Ab initio computational chemistry is finding increased use in the field of surface science, particularly in the study of adsorption. For semiconductor surfaces, such studies typically employ cluster models of the substrate. However, computational expense limits the cluster size and hence the type of adsorbate and range of adsorbate-substrate interactions that it is feasible to investigate. It is therefore desirable to obtain a calculation scheme which yields high accuracy for a minimum computational investment. Here the drawbacks of several approaches to this ideal are discussed, as are some general issues concerning the use of cluster models. In order to reduce computational expense to a level whereby the adsorption of large molecules (for example, fullerenes) may be studied, an empirical density functional, EDF1, has been applied in conjunction with the LANL2DZ pseudopotential basis set. The results of this calculation scheme are compared to those obtained using other functionals in conjunction with both all electron and the LANL2DZ pseudopotential basis sets. These results indicate that EDF1 can serve as a cost effective alternative to B3LYP.