Nucleation and growth models are well developed for nucleation on homogeneous substrates, and they can typically be described-in terms of three energy parameters. Nucleation on substrates containing point-defect traps has been investigated, at the cost of introducing more energy parameters. This paper outlines the quantitative description of such growth models, using rate and rate-diffusion equations, in terms of energies for individual surface processes, with examples taken from metal-metal, metal-insulator and semiconductor growth. The challenge to modelling is to describe the large range of length and time-scales in thin-film fabrication and degradation, without relying on too many (unknown) material parameters, which often occur in combination. Separating them into elementary processes often proves to be a challenge. One typically requires selective nucleation using patterned substrates, in combination with controlled, self-organized, growth for reliable nanotechnology. Reconstructed semiconductor surfaces offer both a further challenge to modelling and an opportunity for future technology; these paradoxes are discussed briefly.