Experimental evidence for the existence of a second partially-ordered phase of ice VI

Ice shows extensive polymorphism due to the geometric flexibility of hydrogen bonds and hydrogen order–disorder transitions, which profoundly affect ice’s static and dynamic properties (e.g., molecular rotation immobilization and ferro/antiferroelectric alignment). A central unresolved question is whether a hydrogen-disordered ice phase can transform into more than one hydrogen-ordered phase, challenging the traditionally inferred one-to-one correspondence in the ice phase diagram. Recent work on high-pressure ice VI identified an unknown hydrogen-ordered form (β-XV) in addition to the known ordered phase, ice XV, but lacked decisive evidence to classify it as a distinct crystalline phase. This study addresses that question by providing in-situ experimental evidence for a second hydrogen-ordered phase of ice VI—ice XIX—stabilized at higher pressures. The pressure dependence of hydrogen ordering indicates that volume changes upon ordering thermodynamically control which ordered state forms, implicating pressure as a key variable to tune hydrogen ordering, centrosymmetry, and potential (anti-)ferroelectricity, with implications for completing the phase diagram of ice.

More than 20 crystalline and amorphous ice phases have been reported. Hydrogen ordering transitions are known to significantly alter ice properties. Theoretical studies have suggested near-degenerate hydrogen configurations in ice due to geometric frustration, raising the possibility of multiple ordered states from a single disordered parent. Experimentally, for ice VI, the known hydrogen-ordered counterpart is ice XV; however, recent experiments hinted at an additional ordered form (β-XV), though evidence was insufficient to assign it a new phase designation. Prior work also indicated that pressure could influence ordering by favoring lower-volume configurations, and theoretical suggestions of pressure-induced multiplicity exist for other ice polymorphs. These studies motivated a focused experimental exploration under pressure to clarify whether ice VI exhibits more than one hydrogen-ordered phase.

Dielectric measurements: Ice VI was initially prepared at room temperature. In-situ dielectric properties were measured during cooling and heating between 100–150 K under 0.88–2.2 GPa using a newly developed piston–cylinder pressure cell enabling simultaneous pressure estimation via ruby fluorescence. HCl (10^−2 M) was used as a dopant to accelerate hydrogen ordering; analogous measurements were performed on DCl-doped D2O (10^−2 M) following the same protocol. The temperature was changed at 2 K/h; frequencies ranged from 3 mHz to 2 MHz. After heating runs, samples were annealed at room temperature, recompressed, and remeasured at different pressures. Pressure changes with temperature were within ~0.1 GPa; reported pressures correspond to those near the transition temperatures. Transition temperatures were determined from sudden weakening of dielectric loss peak intensities, assuming a linear temperature dependence of the ice VI dielectric loss peak (per Supplementary Method 2.2). Neutron diffraction: Time-of-flight neutron diffraction was performed at the PLANET beamline (BL11), J-PARC (Ibaraki, Japan). DCl-doped D2O (10^−2 M) samples were prepared via solid–solid transitions (ice III → V → VI) to obtain fine powder. Data were collected on cooling and heating between 80–150 K at 1.6 and 2.2 GPa; temperature was changed at 6 K/h. Pressure and temperature were controlled using a Mito-system; pressure was estimated from the lattice parameter of Pb added as a pressure marker. Diffraction patterns were collected using new samples for each run to confirm reproducibility. Lattice parameters were analyzed using the ice VI oxygen framework model to enable comparison with ice XIX. Structure analysis: Candidate space groups for ice XIX were generated from the group–subgroup relation of ice VI (P4_2/nmc) using the Bilbao crystallographic server (SUBGROUPS). Considering experimental reflection conditions, 18 candidate space groups were tested by Rietveld refinement using partially hydrogen-ordered models obeying the ice rules (hydrogen occupancies and atomic coordinates as fit parameters). Lower-symmetry subgroups of the best-fitting candidates were not pursued given similar fit quality among higher-symmetry models. Best fits were obtained for P4 and Pcc2, both implying partial hydrogen order (~50% site occupancy); Pcc2 implies a polar (pyroelectric) structure along c.

- Dielectric evidence for two distinct hydrogen-ordered phases from ice VI: a lower-pressure ordered phase (ice XV) and a higher-pressure ordered phase (ice XIX). Transitions occurred near 120–130 K with marked weakening of dielectric response due to suppression of molecular reorientation. - The slope of the VI→ordered phase boundary dT/dP changes sign at ~1.5–1.6 GPa. With ΔS<0 for hydrogen ordering, this implies opposite signs of ΔV for the two ordered phases. Ice XV (lower pressure) has ΔV>0, while ice XIX (higher pressure) has ΔV<0, indicating ice XIX has smaller volume than ice VI and than ice XV and is stabilized by the PV term at higher pressure. - A first-order VI↔XIX transition with thermal hysteresis was observed (dielectric data). - Neutron diffraction at 1.6 and 2.2 GPa revealed new Bragg peaks on ordering that cannot be indexed by ice XV’s unit cell (e.g., peaks at d ≈ 2.20 Å and 2.26 Å), but are consistent with a √2×√2×1 expansion relative to ice VI and a primitive lattice for ice XIX. This unambiguously distinguishes ice XIX from ice XV. - Lattice response on ordering to ice XIX: reduced cell parameters relative to ice VI show expansion of a and contraction of c; c/a changes across the transition. The transition temperature at 2.2 GPa is ~7 K higher than at 1.6 GPa, consistent with the positive dT/dP for VI→XIX. - No significant total volume change was resolved by neutron diffraction (likely due to small contraction), but independent work reported volume contraction on VI→XIX at ambient pressure, supporting ΔV<0. - Rietveld refinements favor space groups P4 or Pcc2; both models indicate partial hydrogen order with ~50% occupancy. Pcc2 implies a pyroelectric structure along c. The multiplicity of configurations is large (1964 symmetry-independent configurations in the ice XIX cell).

The results directly address the central question of whether a hydrogen-disordered phase can transform into multiple hydrogen-ordered phases. For ice VI, pressure selects between two distinct hydrogen-ordered states: ice XV at lower pressures and ice XIX at higher pressures, with the latter stabilized by a smaller molar volume. Dielectric and neutron diffraction data corroborate a genuine new crystalline phase (ice XIX) distinct from ice XV, including unique reflection conditions and supercell features. The pressure dependence of transition temperatures and the inferred negative ΔV for VI→XIX confirm thermodynamic control by the PV term. This demonstrates a new dimension of polymorphism in hydrogen-bonded ice networks and suggests that similar multiplicity may exist in other disordered ice phases. The structural models imply partial hydrogen order and possible differences in centrosymmetry relative to ice XV; in particular, a Pcc2 model would be polar, opening prospects for (anti-)ferroelectric behavior. The vast number of possible configurations highlights the complexity of the low-temperature region (<~150 K) of the ice phase diagram and motivates combined experimental and theoretical approaches. Pressure, possibly in combination with electric fields, emerges as a powerful control parameter for selecting among competing ordered states, suggesting rich P–T–E phase behavior and enabling targeted tuning of functional properties in hydrogen-bonded materials.

This work provides experimental evidence for a second hydrogen-partially-ordered phase of ice VI, designated ice XIX, distinct from ice XV. Dielectric measurements across 0.88–2.2 GPa and neutron diffraction at 1.6 and 2.2 GPa establish that pressure induces different hydrogen-ordering pathways, with ice XIX stabilized at higher pressures due to volume contraction upon ordering. Structural analyses favor P4 or Pcc2 symmetry for ice XIX, indicating partial hydrogen order and, in the latter case, a polar structure. These findings reveal pressure-controlled multiplicity of hydrogen-ordered phases in ice, expand the known ice phase diagram, and suggest broader applicability to other hydrogen-bonded systems. Future research should: (i) perform single-crystal neutron diffraction to definitively resolve hydrogen positions and symmetry; (ii) explore the low-temperature, high-pressure region of the phase diagram more extensively; (iii) investigate combined high-pressure and high-electric-field conditions to access additional ordered states and ferro/antiferroelectric phenomena; and (iv) apply advanced theoretical frameworks capable of evaluating the stability of the large number of possible configurations.

- The precise space group could not be uniquely determined; P4 and Pcc2 both provide good fits, and lower-symmetry options cannot be definitively excluded without single-crystal data. - Neutron diffraction did not resolve a significant total volume change on VI→XIX at high pressure, likely due to the small magnitude of contraction; volume change inference relies on thermodynamics and complementary reports. - The dielectric determination of transition temperatures assumes linear temperature dependence of the ice VI loss peak intensity. - Samples were doped (HCl/DCl, 10^−2 M) to accelerate ordering; dopants could influence kinetics and possibly stabilize certain configurations relative to undoped ice. - Sample pressure varied by up to ~0.1 GPa during temperature changes, introducing small uncertainties in the exact P–T conditions. - The enormous configurational space (1964 symmetry-independent configurations) precluded exhaustive theoretical evaluation.