Characterization and discrimination of periodic nanostructures with scanning-free GEXRF

As nanostructures in the semiconductor industry become smaller and more complex, non-destructive characterization methods capable of measuring buried domains become crucial. Grazing emission x-ray fluorescence (GEXRF) spectroscopy is a measurement technique capable of resolving nanometer-sized features of buried nanostructures while providing information about the sample's elemental distribution. In this work, a study was conducted to realistically assess the uncertainties of this method, considering correlations between geometric parameters. Furthermore, we showed strategies to effectively reduce the measurement time in GEXRF experiments by applying state-of-the-art single photon evaluation and machine learning denoising techniques for two-dimensional detectors. The study was performed on two different sample positions on a HfO₂/TiO₂nanograting, where the GEXRF method was able to resolve geometric differences between them. Based on a finite element method model of the nanograting, the expected fluorescence intensities can be simulated, from which the nanostructure's geometry can be reconstructed. The reconstructed geometric shapes show good agreement with atomic force microscope and transmission electron microscope measurements, highlighting the method's capability for investigating samples within the nanometer regime.