Domain Switching on the Pareto Front: Multi-Objective Deep Kernel Learning in Automated Piezoresponse Force Microscopy

2025-06-09

AI Summary

Systematic characterization of ferroelectric domain switching remains challenging due to its strong dependence on complex, localized microstructural features. Method: This study introduces a multi-objective deep kernel learning framework that automatically extracts physical correlations between domain wall configurations and switching dynamics from high-resolution piezoresponse force microscopy (PFM) images. It innovatively integrates Pareto frontier optimization into PFM-based active learning, enabling the first quantitative mapping of abstract physical rewards—such as switchability and domain symmetry—to interpretable microstructural descriptors (e.g., domain configuration, boundary proximity). Results: Coupled with multi-objective Bayesian optimization, microstructure-aware reward modeling, and automated closed-loop PFM experimentation, the framework significantly improves domain-switching prediction accuracy. It quantitatively uncovers the synergistic regulation of polarization reversal kinetics by geometric curvature of curved domain walls and local defect distributions, thereby advancing mechanism-driven, high-throughput ferroelectric materials design.

Technology Category

• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Search and Optimization: Metareasoning and Metaheuristics
• Constraint Satisfaction and Optimization: Constraint Learning and Acquisition

Application Category

• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Semantics and Knowledge: Methods, algorithms and applications for the development of semantic models, knowledge graphs and other forms of structured data models with machine-interpretable semantics
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

Ferroelectric polarization switching underpins the functional performance of a wide range of materials and devices, yet its dependence on complex local microstructural features renders systematic exploration by manual or grid-based spectroscopic measurements impractical. Here, we introduce a multi-objective kernel-learning workflow that infers the microstructural rules governing switching behavior directly from high-resolution imaging data. Applied to automated piezoresponse force microscopy (PFM) experiments, our framework efficiently identifies the key relationships between domain-wall configurations and local switching kinetics, revealing how specific wall geometries and defect distributions modulate polarization reversal. Post-experiment analysis projects abstract reward functions, such as switching ease and domain symmetry, onto physically interpretable descriptors including domain configuration and proximity to boundaries. This enables not only high-throughput active learning, but also mechanistic insight into the microstructural control of switching phenomena. While demonstrated for ferroelectric domain switching, our approach provides a powerful, generalizable tool for navigating complex, non-differentiable design spaces, from structure-property correlations in molecular discovery to combinatorial optimization across diverse imaging modalities.

Problem

• Understand microstructural rules governing ferroelectric polarization switching
• Identify relationships between domain-wall configurations and switching kinetics
• Provide mechanistic insight into microstructural control of switching phenomena

Innovation

• Multi-objective kernel-learning workflow for PFM
• Infers microstructural rules from imaging data
• Projects reward functions onto interpretable descriptors