Cooling liquids reveal self-limiting particle clusters behind glass transition. According to theoretical physicist Corentin Laudicina, the behavior of materials during the glass transition involves a fourth phase beyond the familiar gas, liquid, and solid states. In the glass phase, a material behaves like a solid but has a disordered structure similar to a liquid. Laudicina and his colleagues from the Soft Matter & Biological Physics group study the viscosity of liquids as they cool. As the temperature drops, the viscosity increases rapidly around the glass transition, yet the internal structure of the material changes little. To understand this phenomenon, Laudicina and his team use complex computer simulations to track the movement of millions of particles over long periods. They simplified the molecular complexity of real liquids into a "model liquid" consisting of perfectly spherical particles interacting in simple ways. Using these simulations, they found that major changes in viscosity are related to particles moving heterogeneously. Some particles barely move, while others form groups that move together. As the liquid cools, the clusters initially become larger. However, below a certain temperature, they start to become smaller again. Older theories predicted that the clusters would continue growing indefinitely, causing the liquid to suddenly become stuck. Laudicina's findings show that the clusters effectively put a natural brake on their own growth, explaining why no abrupt transition is observed in the laboratory or in computer simulations. This feedback mechanism was missing in previous equations, and incorporating it into the theory eliminated the previously predicted sharp transition. Laudicina emphasizes that this discovery could offer new insights into research on the glass transition. He notes that the physics of the glass transition is not limited to liquids. For example, the cells in a tumor are packed together similarly to particles in the simulations. These cells can become stuck in a similar way but can also suddenly start moving again and invade surrounding tissue. Within his group, Laudicina is working with cell biologists to investigate this from a physics perspective. Laudicina also points out that similar physical problems can be found in computer science and machine learning, highlighting the importance of understanding these fundamental concepts beyond theoretical research.