Rotation of crystal seed during early stages of growth reveals the anisotropy of glass matrix

TL;DR

This work demonstrates that a crystal seed embedded in a glass can rotate during the early stages of seeded crystal growth due to nonuniform, anisotropic forces imposed by the surrounding glass at the glass–seed interface. Using molecular dynamics simulations of LiNbO3 and LiNbO3-SiO2 glasses with a fixed seed, the authors link seed rotation to the heterogeneity of the interface and show that rotation magnitude increases with temperature and varies with the glass structure, not solely with density changes. They further reveal that adding SiO2 slows crystal growth but has limited impact on rotation, while Si dynamics during SCG indicate partial Si incorporation into the growing Nb lattice and substantial Si expulsion into the surrounding glass. Together, these findings challenge the view of glass as isotropic during early crystallization and offer insights to tailor glass-ceramic microstructures via interface engineering and composition control.

Abstract

Rotation of crystal seed during the early stages of growth in a glass matrix has been observed due to some torque, contradicting the expectations from the isotropic, uniform structure of the surrounding amorphous matrix. We establish an atomistic origin of this new phenomenon from molecular dynamics simulations using LiNbO3 and LiNbO3-SiO2 glasses as model systems. Effectively, it arises due to non-uniform forces on the seed from the surrounding glass, which appears inhomogeneous and anisotropic on the scale of glass-crystal interface. The seeded crystal growth (SCG) at higher temperatures amplifies this effect due to enhanced atomic dynamics. Silica, when added to LiNbO3 glass, reduces the crystal growth rate due to increased viscosity and restricted atomic mobility across the growth interface, but has minimal effect on the crystal rotation. These findings challenge a general assumption that glass is an isotropic material, especially during the early stage of its crystallization, and provide insights for tailoring the microstructure of widely used glass-ceramics.

Figures (9)

• Figure 1: Time evolution of temperature and potential energy during the melt quench cycle for all models created.
• Figure 2: Structure of a representative seeded glass model (left) and ML algorithm predicted crystal-like atoms at the end of the MQ cycle.
• Figure 3: Radial distribution function of seeded glass models after the melt-quench cycle (bold lines) and after SCG (dashed lines)(left panel). Evolution of the crystal fraction (X) during SCG in different models (right panel).
• Figure 4: Comparison of the time evolution of crystal rotation across various LNO models for NPT (left) and NVT (right) ensemble during SCG.
• Figure 5: Comparison of the evolution of crystal rotation under the NPT ensemble at various temperatures for Model VII.