Making the Connection Between Atomistic Modelling of Interfaces and Real Materials

Abstract

Why is anyone interested in control at the atomic level? One obvious answer is that control at the atomic level enables control at larger scales. Some examples of this are clear enough; nano-engineering is precisely about this. However, in many cases the question of how control at the atomic level results in control at other length scales is far from clear. This is particularly the case in complex materials where many mechanisms are active at once. Putting reactive elements into an alloy to control the growth of the resulting scale is an attempt to control atomistic processes, but the details of how it works are often obscure. Yet control implies that we can identify what the important mechanisms are and then find a procedure that will enhance (or inhibit) the mechanism we want to affect without causing problems elsewhere. The traditional way of controlling complex systems is trial and error. The equally traditional difficulty is the large error to trial ratio. In the hope of reducing this a bit, people have turned to modelling complex processes. However, unless we are in the (unusual) position of being able to model everything the problem of deciding what the important mechanisms are remains. The strategy of mesoscopic modelling is to attempt to identify the individual mechanisms and model these at the appropriate length scale (which will often be atomistic but need not be) and then to build a model of how these mechanisms interact to produce the process. If the result of a process is a material, a final goal will often be to predict the properties of the material as a function of the conditions under which it was made. ‘Properties’ may be of two main kinds. First, we may require the effective bulk properties (elastic constants, thermal conductivity). These depend on the averaged behaviour of the microstructure but are unlikely to be dependent on individual features of the microstructure. Other properties such as the probability of fracture may well depend on such features.