Explaining the difference between brittle and ductile behavior in polymer glasses

October 28, 2026 dialog Explaining the difference between brittle and ductile behavior in polymer glasses Lisa Lock Scientific Editor Robert Egan Senior Editor Glasses are (mostly) brittle, metals can be ductile (plastic), and liquids flow easily. Polymers are interesting because they exhibit all three types of behavior, depending on their chemical nature, temperature, and deformation rate. If polymers are not able to crystallize, their low-temperature state is glassy, and they are hard and transparent. One example is poly(methylmethacrylate) [PMMA], which has been used as an artificial glass since the early 20th century. Surprisingly, polymer glasses can be ductile at temperatures significantly below their glass transition temperature, Tg. For a typical laboratory strain rate (~10⁻³ s⁻¹), the brittle-ductile transition temperature (BDT or TBD) can be 50–100 K below Tg. This result is surprising—naively, one would expect a "glass" to be brittle—yet polymers can deform plastically well below Tg. Developing a simple micromechanical model In collaboration with Profs. Oleg Gendelman (Technion, Israel) and Alessio Zaccone (University of Milan, Italy), I started by developing a simple model that captures the main features of the three types of stress-strain behavior: brittle (Type I), ductile with stress overshoot (Type II), and ductile with no stress overshoot (Type III). For more than a century, scientists have studied viscoelastic materials using simplified representations like the Maxwell element (spring and dashpot connected in series). Our model—now published in Soft Matter—is based on that Maxwell element, but with two big modifications. First, the elastic component is not a pure linear spring, but a sinusoidal function of the elastic strain, reflecting the fact that the potential energy landscape has multiple minima. Second, the dashpot viscosity decreases at large strains, as predicted by the so-called Eyring-Long model. With our model, we can calculate the yield strain corresponding to each strain rate. For small strain rates, the yield strain is small, and the stress-strain curve is of Type III. For intermediate strain rates, the yield strain is larger, and the stress-strain curve is of Type II. Finally, for large strain rates, there is no solution, the material does not flow, and the stress-strain curve is of Type I. Brittle fracture is when everything moves too fast for material to adjust We thus look at the brittle-ductile transition differently from earlier researchers. In our view, a polymer glass can yield uniformly if the strain rate is slower than a temperature-dependent threshold but must break in a brittle fashion if the strain rate is faster than that threshold. How can we compute this threshold? Our hypothesis is that this threshold is inversely proportional to the Johari-Goldstein beta-relaxation time. (This idea was formulated empirically by Dr. S. Wu (Dupont) in the 1970s, but we are now able to come up with a theoretical justification.) In the undeformed state, the relaxation time is large (alpha-relaxation time), and, as the material is deformed, it decreases all the way down to the beta-relaxation time. Knowing the two relaxation times as a function of temperature, we can then compute the BDT as a function of the strain rate. Crucially, the alpha-relaxation time in glassy polymers depends on the aging of the polymer; at temperatures far below Tg, it seldom reaches equilibrium and is generally on the order of the aging time. Thus, the analysis of the BDT requires knowledge of the sample age—a fact known to experimentalists but often underappreciated by theoreticians. From the model, we can get the "universal" map separating the three regions: brittle, ductile with stress overshoot, and ductile with no stress overshoot (see figure below). The x-axis is proportional to the logarithm of the ratio of the alpha- and beta-relaxation times—or, alternatively, the inverse of the absolute temperature. The y-axis represents normalized strain rate. Hence, we see that at high temperatures (small B), only liquid-like (at small strain rates) or brittle (at large strain rates) behaviors are possible. Knowing the temperature dependence of the alpha- and beta-relaxation times for several amorphous polymers, we can translate the "universal" phase map into polymer-specific BDT results. For PMMA, the comparison between the model and experiment was plotted in the top figure. In the paper, we also show similar plots for poly(vinylchloride) [PVC] and polystyrene [PS]. For all three polymers, the model fit is within 5 K of the experimental data. Future directions The model as formulated now is still semiempirical in the sense that there is one adjustable parameter x describing the "relaxation time floor." In reality, this "floor" depends on the structure of the polymer network, including entanglements, crosslinks, and effects of aging and pretreatment. Much of this is being studied, e.g., by Prof. S. Q. Wang (University of Akron), but a comprehensive theory is still elusive. Also, knowing BDT is only part of the story—we still need to know the yield or fracture stress. We are using molecular simulations and experiments to better parameterize our model and provide analytical predictions not just for the BDT, but for the full stress-strain curves. This will enable better prediction of both brittle and ductile failure in plastic parts subjected to stress and thus help us design better materials for the future. This story is part of Science X Dialog, where researchers can report findings from their published research articles. Visit this page for information about Science X Dialog and how to participate. Publication details Valeriy V. Ginzburg et al., Predicting the brittle-to-ductile transition in amorphous polymers, Soft Matter (2026). DOI: 10.1039/d6sm00633g Journal information: Soft Matter Valeriy Ginzburg earned his B. S. (Physics) in 1989, and Ph. D. (Polymer Physics) in 1992 at the Moscow Institute of Physics and Technology in Russia. After two postdocs at the University of Colorado (1993–97) and the University of Pittsburgh (1998–2000), he worked at The Dow Chemical Company (2001–2020). Today, he is a visiting professor at Michigan State University and founder of a consulting company VVG Physics Consulting LLC. Dr. Ginzburg is a co-inventor on 15 US patents and author or co-author of about 100 journal publications. He is a co-editor of two books, Theory and modeling of nanocomposites (published by Springer Nature in 2020), and Handbook of multiscale modeling of soft matter (to be published by Elsevier in 2027). He has been elected Fellow of the American Physical Society (2014). His main research interests are polymer glass transition and polymer statistical physics.