What the camera saw
A metal has a melting point, and copper's sits at about 1,085 degrees Celsius. Heat it slowly and the number holds. Heat it in femtoseconds, faster than the atoms can rearrange, and the number becomes a question: how far past its own melting point can a crystal be pushed before the arrangement gives way? At the MeV-UED instrument at SLAC National Accelerator Laboratory, an experiment sent electron pulses through a thin copper film while a laser dumped energy into it, and the diffraction pattern reported where the atoms were at each instant. This is condensed-matter physics done at the pace of the atoms themselves. Melting began at the surface, slightly below the nominal melting point, and then went through the volume quickly and evenly.[1]
The result that carries the paper is a thing that failed to happen. Proposed limits of superheating put a ceiling for copper at roughly 1.25 times the melting temperature, near 1,424 degrees Celsius, and predict that the lattice gives way suddenly once that ceiling is crossed. At absorbed energy densities 2 to 4 times the melting threshold, the diffraction data show no such collapse. The accompanying molecular dynamics run agrees, and it reproduces the measured structural evolution well enough that the authors also read a weak rate of energy transfer from electrons to the lattice out of it.[1]
Where the claim actually rests
An absence is the hardest kind of measurement to hold, because two very different things produce it. The authors' reading is that the pace at which order is lost is set by how quickly atoms can move, so no ceiling of the proposed kind governs the timing. The prosaic alternative deserves its strongest form: at 2 to 4 times the melting threshold the lattice may simply never have been driven far enough past the limit, for long enough, for a collapse to separate itself from the ordinary melting already under way. Distinguishing the two needs the simulation, which is where the physical claim in this paper is carried.[1]
This desk has been here before. A week ago the fermium-255 result reached its readers through an atomic calculation that supplied the last step between the measured signal and the quoted magnetic moment, and the honest description of that work was that the calculation was load-bearing. The copper measurement puts a molecular dynamics run in the same position: the diffraction data are the observation, the simulation converts an absence into a statement about atomic dynamics, and the two are not interchangeable. That the paper is currently posted as an unedited early-access manuscript, with its full method text still to come, makes the point concrete rather than philosophical.[1], [2]
What would settle it?
There is a clean signal to wait for. When the edited version appears in the journal with its full molecular dynamics parameters, the electron-to-lattice energy transfer rate it implies becomes a number that other experiments on copper can be held against, and the reading of the absence stops depending on a run nobody outside the collaboration can inspect. Until then the useful summary is narrow and worth keeping narrow: at 2 to 4 times the melting threshold, a copper lattice heated in femtoseconds crossed a proposed superheating ceiling without a sudden collapse. Whether that is a fact about ceilings or a fact about how fast atoms can move is the part still travelling through a model. Fusion chamber walls, where copper alloys are candidates for absorbing heat, will care about the difference.[1]