Finite-Temperature Ferroelectric Phase Transitions from Machine-Learned Force Fields

TL;DR

This study demonstrates that machine-learned force fields, trained on ground state structures, can qualitatively predict finite-temperature ferroelectric phase transitions in oxides with high accuracy, offering a computationally efficient alternative to first-principles methods.

Contribution

The paper introduces a method using on-the-fly trained MLFFs to simulate ferroelectric phase transitions, showing qualitative agreement with experiments across multiple oxides.

Findings

MLFFs predict structural phases and transitions qualitatively.

PBEsol functional yields more robust results.

MD simulations reproduce experimental transition behaviors.

Abstract

Simulating finite temperature phase transitions from first-principles is computationally challenging. Recently, molecular dynamics (MD) simulations using machine-learned force fields (MLFFs) have opened a new avenue for finite-temperature calculations with near-first-principles accuracy. Here we use MLFFs, generated using on-the-fly training, to investigate structural phase transitions in four of the most well-studied ferroelectric oxides; BaTiO$_3$, PbTiO$_3$, LiNbO$_3$ and BiFeO$_3$. Only using the 0 K ground state structure as input for the training, the resulting MLFFs can qualitatively predict all the main structural phases and phase transitions, while the quantitative results are sensitive to the choice of exchange correlation functional with PBEsol found to be more robust than LDA and r$^2$SCAN. MD simulations also reproduce the experimentally observed order-disorder character of Ti displacements in BaTiO$_3$, the abrupt first order transitions of BiFeO$_3$ and PbTiO$_3$, and the mixed order-disorder and displacive character of the ferroelectric transition in LiNbO$_3$. Finally, we discuss the potential and limitations of using MLFFs for simulating ferroelectric phase transitions.