Revealing the structure-property relationships of copper alloys with FAGC

TL;DR

This work tackles predicting Cu-Cr-Zr alloy conductivity and hardness from scarce microstructure images by combining EfficientNet-B6 feature extraction with Feature Augmentation on Geodesic Curves (FAGC) in a pre-shape space and a pseudo-labeling strategy. The FAGC pipeline generates augmented features along a Geodesic curve in the pre-shape space, with pseudo-labels assigned via a regressor to expand the training set and improve regression performance. Across regression models, the approach yields high predictive accuracy, notably $R^2=0.978$ for conductivity and $R^2=0.998$ for hardness using a Decision Tree model with about 100 augmented features, and reveals microstructure regions with lower grain/phase boundary density contribute more to conductivity. The method demonstrates data-efficient structure–property mapping for materials design under limited data, and provides actionable insights into how microstructural features translate into performance in Cu-Cr-Zr alloys.

Abstract

Cu-Cr-Zr alloys play a crucial role in electronic devices and the electric power industry, where their electrical conductivity and hardness are of great importance. However, due to the scarcity of available samples, there has been a lack of effective studies exploring the relationship between the microstructural images of Cu-Cr-Zr alloys and their key properties. In this paper, the FAGC feature augmentation method is employed to enhance the microstructural images of Cu-Cr-Zr alloys within a feature space known as the pre-shape space. Pseudo-labels are then constructed to expand the number of training samples. These features are then input into various machine learning models to construct performance prediction models for the alloy. Finally, we validate the impact of different machine learning methods and the number of augmented features on prediction accuracy through experiments. Experimental results demonstrate that our method achieves superior performance in predicting electrical conductivity (\(R^2=0.978\)) and hardness (\(R^2=0.998\)) when using the decision tree classifier with 100 augmented samples. Further analysis reveals that regions with reduced image noise, such as fewer grain or phase boundaries, exhibit higher contributions to electrical conductivity. These findings highlight the potential of the FAGC method in overcoming the challenges of limited image data in materials science, offering a powerful tool for establishing detailed and quantitative relationships between complex microstructures and material properties.

Figures (14)

• Figure 1: An overview of performance prediction for electrical conductivity and hardness of Cu-Cr-Zr based on Geodesic curve feature augmentation.
• Figure 2: Architecture of EfficientNet-B6 for feature extraction. A 2304-dimensional feature vector is obtained after Global Average Pooling and mapped to a 1-dimensional regression output(conductivity or hardness).
• Figure 3: The Feature Extraction and Feature Augmentation process. Cu-Cr-Zr images are projected into the pre-shape space, and FAGC process is performed to generate samples in the pre-shape space.
• Figure 4: The process of assigning pseudo-labels to generated features via FAGC in regression models.
• Figure 5: Microstructural image samples of Cu-Cr-Zr alloys with corresponding electrical conductivity and hardness.