Begin with the energy barrier, because that is the part of the theory that ends up inside calculations. Classical accounting makes it one wall, and only the small fraction of nuclei arriving with enough energy gets over it [13]. What the UCLA measurements describe instead is a sliding scale with intermediate states a nucleus can occupy on the way up, which corresponding author Jianwei Miao likens to a ladder [13]. A rate computed from a single threshold and a rate computed from a sequence of partial steps are not the same number, and the gap is not a coefficient you carry in your head.
The reason this is hard to wave off is that the new equation contains the old one. Miao says the classical theory is a special case: substitute that case into the gradient nucleation pathways model and you recover classical results exactly [12]. So nothing already calibrated against the century-old equation [1] becomes wrong overnight. The awkward part is that the special case has no instance in this data. Across more than 8,000 nuclei, from under ten atoms to more than a thousand [8], a span of over two orders of magnitude in atom count [16], every nucleus had a crystalline core that grew more disordered out to its boundary [9]. That is zero measured examples of the uniform, sharply bounded nucleus the classical equation assumes [17], against thousands of prior experiments that supported the equation without seeing inside a nucleus atom by atom [2].
One result turns the gradient from a static smear into a pathway: as nuclei got larger, their cores grew more orderly [10]. Order accumulates with size rather than switching on at a critical radius.
Now the scope. To see any of this, the team heated alloy nanoparticles above 1,650 C and dropped them to room temperature in a few hundredths of a second [5], roughly 50,000 degrees per second [15]. That was not a stylistic choice: data acquisition takes far longer than nucleation, so the nuclei had to be frozen in place [6]. The specimens were high- and medium-entropy alloys, metallic elements in roughly equal proportions rather than one dominant element as in steel [4]. Cloud droplets and industrial castings, the systems the release invokes for breadth [14], do not form under those conditions. The gradient may well be general, and the experiment as described does not test it outside equal-parts metal alloys quenched far from equilibrium.
Which leaves a narrow, usable reading of a Nature Materials paper making a large claim [3]: the classical barrier is now known to be a limit rather than the mechanism, and anyone whose model depends on that limit has acquired an assumption they did not previously have to defend.
