Migration measurement and modelling from poly(ethylene terephthalate) (PET) into soft drinks and fruit juices in comparison with food simulants

Poly(ethylene terephthalate) (PET) bottles are widely used for beverages. Knowledge about the migration of organic compounds from the PET bottle wall into contact media is of interest especially when post-consumer recyclates are introduced into new PET bottles. Using migration theory, the migration of a compound can be calculated if the concentration in the bottle wall is known. On the other hand, for any given specific migration limit or maximum target concentration for organic chemical compounds in the bottled foodstuffs, the maximum allowable concentrations in the polymer C _P,0 can be calculated. Since a food simulant cannot exactly simulate the real migration into the foodstuff or beverages, a worse-case simulation behaviour is the intention. However, if the migration calculation should not be too overestimative, the polymer-specific kinetic parameter for migration modelling, the so-called A _P value, should be established appropriately. One objective of the study was the kinetic determination of the specific migration behaviour of low molecular weight compounds such as solvents with relatively high diffusion rates and, therefore, with high migration potential from the PET bottle wall into food simulants in comparison with real beverages. For this purpose, model contaminants were introduced into the bottle wall during pre-form production. The volatile compounds toluene and chlorobenzene were established at concentrations from about 20–30 mg kg^?1 to 300–350 mg kg^?1. Phenyl cyclohexane was present at concentrations of 35, 262 and 782 mg kg^?1, respectively. The low volatile compounds benzophenone and methyl stearate have bottle wall concentrations of about 100 mg kg^?1 in the low spiking level up to about 1000 mg kg^?1 in the highly spiked test bottle. From these experimental data, the polymer specific parameters (A _P values) from mathematical migration modelling were derived. The experimental determined diffusing coefficients were determined, calculated and compared with literature data and an A _P′ value of 1.0 was derived thereof for non-swelling food simulants like 3% acetic acid, 10% ethanol or iso-octane. For more swelling condition, e.g. 95% ethanol as food simulant, an A _P′ value of 3.1 seems to be suitable for migration calculation. In relation to PET recycling safety aspects, maximum concentrations in the bottle wall were established for migrants/contaminants with different molecular weights, which correspond with a migration limit of 10 μg kg^?1. From the experimental data obtained using food simulants and in comparison with beverages, the most appropriate food simulant for PET packed foods with a sufficient but not too overestimative worse-case character was found to be 50% ethanol. In addition, it can be shown that mass transport from PET is generally controlled by the very low diffusion in the polymer and, as a consequence, partitioning coefficients (K _P/F values) of migrants between the polymer material and the foodstuff do not influence the migration levels significantly. An important consequence is that migration levels from PET food-contact materials are largely independent from the nature of the packed food, which on the other hand simplifies exposure estimations from PET.