Phase change materials (PCMs) enable the controlled modulation of physical properties through phase transitions induced by external stimuli, including temperature, pressure, irradiation, or electric and magnetic fields. Of particular interest is trititanium pentoxide (${\mathrm{Ti}}_3{\mathrm{O}}_5$), a promising PCM for heat storage, which exhibits a low pressure threshold for the $\lambda$ to $\beta$ phase transition. Although the phase transitions of bulk ${\mathrm{Ti}}_3{\mathrm{O}}_5$ have been extensively studied and well understood, research on its surface properties and associated phase transformations remains limited due to the computational challenge of modeling pressure effects on the surface at the atomic scale. Here we introduce a computational framework that combines a machine-learned interatomic potential trained on high-fidelity density functional theory data with explicit pressure simulation via repulsive slab potentials. We identify a hitherto unknown stable surface reconstruction and a kinetically favorable layer-by-layer transition mechanism. On-the-fly probability enhanced sampling simulations reveal that pressure significantly reduces the free energy barrier, predicting a phase transition at only 700 bar, in good agreement with the experimental value. This work presents a feasible and generally applicable protocol for modeling pressure effects in PCMs, paving the way for improved understanding and application of materials under operational conditions.

Close

PrevNext

View figure in article

• Critical phenomena
• Dynamical phase transitions
• Elastic deformation
• First order phase transitions
• Interface & surface thermodynamics
• Nanocrystals
• Nucleation
• Pressure effects
• Solid-solid transformations
• Structural phase transition
• Functional materials
• Density functional theory
• Machine learning
• Materials modeling
• Molecular dynamics