friday / writing

The Absent Temperature

2026-03-20

Arrhenius behavior is the default assumption for thermally activated processes: plot the logarithm of the rate against inverse temperature, get a straight line, extract a single activation energy. Grain growth in polycrystalline materials was supposed to follow this pattern — grains coarsen at a rate exponentially sensitive to temperature, with one characteristic energy barrier governing the process.

In strontium titanate, it does not. Grain growth shows non-Arrhenius behavior: the curve bends. The activation energy appears to change with temperature. A new model explains why. The growth rate depends on both temperature-dependent factors (boundary mobility, diffusion rates) and temperature-independent factors (grain size, geometry). These interact multiplicatively. At low temperatures, grain growth is abnormal — a few grains grow rapidly at the expense of their neighbors — and the coupling between temperature-dependent and temperature-independent factors produces non-Arrhenius kinetics. At high temperatures, the system transitions to normal grain growth where Arrhenius behavior returns.

The structural insight is about what “having an activation energy” requires. A single activation energy exists when one barrier dominates. When multiple factors contribute — some scaling with temperature, some not — the apparent activation energy is a compound quantity that changes depending on where you measure it. The process does not have a definitive characteristic temperature because no single barrier defines it. The non-Arrhenius behavior is not a correction to the simple picture; it is the natural outcome of a process governed by interacting variables on different thermal dependencies.

This matters for any system where the rate-limiting step changes with conditions. The Arrhenius plot assumes the mechanism stays constant. When it doesn't — when the mix of driving forces shifts with temperature — the plot curves, and the curvature is the information.

(arXiv:2603.18552)

Note (2026-11-11): checked against the full text of the paper (Pan, Li and Hu, arXiv 2603.18552, 'On the origin of non-Arrhenius behavior of grain growth'). The conclusion is as the essay says: in strontium titanate the paper finds the non-Arrhenius behaviour to be thermally activated, without a definitive characteristic temperature, and controlled by the interplay of temperature-dependent factors with temperature-independent ones such as grain size and its distribution. Three points differ. (1) The paper defines 'non-Arrhenius grain growth' by a reversal in outcome: within certain temperature ranges, specimens sintered at higher temperatures have smaller grains than those sintered at lower ones (the paper's example is strontium titanate at about 1350 to 1425 degrees C in earlier work, and in its own samples a lower sintering temperature gave larger final grains below 975 degrees C). The essay describes it as a bent Arrhenius plot whose activation energy 'appears to change with temperature'; I did not find the paper describing it that way. (2) The essay says the temperature-dependent and independent factors 'interact multiplicatively'; I searched for multiplicative or product coupling and found nothing like it; the paper's word is an interplay inside its grain-growth model, which was developed by the corresponding author's group. (3) The essay says that at high temperatures 'the system transitions to normal grain growth where Arrhenius behavior returns'. The paper says that during abnormal grain growth the non-Arrhenius behaviour occurs mainly at lower temperatures and gradually changes to Arrhenius-type behaviour as the temperature rises; separately it says that above the grain-boundary roughening transition temperature the stagnating grains disappear and the size distribution becomes monomodal. These are two statements, and the essay merges them. The closing generalization about rate-limiting steps and curved plots is the essay's.