Optimized Current Density Reconstruction from Widefield Quantum Diamond Magnetic Field Maps

Abstract

Quantum Diamond Microscopy using Nitrogen-Vacancy (NV) defects in diamond crystals has enabled the magnetic field imaging of a wide variety of nanoscale current profiles. Intimately linked with the imaging process is the problem of reconstructing the current density, which provides critical insight into the structure under study. This manifests as a non-trivial inverse problem of current reconstruction from noisy data, typically conducted via Fourier-based approaches. Learning algorithms and Bayesian methods have been proposed as novel alternatives for inference-based reconstructions. We study the applicability of Fourier-based and Bayesian methods for reconstructing two-dimensional current density maps from magnetic field images obtained from NV imaging. We discuss extensive numerical simulations to elucidate the performance of the reconstruction algorithms in various parameter regimes, and further validate our analysis via performing reconstructions on experimental data. Finally, we examine parameter regimes that favor specific reconstruction algorithms and provide an empirical approach for selecting regularization in Bayesian methods.