【作者】 陈建波；

【导师】 孙素琴；

【作者基本信息】 清华大学 ， 化学， 2010， 硕士

【摘要】 复杂混合物体系的分析一直是相关领域的研究热点和重点,而“多级红外光谱宏观指纹分析法”已经逐步成为混合物整体分析的重要手段之一。红外光谱、二阶导数红外光谱和二维相关红外光谱构成的“红外光谱三级鉴别体系”,可以对组成复杂的相似样本进行有效区分。本论文以化学计量学理论为基础,通过模拟数据和实际样本的研究,初步建立了二维相关红外光谱差异分析的系统方法。本论文首先总结了中药、食品等混合物二维相关红外光谱鉴别的系统化方法,包括计算理论、实验操作、谱图预处理、谱图归一化以及特征区域的选择。本论文首次提出了二维相关红外光谱间的“距离”与“相关系数”、混合二维相关红外光谱的“对称距离”与“对称系数”等概念,解决了不同样本二维相关红外光谱的差异量化的问题。本论文首次研究了二维相关红外光谱的差减与求导。对二维相关红外光谱进行差减,可以减弱绝对强度较大的峰,使弱峰更加显著;对二维相关红外光谱进行求导,可以分辨原谱的重叠峰,发掘更多被埋藏的信息;差减与求导均可让不同样本间的差异更加显著的表现出来,解决了高度相似样本的二维相关红外光谱分析和鉴别的问题。本论文首次研究了二维相关红外光谱的统计量与主成分分析,初步解决了对大量样本的二维相关红外光谱进行分析和鉴别的问题。从一组样本的二维相关红外光谱的平均谱可以得知其共同特征,从标准差、偏度和峰度谱可以得知样本在哪些区域内的差异最显著。主成分分析的得分图可以对样本进行有效的分类,研究样本变化规律,从载荷图上则可以找到包含最有价值信息的区域。本论文通过相关研究,建立了更为系统合理且客观量化的二维相关红外光谱间差异分析方法,为使用二维相关红外光谱对中药、食品等复杂混合物体系进行鉴别提供了系统的方法学指导。更多还原

【Abstract】 Analysis of the complex mixture has always been the hotspot in the relevant research fields. Macro-fingerprints of multi-level infrared spectroscopy have been used for the global analysis of mixtures more and more today. Tri-level infrared spectroscopic identification, which employs the Fourier transform infrared spectroscopy (FT-IR), the second derivative infrared spectroscopy (SD-IR) and two-dimensional correlation infrared spectroscopy (2D-IR) to discriminate quite similar mixtures, has proved to be useful. The systematic methods to analysis the differences between 2D-IR spectra were developed by this research, based on chemometrics and both simulated and practical data.The basic theory, experimental details, the preprocessing and normalization of the original spectra and the selection of regions for discriminating mixture samples such as traditional Chinese medicine (TCM) and food was summarized at first. The distance and correlation coefficient between 2D spectra, as well as the symmetry distance and symmetry coefficient of a hetero 2D spectrum, were introduced for the first time to evaluate the differences between 2D spectra quantitatively.The differential and derivative 2DIR were also discussed for the first time by simulated and experimental data. Strong but useless peaks for the identification of samples would fade down on differential 2D spectra while the minor peaks varied much between different samples could be magnified. Overlapped peaks on original 2D spectra could be separated by the derivative processing and more information would come out. Both the differential and derivative 2DIR could make the differences between quite similar samples become much more significant.The statistic and principal components analysis (PCA) of the 2DIR spectra were used for the first time to differentiate plenty of samples. The common features of the samples could be found by the mean 2D spectrum. The spectral regions where different samples could be discriminated effectively could be shown on the standard deviation, skewness and kurtosis of the 2D spectra. Samples could be clustered accurately by the scores plots and the most important spectral regions could be found by the loadings plots of the PCA analysis of the 2DIR spectra.Methods for differentiating similar 2D spectra were well set up by this research. Guidance and techniques for the identification of mixtures such like the traditional Chinese medicine (TCM) and the food by 2DIR were provided.更多还原