Based on the density functional theory,we described here a method to investigate the quantitative relationship between nucleophilicity/basicity and HSAB-theory-based properties of compounds with lone-pair electrons.Descriptors including global softness,Fukui function,local softness and local mulliken charge were calculated at SVWN/DN~* level of DFT with PC Spartan Pro.Nucleophilicity and basicity of 28 selected compounds were classified based on intensity.BP algorithm of artificial neural network(ANN) was employed to study the relationship between the descriptors and nucleophilicity/basicity.Cross-validation was carried out to avoid the over-fitting in training of ANN.A BP network was trained to quantify the relationship between HSAB-theory-based properties and nucleophilicity/basicity of compounds with lone-pair electrons.The results show that the prediction based on the network matches with the experimental results well.The local softness and Fukui function have a better relationship with nucleophilicity and local mulliken charge than with the basicity.The trained BP network could be utilized for predicting the nucleophilicity/basicity of compounds or functional groups with lone-pair electrons.
