Table of Contents

Development of Novel Photocatalysts Using Interfacial Engineering Y. Liang, J. L. Daschbach, S. A. Chambers, Y. Su, Y. Wang, H. Luo,(a.b) and S. Gan(c) Supported by PNNL Level-V Laboratory Directed Research and Development. (a) EMSL User. (b) State University of New York-Buffalo. (c) Postdoctoral Research Fellow. The focus of this project was to use a novel approach to enhance the performance of photocatalysts. The project was designed to address two issues critical to photocatalyst development: 1) increasing the efficiency of photocatalysts, 2) enabling the use of visible light to facilitate reactions on catalysts. We are using interfacial engineering to tailor the catalysts to simultaneously achieve these properties. During FY 1999, the experimental focus was to develop oxide-based materials systems suitable to spatial charge separation and increased optical adsorption. We have demonstrated that synthesis of engineered oxide materials with relevant interfacial properties and stability can be accomplished. The results from two different oxide systems are discussed here. 1. Noncommutative Band Offset at a -Cr2O3/a -Fe2O3(0001) Heterojunctions S. A. Chambers, Y. Liang, and Y. Gao III-V and II-VI semiconductor heterojunctions have proven to be useful materials systems in a wide range of solid-state electronic and optoelectronic devices. In such devices, the valence and conduction band discontinuities, or band offsets, at interfaces of dissimilar semiconductors are important parameters in determining such properties as electron-hole pair separation and confinement, leakage current, and gate voltage swing. In comparison, much less is known about analogous interfacial systems involving epitaxial oxide materials. Nevertheless, recent developments point to the importance and potential unique applications afforded by the use of oxides. For instance, epitaxial oxide films have been used as ferroelectric gates in field-effect transistors (FETs). In addition, the magnetic properties of certain transition metal oxides have given rise to new classes of magnetic devices involving spin-polarized electron transport. As in the case of traditional compound semiconductors, successful device fabrication and operation depends critically on gaining an in-depth understanding of oxide thin-film and interfacial properties. Critical scientific issues range from adhesion of dissimilar oxide materials to interface electronic structure and its effect on electron transport and scattering. A key aspect of interfacial electronic structure that needs to be understood for epitaxial oxides, but for which we have virtually no experimental results, is the band offset at oxide/oxide heterojunctions. We have carried out the first successful synthesis and band offset determination for epitaxial a -Fe2O3/ a-Cr2O3(0001) superlattices, and have observed some unusual, but potentially useful, band offset behavior (Chambers et al. submitted). These structures were grown by oxygen-plasma-assisted molecular beam epitaxy (OPA-MBE) on a-Al2O3(0001) substrates. Valence band offsets were measured by core-level x-ray photoemission. In summary, we have found that once an a-Fe2O3 (0001) epitaxial layer of thickness equal to a few hundred Å or more has been grown, all subsequent a-Cr2O3(0001) layers remain under lateral tension and exhibit the same in-plane lattice parameter as a-Fe2O3(0001). The out-of-plane lattice parameter contracts slightly to offset the in-plane strain. Thus, all heterojunctions formed in this way consist of unstrained a-Fe2O3(0001) and artificially-strained a-Cr2O3(0001). In addition, there is a significant noncommutativity in the band offset. This feature may make superlattices of these materials useful for effective spatial separation of electrons and holes due to the potential gradient that is expected to develop over a few periods of the superlattice. We show in Figure 6.1 a summary of the band offset behavior at these heterojunctions. The valence band offsets were determined to be 0.3 ± 0.1 and 0.7 ± 0.1 eV for Fe2O3 on Cr2O3 and Cr2O3 on Fe2O3, respectively. Using the bulk band gaps of Fe2O3 and Cr2O3, we can also estimate the conduction band offsets, which are 3.0 ± 0.3 and 3.4 ± 0.3 eV for Fe2O3 on Cr2O3 and Cr2O3 on Fe2O3, respectively. Figure 6.1. Energy level diagrams showing: (a) the noncommutative band offsets for a-Fe2O3(0001)/ a -Cr2O3(0001) and a-Cr2O3(0001)/a-Fe2O3(0001) heterojunctions, and (b) the potential gradient that develops over several superlattice periods as a result of the noncommutative band offsets. Based on what is known about traditional semiconductor band offsets, there are three possible physical causes for the noncommutative band offset seen here. The first of these is strain. Strain is known to have a large effect on semiconductor band offsets when the strain configuration at the interface is changed. However, Cr2O3 (Fe2O3) is under tension (unstrained) for both Fe2O3/Cr2O3 and Cr2O3/Fe2O3 heterojunctions. Therefore, we rule out strain as being responsible for the noncommutativity in the present system since the strain configuration in each material is the same for both kinds of heterojunctions. The second possible cause is quantum confinement. Limiting the thickness of an epitaxial layer can have the effect of altering the energy eigenvalues within the valence and conduction bands, thereby changing the band offsets. The change in eigenvalues depends on layer thickness. To test this possibility, we grew Fe2O3 and Cr2O3 overlayers of different thicknesses to see if the band offsets varied with thickness, and they did not. Therefore, quantum confinement is not a factor. The third possible cause is a difference in interface dipole due to variations in the atomic configuration on either side of the interface. It has been shown that in theory, the exact atomic configuration at the interface can have a nonnegligable effect on band offsets at Ge/GaAs and ZnSe/GaAs heterojunctions (Francisco and Van de Walle 1996). Fe is slightly more electronegative than Cr. Therefore, a dipole exists at each heterojunction interface, even if the interface is structurally perfect and completely abrupt. The dipole at a Fe2O3/Cr2O3 interface will be equal but opposite than that at a Cr2O3/Fe2O3 interface if there is no intermixing, or if the extent of intermixing is the same for the two interfaces. However, a net dipole will be established for each full period of the superlattice if the extents of intermixing are different at the two interfaces. This effect is shown schematically in Figure 6.1b. Detecting slight differences in intermixing for the two interfaces is exceedingly difficult. Nevertheless, we are currently designing experiments to do so. A significant implication of the noncommutative band offset behavior noted above is that a potential gradient accompanying the net dipole would be established in superlattices of these materials. This situation is depicted schematically in Figure 6.1b. The potential gradient would ultimately be self-limiting over approximately five superlattice periods as the valence band maximum at one end of the superlattice structure becomes higher in energy than the conduction band minimum at the other end. At this point, carriers would flow and establish a gradient in the direction opposite of that created by the noncommutative band offsets. However, photoexcitation of electron-hole pairs anywhere in the structure would result in two kinds of carrier separation. First, the type-II band offset that exists would naturally drive electrons and holes to opposite ends of the structure. Second, the gradient that develops over several superlattice periods would drive electrons (holes) toward the low-(high-) electron energy end of the structure. The efficiency with which this process occurs would depend on electron and hole mobilities and layer thicknesses, particularly for the Cr2O3 layers at which the larger of the two potential barriers exists. Nevertheless, this materials system shows promise as a useful advanced photocatalyst by virtue of the spatial separation of electrons and holes that would result from irradiation with UV light with hn ³ ~2.2 eV. 2. Synthesis and Characterization of Anatase (TiO2) Films Our second focus of the engineered photocatalyst is Cu2O/TiO2 (anatase). Because of novel photocatalytic behavior of Cu2O and the unique combination of electronic structures between Cu2O and anatase, Cu2O/TiO2 can potentially be a very promising photocatalyst for hydrogen production. Because Cu2O and anatase are not commercially available, our first step was to synthesize anatase films. We successfully grew anatase films on SrTiO3 using MBE. X-ray reflection diffraction (XRD) confirmed that all films were single-crystal anatase phase. Figure 6.2a is a low-energy electron diffraction (LEED) pattern which demonstrates that the anatase (001) surface exhibits (1 x 4) and (4 x 1) reconstructions. Such reconstructions were confirmed by scanning tunneling microscopy (STM) of the corresponding surface. Domains of (4 x 1) and (1 x 4) reconstructions are clearly evident in the image. Growth of Cu2O on anatase and characterization of the materials are currently under way. Figure 6.2. LEED (a) and STM (b) showing the surface and anatase (001) consists of (1 x 4) and (4 x 1) domains. References Chambers, S. A., Y. Liang, and Y. Gao, submitted to Phys. Rev. B (1999). Francioso, A., and C. G. Van de Walle, Surf. Sci. Rep. 25, 1 (1996). William R. Wiley Environmental Molecular Sciences Laboratory Feedback: webmaster@emsl.pnl.gov Revised: June 12, 2001 Security & Privacy PNNL-13147