3D atomic imaging reveals gradient-based crystal formation, challenging century-old theory

For a hundred years, scientists have relied on a single mental model to explain how crystals begin to form—how a liquid becomes solid, how water vapor becomes rain, how disorder transforms into order. Classical nucleation theory, developed in the early twentieth century, painted a clean picture: ordered seeds called nuclei emerge from chaos with sharp, well-defined boundaries, like islands appearing suddenly in a sea. Thousands of experiments seemed to confirm it. The math worked. Textbooks taught it. The theory became bedrock.

But bedrock can shift. A team led by physicist Jianwei Miao at UCLA has now looked directly at the atomic architecture of forming crystals using a technique called atomic electron tomography—essentially a three-dimensional X-ray vision that maps individual atoms in space. What they found does not match the century-old picture. The crystals they observed were not uniform. They did not have sharp boundaries. Instead, each nucleus possessed a highly ordered core that gradually became more chaotic as it extended outward, like ripples fading from a stone dropped in water. The researchers published their findings in Nature Materials in August 2026, proposing what they call the gradient nucleation pathways model.

To make this discovery, Miao's team worked with nanoparticles made from high- and medium-entropy alloys—exotic metallic mixtures where multiple elements exist in roughly equal proportions, unlike familiar alloys such as steel where one element dominates. They heated these particles to more than 3,000 degrees Fahrenheit, then supercooled them to room temperature in fractions of a second. This rapid temperature swing triggered crystal nuclei to form but froze them at various stages of development, like catching frames from a film. The imaging process, which takes far longer than nucleation itself, allowed the researchers to examine more than 8,000 nuclei ranging from fewer than 10 atoms to more than 1,000. The data set was robust enough to reveal a consistent pattern: nature does not build crystals the way the old theory suggested.

The implications reach beyond academic curiosity. Classical nucleation theory described the energy required to form a nucleus as a single, sharp barrier—a wall that only nuclei with sufficient energy could cross. Miao's experiments revealed something more nuanced: nature appears to use multiple intermediate steps, like climbing a ladder rather than vaulting a wall. This gradient approach allows nuclei to overcome energy barriers more efficiently than the classical model would predict. As nuclei grew larger, their cores became increasingly ordered, and when separate nuclei approached each other to merge, most aligned in nearly identical orientations before joining—a phenomenon that suggests nature gravitates toward lower-energy pathways even before crystals fully form.

The gradient nucleation pathways model does not discard classical nucleation theory. Rather, it encompasses it. When you substitute the conditions of the old theory into the new equation, you recover the old results exactly. But experimentally, those conditions never appeared. The sharp boundaries never materialized. What emerged instead was a more complex, more accurate picture of how matter organizes itself at the atomic scale.

The practical reach of this work extends far beyond materials science. Nucleation happens everywhere: in clouds forming rain, in industrial manufacturing, in the crystallization of pharmaceuticals, in semiconductor production. High- and medium-entropy alloys, first developed about two decades ago, promise materials that are simultaneously stronger and more flexible than conventional alloys, with catalytic properties that could drive chemical reactions more efficiently. Understanding how these materials crystallize could accelerate their development and deployment. But the broader implication is that researchers across climate science, electronics, food production, and drug manufacturing may need to reconsider how they model phase transitions in their own domains. Miao expressed hope that textbooks will eventually be revised to reflect this new understanding—a quiet acknowledgment that even century-old certainties can yield to careful observation and rigorous experiment.