Elastomers are among the most familiar and most deceptive solids. They stretch enormously, recover their shape, and appear forgiving in a way that glass or ceramic never would. Yet they are held together by covalent bonds whose intrinsic strength is measured in GPa. The macroscopic strength of rubber-like networks, however, is usually only in the MPa range. How does a material made of very strong bonds become so weak?
The usual reflex, especially for mechanicians, is to think of Griffith. Glass and ceramics are weak because cracks concentrate stress. Perhaps polymer networks are weak for the same reason: a crack tip finds a few bonds, amplifies the stress, and the material fails. This is a natural explanation. According to Mohanty et al., it is also an incorrect one!
Originally posted as a preprint on ArXiv and on mechanicsArXiv and now published in npj Computational Materials, the paper by Shaswat Mohanty, Jose Blanchet, Zhigang Suo, and Wei Cai, titled Why is the strength of an elastomeric polymer network so low?, offers a different picture. The culprit is not necessarily a crack-like flaw. Nor is it simply the shortest polymer strand in a naive bundle of parallel chains. The key object is more global: the shortest path through the network.
This is not simply polymer physics with a few broken bonds added in. It is graph theory, statistical mechanics, molecular simulation, and solid mechanics all colluding inside a deforming network.
The idea is simple once stated. Take two monomers on opposite ends of the network. Among all possible routes of covalent bonds connecting them, find the one with the fewest bonds. That is the shortest path. Now do this for many such pairs. The network has a distribution of shortest paths, and the left tail of this distribution matters most.
As the elastomer is stretched, most strands still deform mainly by entropic elasticity. They are not straight; they still have slack. But a small number of paths in the left tail become nearly straight. These paths are special not because every bond in the network is highly stressed, but because they are the rare routes that have exhausted their entropy. Their tension is set by covalent bond stretching.
Then one bond on one of these paths breaks. The network does not immediately disintegrate. Instead, the load shifts to other surviving paths in the left tail. Another path straightens. Another bond breaks. The process repeats.
This is the central message of the paper: the network ruptures by a sequence of rare bond-breaking events, not by the simultaneous failure of a large fraction of bonds. In the simulations, the peak stress is about 21 MPa, roughly 200 times lower than the strength of the coarse-grained covalent bond. At the peak stress, less than 5% of the strands and less than 2% of the cross-links have broken. The network fails while almost all of its bonds are still intact.
The visual evidence is especially striking. Up to the peak stress, the broken bonds are distributed throughout the network, not concentrated near a crack-like region. Large holes appear only later, after the stress has already declined. In other words, the crack is not the protagonist of the story. At least for the intrinsic strength of a homogeneous elastomeric network, the action is elsewhere.
What I found compelling about this paper is that it challenges a familiar intuition. We might guess that short strands should break first. The authors test this directly and find otherwise: the length distribution of broken strands nearly coincides with the length distribution of all strands. A short strand is not necessarily a highly loaded strand. A network is not a collection of independent strings between two rigid plates. Load is transmitted, redistributed, and rerouted. The relevant quantity is not local strand length, but connectivity across the network.
This perspective also suggests a design principle. To make elastomeric networks stronger, it may not be enough to make individual bonds stronger, or even to increase the average connectivity. We must engineer the distribution of shortest paths, especially its left tail. The enemy is not the average strand. The enemy is the rare, nearly straight path that reaches the covalent limit, while the rest of the network is still comfortably deforming entropically.
That is a beautiful mechanics story: a macroscopic strength, three orders of magnitude below bond strength, emerging from a small number of extreme paths hidden in a disordered network.