The everlasting hunt for new ice phases

Water ice exists in hugely different environments, artificially or naturally occurring ones across the universe. The phase diagram of crystalline phases of ice is still under construction: a high-pressure phase, ice XIX, has just been reported but its structure remains ambiguous.

The ordering from ice I h to XI, from V to XIII, from XII to XIV needs to be induced through basic or acidic doping, as the ordering temperatures are too low to observe spontaneous ordering in laboratory time scales. However, in the case of ordering of ice VI to XV a more complex kinetics has been observed. The ordering temperature tails from a relatively high 129 K down to 100 K, and the process depends on the applied pressure. Also, the experimentally found (partial) ordering of hydrogen differs from what is predicted by computer simulation 18 .
Back in 2018, Gasser et al. reported on a 'β-ice XV' phase 19 , obtained, like the hydrogen ordered phase ice XV, upon cooling ice VI. At that time, experimental evidence was insufficient to attribute a Roman numeral to this phaseas is the convention for crystalline phases of icebecause its structure remained to be elucidated. Now, in a recent paper, Yamane et al. 9 investigate this highpressure phase and describe it as a hydrogen-ordered phase of ice VI, with a different hydrogen ordering as compared to ice XV, and assign to it a new Roman numeral, XIX. While they can show convincingly that ice XIX is different from ice XV and ice VI, by means of dielectric measurements and in situ neutron powder diffraction, its crystal structure remains ambiguous: several space groups remain possible according to the authors' data on the deuterated in situ high-pressure samples of ice XIX… Ice XIX definitely forms a ffiffi ffi 2 p ffiffi ffi 2 p 1 super-cell with twice the unit cell volume of ice VI (and ice XV). Neither orthorhombic nor tetragonal unit cells can be ruled out convincingly against each other. As we will learn later, it is not even clear whether it is the hydrogen ordering, which is different compared to ice VI and ice XV, although it remains the working hypothesisand the selling argument, or headline, of their manuscript. The dielectric measurements indicate an ordering of hydrogen as compared to ice VI, but whether the topology of this order is different from ice XV, as the title suggests, may be contested.
At the same time, Gasser et al. have investigated the same phase of ice, and published a paper on the structure of ice XIX 10 , again announced as a second (different from ice XV) hydrogenordered polymorph to ice VI. They performed ex situ high resolution neutron powder diffraction on recovered samples, providing better quality data as compared to in situ high-pressure data. The work also sheds more light on the phase boundaries, such as the order-order transition between ice XV and ice XIX. Yet, by a crystallographically different approach (filtering of eligible subgroups by compatibility with the oxygen lattice), an ambiguity between different possible space groups remains. To reduce the 109 possible solutions in five space groups to a smaller set of likely solutions, they referred to the preprinted paper of Yamane et al. and limited the quantitative tests to Yamane's structure models in the three space groups both groups found to be eligible. Again, even a distinction of orthorhombic and tetragonal solutions remains ambiguous. The oxygen topology is found to be the same in ice VI, XV and XIX. The hydrogen order is only partial in both, ice XV and XIX. And in both cases, the ordering is found to be anti-ferroelectric (in contrast to initial suspicions to find a ferroelectric ordering like in ice XI). Two space groups are most likely, tetragonal P4 or orthorhombic Pcc2. It would be easy to distinguish the solutions as Pcc2 should exhibit pyroelectricity and non-polar P4 piezoelectricity, but for this one would need single crystals which are very likely inaccessible.
It is not surprising that the phase, which had been called β-ice XV previously, which is found different from ice XV and therefore called ice XIX, is suspected to show hydrogen ordering, and a different one as compared to ice XV. How H atoms order and whether an ordered structure is ferroelectric or anti-ferroelectric remains difficult to predict, a fact which nourished the suspicion that not only one H-ordering could be possible. When, in the ordering of ice VI to ice XV, calorimetry suggested a second, underlying, process, and that both ordering processes would have entropy changes corresponding to only partial H-ordering, a different hydrogen ordering became the likely candidate. Yet, this expectation biases the scientist's look at experimental results and may limit the possible interpretations of the experiment.
In addition, as for the formation of other hydrogen-ordered phases, certain doping strategies were needed to obtain the new ice phase. And here we eventually run in some danger, as every group has its own recipe to obtain ice XIX, can we really be sure to look at the same phase in all three papers?
The third paper on ice XIX 8 , published in Nature Communications only a few weeks later than the other two, comes from a team, Salzmann et al. 16,17 , which has quite some experience in the hunt for hydrogen-ordered ice phases. The structures of ice XIII, XIV and XV have already been determined as well by Salzmann et al. As in the previous investigations, upon cooling a doped sample at higher pressure, additional Bragg peaks appear which imply an increase of the ice VI unit cell to a ffiffi ffi 2 p ffiffi ffi 2 p 1 supercell. However, the authors consider local distortions of the two individual networks of the 'self-clathrate' structure of ice VI. They consider all possible permutations ofrespectivelytilting, shearing or'squishing' the hexameric clusters of water molecules, the characteristic building unit in the ice VI structure which leads torespectivelythree, two or again three different space groups, all subgroups of the space group symmetry of ice VI and compatible with the super-cell. Only one space group, Pbcn, allows for the observed additional Bragg peaks (and all forms of distortion). The space group allows a good fit of the diffraction datatogether with continued total hydrogen disorder in contradiction to the two previous papers. The lower symmetry with respect to the one of ice VI justifies the assignment of the Roman numeral 'XIX' to this phase, but concerning the hydrogen ordering, ice XIX remains a deep glassy state of ice VI with pressure-induced distortions as already suspected earlier by Rosy-Finsen et al. 20 .
Although some weak hydrogen ordering cannot be excluded, it is not the main structural feature distinguishing it from ice VI (and ice XV). However, slightly different doping strategies have been applied, are we really looking at the same ice phase in all cases? Are the conclusions of Salzmann  Surely, we have not seen the last paper on this new crystalline polymorph of ice yet. There is still additional proof to provide to support the solution found by the paper hereafter. If the last solution offered, the deep glassy state, is the right one, one may wonder whether there is a hydrogen-ordered phase of this distorted phase. Also, might a second hydrogen-ordered phase, as it was claimed by Yamane et al. and Gasser et al. a few weeks ago, exist for any hydrogen-ordered phase?
A remarkable common feature of all three papers is that they are based on neutron powder diffraction (at different facilities -J-PARC and ISISwith different instruments and at different resolutions -HRPD and PEARLwith different experimental approachesex and in situ). I may conclude with this biased observation, the role of neutron diffraction in the investigation of ice structures remains crucial.