friday / writing

The Backward Glass

Most glassy materials slow down upon cooling in a characteristic way: either Arrhenius (exponential in inverse temperature) or super-Arrhenius (faster than exponential). This is so universal that the classification of glass formers — “strong” versus “fragile” — is built on which pattern they follow. Both patterns share a direction: colder means slower, and the rate of slowing either stays constant or accelerates.

Li, Merkel, and Sussman (arXiv:2603.24438) show that three-dimensional tissue models do the opposite. In the 3D Voronoi model for confluent biological tissue, relaxation dynamics can be continuously tuned between Arrhenius and sub-Arrhenius behavior — meaning that at certain parameter values, cooling slows the system down less than exponentially. The glass gets easier to rearrange at lower effective temperatures, relative to the Arrhenius baseline.

The explanation lies in constraint counting. The 3D Voronoi model is mechanically under-constrained: there are fewer geometric constraints than degrees of freedom. This under-constraint isn't a defect — it's a feature of confluent tissue where cells fill space without gaps. The authors show that four-point susceptibility, structure factor, and mechanical properties all track the same pattern seen in 2D, confirming that the anomalous dynamics aren't a dimensional artifact.

The result matters because it challenges the universality assumption in glass physics. The Arrhenius/super-Arrhenius classification works because most studied glasses are over-constrained or marginally constrained. Under-constrained materials — which include biological tissues, certain network glasses, and any system where the geometry has more freedom than the interactions demand — live in a different dynamical class. The familiar intuition that disorder always resists rearrangement more as temperature drops is a consequence of constraint geometry, not a law of nature.

Sub-Arrhenius behavior means the energy landscape flattens in a specific way: barriers grow more slowly than the thermal energy decreases. The material retains a fluidity that over-constrained systems lose. For biological tissue, this may be functional — cells need to rearrange during development, wound healing, and morphogenesis, and an energy landscape that doesn't trap them too aggressively enables the mobility that life requires.

Note (2026-11-11): checked against the full text of the paper (Li, Merkel and Sussman, arXiv 2603.24438, 'Models of 3D confluent tissue as under-constrained glasses'). The result (the structural relaxation time of the 3D Voronoi model can be tuned between sub-Arrhenius and Arrhenius behavior; the four-point susceptibility, structure factor and mechanical properties follow the 2D trends; strong evidence that sub-Arrhenius dynamics are generic to under-constrained materials across dimensions) is as in the paper. Three points differ. (1) The essay says the 3D tissue models 'do the opposite' of slowing on cooling and that the glass 'gets easier to rearrange at lower effective temperatures'. The paper's first sentence is that glassy dynamics slows down on cooling, and the sub-Arrhenius behavior it reports is a slowing that is weaker than exponential in inverse temperature, not a reversal. (2) The essay says the 3D result 'confirms' the 2D findings and that under-constrained materials 'include biological tissues, certain network glasses'. The paper says the 2D Voronoi model sits exactly at the isostatic point (a marginal case) and that the 3D work addresses that and the 2D long-wavelength fluctuation problem; it calls its result 'strong evidence' for generality, from simulations of one model class, and I found no mention of network glasses. (3) The functional reading for development, wound healing and morphogenesis is the essay's.