Systematic Modelling of Disorder In B/C/N and Si/C Composites

Boron nitride, carbon and silicon are closely related, displaying similar crystal structures (for example, based on both 3- and 4-coordinate local coordination geometries under ambient conditions). The crystalline phase diagrams are relatively well understood to both high pressures and temperatures. In disordered (liquid, amorphous or glass) states structural characterisation is often less clear. This is more so for BN for which the structures are inherently more complex simply owing to the presence of two species. As a result, any structural disorder may be deconstructed into a topological contribution and a site disorder – the B and N atoms can arrange themselves in a huge number of different ways across the same basic network topology.

At the most ordered level both BN and C form sheets constructed exclusively from percolating hexagons. Topological disorder can be introduced by changing the ring size distribution (i.e. introducing non-hexagons). Whilst the mean ring size is fixed (<n>=6; Euler’s law), the rings in a given distribution may be arranged in a huge number of ways. For BN the disorder is potentially more complex in terms of both potential topological and site disorder. For the latter, clearly the B, N or C atoms may occupy different sites. However, the differing electronegativities of the three elements may affect the network topology itself. In the limiting case, in which homopolar bonds are energetically forbidden, then the network must contain even-membered rings only. In less extreme cases the ring size distribution may be simply biased towards even-membered rings. For BN/C composites the disorder is potentially even more complex, depending also on the stoichiometry.

Si and C share a highly stable 4-coordinate diamond crystal structure. Analogous comments to the B/N/C composites then apply regarding the balance of topological and site disorder, although here in the context of 3-dimensional crystalline and amorphous systems.

These concepts of control of topological and site disorder very much translate into other dimensionalities covering, for example, liquid and amorphous states as well as thin films, nanotubes and fullerene-like morphologies. The nanotubular structures formed are of direct relevance to on-going experiments performed in the Grobert group. Furthermore, these materials may be spin into complex ceramic fibres with potentially useful mechanical properties.

Density-functional theory-based simulations, although potentially useful, display very different levels of site disorder dependent on the subtle details of the underlying calculation. In this project we will develop potential models to study BN/C and Si/C composites across a wide range of phase space and dimensionalities including liquid, amorphous and glassy states, as well as thin films, sheets and nanotubes. Physical insight, driven by a fundamental understanding of the subtle balance of interactions which control the resulting morphologies, is key is the useful properties of these systems are to be fully exploited and controlled. We shall combine recent developments in machine-learning, such as the application of gaussian approximation potentials (GAP), will be combined with the development of state-of-the-art potential models which incorporate the (physically-motivated) interaction terms. The development of relatively simple models also allows for the atomistic simulations over the long length- and time-scales required to model fibre formation and their fundamental properties.