【摘要】 采用密度泛函理论方法,在B3LYP/6-31+G(d,p)水平上,对任意选定的32个有机化合物或药物进行最低能量构象优化和结构参数理论计算.建立了四极矩参数Qii与半数摩尔热分解函数Yd(1/2)的相关方程,其定量构性关系(QSPR)方程为Yd(1/2)=-8.65747-3.8954Qii,相关系数为r2=-0.99297,交叉验证相关系数为XV-r2=0.99188,F检验结果为4237.343321.训练集化合物的半数分解温度Td(1/2)的平均绝对预测误差(AVEDEV)为14.70 K.进一步利用该方程对测试集中43个分子进行预测验证,Td(1/2)的预测值与实验值的相关系数为0.92304,Yd的预测值与实验值的相关系数为0.99345,证实了所建立方法的可靠性.结构差异性分析表明,训练集和测试集中的化合物均较均匀地分布在结构参数的3D空间中,化合物结构具有较好的多样性和差异性.更多还原

【Abstract】 Density functional theory(DFT) calculations at the B3LYP/6-31G+(d,p) level were carried out for 32 random selected compounds in training set and 43 ones in testing set after determining the experimental half decomposition temperature T_d(1/2) or partial ones from literature.The calculated quantum chemical descriptors were used to construct quantitative structure-property relationship(QSPR) models of quadrupole moments(Q_ii) parameters with the molar thermal decomposition function Y_d(1/2).The equation of QSPR is Y_d(1/2)=-8.65747-3.8954Q_ii for training set possessing a correlation coefficient of r2=-0.99297,cross validation XV-r2=0.99188 and F-test=4237.343321.Further,the proposed model is validated by examining 43 compounds in testing set with a correlation coefficient of 0.92304 between prediction and experiment T_d(1/2) values and a correlation coefficient of 0.99345 between prediction and experiment Y_d values.The selected compounds rank well with a certain space distribution of structure descriptors.The developed equation can be used to predict the heat decomposition stability for new designed or synthesized compounds or drugs in virtual screening.更多还原