Considerations for electromagnetic simulations for a quantitative correlation of optical spectroscopy and electron tomography of plasmonic nanoparticles

Abstract

The optical cross sections of plasmonic nanoparticles are intricately linked to their morphologies. Accurately capturing this link could allow determination of particles' shapes from their optical cross sections alone. Electromagnetic simulations bridge morphology and optical properties, provided they are sufficiently accurate. This study examines key factors affecting simulation precision, comparing common methods and detailing the impacts of meshing accuracy, dielectric function selection, and substrate inclusion within the boundary element method. To support the method's complex parameterization, we develop a workflow incorporating reconstruction, meshing, and mesh simplification, to enable the use of electron tomography data. We analyze how choices of reconstruction algorithm and image segmentation affect simulated optical cross sections, relating these to shape errors minimized during data processing. Optimal results are obtained using the total variation minimization (TVM) reconstruction method with Otsu thresholding and light smoothing, ensuring reliable, watertight surface meshes through the marching cubes algorithm, even for complex shapes.