Integrating statistical and experimental protocols to model and design novel Gemini surfactants with promising critical micelle concentration and low environmental risk.

A paradigmatic study of integrating statistical modeling and experimental analysis to investigate the critical micelle concentration (CMC) and environmental risk of 120 structurally diverse Gemini surfactants is performed. In this procedure, the structural profiles of studied compounds are characterized using hundreds of constitutional, topological, geometrical and electrostatic descriptors, and the resulting variables of the characterization are then calibrated on the basis of experimentally measured properties via a variety of regression techniques, including MLR, PLS, SVM, RF, and GP, in conjunction with two sophisticated variable selection methods, i.e. empirical heuristic strategy and nonnumerical genetic algorithm. Among all the built models the most predictable one is constructed based on the simplest combination of heuristic variable selection and MLR modeling, with its predictive coefficient of determination (r(pred)(2)) and root-mean-square error of prediction (RMSP) on external independent test set of 0.90 and 0.39, respectively. Subsequently, this model is used to explain the structural factors that fundamentally govern the self-assembly behavior of Gemini surfactant molecules in solution and to design several new Gemini surfactants with potentially high CMC activity and low environmental risk. Further, these designed compounds are synthesized by diquaternary ammonium reaction and characterized by elemental analysis, (1)H NMR, (13)C NMR and mass spectrum. Found a promising candidate that possesses particularly high CMC potency as 0.83 mmol L(-1) at 25°C. This experimentally measured value is in agreement with the model-predicted 0.89 mmol L(-1) fairly well.