The solution of crystal structures from thin films is a contemporary problem in crystallography. A combined experimental / theoretical approach is used where the first step is a grazing incidence x-ray diffraction experiment. The diffraction patterns have to be indexed, which requires the assignment of Laue indices to individual Bragg peaks. The result of the indexation procedure is the geometry of the crystallographic unit cell in terms of lattice constants a, b, c, and the angles α , β , γ. For that purpose, a computer code has to be developed which performs an indexation procedure for grazing incidence diffraction pattern.
Central question of the master thesis:
How is it possible to index a diffraction pattern which arises from two different types of crystal lattices?
There are two avenues to answer this question:
A) Separation of two crystal lattices by neuronal networks. A large number (> several thousand) of diffraction patterns will be calculated and combined into a superposition of two crystal lattices. Neuronal networks will use these examples for for training, and then be applied to (experimentally obtained) diffraction patterns
B) indexation of a single phase by using a specular diffraction peak. Result: 3 real values and two integers besides the individual Laue Indices for each peak Boundary conditions: restricted volume, restricted lattice constants, find solutions for a reduced number of reflections, neglecting of one or more peaks which cannot be explained. In a subsequent step indexation of a single phase without using a specular diffraction peak will be performed.
Contact: Oliver Hofmann (email@example.com) / Roland Resel (firstname.lastname@example.org)