Polyamorphism in silica glass under pressure : from percolation transitions to mechanical responses, a molecular dynamics study

Julien PERRADIN

Understanding amorphous–amorphous transitions (AATs) and polyamorphism in glasses remains challenging because conventional structural and thermodynamic indicators tend to smooth out the reorganization of disorder into apparent continuous evolutions rather than sharp transitions. This thesis advances a unified framework for vitreous silica (v‑SiO2) that combines large‑scale atomistic simulations, a dedicated clustering and percolation toolkit (Nexus‑CAT), and finite‑size scaling analysis to uncover both the critical nature of pressure‑induced percolation transitions and their impact on macroscopic mechanical responses. First, the structural evolution under hydrostatic compression is examined across multiple interaction models and benchmarked against experimental observables. In addition, the study tracks the formation of SiO5 and SiO6 units and the development of edge-sharing connections, which drive the densification of the tetrahedral network toward higher coordinations units and the development of edge-sharing connections, which drive the densification of the tetrahedral network toward higher coordinations. Second, the package Nexus‑CAT developed during this work, enables robust identification of connected polyhedral clusters and implements percolation properties calculations to monitor the formation and system‑spanning of polyhedral clusters throughout compression. Third, finite‑size scaling shows that the low‑ to high‑density transformation proceeds via distinct percolation transitions including intermediate (SiO5) and octahedral (SiO6) polyhedra with scale‑free fluctuations. This allows extraction of critical exponents and fractal dimensions, assessment of universality by contrasting bonded and non‑bonded criteria, and qualitative comparison to exponents reported for amorphous water. Finally, the work links these microscopic percolation thresholds to macroscopic anomalies including bulk modulus slope changes, the elastic‑to‑plastic transition above ~ 10 GPa, the permanent densification, and the double hysteresis observed in Brillouin light-scattering across compression–decompression cycles. Altogether, the results establish percolation as a physically grounded framework to analyse AATs in disordered materials, align atomistic signatures with experimental data, and provide predictive tools to map structural transitions onto mechanical responses under compression, delivering a transferable approach to other amorphous materials.