A new approach to automated structure optimization in LEED intensity analysis

Abstract

We present a new structural search concept for low-energy electron diffraction (LEED) intensity analyses. It combines the advantages of an efficient automated optimization scheme with rapid access to model intensity data by tensor LEED. The geometrical, chemical and thermal versions of the latter method are employed to calculate the diffraction amplitudes for structures deviating from a certain preselected reference by atomic shifts, chemical occupation of lattice sites and both isotropic and anisotropic atomic vibrations. The automated optimization of the theory-experiment fit results from a modified random sampling algorithm, which scales as N2.5 with N the number of parameters, turning out to be a good compromise between global and local search methods. Using simulated intensity spectra for the Ir(110)-(2 × 1) missing-row reconstructed surface as pseudo-experimental data, we test and demonstrate the reliability of the new procedure and its efficiency. Also, we successfully apply the algorithm to real experimental data of two different surface phases of FeAl(100) and to data of Mo0.95Re0.05(100)-c(2 × 2)-C as examples for multi-parameter fits.