The Application Of Different Analysis Methods In The Study Of Quantitative Structure-Retention Relationship (QSRR) In Gas Chromatography

Chromatography is an effective analytical method for the separation of complex mixtures. Quantitative structure-retention relationship (QSRR) of compounds is an important subject in chromatographic study. It bases on the weakly intermolecular action that can be described by the structure parameters of solutes.After calculating the structure parameters, the quantitative structure-retention relationships of the series of alkylbenzenes, alkyl-nitrophenols and sulfur alcohol compound are studied.The work is divided into four sections.By summarizing the 141 referenced literatures, in the first chapter, we discussed the development, structure parameters, analyze methods and applications of QSRR. Based on this, we proposed the task of this dissertation.In chapter 2, chemical structures of alkylbenzenes were established by Materials Studio 4.0 (MS) software. Optimization of the structures was performed with the COMPASS force field in DISCOVER module in MS. Structure parameters can be calculated in QSAR module. The optimized parameters of compounds were correlated with alkylbenzene retention index by multiple linear regression method. Related to the comparison of factor analysis, the efficient QSRR models on different polar stationary phase were found. The regression model show that the main factors affecting alkyl benzene compounds isolated is the same,Alkyl-nitrophenols are studied in Chapter 3. Optimization of the structures was performed with the AMI method in VAMP module. The different QSRR equations were built with genetic algorithm and multiple linear regression method. Because Alkyl-nitrophenols are studied in Chapter 3. Optimization of the structures was performed with the AMI method in VAMP module. The different QSRR equations were built with genetic algorithm and multiple linear regression method. Because genetic algorithm is suitable for large samples and variables, the forecast accuracy and stability of QSRR model are better than the result in multiple linear regressions.In Chapter IV, QSRR models of sulfur alcohol compounds were established. Optimization of the structures was performed with the COMPASS force field in DISCOVER module. Optimized parameters of the molecular structure are divided into two types of quantum parameters and topological index. Regression equations were built with different types of parameters with multiple linear regression method. Further more quantum parameters were correlated with retention index by genetic algorithm. Although the QSRR models of genetic algorithm have good linear relationship, but it needs more structural parameters and it is more difficult to calculate for small samples. QSRR equations of multiple linear regression method have fewer parameters and able to reflect elements of the structural information. These equations also retain the better accuracy and stability.The QSRR models built can be used to predict the chromatographic retention values and select the best separation conditions. It is efficient to reduce the experimenter blindness and save a lot of manpower.