Molecular scale understanding of the structure and properties of aqueous interfaces with clays, metal (oxy-) hydroxides, layered double hydroxides, and other inorganic phases is strongly affected by significant degrees of structural and compositional disorder of the interfaces. ClayFF was originally developed as a robust and flexible force field for classical molecular simulations of such systems (Cygan, R. T.; Liang, J.-J.; Kalinichev, A. G. J. Phys. Chem. B 2004, 108, 1255–1266). However, despite its success, multiple limitations have also become evident with its use. One of the most important limitations is the difficulty to accurately model the edges of finite size nanoparticles or pores rather than infinitely layered periodic structures. Here we propose a systematic approach to solve this problem by developing specific metal-O-H (M-O-H) bending terms for ClayFF, Ebend = k (theta -theta(0))2 to better describe the structure and dynamics of singly protonated hydroxyl groups at mineral surfaces, particularly edge surfaces. On the basis of a series of DFT calculations, the optimal values of the Al-O-H and Mg-O-H parameters for Al and Mg in octahedral coordination are determined to be theta(0)AlOH = theta(0)MgOH = 110°, kAlOH = 15 kcal mol-1 rad-2 and kMgOH = 6 kcal mol-1 rad-2. Molecular dynamics simulations were performed for fully hydrated models of the basal and edge surfaces of gibbsite, Al(OH)3, and brucite, Mg(OH)2, at the DFT level of theory and at the classical level, using ClayFF with and without the M-O-H term. The addition of the new bending term leads to a much more accurate representation of the orientation of O-H groups at the basal and edge surfaces. The previously observed unrealistic desorption of OH2 groups from the particle edges within the original ClayFF model is also strongly constrained by the new modification.