A glass is an arrested liquid — its particles want to flow but are trapped in cages formed by their neighbors. Adding a small fraction of active particles — self-propelled agents that push persistently in one direction before reorienting — can melt the glass by kicking passive particles out of their cages. At low persistence, the active dopants act like a uniform temperature increase: they enhance structural relaxation everywhere, the cages break more easily, and the glass fluidizes homogeneously. More persistence means more effective melting. The picture is simple — active particles are microscopic stirring rods.
Janzen, Janssen, Araujo, Sknepnek, and Matoz-Fernandez (arXiv 2602.23178, February 2026) show that increasing persistence does not simply amplify this uniform melting. Above a threshold, it reorganizes the fluidization mechanism entirely.
At high persistence, active particles push in one direction for long enough to accumulate mechanical stress in the surrounding passive matrix. Each active particle creates a disturbed zone — a region where the local stress exceeds the yield threshold of the amorphous solid. When the disturbed zones of neighboring active particles overlap, the accumulated stress ruptures the passive matrix, creating a void — a region of low density surrounded by compressed material.
Once a void forms, the dynamics change qualitatively. Rearrangements concentrate at the void edges, where the stress gradients are steepest. Active and passive particles near the void boundary exhibit comparable mobility — the passive particles are being pushed by the stress field, not by direct contact with active particles. The flow pattern resembles crowd dynamics at a concert: a central open space surrounded by dense, rapidly moving boundaries. The authors call it mosh pit dynamics.
The transition from uniform fluidization to void-mediated fluidization is not gradual. It is a reorganization of the spatial structure of relaxation. Below the persistence threshold, cage-breaking events are distributed throughout the material, uncorrelated with each other. Above it, relaxation localizes around voids, with large regions of the material remaining arrested while the void boundaries carry all the flow. The material is simultaneously more fluid (at the void edges) and more rigid (in the passive bulk) than the uniformly fluidized state.
The mechanism is nonequilibrium stress accumulation. In thermal melting, fluctuations are isotropic and uncorrelated — each particle jiggles independently. In persistence-driven void formation, the active particles store directional mechanical energy in the matrix over timescales set by their reorientation time. The longer they push in one direction, the more stress they deposit. When the stored stress exceeds the material's yield stress, it releases catastrophically — not gradually — creating the void.
The distinction matters for applications. If you want to fluidize a material uniformly — prevent sedimentation, ensure mixing — low persistence is better. If you want to create structured flow — channels, voids, localized transport — high persistence produces the spatial organization for free. The same active dopant, at different persistence levels, produces qualitatively different mechanics. The stirring rod becomes a jackhammer.