Partition Coefficients of Aroma Compounds Between Polyethylene and Aqueous Ethanol and their Estimation Using UNIFAC and GCFEOS

Partition coefficients were measured for n-alkanes (pentane to docosane) and 13 different aroma compounds (isoamylacetate, d-limonene, camphor, linalylacetate, L-menthol, dimethylbenzylcarbinol, citronellol, phenylethylalcohol, diphenylmethane, diphenyloxide, eugenol, r-undelactone) at dilute concentrations between aqueous ethanol solutions (100%, 66% and 33% for n-alkanes; 100%, 75%, 50%, 35% w/w) and nitrogen at 25 °C using a gas stripping column method. Partition coefficients for n-alkanes (octane, nonane, decane, dodecane, tetradecane, hexadecane, octadecane, eicosane, docosane) and the aromas were also measured between low density polyethylene, high density polyethylene and ethanol and aqueous ethanol liquid phases (100%, 75%, 50% and 35% ethanol w/w) at 10°, 25° and 40 °C using an equilibrium sorption method. Within the experimental uncertainty no significant differences were found for polyethylene samples with different crystallinities and very little temperature effect was seen for the polymer/liquid partition coefficients. The polymer/liquid partition coefficients were most affected by the chemical nature of the mixture. The liquid/gas and polymer/liquid partition coefficients were estimated using UNIFAC with UNIFAC-FV and using the Group-Contribution Flory Equation-of-State (GCFEOS). UNIFAC and GCFEOS are useful for qualitative estimations. Significant quantitative differences between the experimental data and the estimations were found for the liquid/gas and polymer/liquid partition coefficients of some solutes, in particular middle polarity aroma compound. A correlation of the size of estimation error with increasing molecular weight was observed for the n-alkanes and phenols. The variances between experimental and estimated values are explained in terms of the methods' group-contribution additivity and solution of groups assumptions and the methods' semi-empirical nature. UNIFAC gave more consistent and on average better quantitative estimations than GCFEOS.