Classical atomistic simulations and tight binding electronic structure calculations were combined in order to study the (110), (001) and (100) surfaces of rutile. The atomistic simulations demonstrated that they could provide accurate surface relaxation data, as compared with those obtained from first-principles pseudopotential calculations. This was true of both ideal surfaces and those which contained O vacancies. The calculated surface energies indicated that the (110) surface was the most stable and that (001) was the least stable; in agreement with experimental observations. Tight-binding calculations of the relaxed geometry were used to obtain some insight into the electronic structure of the rutile surfaces. An advantage of the present calculations was that, unlike the case of pseudopotential calculations, slabs could be treated which were sufficiently large to ensure that the potential at the center of the slab had converged to the bulk value. It was noted that the electronic structure for bulk rutile was in good agreement with that determined by using pseudopotential and Hartree-Fock methods. With regard to defective surfaces, a surface state which was predominantly Ti3d in character was observed at the bottom of the conduction band. This state was about 1.0eV below the Fermi level in X-ray photo-electron spectrographic experiments.

Electronic Structure and Atomistic Simulations of the Ideal and Defective Surfaces of Rutile. Purton, J., Bullett, D.W., Oliver, P.M., Parker, S.C.: Surface Science, 1995, 336[1-2], 166-80