An international team has confirmed a phase of water ice that theorists predicted more than three decades ago but nobody had observed cleanly: a hexagonal close-packed (hcp) form of "superionic" ice, at pressures and temperatures matching the deep interiors of Uranus and Neptune.

The result, published in Physical Review Letters by Alexis Forestier and colleagues, is the newest entry in a catalogue of more than 20 crystalline ice phases. Superionic ice is the strangest of them. The heavy oxygen atoms stay locked in a rigid crystal lattice while the light hydrogen nuclei — bare protons — break free and wander through that lattice like a liquid. The material is simultaneously a solid and, for its hydrogen, a fluid, and because the moving protons carry charge, it conducts electricity.

Under such conditions water does not melt the way it does on a kitchen counter. At 219 gigapascals and roughly 2,630 kelvin, the sample stayed solid: the pressure squeezes the oxygen atoms so tightly that the lattice's binding forces outweigh thermal motion, and the heat is only enough to set the protons moving.

To get there, the researchers used laser-heated diamond anvil cells with boron-doped diamond absorbers that soak up infrared light and heat the sample evenly from both sides. The structure was read out by X-ray diffraction at the European Synchrotron Radiation Facility's ID27 beamline, with a beam narrower than one micrometre. Above 200 gigapascals the water sample measured about 12 micrometres across.

The experiment showed that above 200 GPa and around 1,800 K, the previously known face-centred cubic (fcc) superionic phase — ice XVIII — gives way to the hcp arrangement, in which oxygen layers stack ABAB instead of ABCABC. The switch happens through a martensitic transition, a coordinated shear of the whole lattice rather than individual atoms diffusing, the same class of transformation metallurgists use to harden steel. The two structures differ only in stacking order, so a mechanical slip is enough.

The finding matters beyond crystallography. Uranus and Neptune have lopsided, non-dipolar magnetic fields, and the leading explanation has been a dynamo in their superionic ice layers, where models typically assume a single fcc layer. If hcp dominates deeper down, then density, viscosity and electrical conductivity would follow different depth profiles — and magnetic-field simulations may need revising.

Two caveats are worth keeping in view. X-ray diffraction is far more sensitive to oxygen than to hydrogen, so the team observed the hcp oxygen lattice directly and inferred superionicity from an anomalous thermal expansion; the proton motion itself, the conductivity, and any direction-dependence remain unmeasured. And the field is contested: different groups report transition temperatures that differ by hundreds of kelvin, and an earlier shock-compression study saw mixed close-packed signatures.

One oddity in the work: the team suspects it had already produced hcp ice in earlier runs without recognising it, after finding an unassigned diffraction peak above 130 GPa in archived data.

Sources