关键词: 污染, 热力学性质, 计算机模拟, 有机化合物, 预测

Abstract:

The thermodynamic properties of 150 n-alkyl phenols are chosen for investigation. To develop quantitative structure-property relationship models of the thermodynamic properties, topological indexes of principal quantum number ⁰P and ¹P are introduced and simplified. Considering the influence of molecular symmetry on thermodynamic properties, the alkyl is regarded as atom alike to facilitate the symmetry quantization. Based on defining the equivalent valences and similar degrees of alkyls, symmetrical feature index (Q) of n-alkyl phenol is calculated by matrix, which is established with elements of similar degrees. Molecular descriptors, including topological indexes of principal quantum number (⁰P, ¹P), symmetrical feature index (Q), number of alkyl (T) and number of pairs of ortho alkyls (B), are screened with variance inflation factors, which confirms ¹P, T, B, and Q as the independent variables in quantitative structure-property relationship models. With leaps and bounds regression method and applying Akaike's information criterion and Kubinyi function, a series of relationship models between the molecular structures and the thermodynamic properties are constructed, and all the correlation coefficients are larger than 99% even reach 100%. The leave one out cross validation method is adopted to test the stability and the prediction ability of each model, and the result is quite satisfactory. It is demonstrated that the quantitative structure-property relationship models present satisfactory stability and good prediction ability.

Key words: pollution, thermodynamic properties, computer simulation, organic compounds, prediction