A rare event: Understanding, inhibiting, and optimising nucleation in colloidal systems

Colloidal systems are composed of big particles — colloids, whose size ranges from 1 nm to 1 μm approximately — dispersed in a medium of smaller particles. As a consequence, colloids interact not only with the solvent particles, resulting in a stochastic dynamics called Brownian motion, but also with each other. The emerging collective behaviour is surprisingly rich and diverse. Remarkably, the phase behaviour and the self-assembly processes of colloidal systems are strikingly analogous to those of atoms and molecules, but on larger scales. Among the many self-assembly processes displayed by colloidal systems, one of the most common and certainly one of paramount importance is nucleation. Nucleation is the process through which a first-order phase transition happens, namely with the birth of a microscopic nucleus of the embryo phase inside the parent phase, which in some conditions is able to grow out and become macroscopic. Nucleation is an ubiquitous event in nature, and hence understanding its mechanism would generate remarkable breakthroughs in very diverse areas of science such as optics, pharmaceutics, and climate science. In the field of soft matter, nucleation constitutes an extremely broad topic, with many open, challenging, and fascinating questions. With the current thesis, we aim to address some of these questions by means of numerical simulations and machine learning algorithms. We try to understand its underlying mechanisms for several systems, and we pay special attention to the inhibitory factors as well as the optimal conditions for nucleation to happen.