An approach for measuring the magnetic susceptibility of diamond using a tunnel magnetoresistance (TMR) sensor array

Low-frequency magnetic susceptibility is a critical parameter for evaluating the magnetic properties of diamonds and is closely related to their physical characteristics, such as compressive strength and thermal stability. To overcome the limitations of conventional susceptibility measurement methods, this study proposes a novel approach for quantifying diamond magnetic susceptibility using a tunnel magnetoresistance (TMR) sensor array. To address the ill-posed nature of image reconstruction and the limited accuracy of linear mapping, two key optimization strategies are introduced: (1) a sparse autoencoder-based reconstruction algorithm (SAE-EMT) that incorporates nonlinear prior constraints to enhance image reconstruction accuracy, and (2) a gradient-optimized mapping strategy that enables accurate conversion from relative susceptibility distributions to absolute susceptibility values. Experimental results demonstrate that the SAE-EMT algorithm outperforms conventional Tikhonov and Landweber methods in terms of image quality and robustness, maintaining high reconstruction accuracy even under noisy conditions. Moreover, the gradient-optimized mapping strategy reduces the relative error by 34% and improves the signal-to-noise ratio by 6 dB, while more accurately capturing susceptibility variations and reducing the gradient magnitude similarity deviation by 40%.