Abstract
Several methods currently used to calculate isotropic dispersion energy coefficients, C 2n , are investigated: (i) a version of the single excitation energy Unsöld scheme, (ii) a ‘bond’ oscillator model (which employs Frost model wavefunctions), and the generalized ab initio (iii) Unsöld and (iv) Kirkwood schemes. The large numerical discrepancies, which often occur in the results for β=C 8/C 6 and γ=C 10/C 6 obtained by the various methods, are ascribed to difficulties with (i) and (ii). Corrected versions of (i) and (ii) are discussed and they yield larger results for β and γ which, especially in the case of (i), are in much better agreement with those of the ab initio schemes (iii) and (iv) which show the expected marked increase in β and γ as the size of the interacting species increases. Results from the ab initio schemes, using SCF wavefunctions, are presented for C 6, C 8 and C 10 for interactions arising from the molecules BH3, CH4, NH3, H2O, HF, CO, CO2 and N2O. The numerical values for the C 2n are still subject to rather large uncertainties although for some interactions the results are apparently (fortuitously) very good. The results for β and γ are much more reliable than the values of the individual C 2n and hence can provide a reasonably reliable description of isotropic long range interactions when combined with accurate values for C 6.