The present work describes an intellectual computational approach to X-ray resonant reflectometry – a synchrotron-based method aimed at non-destructive characterization of epitaxial nanoscale multilayers with low optical contrast between the sublayer materials. Resonant reflectance from specially designed bilayer structures was measured as a function of grazing angle and photon energy across the L absorption edges of rare earth elements and then modeled with the aid of a specially developed software utilizing OpenCL high speed computations performed on a graphical processing unit. The map fitting was carried out in the blind mode in which the spectral shapes of the optical constants of the sublayers were reconstructed by the fitting routine without initial knowledge of the chemical composition and without using any reference spectra. The model uses stepped Lorentzian line shape for the imaginary part of the scattering length density and the Kramers-Kronig consistent line shape for the corresponding real part. Epitaxially grown Eu2O3 / Gd3Ga5O12 bilayers were used as a test bench system to demonstrate the benefits of the proposed 2D mapping technique over conventional X-ray reflectometry. To increase reliability, the map fitting was carried out simultaneously at the absorption edges of two different chemical elements. The derived shapes and positions of the absorption peaks were used to uniquely identify rare earth elements within the corresponding sublayers and provide information on their oxidation states. The method was experimentally tested on the bilayers having different material order to show that not only the chemical composition but also the layer sequence can be effectively reconstructed in the automatic way from the 2D resonant reflectance maps.
Read full abstract