Third Virial Coefficients of Ar, Kr, and Xe

In view of serious discrepancies in polynomial fitting of PVT measurements over different density ranges, we have reanalyzed the data of the Amsterdam group to obtain improved third virial coefficients C_expt1(T) for Ar, Kr, and Xe at low temperatures. The third virial coefficients assuming additivity C_A(T) and the nonadditivity corrections ΔC(T), assuming the triple‐dipole potential to be valid at arbitrarily small distances r_ij, have been calculated for some of the better pair potentials, and in all cases it is found that $C_A\mathrm{(T)+\Delta \; C(T)}$ agrees well with the reanalyzed experimental value C_expt1(T). We consider the values of ΔC when a cutoff at r^* is made such that the three‐body interaction vanishes when any $r_{\mathrm{ij}}{\mathrm{<r}}^{*}$ . This shows that the major contribution to the ΔC integral comes from triangles with one or more sides r_ij smaller than r_m, the separation at the pair‐potential minimum. Since one expects both higher multipole and exchange‐overlap three‐body interactions to be appreciable for these small triangles, it is difficult to understand the consistently good agreement between $C_A\mathrm{+\Delta \; C}$ and the experimental values. The large discrepancies between C_A(T) and C_expt1(T) cannot be reasonably attributed to uncertainties in the pair potential or in the revised polynomial fitting of the PVT data.