Slice-based Multivariate Calibration Strategy for Quantification of Polycyclic Aromatic Hydrocarbons in Oil Fractions by Means of GC×GC-TOFMS

Combining GC×GC with mass spectrometry (MS) enhances the power of the instrumentation beyond conventional chromatography methods. The GC×GC–TOFMS method offers several practical advantages, including higher peak capacity, greater separation power, high sensitivity, and improved selectivity [1]. Nevertheless, the complexity of multidimensional chromatography presents a significant challenge in handling and interpreting the vast amount of data generated [2]. In this study, the focus was on quantitation of a mixture of polycyclic aromatic hydrocarbons (PAHs) including dibenzoanthracene, fluorene, dibenzothiophene, naphthalene, 1- methylnaphthalene, 3,6-dimethylnaphthalene, pyrene, 1-methylphenanthrene, phenanthrene, and anthracene. To address these issues, a novel slice-based multivariate calibration strategy was introduced, utilizing the total ion chromatogram (TIC). The data related to distinct segments of the target analytes injected from the first column into the second chromatography column were organized, creating a new matrix. The concentration of each segment was then determined based on the ratio of the total concentration of that analyte in the first column. Subsequently, multivariate calibration techniques of partial least squares regression (PLSR), support vector machine (SVM) and radial basis function-artificial neural network (RBF-ANN) method, were employed to construct appropriate models, and analytical figures of merit (AFOMs) were obtained for each method. The PLSR model outperformed the other ones according to the two criteria: R2 (0.991- 0.999) and RMSE (0.02-0.07). Compared to the conventional approach of using the entire chromatogram in the form of pixels and applying data volume reduction methods, the advantage of this approach lies in its reliance on the concept of two-dimensional chromatography itself, eliminating the need for pre-processing methods to reduce data volume. This leads to a more efficient analysis of complex samples with less loss of important information [3]. For example, the values of sensitivity, analytical sensitivity and LOQmin obtained for analytes in the PLSR method are in the range of 2.64×105-2.41×106, 7.18-102.88 and 0.43-2.42, respectively. Finally, to prove the potential of the proposed strategy in real samples, quantification of PAHs in the heavy oil sample was successfully performed.