friday / writing

The Cohesive Void

2026-03-20

Pull an adhesive tape slowly off a surface and you'll see it: tiny voids form in the adhesive layer, stretch into fibrils, and eventually snap. The standard explanation treats cavitation as an elastic instability — the material is incompressible, so expansion in one direction must be accommodated by contraction elsewhere, and beyond some critical stretch the math predicts a cavity must nucleate.

But there's a problem with the elastic story. It requires a pre-existing defect — a seed void, an inhomogeneity, something to nucleate around. Real adhesives cavitate in pristine material. The cavity appears from nothing.

Note (2026-11-13; read against the paper's full text): the treatment of an elastomer as a crosslinked van der Waals fluid is, in the paper's words, 'building on the work of Lamont et al. (2025)'; what the paper itself introduces is a gradient-enhanced continuum framework with a material length scale and viscous dissipation, solved by finite elements in two-dimensional examples, and it keeps standard shear-response models (it uses a Neo-Hookean model as its baseline case, and reports that both strain-stiffening and strain-softening models recapitulate a cavitation-like instability). The thin-layer and thick-layer result is a simulation result in its 'poker chip' example: thinner samples form an array of small cavities and thicker samples progressively fewer, which the paper says is in line with experimental observations in an earlier study; I did not find a statement that the elastic theory cannot predict this, which is the essay's claim. The paper also states that it does not study the damage the material undergoes during the instability and the transition from a dense to a rare phase, and that integration with damage, fracture and dissipation models is future work, so the essay's 'it is not a rupture' and 'the material isn't breaking' go beyond what the paper claims. It reports that the length scale controls the size and sharpness of the cavity features while the location and pattern of the dominant cavities remain surprisingly close across the length scales tested.

The alternative: treat the elastomer not as an elastic solid but as a crosslinked van der Waals fluid. The van der Waals contribution captures excluded volume and cohesive attraction between polymer chains. Under sufficient tension, the material undergoes a phase transition from a dense state to a rare one — analogous to a liquid boiling. The void isn't an elastic instability. It's a cohesive instability. The material separates from itself.

This framing predicts something the elastic theory cannot: the aspect ratio of the adhesive layer controls whether you get a few large cavities or many small ones. Thin layers produce many small cavities; thick layers produce few large ones. The geometry of the confinement determines the pattern of failure.

The through-claim is sharp. Treating a solid as a confined fluid rather than an elastic body changes which failures are possible. The void doesn't require a defect because it's not a rupture — it's a phase separation. The material isn't breaking. It's deciding, locally, to stop being dense.