A quantitatively constrained framework for defect identification in oxides: application to Cr³⁺ centers in PbTiO₃

Area of Science:

• Materials Science
• Solid State Physics
• Computational Chemistry

Background:

• Identifying paramagnetic defect centers in ferroelectric oxides like PbTiO₃ is complex due to vast structural possibilities and sensitive spectroscopic data.
• Chromium (Cr³⁺) defects in these materials significantly influence their properties, necessitating accurate identification.

Purpose of the Study:

• To develop a unified framework for assigning Cr³⁺ defect centers in PbTiO₃.
• To combine machine learning (ML) for structural optimization with constrained spectroscopic analysis (zero-field splitting) for defect identification.

Main Methods:

• Utilized ML-based structural optimization (CHGNet) to relax various Cr defect configurations.
• Employed superposition model calculations for zero-field splitting (ZFS) parameters, constrained by experimental electron paramagnetic resonance (EPR) data.
• Developed objective criteria integrating ZFS mismatches, parameter stability, and structural metrics to validate defect models.

Main Results:

• Successfully assigned experimental EPR centers (C1-C4) to a subset of plausible Cr³⁺ defect models in PbTiO₃.
• Identified the dominant bulk center (C1) with low-distortion Ti-site substitution and long-range interactions.
• Suggested a size-dependent, potentially surface-related origin for the nanopowder-specific center (C4), as bulk models were inconsistent.

Conclusions:

• The developed methodology provides a robust and transferable approach for identifying defects in functional oxides.
• Integration of ML structural fidelity with physically constrained spectroscopic modeling enables accurate defect assignment.
• The framework offers objective criteria for discriminating between valid and invalid defect configurations.