Structure of Fe 3 Si Õ GaAs(001) epitaxial films from x-ray crystal truncation rods Vladimir M. Kaganer, Bernd Jenichen, Roman Shayduk, and Wolfgang Braun Paul-Drude-Institut für Festkörperelektronik, Hausvogteiplatz 5-7, D-10117 Berlin, Germany Received 21 November 2007; published 19 March 2008 Thin 10–15 nm thick Fe 3 Si films are grown on GaAs001 by molecular beam epitaxy and studied in situ by grazing incidence x-ray diffraction. We find two interfacial structures in different samples, with the first atomic layer of Fe 3 Si consisting of either iron atoms only or both Fe and Si atoms. In both cases, the top atomic layer at the surface contains both Fe and Si atoms. The films are fully ordered, except 1 or 2 monolayers at the surface, where Fe and Si atoms within one and the same atomic layer are intermixed. DOI: 10.1103/PhysRevB.77.125325 PACS numbers: 68.35.-p, 61.05.cp, 68.35.Ct I. INTRODUCTION Novel electronic and spintronic device concepts require various combinations of metals, semiconductors, magnetic materials, insulators, etc., with highly perfect interfaces. Ma- terials with different crystal structures and bonding can be grown epitaxially on each other if their lattices are appropri- ately matched. 1 X-ray diffraction is a powerful tool to study both the structure of an epitaxial film and its arrangement on a substrate, thanks to the interference between waves scat- tered by the film and the substrate. Fe 3 Si epitaxial films on a GaAs substrate, studied in the present work, are a combination of a ferromagnetic film, with Curie temperature well above room temperature, and a semiconductor. Such a system can be used to inject a spin- oriented electrical current into the semiconductor. 2 An ideal lattice match is achieved by varying the Fe and Si deposition fluxes, close to the stoichiometric composition. 3 The stoichi- ometric films possess the smallest sheet resistance. 4 A per- fectly coherent dislocation-free interface is observed by transmission electron microscopy. 5,6 A thin crystalline film is a planar object with the scatter- ing pattern consisting of lines called crystal truncation rods CTRs normal to the interface. The intensity distribution along a CTR results from the interference of the waves scattered by both crystal lattices and hence is highly sensitive to the relative positions of the atoms in crystals. The sensitivity of the CTR scattering to interface structures was first demonstrated by Robinson et al. 7,8 in CTR studies of Si111 / SiO 2 and NiSi 2 interfaces. CTR measure- ments have since been used to study interfacial structures of various lattice matching epitaxial systems, such as CaF 2 / Si111, 9–12 CaSrF 2 / GaAs111, 13,14 Ge  layers on Si001, 15–17 Pd / MgO001, 18 and several semiconductor heterostructures. 19–21 Still, x-ray diffraction is much more rarely applied to study interfaces, as compared to surfaces. Epitaxial films with thicknesses up to several tens of monolayers, the subject of our study, lie in between two well established fields of research. On one end, crystalline sur- faces are commonly analyzed in the kinematical single scat- tering approximation. 22,23 On the other end, thicker films and multilayers are studied with dynamical diffraction theory. 24 In surface structure analysis, the regions of strong scattering close to the bulk Bragg reflections, where dynami- cal calculations are mandatory, are excluded from the analy- sis. The measured intensity is scaled to the calculated one by using an arbitrary fit parameter. On the other hand, dynami- cal calculations are very accurate at the Bragg peak but re- stricted to the close vicinity of the peak, since the two-beam dynamical theory loses its applicability away from the Bragg peak. We have used dynamical calculations to characterize the order in the Fe 3 Si / GaAs001 films. 25 The films were several times thicker, compared to the ones used in the present study, and the vicinity of the Bragg peaks contained all necessary information. The problem of a dynamical calculation of the diffracted intensity in a wide wave vector range has been the subject of a number of investigations. 26–35 Recently, we have shown 36 that the dynamical calculation can be extended to the whole CTR by summing up the amplitudes of the diffracted waves of the two-beam diffraction problems for all Bragg reflec- tions along the CTR. Dynamical and kinematical scattering intensities quantitatively agree everywhere except in the vi- cinity of the Bragg peaks, where the kinematical intensity diverges. The dynamical calculation gives, within the Darwin width, a reflectivity close to 1 and somewhat smaller than 1, which provides an absolute scale for the measured inten- sity. The absolute intensity is especially important if the structure factors of the film are not known in advance. Fe 3 Si is an example: if the long-range order of the Fe and Si sub- lattices is disturbed, Fe and Si atoms intermix producing an- tisite defects, and the structure factors of the superstructure reflections decrease or may even vanish. The dynamical Bragg reflections, independent of the long-range order in the film, provide a reference to obtain the structure factors of a partially ordered film. By comparing the measured CTRs with the calculated ones, we find that the Fe 3 Si lattice takes two out of possible four high-symmetry positions with respect to the GaAs lat- tice. The first Fe 3 Si atomic layer at the interface contains only Fe atoms in one case and both Fe and Si atoms in the other. These two epitaxial positions are realized in different samples. We find that the Fe 3 Si film exhibits full long-range order in the Fe and Si sublattices, except for atomic layers immediately adjacent to either the surface or the interface, where disorder is observed. The fits imply a relaxation of 0.2 Å of the Fe 3 Si film toward the substrate. II. EXPERIMENT Fe 3 Si films were grown by molecular beam epitaxy MBE on GaAs001 substrates in an MBE chamber inside PHYSICAL REVIEW B 77, 125325 2008 1098-0121/2008/7712/1253258 ©2008 The American Physical Society 125325-1