Glancing and normal incidence CEMS study of hyperfine interactions in iron-57/gallium arsenide(110) single-crystal films

CEMS spectra were gathered at room temperature for a series of Al overcoated $\sp{57}$Fe/GaAs(110) thin films (32-123 A) utilizing both normal and glancing incidence geometries. This research was conducted in order to investigate the effects which impurity diffusion, chemical reaction, and the Fe/substrate interface have on the hyperfine interactions within this system. All of the films were found to contain Fe nuclei with the same hyperfine field and isomer shift as is found in bulk Fe. Approximately 1-3 ML of paramagnetic Fe were also present, along with Fe atoms belonging to a solid mixture. None of the Mossbauer spectra of the bulk compounds FeAs, Fe$\sb2$As, or FeAs$\sb2$ were observed in the spectra of these films. Both the bulk-like Fe and mixture region were discovered to increase in thickness as the number of deposited Fe layers increased. In addition, the correlation between the average isomer shift and hyperfine field for part of the mixture in the thickest films was in good agreement with that found in dilute Fe-As alloys. This was not the case however, for the thinner films and that part of the mixture made up predominantly of Fe atoms which had lower hyperfine fields. Furthermore, the data revealed that the reduced magnetization previously measured in Fe/GaAs(110) thin films was not due exclusively to a decrease in the magnitude of the moment per Fe atom. Additional spectra were gathered for the 123 A film in applied magnetic fields with two different glancing geometries. The relative line intensities measured with the two geometries were compared at similar fields using a method which also incorporated the normal incidence results. From this comparison, it was discovered that 5.5% of the moment could be modeled oriented normal to the film plane with the rest confined to the film plane. Another model, which allowed for 16% of the moment to be randomly oriented while the remaining fraction was in-plane, also yielded better results than a model which assumed an entirely in-plane moment.