After finishing his Ph.D. on glass formation, chemical physicist Patrick Charbonneau thought he’d never study the material again. But something kept nagging him: In some experiments, materials would unexpectedly morph into glass, solid as a rock but molecularly disordered like a liquid. The results didn’t match with glass-formation theory, but they were easy to dismiss as a fluke. “If I want to have a career,” Charbonneau remembers thinking, “there’s no way I should work on this problem. It’s ridiculous.”
Other researchers had found mismatches between the theory of how glasses form and actual glassmaking (real or simulated on a computer). But most blamed the discrepancies on experimental conditions, impurities, inherent material properties — such as the shapes of molecules — or other factors. Charbonneau wasn’t satisfied with those explanations, though, and started moonlighting to learn more. He found a collaborator in his cousin’s husband, a mathematician who “owed him,” Charbonneau jokes, for the matchmaking with his cousin. The friends settled on repaying the debt through solving math problems together: “We meet at family gatherings, and when the aunts and uncles find us to be a little too boring, we sit down in the corner and start doing math.”
After months of frustration, Charbonneau, now at Duke University, realized that the pair didn’t really understand what was happening in glass mathematically. Researching the literature, he found no experiments that tested a decades-old theory that described glass formation in three-dimensional space but not for higher dimensions. Then came a moment of clarity. “People have been looking at the consequences of that theory without having carefully checked the assumptions,” he concluded.
Finally putting theory to the test, Charbonneau and his team simulated glass formation in four-dimensional space and higher, eventually proving that a crucial assumption of that early glass-formation theory was wrong: The molecular structure within glass is much more disorderly than had been thought. The results and Charbonneau’s methods — simulating each assumption in the lab before theorizing further — are spreading across the glass-research community like a tiny crack across a windshield. That’s prompting some glass researchers to think about repairing their glass-formation theories, others to consider replacing them and still others to ignore the crack, perhaps hoping it won’t spread further.