Linking three-dimensional microstructure and nanoindentation hardness using machine learning

Understanding the relationship between microstructure and mechanical properties is crucial for the development of advanced materials. Unlike conventional two-dimensional (2D) characterization, three-dimensional (3D) techniques enable a more comprehensive understanding by revealing volumetric phase morphologies, spatial distributions, and interphase connectivity. This study proposes a workflow that integrates 3D visualization with nanoindentation to establish microstructure-mechanical property correlations. Two V-Si-B alloys were investigated. Automated serial sectioning was employed to reconstruct 3D microstructures for quantitative analysis. A machine learning (ML) model, trained on experimental nanoindentation data, was developed to predict localized hardness from microstructural images and to generate a 3D hardness distribution map. By correlating the 3D microstructural features with spatially resolved hardness, the approach uncovers how phase morphology and spatial arrangement govern overall material performance. This ML-based approach offers a data-driven pathway to link microstructure and mechanical property, supporting both material design and simulation-based studies.