Acetylpropionyl (2,3-pentanedione) and diacetyl (2,3-butanedione) are by-products of sugar manufacture. Both diketones have many uses, mainly food related. Vapour-liquid equilibrium data and infinite dilution activity coefficients are required to design purification processes for these chemicals. A review of available experimental methods revealed that the vapour and liquid recirculating still is most appropriate when both isobaric and isothermal VLE are required. The low-pressure dynamic still of Raal and Muhlbauer (1998) used in this study incorporates many features to ensure that measurements are of excellent quality (as demonstrated by Joseph et al., 2001). VLE measurements were made for the following systems: • Acetone with diacetyl at 30 C, 40 C, 50 C and 40 kPa • Methanol with diacetyl at 40 C, 50 C, 60 C and 40 kPa • Diacetyl with 2,3-pentanedione at 60 C, 70 C, 80 C and 40 kPa • Acetone with 2,3-pentanedione at 50 C, 30 kPa and 40 kPa. All the systems, except for methanol with diacetyl, displayed close to ideal behaviour. This was expected as they are mixtures of ketones. Solution thermodynamics allows one to perform data reduction of the measured VLE data to ensure accurate extrapolation and interpolation of the measurements. Furthermore, the quality of the data can be judged using thermodynamic consistency tests. The data were represented by the Gamma-Phi approach to VLE (the preferred method for low-pressure VLE computations). The two-term virial equation of state was used to account for vapour phase non-ideality. Second virial coefficients were calculated by the method of Hayden and 0'Connell (1975). The liquid phase non-ideality was accounted for by the Wilson, NRTL or UNIQUAC models. The best fit models are proposed for each system, as are parameters as functions of temperature for the isobaric data. The data were judged to be of high thermodynamic consistency by the stringent point test (Van Ness and Abbott, 1982) and the direct test (Van Ness, 1995) for thermodynamic consistency. The data sets were rated, at worst, "3" on the consistency index proposed by Van Ness (1995). A rating of "I" is given for a perfectly consistent data set and "10" for an unacceptable data set. For the system acetone with 2,3-pentanedione, isobars at 30 kPa and 40 kPa were measured. The results from the reduction of the 30 kPa set were used to accurately predict the 40 kPa data set. Infinite dilution activity coefficients were measured by the inert gas stripping method (based on the principle of exponential dilution). In order to specify the appropriate dilutor flask height (to ensure equilibrium is achieved), mass transfer considerations were made. These computations ensured that the gas phase was in equilibrium with the liquid phase at the gas exit point. The following infinite dilution activity coefficients were measured: • Acetone in diacetyl at 30 C • Methanol in diacetyl at 40 C • Diacetyl in 2,3-pentanedione at 60°C • Acetone in 2,3-pentanedione at 50 C. The ketone mixtures, once again, displayed close to ideal behaviour.

Thesis (M.Sc.)-University of Natal, Durban, 2003.