Abstract:

Some liquid alkanes bearing 5—24 carbon atoms were optimized in structure and the topological distance matrix was built with the space topological distance among atoms calculated by using density functional theory (DFT)at the B3LYP/6-31+G(d)level. A new group of topological index PX_m(m=1, 2, 3)was introduced based on the topological distance matrix, and branch vertex of atoms in a molecule was put forward by coloring atoms in the molecular graph with equilibrium electro-negativity. The topological model of quantitative relationship between thermal conductivity λ and PX_m of liquid alkanes was proposed by the multivariate linear regression (MLR).The results from the formula indicated that the deviation between calculated values and experimental data and that between predicted values and experimental data were small and within the experimental errors. To validate the estimation stability for internal samples and the predictive ability for external samples of the models, the method of leave-one-out (LOO)cross validation (CV)and external validation were performed. Compared with other results reported, the number of parameters in this model is much fewer and the calculation is much easier. It provides a new method for predicting the thermal conductivity of liquid alkanes.