heralded by the residual entropy of ice I_h. However, owing to combinatorics, a certain number of these six rings is proton ordered in macroscopic crystals. These chiral hexameric rings might support coherent tunnelling of the hosted protons. Indeed, there is some evidence in the recent literature, both experimental and simulational, that correlated tunnelling of all six protons might be possible in proton-ordered six rings in ice I_h if temperatures are low enough. In this Perspective, the key ideas and previous findings will be reviewed in the light of relevant experiments with a focus on available ab initio path integral simulation work supplemented with additional data provided herein.