The nature of geometric frustration in the Kob-Andersen mixture.

Geometric frustration is an approach to the glass transition based upon the consideration of locally favoured structures (LFS), which are geometric motifs which minimise the local free energy. Geometric frustration proposes that a transition to a crystalline state is frustrated because these LFS do not tile space. However, this concept is based on icosahedra which are not always the LFS for a given system. The LFS of the popular Kob-Andersen (KA) model glassformer are the bicapped square antiprism, which does tile space. Such a LFS-crystal is indeed realised in the Al2Cu structure, which is predicted to be a low energy state for the KA model with a 2:1 composition. We, therefore, hypothesise that upon changing the composition in the KA model towards 2:1, geometric frustration may be progressively relieved, leading to larger and larger domains of LFS which would ultimately correspond to the Al2Cu crystal. Remarkably, rather than an increase, upon changing composition we find a small decrease in the LFS population, and the system remains impervious to nucleation of LFS crystals. We suggest that this may be related to the composition of the LFS, as only a limited subset is compatible with the crystal. We further demonstrate that the Al2Cu crystal will grow from a seed in the KA model with 2:1 composition and identify the melting temperature to be 0.447(2).