Determination of the soil–water partition coefficients (log K_OC) of some mono- and poly-substituted phenols by reversed-phase thin-layer chromatography

Abstract

In order to determine the soil–water partition coefficient for eleven mono- and poly-substituted phenolic compounds, for which there is still no literature data available, the possibility of using thin-layer chromatography (TLC) as a means for rapid and reliable log K_OC estimation was examined. A series of chromatographically derived descriptors: $R_{\text{M}}^0$, b, C₀ and PC1 (first principal component), calculated from retention data obtained under reversed-phase conditions, were used for the assessment of models as well as for a direct calibration procedure. The final calibration models are discussed with regard to the achieved accuracy and statistical quality, the type of descriptors used and the corresponding chromatographic conditions. The estimated log K_OC values of the studied phenols were compared with those obtained by other means: (a) the present OECD guideline based on an HPLC technique; (b) the KOCWIN software package, available free of charge from the US Environmental Protection Agency web site and (c) general LSER models established by Nguyen and coworkers, and Poole and coworkers. The proposed method showed the best agreement with the results obtained by the OECD procedure, followed by the LSER models of Poole and Nguyen. Lower quality correlations were achieved with the KOCWIN calculated values, especially those predicted by molecular connectivity indices.

Introduction

The environmental fate of organic compounds is, among others, firmly determined by their sorption interactions with soil organic matter or soluble humic acids. Therefore, the determination of soil–water partition coefficient normalized to the content of organic carbon as one of the primary parameters for the evaluation of chemicals movement is very important (Babich and Davis, 1981, Reinhard et al., 1984). Phenol is one of the most widely used compounds in the world and the US Environmental Protection Agency lists it as a priority pollutant (Electronic Code of Federal Regulations, Appendix A to Part 423, 2010). It adversely affects the indigenous biota, including algae, protozoa, invertebrates and vertebrates. Apart from its overt toxicity, phenol causes many subtle effects, such as reduced fertility, decreased survival of the young and inhibition of growth (Babich and Davis, 1981). In addition, phenols are naturally present in surface and ground waters as degradation products of lignin and other plant materials (Reinhard et al., 1984, Stone, 1987).

There are several techniques for the direct determination of soil-sorption coefficients (Delle Site, 2001), but most of them are time consuming, tedious and suffer from serious disadvantages. They are mostly inapplicable to compounds sparingly soluble in water or those of greater binding affinity towards the soil organic or silicate matter, which can increase retention in flow based systems, or produce extremely low, barely detectable, analytical signal in the aqueous phase of a batch soil–water partitioning system. For the same reasons, they are unsuitable for highly volatile compounds. In order to overcome these problems, numerous rapid K_OC estimation methods have been effectively developed in the past decade, mainly relying on correlations between log K_OC data and different structural, electronic and other molecular and chemical descriptors, that is based on the Quantitative Structure Property Relationship (QSPR) approach (Sabljić et al., 1995, Baker et al., 2001, Liu and Yu, 2005). In addition, good Linear Free Energy Relationship (LFER) models were established between log K_OC and the Abraham solvato-chromatographic parameters (Nguyen et al., 2005). Almost simultaneously with the QSPR methodology, another approach was developed. Replacing the soil with commercially available chromatographic stationary phases, High Performance Liquid Chromatography (HPLC) was introduced. Hitherto, many commercially available stationary phases have been investigated, such as CN – (Vowles and Mantoura, 1987) and C18-modified silica (Koerdel et al., 1995, Gawlik et al., 2000), which were followed by studies on natural and artificial soils (Koerdel et al., 1995, Guo et al., 2004). Finally, this approach has resulted in many successful procedures and some of them are being implemented in international laws and guidelines for testing chemicals (OECD, 2001). While HPLC is a very robust, precise and controllable technique that is also faster and with lower costs than many other methods involved in direct K_OC measurements, thin-layer chromatography (TLC), which is much cheaper and simpler to perform, is not at all present in this field. The exception is soil thin-layer chromatography, which is mostly used for a rough assessment of the mobility of solutes in the microstructure of soil (Ravanel et al., 1999). This is quite surprising when the success of this technique in the estimation of the similar physico-chemical properties, i.e. lipophilicity (log K_OW) (Sarbu et al., 2001) and biological activity over many years are considered (Giaginis and Tsantili-Kakoulidou, 2008). Moreover, TLC has several advantages, such as short duration of analysis, very low costs and the possibility of the simultaneous analysis of several (over 20) samples under the same chromatographic conditions, which make it a method of choice.

Since no systematic study has been performed on the application of reversed-phase thin-layer chromatography for the assessment of the soil–water partition coefficients normalized to the content of organic carbon, the present work is focused on: (a) the estimation of the applicability of thin-layer chromatography as a means for log K_OC assessment of the phenols and phenolic type compounds and (b) determination of log K_OC values for some phenolic compounds for which there is no literature data available. In order to assess the applicability of the proposed TLC approach, first the method was compared with the current OECD procedure for the determination of the log K_OC parameter, and also with different QSPR methodologies, such as the general LSER approach developed by Nguyen and coworkers (Nguyen et al., 2005) and the LSER model of Poole and coworkers (Poole and Poole, 1999). In addition, the log K_OC values obtained through the freely available KOCWIN software package (EPA), using models based on molecular connectivity indices (MCIs) and the octanol–water partition coefficient (log K_OW) were used for comparison. In order to establish calibration models, two strategies were adopted. One relying on previously described chromatographic descriptors while the other is based on the direct calibration procedure (regression of R_M upon log K_OC parameter, which results in one calibration curve for each chromatographic system).