This research proposes a physics-informed deep learning framework for the inverse design of novel solid-state ferroelectric perovskite materials. The approach integrates a Crystal Diffusion Variational Autoencoder, a Crystal Transformer Graph Neural Network, a multi-objective optimization module, and an active learning loop with DFT validation to address data scarcity, composition-structure-property relationships, and synthesizability constraints.
Key findings
Proposes a comprehensive generative design framework for ferroelectric perovskite discovery.
Integrates crystal diffusion models, graph neural networks, and physics-informed machine learning for efficient inverse design.
Addresses challenges in materials discovery including data scarcity and composition-structure-property relationships.
Limitations & open questions
The framework's effectiveness is contingent upon the quality of the initial data and the accuracy of the DFT validation.