Investigating the HCBM – GCxGC relationship: an elution model to interpret GCxGC retention times of petroleum substances
Risk assessments on petroleum substances (PS) are challenging due to the fact that these complex materials contain thousands of individual chemical components having widely differing physico-chemical characteristics, environmental degradation rates, and toxicities. However, comprehensive two-dimensional gas chromatography (GCxGC) can be used to separate hydrocarbon complex mixtures, thereby supporting studies of environmental risk associated with PS.
To facilitate the development of GCxGC for the evaluation of PBTproperties (persistence, bioaccumulation, toxicity) of PS, Concawe developed the hydrocarbon block method (HCBM). The HCBM delineates the GCxGC chromatogram into hydrocarbon blocks (HCBs) of closely related hydrocarbon compounds: a PS containing thousands of individual constituents can be described using about 200 HCBs. Nonetheless, the mass assignments attributed by the HCBM may contain uncertainties. The objective of the present study is to investigate the assumptions and approximations that underlie the application of the HCBM to GCxGC data for the purposes of supporting PBT assessment.
To investigate the HCBM-GCxGC relationship, a theoretically-based elution model of GCxGC retention times was developed and tested. According to the elution model, the GCxGC retention times of known or hypothesized analytes can be predicted from their chemical structures, formatted as SMILES input. The retention time model was calibrated with a set of previously measured retention times for 56 hydrocarbon compounds having diverse chemical structures. The calibrated model was then applied to the entire Concawe GGraph library of 15397 individual hydrocarbon structures, providing an unprecedented theoretical prediction of the elution patterns of petroleum hydrocarbon compounds spanning the GCxGC chromatogram of diverse PS. We plotted the simulated GCxGC retention times for the 14190 chemical structures that were predicted to fall in the C10n-C30 elution window. To visually differentiate the chromatographic region occupied by the structural members of each structural class and carbon number, a solid-line coloured polygon was overlaid onto the plots of modelled retention times. Each coloured polygon represents the convex hull that envelops the two-dimensional retention times of all members of a single class and carbon number.
By visualizing the location of each polygon for each class and carbon number in the n-C10 to n-C30 elution window, we conclude that the majority of modelled retention time polygons exhibit overlap with neighbouring groups of differing compound classes having the same carbon number. This approach made it possible to quantify the extent of overlap among the different chemical classes. As an example, we assumed a worst case of a PS that contains all 14190 library constituents in the n-C10n-C30 elution window at relevant concentrations. Ignoring differences in carbon number within each class, we found that 79% of individual compound structures were enveloped by polygons representing two or more distinct classes, whereas 21% of individual compound structures fell into an area of the chromatogram occupied by only a single class. However, in practice, the majority of PS contain far fewer constituents in relevant concentrations, and many PS contain only a subset of the classes that are encompassed by the library.
In conclusion, the GCxGC retention time model enables an improved understanding of the elution patterns of diverse hydrocarbon compound structures on the GCxGC chromatogram. In particular, the retention time model reveals substantial overlap among the elution regions of most of the hydrocarbon classes that the HCBM was designed to quantify. This finding contrasts with current implementations of the HCBM, which assume that the designated hydrocarbon classes do not overlap in the GCxGC chromatogram. Further work would be needed to elucidate the uncertainties of the HCBM that arise from overlapping elution patterns of compound classes.