物理化学学报 >> 2005, Vol. 21 >> Issue (02): 214-217.doi: 10.3866/PKU.WHXB20050221

研究简报 上一篇    下一篇

微观化学反应产物分子能量分配的模式识别

杜凤沛;陈兰;蔡遵生;赵学庄   

  1. 中国农业大学理学院应用化学系,北京　100094;　南开大学化学学院 ,天津　300071
  • 收稿日期:2004-06-16 修回日期:2004-09-23 发布日期:2005-02-15
  • 通讯作者: 杜凤沛 E-mail:dufp@cau.edu.cn

Energy Partition Among the Product Molecules in Microcosmic Chemical Reaction Determined by Pattern Recognition

DU Feng-Pei;CHEN Lan;CAI Zun-Sheng;ZHAO Xue-Zhuang   

  1. DepartmentofAppliedChemistry,ChinaAgricultureUniversity,Beijing　100094;　CollegeofChemistry,NankaiUniversity,Tianjin　300071
  • Received:2004-06-16 Revised:2004-09-23 Published:2005-02-15
  • Contact: DU Feng-Pei E-mail:dufp@cau.edu.cn

PDF (PC)

3178

赞

摘要： 在不同拓扑特征的London-Eying-Polanyi-Sato(LEPS)型势能面上,应用经典轨线法计算了不同动态条件下的样本集,应用模式识别研究了势能面特征及动态因素对产物分子能量分配的影响. 结果表明: 对A＋BC→AB＋C类反应,模式识别方法可验证若干已知的主要影响因素,并发现不同影响因素之间的乘积对产物分子能量分配有更重要的作用.根据模式识别的判别分析方法建立了产物分子能量分配的定性预测模型.

关键词: 分子反应动态学, 能量分配, 模式识别

Abstract: Classical trajectory calculation based on extended London-Eying-Polanyi-Sato (LEPS) potential energy surface (PES) is carried out to study the dependence of the energy partition among the reaction product molecules on the properties of PES and the reaction dynamic conditions. All the PESes have different topological-structures, which can be described by 6 features, and the reaction conditions can be described by 7 features. By this method, 3 sample collections are obtained. Then, the methodology of pattern recognition is used to analyze the relationship between the observables and the PESes and the reaction conditions. After the analysis, some known rules are obtained. We also found that the multiplication between different features has more important effect than single features. By pattern recognition, 3 discrimination equations are obtained, which only have 3,4,2 different features but not the whole 13 features. The 3 discrimination equations can be used to predict the behavior of unknown samples.

Key words: Molecule reaction dynamics, Energy partition in product molecule, 　Pattern recognition