Generative inverse design for microstructure control in precursors for Li- and Mn-rich layered-oxide cathodes

Area of Science:

• Materials Science
• Chemical Engineering
• Artificial Intelligence

Background:

• Material microstructure significantly impacts performance but is challenging to control.
• Current methods lack efficient quantification, prediction, and optimization of microstructures.

Purpose of the Study:

• To create an AI-driven framework for predictive design and optimization of battery cathode precursor synthesis.
• To enable microstructural morphology as a controllable design objective.

Main Methods:

• Integrated diffusion-based image generation, quantitative image analysis, and particle swarm optimization (PSO).
• Extracted morphological descriptors (texture, sphericity, particle size) from SEM images.
• Developed a closed-loop system for image-centric material design.

Main Results:

• The framework accurately predicts SEM morphologies based on co-precipitation conditions (time, concentration, pH).
• Experimental validation confirmed optimization yields target Li- and Mn-rich (LMR) layered oxide morphologies.
• High agreement between predicted and synthesized structures was achieved.

Conclusions:

• The AI framework offers a practical, data-driven strategy for materials design.
• Enables both forward prediction and inverse design of synthesis parameters.
• Paves the way for autonomous, image-guided microstructure engineering.