American Mineralogist, Volume 91, pages 1761–1772, 2006 0003-004X/06/1112–1761$05.00/DOI: 10.2138/am.2006.2180 1761 INTRODUCTION The surface structure and reactivity of different crystallo- graphic faces of uranium dioxide are of interest from a mineral- ogical and nuclear waste management perspective. Spectroscopic and microscopic techniques have been used to characterize the (111), (110), and (100) surface structures of UO 2 (Ellis 1968, 1974; Taylor and Ellis 1978; Castell et al. 1996, 1998a, 1998b; Muggelberg et al. 1998, 1999), which represent the three basic terminations of this ionic solid (Tasker 1979a, 1979b). Empiri- cal potential modeling has been applied to atomic-scale studies involving bulk UO 2 (Catlow 1977; Jackson and Catlow 1985; Grimes and Catlow 1991; Meis and Gale 1998; Abramowski et al. 1999 and references therein), and more recent studies have involved clean and hydroxylated UO 2 surfaces (Abramowski 1999, 2001; Tan et al. 2005a, 2005b). Quantum mechanical calculation techniques have been used to investigate bulk elec- tronic properties of UO 2 (Dudarev et al. 1997; Kudin et al. 2002; Boettger 2003 and references therein), which is of interest as UO 2 is a weakly semi-conducting material with a band gap of 2.14 eV (Killeen 1980). Quantum mechanical calculations have also been performed to better understand the interaction of hydroxyls with the UO 2 (111) surface (Boettger and Ray 2002). Although quantum mechanical methods are not dependent on bulk-derived potential sets and have the advantage of providing information on bulk and surface electronic structure, they are currently limited to systems with a relatively small number of atoms (e.g., tens to a few hundreds of atoms). Thus, large-scale simulations of complicated surface morphologies and molecular dynamics simulations rely heavily on the less computationally expensive force-field methods. However, since empirical poten- tials are typically developed and tested using bulk structures, transferability to the application of surfaces needs to be carefully evaluated. The size of the system chosen in this study allows the application of both types of methods, such that comparisons of surface relaxation and surface energies are directly possible. In this study, we use quantum mechanical and empirical potential modeling techniques to calculate and compare surface energy trends on the (111), (110), and (100) crystallographic faces * E-mail: fskomurs@umich.edu Quantum mechanical vs. empirical potential modeling of uranium dioxide (UO 2 ) surfaces: (111), (110), and (100) FRANCES N. SKOMURSKI, 1, * RODNEY C. EWING, 1 ANDREW L. ROHL, 2,3 JULIAN D. GALE, 2 AND UDO BECKER 1 1 Department of Geological Sciences, University of Michigan, 2534 C.C. Little, Ann Arbor, Michigan 48109-1005, U.S.A. 2 Nanochemistry Research Institute, Department of Applied Chemistry, Curtin University of Technology, Perth, Western Australia 6845 Australia 3 iVEC, 26 Dick Perry Avenue, Technology Park, Kensington, Western Australia 6151, Australia ABSTRACT To evaluate the stability, potential reactivity, and relaxation mechanisms on different uraninite surfaces, surface energy values were calculated and structural relaxation was determined for the (111), (110), and (100) crystallographic faces of uranium dioxide (UO 2 ) using quantum mechanical (density functional theory) and empirical potential computational methods. Quantum mechanical results are compared with empirical potential results, which use surface slab models with two different geometries, as well as various different empirical force fields. The strengths and weaknesses of the different ap- proaches are evaluated, and surface stabilizing mechanisms such as relaxation, charge redistribution, and electronic stabilization are investigated. Quantum mechanical (q.m.) surface energy results are in agreement with the relative surface energy trends resulting from calculations using three different empirical potential sets for uranium and oxygen (two from Catlow 1977; one from Meis and Gale 1998), and with empirical force-field values published in the literature (Abramowski et al. 1999, 2001). The (111) surface consistently has the lowest surface energy (0.461 J/m 2 from q.m. calculations) and the smallest amount of surface relaxation, followed by the (110) surface (0.846 J/m 2 ; q.m.), and the (100) surface (1.194 J/m 2 ; q.m.) (quantum mechanical surface energy values in parentheses are for surface slabs with a thickness of four UO 2 units). Differences exist, however, in the absolute values of surface energies calculated as a function of potential set used. Quantum mechanical values are consistently lower than values calculated using empirical potential methods. A fourth potential set is presented that is derived from fitting electrostatic and short-range repulsive parameters to experimental bulk properties and surface energies and relaxations from quantum mechanical calculations. Keywords: Uraninite, UO 2 , surface energy, quantum mechanical, empirical potentials, force fields, uranium dioxide