Stochastic Description of Agglomeration and Growth Processes in Glasses

In a series of papers published during the past ten years (, , , ), new models of growth by agglomeration of smaller units have been elaborated, and applied to many important physical systems, such as quasicrystals (), fullerenes (, ), and oxide and chalcogenide glasses, (, ,, ). Here we shall present the main ideas on which these models are based, and briefly discuss the latest developments. In order to make our presentation concise, the example we choose is the simplest covalent network glass known to physicists, the binary chalcogenide glass AsxSe(1−x), where x is the concentration of arsenic atoms in the basic glass-former, which in this case is pure selenium. The generalization to other covalent networks, e.g. GexSe(1−x), is quite straightforward. These glasses (in the form of thin and elastic foils) are used in photocopying devices. Whether the formation of a solid network of atoms or molecules occurs in a more or less rapidly cooled liquid, or as vapor condensation on a cold support, the most important common feature of these processes is progressive agglomeration of small and mobile units (which may be just single atoms, or stable molecules, or even small clusters already present in the liquid state) into an infinite stable network, whose topology can no longer be modified unless the temperature is raised again, leading to the inverse (melting or evaporation) process. To describe such an agglomeration with all geometrical and physical parameters, such as bond angles and lengths, and the corresponding chemical and mechanical