The diamond that took twenty years to melt

Twenty years ago, physicists at Lawrence Livermore shocked a sliver of diamond to the pressures found deep inside Neptune and Uranus and tried to catch the moment it melted. The number they measured and the number every theory predicted disagreed by roughly 20%, more than 1,000 degrees apart, and neither side could explain why. “No matter what the theorists did,” physicist Marius Millot later put it, “they could not reproduce the experiments.” A separate run of shots at Sandia's Z machine complicated things further, hinting at some intermediate crystal phase between solid and liquid that nobody could actually see (ScienceDaily, on the new paper; LLNL's own writeup).

A paper in Nature Physics, published August 20, 2026, finally settled it — not by refining either side's model, but by finding a way to look instead of infer. At the Omega Laser Facility, Millot's team crushed diamond to three times Earth's core pressure, hotter than the sun's surface, for billionths of a second, and ran x-ray diffraction straight through the melt for the first time: reading the actual atomic structure as the lattice let go, rather than guessing from temperature and brightness alone. There was no hidden intermediate phase. Diamond simply melts into a carbon liquid dense enough that the solid floats on top of it, the same way ice floats on water — and the twenty-year mismatch dissolved with it. Both the old experiments and the old theory had been circling the same real answer; nobody had a picture to check either one against.

The finding comes with two consequences that share nothing but the physics. Bury that liquid-carbon layer deep enough inside an ice giant and it explains the diamond rain long suspected under Neptune and Uranus's clouds, and the odd, off-axis magnetic fields that come with it. Run the same melt curve toward a fusion capsule instead of a planet, and it says something else entirely: a slower initial shock lets a diamond-carbon capsule melt all the way through rather than partway, keeping the fuel more compressible for the same laser energy — a change modeled to triple the yield of an inertial-fusion shot.

Twenty years is a long time for two careful groups of scientists to disagree about how a single, well-studied material behaves under pressure. It didn't end because either side gave in. It ended because someone built an experiment that could finally show its own answer, instead of asking to be believed.