Genetic algorithm for solution of the inverse problem in high-resolution X-ray reflectometry

The method of high-resolution X-ray reflectometry with application of the genetic Differential Evolution algorithm for global minimization of the error function is developed. The optimization of the computational process is performed by combining the genetic algorithm with the Levenberg-Marquardt method. This optimization reduced the computational time by a factor of 3-4. The structural properties of multilayer systems AlxGa1-x As/GaAs/AlxGa1-x As (x similar to 0.2) grown on GaAs(001) are investigated. The density profiles along the normal to the surface are obtained and the sizes of the transition layers are determined with a resolution of 0.1-0.2 nm.