Measuring material hardness without contact

Different heat treatment processes used during the manufacturing of metallics result in varying hardness, a property that affects a material’s performance. Researchers and engineers need to be able to measure material hardness for industrial, aerospace, and nuclear power applications. However, conventional techniques for characterizing hardness, such as indentation testing, often cause surface damage.

Though ultrasonic techniques can measure hardness nondestructively and with high precision, they require contact between the probe and sample, which limits their application. Wen et al. developed a contactless approach for characterization of material hardness. This approach overcomes the limitations of conventional techniques by combining electromagnetic acoustic resonance (EMAR), a way to generate and receive acoustic waves without direct contact, with a one-dimensional convolutional neural network (1D-CNN), a type of deep learning model that extracts features from measured signals.

The authors used their method to successfully determine changes in hardness in GCr15-bearing steel induced by different heat treatments. Though variations in sample surface roughness and thickness typically cause interference in EMAR signals, the feature extraction capabilities of 1D-CNN addressed this challenge to accurately measure material hardness.

The high temperature tolerance of the transducer that transmits and receives the EMAR signals suggests this method could be used to nondestructively characterize industrial components, and the precision of this method may suit aerospace applications.

“Considering the accuracy and contactless capability of the proposed approach, it is of great potential for online monitoring of material degradation and residual life assessment of industrial structures under high-temperature service conditions,” said author Weibin Li. “The promotion of this approach on a wider range of ferromagnetic or non-ferromagnetic metallic materials is another interesting plan.”

Source: “Non-contact characterization of material hardness by deep learning-assisted electromagnetic acoustic resonance method,” by Jinshan Wen, Mingxi Deng, and Weibin Li, Applied Physics Letters (2025). The article can be accessed at https://doi.org/10.1063/5.0265254 .