Journal of Applied Crystallography vol:34 Part 3 pages:336-342
The complicated inverse scattering problem of reconstructing depth-dependent lattice parameters from high-resolution X-ray diffraction spectra is analysed by using neural networks. Attention is paid to the practically important case of structural modifications in the near-surface layers of ion-implanted single crystals. The feasibility of a neural network algorithm is assessed on the basis of the performance statistics on a large number of simulated examples. The performance of the method on experimental data is tested using high-resolution X-ray diffraction spectra taken from He-implanted lithium niobate crystals.