Using Aberration-corrected STEM Imaging to Explore Chemical and Structural Variations in the M1 Phase of the MoVNbTeO Oxidation Catalyst W.D. Pyrz*, D.A. Blom**, V.V. Guliants***, T. Vogt**, D.J. Buttrey* * Center for Catalytic Science and Technology, University of Delaware, Newark, DE 19716, USA ** University of South Carolina, Columbia, SC 29208, USA *** Department of Chemical and Materials Engineering, University of Cincinnati, Cincinnati, OH 45221, USA Selective oxidation catalysis is crucial to society, resulting in about 25% of all important organic chemicals and intermediates used to make consumer and industrial products [1]. Current processes used to produce high-demand C3 derivatives, namely acrylic acid and acrylonitrile, require the use of multicomponent bismuth molybdates and the starting material propene [1-2]. Significant cost savings exist by replacing the expensive propene with propane as a feedstock. A top candidate for this replacement is based on the multiphase MoVTeNbO complex oxide system [1-2]. The best MoVTeNbO catalysts with respect to selectivity and activity are two-phase mixtures comprised of an orthorhombic network bronze phase (M1) and a hexagonal tungsten bronze (HTB)-type phase (M2) [1-2]. Structural models for M1 and M2 currently exist based on simultaneous Rietveld refinement of high-resolution synchrotron X-ray and neutron powder diffraction data [2]. In the present study, we use C s -corrected STEM methods to study the M1 phase for two different synthetic preparations. Structural models based on high-angle annular dark-field (HAADF) images [3] are developed and compared to the refined model developed by DeSanto et al. (Figure 1(a)) [2]. Understanding the relationship between crystal chemistry, structure, and catalyst performance is central to the continued development of these catalysts. Two different approaches were used in the synthesis of the MoVNbTeO M1 specimens examined in this study. The first sample was prepared using previously published slurry methods under ambient pressures (referred to as M1-Amb) [2], and the second by hydrothermal synthesis at 175°C under autogenous pressure (referred to as M1-Hydro). HAADF images from a M1 in [001] orientation are shown in Figure 1(a-c). A comparison between the Rietveld-refined model and the HAADF image of a M1-Hydro particle is shown in Figure 1(a) and shows a good qualitative fit. HAADF images of M1-Hydro (Figure 1(b)) and M1-Amb (Figure 1(c)) appear similar with the exception of metal site 13, where M1-Amb shows partial Te occupancy in the heptagonal channels. This direct observation provides evidence to help clarify the debate over whether a metal can occupy the heptagonal channel site [5]. Good quantitative agreement between Rietveld-derived [2] and STEM-derived x,y coordinates was observed for both the M1-Amb (Figure 1(d)) and M1-Hydro samples (Figure 1(e)). Quantification of contrast in HAADF images using a simple Z 2 Rutherford scattering relationship provides a check on the Rietveld-obtained occupancies. Comparison of these HAADF results with the Rietveld model is presented in Figure 1(f) and shows good overall agreement between the models derived from the HAADF images and the Rietveld model. Some discrepancies exist (namely the Mo4 and Te sites) and these will be discussed in detail. These variations suggest model improvements are necessary and will be addressed through re-refinement of the powder data and simulation of the HAADF images. The ability to quickly develop structural models directly from HAADF images should vastly improve the variety and complexity of crystal structures that can be solved using powder methods. Microsc Microanal 14(Suppl 2), 2008 Copyright 2008 Microscopy Society of America 2 DOI: 10.1017/S143192760808327X