講演情報
[9a-N304-6]Electric-Field-Induced Switching Behavior in Antiferroelectric PbZrO3 via Machine Learning Potential Molecular Dynamics
〇ChihLun Hsu1, Ryotaro Sahashi1, Po-Yen Chen1, Teruyasu Mizoguchi1,2 (1.Univ. of Tokyo Eng., 2.Univ. of Tokyo IIS)
キーワード：
lead zirconate、antiferroelectrics、Machine Learning Potentials
Antiferroelectric (AFE) materials like PbZrO3 (PZO) are promising for high-density energy storage due to field-induced phase transitions. However, their atomic-scale switching mechanisms remain elusive. Here, we investigate this process using large-scale molecular dynamics (MD) simulations. To overcome computational bottlenecks, we developed a PZO-specific MACEField machine learning potential, fine-tuned from the MACE-MP-0 foundation model. We constructed a on-the-fly learning database using the Pbam, Pm3m, Pnma, Pba2, and Amm2 phases, calculating reference properties via PBEsol function and density functional perturbation theory (DFPT). Using this model, electric-field-induced MD simulations on a 1080-atom supercell successfully reproduced PZO's characteristic double hysteresis loop. By tracking polarization vectors from Pb displacements, we clarified the atomic-scale dynamics governing the AFE-to-FE switching process.