摘要
运用量子化学软件包Gaussian-16对无环烷烃分子进行了密度泛函理论计算,考虑取代基的位置对分子能量的影响,获得最稳定的构型,用于计算分子拓扑指数,计算软件为AlvaDesc,共计算了18大类2438个分子拓扑指数。随机选择50个烷烃分子和26个烷烃分子分别作为训练集和外部测试集。经逐步多元回归分析,得到七个分子描述符piPC01、MATS3p、IVDE、E1s、Av、Mor30p和Mor20u对烷烃临界温度具有显著的影响,相关系数达到0.9997,标准偏差为3.11K,经t检验、F检验、D-W检验、留一法交叉验证和外部测试集验证,表明QSPR模型稳健并且具有良好的预测能力,可用于预测烷烃的临界温度。
Quantitative Structure-Property Relationship(QSPR)is developed between critical temperature and molecular topological indices of acyclic alkanes.28 normal alkanes and 48 branching alkanes are randomly split into two subsets,50 alkanes in training set and 26 alkanes in test set,respectively.The different configuration of each alkane molecule is considered,and its geometrical optimization and frequency analysis is carried on density functional theory(DFT)under 6-311 g(d,p)basis set via Gaussian-16 software.The Gaussian output file with the lowest energy and no imaginary frequency for each alkane molecule is applied to produce topological indices with AlvaDesc software.A seven parameters linear equation is obtained with stepwise multiple regression method.The seven variables include 2 D and 3 D topological indices,such as piPC01,MATS3p,TVDE,E1s,Av,Mor30p and Mor20u,have no obvious autocorrelation and have been confirmed to make significant contributions to the critical temperature of acyclic alkanes via student t-test,Fisher F-test,Durbin-Watson test,Leave-One-Out Cross Validation and external validation test.The MLR coefficient(R)is 0.9997 and standard deviation is only 3.11 K.The calculated critical temperatures of 76 alkanes were very close to their experimental ones.This QSPR model will be useful for estimating the critical temperature of acyclic alkanes.
作者
饶火瑜
祝志强
徐珍珍
RAO Huo-yu;ZHU Zhi-qiang;XU Zhen-zhen(Department of Applied Chemistry,East China University of Technology,NanChang 330013,China)
出处
《化学研究与应用》
CAS
CSCD
北大核心
2022年第4期738-745,共8页
Chemical Research and Application