Politecnico di Torino (logo)

Connecting Structure and dynamics in model network glasses

Simone Monaco

Connecting Structure and dynamics in model network glasses.

Rel. Alfredo Braunstein, Frank Smallenburg. Politecnico di Torino, Corso di laurea magistrale in Physics Of Complex Systems (Fisica Dei Sistemi Complessi), 2020

PDF (Tesi_di_laurea) - Tesi
Licenza: Creative Commons Attribution Non-commercial No Derivatives.

Download (8MB) | Preview

In recent literature, increasing experimental and numerical e ort has been de- voted to understanding and predicting the behavior of glassy systems. Among those classes of materials, of particular relevance are network glasses. In such system the composing particles can form a small number of bonds with their neighbors, leading then to an overall network structure. Among everyday mate- rials, common window glass is perhaps the most obvious example of a network glass, but many others have, on the microscopic scale, a disordered network structure. To better understand these materials, it is of primary importance to gure out the role and the local organization of the bonds in the network structure. In literature, a common way to study such systems is via the use of patchy colloids as a simpli ed model. In this thesis we analyze the behavior of patchy colloids by simulating them as patchy particles. These patchy particles consist of hard-spheres with a xed number of patchy sites on their surfaces, through which the bonds can form. Starting from a simulated 2d system of such particles, we compare its behavior with analytical results from Wertheim theory. We then analyze the e ect of the external conditions of the system by constructing an approximated phase diagram. Once the global behavior is fully characterized, the main goal of this thesis is to investigate how di erent local conditions drive the breaking and forming of bonds. In particular, we apply supervised learning techniques to predict which bonds are more likely to break in a certain con guration, looking only at the information given by the positions and connections of the particles. We essentially show which local parameters associated with each bond, from a picture of the system taken at equilibrium, are su??cient to train a neural network and infer the breaking time. Even though the dynamics of the particles cannot be completely predicted by their position at a certain time, this approach can be a good starting point for understanding the interplay between local structure and dynamics in these complex materials.

Relators: Alfredo Braunstein, Frank Smallenburg
Academic year: 2019/20
Publication type: Electronic
Number of Pages: 113
Corso di laurea: Corso di laurea magistrale in Physics Of Complex Systems (Fisica Dei Sistemi Complessi)
Classe di laurea: New organization > Master science > LM-44 - MATHEMATICAL MODELLING FOR ENGINEERING
Ente in cotutela: Universite de Paris-Sud (Paris XI) (FRANCIA)
Aziende collaboratrici: CNRS LPS
URI: http://webthesis.biblio.polito.it/id/eprint/15310
Modify record (reserved for operators) Modify record (reserved for operators)