Nucleation and growth in small systems

The unique properties of matter at the nanoscale, mainly arising from a large surface to volume ratio, offer the exciting possibility to create new materials with designed properties and functions based on the characterstics of their nanoscopic building blocks. During the last decades, tremendous progress has been made in this field and today nanoparticles, synthesized with precise control of size and shape, are the basic components of many significant applications ranging from photovoltaics to electronics and biomedicine. Such nanoparticles, however, are not only of technological interest, they also provide the opportunity to study fundamental condensed matter processes. Semiconductor and metal nanocrystals, for instance, have been used extensively to investigate the thermodynamics and kinetics of pressure and temperature driven phase transitions, revealing strong effects of size and shape on the phase stability and the mechanism of phase transformations. While experimental studies of nanoscale systems, carried out with advanced equipment, yield important information about such processes, they lack the time and space resolution required to obtain detailed insights into the transformation mechanism at the atomistic level, which is key for understanding and, eventually, controlling the dynamical properties of nanomaterials. In this project, we will use modern computer simulations methods to study phase transitions in nanoparticles with the objective to reveal the underlying nucleation and growth mechanism on the atomistic scale. Such computer simulations are challenging because of the wide range of involved time scales and require massive computational resources. We will concentrate our efforts mainly on the simulation of a structural transformation occurring in copper sulfide nanocrystals, as this process has been studied in great detail in a series of recent experiments. To carry out these computationally demanding simulations, we will develop new algorithms including an enhanced procedure for the simulation of rare events, a neural network based empirical potential for the calculation of energies and forces, as well as new approach for the classification of local crystal structures. The results of these simulations will throw light on the effects of finite size and surfaces on the kinetics and, in particular, on the mechanism of first order phase transitions occurring in nanoparticles. This knowledge is a precondition for the controlled stabilization of metastable structures and might open routes for the design of nano- structured materials with new properties and phases not accessible in the bulk.

It is an everyday observation that when water is cooled below the freezing point it suddenly freezes, transforming from a liquid into a solid with completely different properties. While in the liquid the molecules can move around freely without an particular order, in the solid they are arranged in a very regular pattern, which is reflected in the beautiful and symmetric shape of snow crystals. What, then, are the forces that lead to this order and how do the molecules move around in concerted ways during this ordering process? In this project we used atomistic computer simulations to find out how crystals form in a liquid and other phase transitions occur on a microscopic level. From a computational point of view, the simulation of phase transitions is challenging for several reasons. As has been known for centuries, a glass of water cooled below zero degrees Celsius does not freeze immediately if the cooling is done carefully. In fact, a liquid can be kept in such a metastable state almost indefinitely. Only if a small crystal nucleus forms, either by chance or by the effect of an external perturbation, will the freezing start transforming the entire glass into solid ice. The long waiting times associated with the formation of such nuclei makes the simulation of nucleation processes difficult. Another challenge consists in tracking and making sense of the irregular dance of thousands of molecules as they conspire to form order from disorder. Here we have developed and improved computer simulation algorithms designed to solve exactly these problems. Executing these programs on high performance computers, we have studied how crystalline seeds of different structures affect the crystallization process, finding that only seeds of a particular structure enhance the crystallization. In another study, we have investigated the process of cavitation, i.e. the formation of vapor bubbles at negative pressures. Cavitation may occur in plants under dry conditions interrupting the flow of water and is also of technological relevance, because bubbles forming and collapsing near fast moving ship propellers or turbine blades can cause severe damage to the material. Using the methods developed in this and previous projects, we have studied this process in detail. Based on the results of simulations, we have now devised a complete theory that sheds light on the mechanism for bubble formation and growth and allows predicting the rate at which bubbles form depending on external conditions. Building on the methods developed in this project, future studies will be directed towards the simulation of heterogeneous nucleation in which the formation of a nucleus is enhanced by the presence of surfaces and/or impurities.