Advanced separation of mineral oil aromatic hydrocarbons by number of aromatic rings using donor-acceptor-complex chromatography to extend on-line coupled liquid chromatography-gas chromatography.

ABSTRACT

An automated implementation for a subfractionation of mineral oil aromatic hydrocarbons (MOAH) into a mono-/di-aromatic fraction (MDAF) and a tri-/poly-aromatic fraction (TPAF) is presented, which is highly demanded by the European Food Safety Authority (EFSA) respecting the genotoxic and carcinogenic potential of MOAH. For this, donor-acceptor-complex chromatography (DACC) was used as a selective stationary phase to extend the conventional instrumental setup for the analysis of mineral oil hydrocarbons via on-line coupled liquid chromatography-gas chromatography-flame ionization detection (LC-GC-FID). A set of six new internal standards was introduced for the verification of the MOAH fractionation and a quantification of MDAF and TPAF, respectively. The automated DACC approach was applied to representative petrochemical references as well as to food samples, such as rice and infant formula, generally showing well conformity with results obtained by state-of-the-art analysis using two-dimensional GC (GCxGC). Relative deviations of DACC/LC-GC-FID compared to GCxGC-FID methods regarding the ≥ 3 ring MOAH content ranged between -50 and +6 % (median -2 %, all samples, only values above limit of quantification). However, crucial deviations mainly result from "border-crossing" substances, e.g., dibenzothiophenes or partially hydrogenated MOAH. These substances can cause overestimations of ≥ 3 ring MOAH fraction during GCxGC analysis due to co-elution, which is mostly avoided using the DACC approach. Furthermore, the DACC approach can help to minimize underestimations of toxicologically relevant ≥ 3 ring MOAH caused by an unavoidable loss of MOAH during epoxidation, since natural olefins, such as terpenes, predominantly elute in MDAF, which was exemplarily shown for an olive oil and a terpene reference. The presented approach can be implemented easily in existing LC-GC-FID setup for an automated and advanced screening of MOAH to lower the need for elaborate GCxGC analysis also in routine environments.