The energies of stacking faults arising from shear along <331> were calculated by means of ab initio and modified embedded-atom method calculations. The results were used to investigate the configurations of ½<331> dislocations, and their mobilities. A shear of 1/6<331>, in a {103} plane, produced an antiphase boundary whose geometry (APB1) was different to that produced by 1/6<331> shear in the opposite direction (APB2). Modified embedded-atom method calculations calculations showed that APB1 was stable, while both types of calculation showed that APB2 was unstable. Both ab initio and modified embedded-atom method calculations showed that there was a stable fault close to APB2, with a displacement of about 1/8<331> in the same direction. The calculations also showed that there was a stable fault in {110} planes, with a displacement of ¼<111>. An identical fault was produced by a shear of 1/4<331>. There was found to be good agreement between the fault energies that were calculated by using the 2 methods, and also with experimentally determined values (200 to 370mJ/m2). The agreement between the calculated fault energies in the {013} plane was not so good. One cause was suggested to be that the relaxation procedures were different. The modified embedded-atom method was more flexible, as well as involving a larger number of atoms. It was suggested that this was why it predicted lower stable fault energies. The {103} planes had an unusual 5-layer A-B-C-D-E stacking sequence, with successive planes offset by 1/5<301>. A shear of 1/10<351> in the correct direction produced a low-energy fault, with Mo atoms surrounded by the correct number (10) of Si nearest-neighbours. This vector was close to the 1/8<331> shear which produced a stable fault, and was suggested to explain its low predicted energy. Various dissociated configurations of ½<331> dislocations were considered on the basis of 1/6<331>, 1/8<331>, ¼<331> and 1/10<351> partials. All of them could have asymmetrical arrangements which could respond differently to the direction of the applied stress; thus explaining why {103}<331> slip was much easier for crystals which were compressed along [100], rather than [001].

Stacking-Fault Energy and Yield Stress Asymmetry in Molybdenum Disilicide. T.E.Mitchell, M.I.Baskes, S.P.Chen, J.P.Hirth, R.G.Hoagland: Philosophical Magazine A, 2001, 81[5], 1079-97