WRAP: A wavelet-regularised reconstruction algorithm for magnetic vector electron tomography

Abstract

Magnetic vector electron tomography (VET) is a promising technique that enables better understanding of micro- and nano-magnetic phenomena through the reconstruction of 3D magnetic fields at high spatial resolution. Here we introduce WRAP (Wavelet Regularised A Program), a reconstruction algorithm for magnetic VET that directly reconstructs the magnetic vector potential A using a compressed sensing framework which regularises for sparsity in the wavelet domain. We demonstrate that using WRAP leads to a significant increase in the fidelity of the 3D reconstruction and is especially robust when dealing with very limited data; using datasets simulated with realistic noise, we compare WRAP to a conventional reconstruction algorithm and find an improvement of ca. 60% when averaged over several performance metrics. Moreover, we further validate WRAP's performance on experimental electron holography data, revealing the detailed magnetism of vortex states in a CuCo nanowire. We believe WRAP represents a major step forward in the development of magnetic VET as a tool for probing magnetism at the nanoscale.