Abstract
The fractional changes of Knight shift ΔSS for solvent and solute atoms in liquid In-Tl, Pb-Sn, Hg-In, and Ga-In alloys have been calculated from the pseudopotential formalism. It is shown that the Faber theory developed for the substitutional dilute liquid alloys can be extended to include the nonsubstitutional liquid alloys of any concentration within the first-order approximation. The experimental and hard-sphere-model interference functions I(K) have been used, and it is pointed out that small errors in I(K), particularly with respect to its peak shape and peak position, are rather unimportant as far as the qualitative results for ΔSS are concerned. The calculations predict the right sign and right trend of the ΔSS-versus-c plots of the above liquid alloys. Quantitatively, however, there exists some disagreement, which may easily be interpreted in terms of the variation of the spin paramagnetic susceptibility and perhaps the Fermi diameter with the solute concentrations; these effects were not considered in the basic theory. The present findings strongly suggest that the psedudopotential approach leads to encouraging results if the pseudopotentials are correctly evaluated and necessary corrections are applied.