Table of Contents

Lattice Mismatched Oxide Heteroepitaxy – a -Fe2O3/a -Al2O3(0001) S. I. Yi, Y. Liang, S. Thevuthasan, D. E. McCready, and S. A. Chambers Supported by OBER/OBES EMSP and EMSL Operations. A significant effort has been made within the Interfacial and Processing Sciences directorate of the EMSL to design, procure, commission, and utilize state-of-the-art epitaxial growth systems for the purpose of synthesizing well-defined oxide surfaces. Such surfaces are exceedingly important in the fields of environmental surface science and oxide catalysis. Their availability opens entirely new avenues of research in that crystalline oxides can be synthesized as thin films that are not available in bulk form. There are many external collaborators and internal EMSL scientists who are taking advantage of this capability to do research that would not otherwise be possible. However, obtaining the desired goal of atomically flat, single-crystal surfaces of unique surface composition and structure, for a wide range of materials, is not at all easy. The biggest problem is that of lattice mismatch. There are relatively few readily available, inexpensive substrates of adequate surface quality on which these oxides can be grown. As a result, the lattice mismatch between the oxide of interest and the substrate is often rather large (a few percent or more). Lattice mismatch causes strain energy to accumulate in the film up to some critical thickness, above which the film undergoes substantial structural and/or morphological changes to relieve the strain. The effect appears to be most severe when the in-plane lattice parameter of the film material exceeds that of the substrate. In this case, the film material is in lateral compression on the substrate below the critical thickness, and the structural changes are significant when they occur. The heteroepitaxial system most thoroughly studied thus far in the EMSL is a -Fe2O3/a -Al2O3, in large part because of the broad importance of a -Fe2O3 (hematite) in geochemistry, catalysis, corrosion and magnetism. It was shown early on that a -Fe2O3 crystalline could be grown on a -Al2O3(0001) by molecular beam epitaxy with some form of activated oxygen (Fujii et al. 1996; Kim et al. 1997), despite the compressive 5.8% lattice mismatch. The feasibility of growing hematite on sapphire in other orientations was subsequently demonstrated (Gao et al. 1997). More recently, we have investigated interface formation and the associated effects on film structure and morphology in much more detail (Yi et al. 1999a, 1999b, 1999c). While it is relatively easy to grow oriented, nanocrystalline hematite on sapphire(0001), it is much more difficult to generate truly monocrystalline hematite with a flat surface morphology and large terraces. Lattice relaxation after 2-3 ML of growth is accompanied by 3-D island or nanocrystal formation as the primary mechanism for strain relief. These nanocrystals then gradually coalesce into what looks by all probes except high-resolution x-ray diffraction to be a single crystal with a reasonably flat surface. In Figure 2.5 we show an STM image of one such film that was doped with Fe(II) to enhance conductivity so that low but stable tunneling currents could be obtained for imaging (Chambers and Yi 1999). The film thickness was 2500 Å. Also shown is a line scan taken to determine the step height distribution. The typical terrace width is a few tens of Å, and terraces are separated by ~2-Å-high steps. This step height is the difference between structurally equivalent planes in the corundum(0001) structure. Thus, the surface consists of a single termination. X-ray photoelectron diffraction results established that the terminating plane is that of a single 1/3 ML of Fe, which is the only autocompensated and, therefore, stable termination of corundum(0001) (Chambers and Yi 1999). Surfaces such as that shown in Figure 2.5 represent the best we have been able to generate thus far. We have found that it is critical that the first several tens of Å be grown at an exceedingly slow rate (~1 Å/min) (Yi et al. 1999a, 1999b). Doing so produces a heavily-strained wetting layer of a -Fe2O3 that is buckled along the direction. This strained layer then transforms to 3-D nanocrystals above a coverage of a few ML. A higher growth rate (such as ~0.1 Å/sec) kinetically impedes the critically important transition from strained layer-by-layer to relaxed 3-D island formation that leads to the nucleation of well-ordered epitaxial a -Fe2O3(0001). Instead, a g -Fe2O3–like epilayer is nucleated which exhibits domains with 180° rotational twinning and surface orientations that deviate substantially from (111) (Yi et al. 1999c). The formation of single-phase a -Fe2O3 never occurs once this g -Fe2O3–like layer has nucleated. Even when all other structural probes available to us (RHEED, LEED, XPD, and RBS/channeling) tell us that we have made high-quality crystalline films, high-resolution XRD reveals that the mosaic spread is sufficiently large that certain critical experiments, such a crystal truncation rod (CTR) diffraction, will not be possible with the current level of film quality. CTR is one of the few surface-structural techniques that allows the liquid/solid interface to be probed. As it is currently, the nanocrystals formed early in the growth never fully coalesce to make what is truly a single crystal, even under optimized growth conditions. So, domains of nearly identically oriented a -Fe2O3 grow together in the film. Figure 2.5. STM image and line scan for 2500 Å a -Fe2O3/a -Al2O3(0001) doped with Fe(II) for enhanced conductivity.. Early on, we did a few experiments using an ~200-Å-thick interlayer of Pt between the sapphire substrate and the hematite film to take up some of the strain (Gao and Chambers 1997). The in-plane lattice parameter of Pt(111) is the same as that of a -Al2O3(0001), and Pt grows epitaxially on a -Al2O3(0001). Being more ductile that a -Al2O3, Pt is expected to absorb some of the strain energy, and thereby lower the strain in the film. Inadequate time and funding at that time prevented us from doing a more thorough study of this system. Over the next year, we plan to revisit this topic and examine more carefully the effect of Pt interlayers on the quality of a -Fe2O3(0001) films. References Chambers, S. A., and S. I. Yi, Surf. Sci. 439, L785 (1999). Fujii, T., D. Alders, F. C. Voogt, T. Hibma, B. T. Thole, and G. Sawatsky, Surf. Sci. 366, 579 (1996). Gao, Y., and S. A. Chambers, J. Crystal Growth 174, 446 (1997). Gao, Y., Y. J. Kim, G. Bai, and S. A. Chambers, J. Vac. Sci. Technol. A 15, 332 (1997). Kim, Y. J., Y. Gao, and S. A. Chambers, Surf. Sci. 371, 358 (1997). Yi, S. I., Y. Liang, and S. A. Chambers, J. Vac. Sci. Technol. A 17, 1737 (1999a). Yi, S. I., Y. Liang, S. Thevuthasan, and S. A. Chambers, Surf. Sci. to appear (1999b). Yi, S. I., S. Thevuthasan, D. E. McCready, and S. A. Chambers, Thin Solid Films submitted (1999c). William R. Wiley Environmental Molecular Sciences Laboratory Feedback: webmaster@emsl.pnl.gov Revised: June 12, 2001 Security & Privacy PNNL-13147