Quantitative relationships between molecular structures, environmental temperatures and octanol–air partition coefficients of polychlorinated biphenyls

https://doi.org/10.1016/S1476-9271(02)00089-0Get rights and content

Abstract

A quantitative model that incorporates information on both environmental temperatures (T) and molecular structures, for logarithm of octanol–air partition coefficient to base 10 (log KOA) of polychlorinated biphenyls (PCBs) was developed. Partial least squares (PLS) analysis together with 16 theoretical molecular structural descriptors was used to develop the Quantitative relationships between structures, environmental temperatures and properties (QRSETP) model. The cross-validated Qcum2 value for the optimal QRSETP model is 0.976, indicating a good predictive ability for log KOA of PCBs at different environmental temperatures. T, ELUMO (the energy of the lowest unoccupied molecular orbital), molecular size or average molecular polarizability (α), and the net atomic charges on chlorine, hydrogen and carbon atoms of PCB molecules, are major factors governing log KOA. The lower the ELUMO, the greater the intermolecular interactions between octanol and PCB molecules, and thus the greater the log KOA values. Because of intermolecular dispersive forces, the more chlorine atoms in PCB molecules, the greater the molecular size or α, the greater the log KOA. The largest negative net atomic charge on a carbon atom (qC) and molecular size or average molecular polarizability (α) are major factors governing temperature dependence of log KOA. PCB molecules with low qC values and more chlorines (big size or α) tend to have strong temperature dependence, due to intermolecular electrostatic interactions and dispersive forces, respectively.

Introduction

Polychlorinated biphenyls (PCBs), a family of 209 congeners each of which consists of two benzene rings and one to ten chlorine atoms, are ubiquitous in the global environment because of their biological and chemical stability and their historical widespread use in the power-generation industry (Giesy and Kannan, 1998, King et al., 2000). Toxicological effects of exposure to PCBs include hepatotoxicity, immunotoxicity, and reproductive problems, as well as respiratory, mutagenic and carcinogenic effects (Giesy and Kannan, 1998, Sánchez et al. (2000)). Being semivolatile organic compounds, PCBs can undergo long-range atmospheric transport and partition between air and other environmental compartments.

The octanol–air partition coefficient (KOA) is recognized as a key descriptor of chemicals partitioning between the atmosphere and organic phases (Harner et al., 2000). Recently, KOA based approaches have been successfully employed to model surface–air partitioning of persistent organic pollutants to aerosols (Finizio et al., 1997, Harner and Bidleman, 1998, Kaupp and McLachlan, 1999, Lee and Jones, 1999, Lohmann et al., 2000, Falconer and Harner, 2000), soil (Hippelein and McLachlan, 1998, Cousins et al., 1999), vegetation (Thomas et al., 1998, Weiss, 2000), and even indoor carpet (Won et al., 2000). Therefore, KOA is indispensable for exposure assessment of persistent organic pollutants. The development of KOA determination and prediction methods is of great importance.

Due to the large enthalpy change involved in octanol to air transfer, KOA has strong temperature dependence, which is very important for assessing the global transportation of persistent organic pollutants. In principle, KOA can be estimated using the n-octanol–water partition coefficient (KOW) and Henry's law constant (H), i.e.:KOA=KOWRTHwhere R is the universal (ideal) gas constant; T, the absolute temperature. However, KOW represents octanol saturated with water and water saturated with octanol whereas H represents pure water, and the literature values of KOW and H for some persistent organic pollutants vary by more than one order of magnitude (Mackay et al., 1991). These errors will be propagated in the calculation. Another problem is the absence of H, KOW and their temperature-dependence data for many persistent organic pollutants. It is, therefore, desirable to determine or predict KOA values directly.

Based on the log KOA data for selected PCBs determined by Harner and Mackay, 1995, Harner and Bidleman, 1996, we successfully developed quantitative structure–property relationship (QSPR) models for logarithm of KOA to base 10 (log KOA) at 293 K, which can be used for prediction (Chen et al., 2002). However, such QSPR models can only predict log KOA at a fixed temperature, 293 K. Is it possible to develop quantitative KOA predictive models that incorporate information on both molecular structures and environmental temperatures? In essence, such models are not QSPR models, because environmental conditions will be included as independent variables. We termed such models quantitative relationships between structures, environmental conditions and properties (QRSECP). Strictly in the present study, the models to be developed are quantitative relationships between structures, environmental temperatures and properties (QRSETP). Therefore, it is the purpose of this study to develop QRSETP models for log KOA of PCBs.

As theoretical molecular structural descriptors such as quantum chemical descriptors, molecular volume and molecular surface area, can be easily obtained by computation, can clearly describe defined molecular properties, and are not restricted to closely related compounds, the development of QRSETP models in which theoretical molecular structural descriptors are used is of great importance.

As many relevant data as possible should be considered in QRSETP studies because this increases the probability of a good characterization of compounds. As a consequence of the increase of the number of descriptors, the problem of intercorrelation of independent variables (multicollinearity) will increase. Especially when the number of observations in the training set is less than four to five times of the number of the independent variables in a model, regression analysis (a method that was frequently used in QSPR studies) will not be useful. To overcome these problems, the partial least squares (PLS) method, a widely used chemometric method first developed by Wold et al. (1984), will be used in this study. PLS finds the relationship between a matrix Y (containing dependent variables—often only one for QRSETP studies) and a matrix X (containing predictor variables) by reducing the dimension of the matrix X while concurrently maximizing the correlation between them.

Section snippets

Materials and methods

All 209 PCB congeners along with biphenyl were included in the study. For 19 PCBs, log KOA values (Table 1) at temperature range −10 to 30 °C were determined using a generator column method (Harner and Bidleman, 1996). The 19 PCBs for which log KOA values were determined directly served as the training set of the study, which was composed of 87 observations when taking into consideration, the factors of environmental temperatures. The Kelvin temperature (T (K)=273.15+t (°C)) was adopted in the

Theory

Many previous studies (Harner and Bidleman, 1998, Harner et al., 2000) indicated that log KOA varied linearly with 1/T and can be described by:logKOA=a+b/T,b=ΔHOA2.303Rwhere a and b are regression parameters; R, the ideal gas constant; ΔHOA, the enthalpy of phase change from octanol to air (ΔHOA=2.303bR). The regression constant a is a weak function of temperature (Harner and Bidleman, 1998), thus mainly the slope b or ΔHOA characterizes the temperature dependence. Different compounds have

The QRSETP models obtained

Based on Scheme 1, the PLS analysis resulted in QRSETP model (9), for which the results are listed in Table 2. In Table 2, RX(adj)(cum)2 and RY(adj)(cum)2 stand for cumulative variance of all the X's and Y's, respectively, explained by all extracted components. It can be concluded that four PLS components are selected in Model (9), and the PLS components explain 94.8% of the variance of the independent variables, and 97.4% of the variance of the dependent variable.

Variable Importance in the

Acknowledgements

The study was supported by the Teaching and Research Award Program for Outstanding Young Teachers in Higher Education Institutions of MOE (TRAPOYT), People's Republic of China. The research results were attained with the assistance of the Alexander von Humboldt (AvH) Foundation.

References (28)

  • M.L. Connolly

    J. Mol. Graphics

    (1993)
  • I.T. Cousins et al.

    Sci. Total Environ.

    (1999)
  • R.L. Falconer et al.

    Atmos. Environ.

    (2000)
  • A. Finizio et al.

    Atmos. Environ.

    (1997)
  • H. Kaupp et al.

    Chemosphere

    (1999)
  • C.M. King et al.

    J. Organomet. Chem.

    (2000)
  • R. Lohmann et al.

    Atmos. Environ.

    (2000)
  • E. Sánchez et al.

    Chem-Biol. Interact.

    (2000)
  • P.U. Biedermann et al.

    Environ. Sci. Pollut. Res.

    (1999)
  • Chen, J.W., Xue, X.Y., Schramm, K.-W., Quan, X., Yang, F.L., Kettrup, A., 2002. Chemosphere, in...
  • J.P. Giesy et al.

    Crit. Rev. Toxicol.

    (1998)
  • T. Harner et al.

    Environ. Sci. Technol.

    (1995)
  • T. Harner et al.

    J. Chem. Eng. Data

    (1996)
  • T. Harner et al.

    J. Chem. Eng. Data

    (1998)
  • Cited by (0)

    View full text