I will shortly review transferability issues with classical potentials. I will therefore start with implications of the choice of formulating atomic interactions in a pairwise manner and show what density dependent potentials achieved. Transferability will be assessed by investigating materials properties in different local environments. Going from bulk structures to surfaces and from equilibrium properties to bond breaking processes will help to assess the range of properties a potential can reliably reproduce.
Then I will focus more specifically on bond order potentials and on including some of the electronic information into classical potential formulations. While this helps to gain confidence in the ability of the potential to represent certain properties it does not make fitting and testing obsolete. I will close with a few examples of the ability of current bond order potentials to represent materials mechanical properties of covalently bonded metals and semiconductors.
Author entry (protected)