Chromium in steel protects against corrosion. But under neutron radiation — inside a nuclear reactor — chromium migrates to or away from grain boundaries, and the direction depends on temperature. Below a critical temperature, chromium enriches at boundaries. Above it, chromium depletes. The transition is sharp, and its location determines whether the steel corrodes from the inside.
Oplinger et al. (arXiv:2603.14574) build a rate-theory model that reproduces this flip. The mechanism is a race between two defect-mediated transport processes. Radiation creates vacancies and interstitials in roughly equal numbers. Vacancies drag chromium toward boundaries (enrichment) because chromium-vacancy binding is favorable. Interstitials push chromium away (depletion) because self-interstitial atoms preferentially replace chromium at boundary sites.
Note (2026-11-13; read against the paper's full text): the essay has the roles of the two defects the wrong way round. In the paper, below about 550 K the self-interstitial flux dominates and preferentially transports chromium toward the sinks, causing enrichment; above about 550 K the vacancy flux dominates and the chromium is depleted. The essay says vacancies drag chromium to the boundary and cause enrichment, and that interstitials push it away and cause depletion at high temperature. The paper computes this from partial diffusion coefficients obtained from Onsager transport coefficients, for dilute Fe-0.1 at.% Cr, not for steel generally; I did not find the essay's statement that interstitials replace chromium at boundary sites. The model, which the paper calls physics-based rate theory, is a model of a transition seen in experiments, with the crossover near 550 K when defect production and absorption are unbiased.
Temperature controls which process wins. At low temperatures, vacancy mobility is limited — vacancies accumulate near where they form and interact strongly with nearby chromium. The drag is local and effective. At high temperatures, vacancies are mobile enough to reach sinks before binding with chromium, weakening the drag mechanism. Meanwhile, interstitial transport becomes dominant, and the flow reverses.
The elegant detail: dose rate and dislocation density change only the magnitude of segregation, not its direction. Temperature alone controls the sign. But the transition temperature itself shifts when you account for realistic conditions — biased defect production at cascades, selective absorption at dislocations. The simple transport model predicts the flip at one temperature; the full model shifts it lower. The physics is qualitatively right but quantitatively wrong without the secondary effects.
One variable controls the direction. Every other variable controls only the speed. The material's corrosion fate is set by a single parameter crossing a threshold that itself depends on the details you thought were negligible.