Why is the slip direction different in different B2 alloys?

The dominant slip directions in different intermetallic alloys with B2 structure are different, either 〈0 0 1〉 or 〈1 1 1〉. The elastic energy of 〈1 1 1〉 dislocations is usually significantly higher than that of 〈0 0 1〉 dislocations and it is commonly assumed that 〈1 1 1〉 slip occurs if 〈1 1 1〉 dislocations can dissociate on {1 1 0} planes into 1/2〈1 1 1〉 superpartials. However, we show in this paper that 1/2〈1 1 1〉 anti-phase boundaries may not be metastable faults on {1 1 0} planes and the displacement vectors of metastable stacking-fault-like defects on these planes vary from material to material. This analysis involves calculations of {1 1 0} γ-surfaces for eight B2 alloys (CuZn, FeAl, NiAl, FeTi, CoTi, NiTi, FeGa, PdAl) using a density functional theory-based method. Since both 〈1 1 1〉 and 〈0 0 1〉 screw dislocations may possess non-planar cores if undissociated and will then control plastic properties analogously as 1/2〈1 1 1〉 screw dislocations in body-centered cubic metals, the dissociations have been analyzed for screw dislocations. Subsequently, we assume that if the width of splitting in {1 1 0} planes exceeds the Burgers vector of the corresponding dislocation, the dislocation spreads in this plane and is glissile while undissociated dislocation is sessile. The ability to glide is then regarded as the determining factor for the choice of the slip direction. If no planar spreading occurs the dominant dislocations are determined by their energy. This analysis predicts the slip directions for all alloys studied and demonstrates that an interplay of elastic anisotropy, displacement vectors of metastable stacking-fault-like defects and their energies govern the choice of the slip direction in any specific B2 alloy.