Multi-Axis Acquisition Schemes for Scalar and Vector Electron Tomography

Electron tomography (ET) allows us to probe nanoscale morphologies and compositions in 3D with high spatial resolution In a typical scalar ET experiment, this is achieved by rotating the sample about a single axis and acquiring projection images at regular intervals, which can then be reconstructed to form a 3D representation. In magnetic vector ET it is common to rotate about a second perpendicular axis to allow for reconstruction of the magnetic field B, since each projection is insensitive to the component of B parallel to the beam direction However, this regular sampling is not necessarily the optimum way to carry out an ET experiment, with reasons to explore alternative acquisition schemes including: anisotropic distributions of sample complexity; the need to reconstruct three perpendicular components in vector ET; and the fact that near-random sampling approaches may benefit reconstructions using compressed sensing regularisation. In this work we first build upon the well-known Crowther criterion to create a directional, vector-resolved resolution (DVR) prediction metric, and subsequently compare the performance of different acquisition schemes for both scalar and vector simulated ET experiments. Our results show that the DVR metric helps when designing tomography experiments, and that particularly for magnetic vector ET, multi-axis acquisition schemes may improve reconstruction quality.