材料期刊网

气相中有机化学品与羟基自由基(·OH)反应速率常数(kOH)是评价化学品大气持久性的重要参数.而化学品种类繁多,现有的kOH实验数据不能满足其生态风险评估的需求.因此,需要建立一种能够快速预测有机化学品kOH的方法,填补现有数据的缺失.本研究搜集整理了917种有机化合物的kOH实验值,采用逐步多元线性回归法(MLR)和支持向量机法(SVM)分别构建了用于预测多种类化合物lgkOH的线性和非线性定量结构-活性关系(QSAR)模型.基于MLR的预测模型具有良好的拟合度(经校正的相关系数的平方Radj.tr2=0.862、均方根误差RMSEtr=0.455)、稳健性(交叉验证系数QLOO-0.856)和预测能力(外部验证系数Qext2=0.850).基于SVM的预测模型也具有良好的拟合度(Radj.tr2=0.915,RMSEtr=0.358)和预测能力(Qext2=0.860).机理分析表明,最高占据分子轨道能(EHOMO)、卤素原子在分子中所占的百分比(X％)和分子中具有=CH—结构的数目(NdsCH)是最重要的3个描述符,解释了数据集78.3％的方差.采用Williams法表征模型的应用域.所建立的模型可用于预测烷烃、烯烃、炔烃、芳香族化合物、醇类、酮类、醚类、醛类、酸类、酯类、卤代化合物、含氮化合物、含硫化合物等室温下的kOH.

Reaction rate constant of hydroxyl radical with organic chemicals (kOH) is an important parameter for assessing the atmospheric persistence.However,for the large number of chemicals,itis difficult to meet the demand of ecological risk assessment with existing experimental kOH values.Thus,it is significant to develop an effective prediction method for kOH.In this study,we collected kOH values of 917 chemicals,and used Multiple Linear Regression (MLR) and Support Vector Machine (SVM) to develop linear and non-linear QSAR model for lgkOH prediction.Results of linear QSAR model show that it has satisfactory goodness-of-fit (Radj.tr2 =0.862,RMSEtr =0.455),robustness (QLOO2 =0.856) and good predictability (Qext2 =0.850).Results of non-linear QSAR modelshow that it also has good fitness (Radj.tr2 =0.915,RMSEtr =0.358) and predictability (Qext2 =0.860).Mechanistic analysis reveals that the energy of the highest occupied molecular orbital (EHOMO),the percentage of halogen atoms in a molecule (X％),and the number of =CH— structure in a molecule (NdsCH) are the most important descriptors in the model,which explain the 78.3％ variance of data set.The applicability domain of the models was characterized by the Williams plot.The developed models can be used to predict kOH values (298 K) of alkanes,alkenes,alkynes,ketones,ethers,aromatic compounds,alcohols,acids,esters,halogenated compounds,nitrogen compounds,sulfur compounds.