Rodriguez-Lopez and Ferrero (2602.22198) show that heat alone — no external stress, no driving force — can trigger catastrophic avalanches in amorphous solids.
The conventional picture: amorphous materials like glasses fail through mechanical loading. You push on them, some local region yields, the stress redistributes, maybe that triggers a neighbor, and you get an avalanche of rearrangements. The avalanche statistics are scale-free with exponential cutoffs — power laws with a cap set by the system size. Temperature is a footnote, something that assists the loading-driven dynamics.
Rodriguez-Lopez and Ferrero flip this. Using elastoplastic models with purely local thermal activation rules and no external driving, they show that as temperature increases, the avalanche statistics transition from the familiar scale-free regime to one dominated by system-spanning runaway events. There's a critical temperature T_c(L) that separates intermittent dynamics from wholesale fluidization.
The most striking finding: T_c decreases algebraically with system size. In the thermodynamic limit, arbitrarily small finite temperatures may destabilize the intermittent regime. This means the scale-free avalanche behavior seen in finite simulations might not survive infinite-size extrapolation — at any finite temperature, the system eventually accesses a runaway mode.
The mechanism is the spatial reorganization of marginal stability. Thermal activation doesn't just nudge individual sites over their yield thresholds — it reshapes the distribution of distances-to-yielding across the entire system. Sites that were safely below threshold get pushed into a narrower band of stability, making cascading failure easier. The temperature acts not as an additive perturbation but as a restructuring of the energy landscape's cliff edges.
This is a finite-size-controlled instability: not a bug of small simulations but a genuine transition that small systems approximate and large systems sharpen.