Development and application of computational methods for the investigation of the structures of metal nanoparticles

This Ph.D. thesis is dedicated to the development and application of computational techniques for the study of metallic and bi-metallic nanoparticles. Metallic nanoparticles have been studied in the recent years for their unique properties that differ a lot from the properties of the bulk materials from which they are produced. These properties include catalysis, data storage, plasmonics, nanomedicine, water purification and many others. For this reason, computational meaningful models and clever methods for studying the structural and dynamical properties of these nanoscale systems have been elaborated and established, but still today they need further development and refinement. The scope of this thesis was to improve the set of tools already present in the literature and also to add new ones that are meant to increase our ability to study these objects. In the introduction, metal nanoparticles are defined and some of the experimental methods to produce them, as well as some characterization techniques are described. In Chapter 1, the most common geometrical structural motifs and chemical arrangements of atoms are presented and described in detail. In Chapter 2, the theoretical and computational models and methods used throughout this thesis are discussed. In Chapters 3 and 4, the new methods that were developed are described. In particular, Chapter 3 deals with the problem of clustering nanoparticles structures. Usually, in the output of numerical simulations of nanoparticles, several different structures are produced. We designed a technique for a meaningful and automatic partition of these outputs, which is based on the use of a geometrical description derived from the Common Neighbour Analysis and some Machine Learning unsupervised learning algorithms. In Chapter 4, two new algorithms for the approximate solution of the global optimization of nanoparticles structure problem are presented. This problem is of utter importance since the search of the lowest-energy structures for a given fixed number of atoms of some metal (or metals) determines in principle what could be the most probable structures found in experiments that produce nanoparticles at low-temperature thermodynamic equilibrium. Applications of the computational models and methods described in this thesis are given in Chapters 5 and 6. In Chapter 5, we show that a specific geometrical structure of tetrahedral symmetry, that is generally unusual for pure metal nanoparticles, becomes more stable when alloying two metals. These tetrahedral clusters, that were obtained from global optimization algorithms such as those described in chapter 4, correspond to a new series of magic numbers that we derive in an explicit formula. In Chapter 6, the results of a collaboration with an experimental group about the growth of silver nanoparticles are showed, demonstrating the importance of the cooperation between simulations and experiments. Finally, the material of this thesis is summarized in the Conclusions, where also some future perspectives for possible extensions of this work are listed.