Predicting gas chromatography relative retention times for polychlorinated biphenyls using chlorine substitution pattern contribution method
Introduction
High resolution gas chromatography (GC) is one of the dominant instrumental techniques of separation in the analyses of organic compounds in various environmental, food, pharmaceutical and other science disciplines and industries. In past decades, a great deal of efforts has been made to establish quantitative structure retention relationships (QSRRs) which link the chemical structures of the analytes to chromatography retentions. Comprehensive reviews on QSRRs are available and cover a wide spectrum of analytes [1], [2], [3].
For polyhalogenated organic chemical families such as polychlorinated biphenyls (PCBs), different number and substitution positions of halogens in molecules are the root of differences among congeners in physicochemical properties including chromatographic retention. Simple structural parameters such as the numbers of ortho, meta, para and/or total halogen substituents have been frequently used as independent variables for the purpose of predicting PCBs’ GC retentions [4], [5]. The use of these simple variables is convenient, but with only limited success due to the negligence of interactions among atoms and between rings. Correlations of GC retention with boiling point temperature, vapor pressure, gas-octanol partition coefficient, etc., make theoretical sense, but the experimental values of these physicochemical properties are scarce. More sophisticated approaches to quantifying GC retention involve various constitutional, topological, geometrical, electrostatic, and semi-empirical molecular descriptors [6], [7], [8], [9], [10], [11], [12], [13], [14], [15]. Quantum chemistry computations are often needed to generate these independent variables.
In PCBs, the chlorine atoms are present in a total of 19 substitution “patterns” such as 2-, 2,4-, 2,3,6-, etc. These patterns represent all chlorine atoms in a phenyl ring as a group. Each PCB molecule is represented by 1 or 2 patterns, as shown in Appendix A. For example, PCB-8 (2,4′-dichlorobiphenyl) is presented by two patterns, 2- and 4-, on different rings. Similar to PCB-8 are a total of 171 congeners in which the two rings are differently substituted with at least 1 chlorine atom. There are 19 other congeners with one unsubstituted ring thus having only one chlorine substitution pattern. For example, PCB-7 (2,4-dichlorobiphenyl) has only one pattern 2,4- on one ring with the other ring having no chlorine. The remaining 19 congeners are substituted with the same chlorine pattern on both rings. For example, PCB-4 (2,2′-dichlorobiphenyl) and PCB-47 (2,2′,4,4′-tetrachlorobiphenyl) possess patterns 2- and 2,4- on both rings, respectively. Counting through 209 congeners, each of the 19 chlorine substitution patterns appears equally in 20 congeners for a total of 21 times (Appendix A). If sufficient data is available for simultaneous estimation of an independent “substitution pattern contribution factor” for each pattern using multiple linear regression (MLR) analysis, the retention behavior could be quantified as the weighted sum of contribution factors plus the intercept.
This approach was presented by Sissons and Welti in 1971 [16] to determine the congener compositions of three Aroclors, and later expanded to all PCBs in 13 packed GC columns [17]. Similar methods were used for polychlorinated diphenyl ethers (PCDEs) [18] and polybrominated biphenyls (PBBs) [19]. However, this simple yet potentially highly accurate approach has not been widely applied and, based on our search of the literature, has not been used further for PCBs. The motivation for our current work stemmed from a practical need in research involving dehalogenation product identification. The objectives were to examine the feasibility of using a halogen substitution pattern contribution (SPC) model for the prediction of relative GC retention time, evaluate the model performance and stability, and use the modeling results to investigate the effects of halogen substitution positions and the column stationary phase on chromatographic retention. Our efforts of model development started with PCBs due to the availability of large experimental GC retention time databases.
Section snippets
The experimental database
Among many experimental databases for GC relative retention times (RRT) for PCBs, the one assembled by Frame [20] is one of the most comprehensive. It contains RRT data obtained from 27 GC columns with 20 distinct stationary phases. The column stationary phases, column length, detector used, and the number of co-eluting congeners are presented in Appendix B. Additional information including injection volume and port temperature, column diameter, film thickness, carrier gas and flow rate, oven
Model performance
Table 1 summarizes model statistics and pattern contribution factors βk with standard errors for the 27 Frame's GC systems using full data sets of Mixes 1–9 (N = 209, except for S15 for which N = 208). All 27 multiple linear regressions are strongly statistically significant, demonstrating the efficacy of the Cl-SPC approach. The R2 of the 27 regressions ranged from 0.961 to 1.000, had an average and a median of 0.994, and all p-values were <0.0001. Based on the rank in statistics including R2, F-
Acknowledgments
This work was supported by the United States National Science Foundation under Grant CBET 0756428.
References (31)
Quantitative structure-(chromatographic) retention relationships
J. Chromatogr. A
(2007)- et al.
Investigation of retention behavior of polychlorinated biphenyl congeners on 18 different HRGC columns using molecular surface average local ionization energy descriptors
J. Chromatogr. A
(2012) - et al.
Computer-assisted prediction of gas chromatographic retention times of polychlorinated biphenyls by use of quantum chemical molecular properties
Chemosphere
(1992) - et al.
Cross-column prediction of gas-chromatographic retention of polychlorinated biphenyls by artificial neural networks
J. Chromatogr. A
(2011) - et al.
Structural identification of polychlorinated biphenyls in commercial mixtures by gas-liquid chromatography, nuclear magnetic resonance, and mass spectrometry
J. Chromatogr.
(1971) - et al.
Identification of the individual polychlorinated biphenyls in a mixture by gas-liquid chromatography
J. Chromatogr. A
(1977) - et al.
Requirement for a standardized nomenclature criterium for PCBs – computer-assisted assignment of correct congener denomination and numbering
Chemosphere
(1993) - et al.
Gas chromatographic retention of 180 polybrominated diphenyl ethers and prediction of relative retention under various operational conditions
J. Chromatogr. A
(2010) - et al.
Recent developments in the high-resolution gas chromatography of polychlorinated biphenyls
J. Chromatogr. A
(1999) - et al.
Improved separation of the 209 polychlorinated biphenyl congeners using comprehensive two-dimensional gas chromatography-time-of-flight mass spectrometry
J. Chromatogr. A
(2004)
Comprehensive two-dimensional gas chromatography of the 209 polychlorinated biphenyls
J. Chromatogr. A
Structure and Retention in Chromatography: A Chemometric Approach
Quantitative Structure: Chromatographic Retention Relationships
Prediction of gas chromatographic retention times of polychlorinated biphenyls by mono-dimensional molecular descriptors and multivariate techniques
Chromatographia
Development of a multiple-class high-resolution gas chromatographic relative retention time model for halogenated environmental contaminants
Anal. Chem.
Cited by (3)
Potential of topological descriptors to model the retention of polychlorinated biphenyls in different gas chromatography stationary phases, including ionic liquid-based columns
2020, Journal of Chromatography ACitation Excerpt :However, the application of this model involved 19 topological descriptors (Table S.1), which is a number relatively high compared with that of experimental PCB data used in the present study. This could result in artificially high values of the fit quality (Table S9 of the SI), something that prevents from direct comparison with previously evaluated models [7]. For this reason, in this section, the values and significance of the descriptor contributions were used for results discussion instead of the fit quality.
A simple and reliable method for separation of mineral oil/polychlorobiphenyl mixtures
2023, Waste Management and ResearchGibbs energy additivity approaches to QSRR in generating gas chromatographic retention time for identification of fatty acid methyl ester
2017, Analytical and Bioanalytical Chemistry