Pressure-induced structural changes in liquid SiO2 investigated by molecular dynamics and machine learning approaches

Dung Tri Pham, Lan Thi Mai, Hong Van Nguyen,
Pressure-induced structural changes in liquid SiO2 investigated by molecular dynamics and machine learning approaches,
Solid State Communications,
Volume 406,
2025,
116217,
ISSN 0038-1098,
https://doi.org/10.1016/j.ssc.2025.116217.
(https://www.sciencedirect.com/science/article/pii/S0038109825003928)
Abstract: The structural changes of liquid SiO2 under compression pressure are investigated using machine learning algorithms based on molecular dynamics simulation data. In this work, we apply the k-nearest neighbors algorithm to identify differences in Si-O bond distances, thereby classifying the system into three regions with distinct structures and densities at 15 GPa. A density-based spatial clustering algorithm was used to clarify the spatial distribution of these regions, which include high-density, intermediate-density, and low-density regions. The detailed structural analysis results show that these regions exhibit distinct differences in their short-range and intermediate-range order structures. Furthermore, we observed the formation of SiO6 clusters in the high-density region through clustering techniques. The results show that clusters in the high-density region are surrounded by SiO5 units in the intermediate-density region. The computational methods and findings of this study offer a new approach and valuable insights into the structural and property analysis of silica from simulation data.
Keywords: Liquid SiO2; Clustering algorithm; K-nearest neighbors; Data mining