The sorption and desorption of heavy metals in soil are important processes influencing their transport through soil and their ultimate fates. Many attempts have been made to predict the solid–liquid portioning coefficients, often required for modeling the transport and fates of heavy metals in a soil system. In the conventional approach, several soil properties, including pH and cation exchange capacity, are used as predictors to estimate sorption constants. While there is empirical evidence that the environmental and solution parameters, often varying with the experimental conditions of the sorption test, influence the measured soil sorption, previous studies have not considered the effects of these experimental parameters on the reported results thoroughly. Disregarding the experimental conditions may result in biased predictions of the sorption constants, resulting in inaccurate risk assessments. Herein, multiple linear regression models are developed to predict the Freundlich equation coefficients of Cd and Pb based on both the soil properties and experimental conditions as explanatory variables. Results show that the proposed model provided better performance than the conventional techniques that only consider the soil properties. It is demonstrated that the impacts of certain experimental parameters, such as liquid–solid ratio and background electrolyte concentration in solution, on the sorption constants were comparable in magnitude to the effects of the soil properties. These findings indicate that different experimental conditions should be accounted for when comparing sorption constants determined in different studies. The proposed approach could be applicable to the estimation of soil sorption constants with range of environmental and solution parameters.
Read full abstract