Prediction of ternary liquid-liquid equilibrium from binary data: (for the system furfural—iso-octane—benzene)

Abstract

The distribution of benzene in the ternary system furfural — iso-octane — benzene was predicted from binary data, obtained by total pressure measurements on the systems benzene — iso-octane, benzene — furfural and iso-octane — furfural. The distribution was predicted using the two — suffix equations of vanL aar [2] and the semiempirical methods of H ildebrand [5] and S cheibel and F riedland [6]. The system carbon tetrachloride — acetic acid was also investigated by the total pressure method. The binary data obtained from these total vapour pressure measurements can be taken as fairly reliable, since azeotropes were picked up quite accurately and R edlich and K ister plots showed that the results were thermodynamically consistent. The method of calculation ensures that the final activity coefficients are a· very good representation of the original pressure measurements. This binary data, when applied in the van L aar equations for the prediction of ternary equilibrium, gives a distribution which is consistently 1—2 mole % of solute low in the iso-octanerich phase when compared with H enty's experimental distribution. The distribution predicted by H ildebrand's method is more accurate than the van L aar prediction in the region near the plait point, but is fallacious in the near binary regions in that it indicates a molar solutrope. The ternary van L aar distribution obtained using the binary data and the binodal curve is accurate in the region near the plait point, but remains 1–2 mole % of solute in error near the binaries. S cheibel and F riedland's correlation is generally more satisfactory than van L aar's except near the binaries, where it indicates a molar solutrope, which does not in fact exist.