Studies and Application of Chemometrics in Molecular Spectroscopic Analysis

分子光谱手段是化学分析家们常用的一种获得化合物信息的方法，如红外光谱、拉曼光谱、紫外光谱等等。这些方法在化学常规分析中有着非常广泛的应用。化学计量学（Chemometrics）是应用数学、统计学与计算机的工具与手段设计或者选择最优化学测量方法，并通过解析化学测量数据，最大限度的获取化学及其他相关信息。20世纪70年代以来，随着计算机技术的迅速普及，数学和计算机科学在分析化学中的应用日益广泛。在分子光谱分析中，化学计量学方法被用来给一些复杂体系的样品作定量和定性分析。对于一些成分简单的样品，我们可以直接根据该样品的红外、紫外等光谱来得到分析结果，但是对于复杂的样品，尤其是天然产物，根本无法解释它的红外谱图。因此，如果不采用化学计量学方法，根本无法做出准确的定性定量分析。本文更多的是以近红外（Nearhifrared，NIR）光谱方法为主，结合化学计量学方法，包括模式识别、偏最小二乘（PLS）等，对各种复杂体系，如中药，农产品等，进行定量定性分析。对中药的分析结果表明，分子光谱分析技术完全可以实现对中药样品的鉴别分析，尤其是近红外光谱分析技术，分析更为简单、快速，完全可以取代传统方法。用近红外光谱分析技术对大豆、酒醅各项指标进行分析，结果表明近红外分析技术也可用于在实际工作对大豆、酒醅的分析。研究结果表明，分子光谱方法简单快速。同时借助于各种化学计量学方法，能快速得到准确的分析数据，完全可以替代标准方法。只要在实验初期收集足够的具有代表性的样品，然后采用标准方法得到各种化学分析数据，应用计量学建立模型。在对未知样品进行分析时，只需利用采集的谱图数据和建立的模型，就能实现对样品化学分析数据的准确预测，包括定性和定量分析。相比于常规方法，分子光谱分析方法更加简单、快速，能大大提高分析效率。Molecular spectroscopic analysis methods, including IR spectroscopic analysis, Raman spectroscopic analysis, are the most important methods that chemists can use to get information from compounds. Those methods are much widely applied to regular analysis. Chemometrics can be used to design or choose the optimized measure method by combination of mathematics, statistics, and computer techniques, so as to improve the understanding of chemical and other information from the data. Since the 1970s', mathematics and computer science have been widely used in the analytical chemistry with the popularization of computer techniques. Chemometrics can be applied to qualitative analysis and quantitative analysis of complicated samples in molecular spectroscopic analysis. To simple samples, we can get the useful information directly from the IR and UV spectra of samples. For complex samples, especially for natural products, however^ it is quite hard to obtain the useful information directly from their IR spectra. We can't get the right result of qualitative and quantitative analysis without chemometrics for this kind of samples. In this paper near IR(NIR) spectroscopic analysis, combating with chemometrics methods including pattern recognition and partial least square(PLS), was applied to the qualitative and quantitative analysis of complex samples, such as Chinese traditional medicine, agricultural products etc. The analytical results of Chinese traditional medicine showed that molecular spectroscopic analysis, especially NIR spectroscopic analysis, could be applied to identification of Chinese traditional medicine. The results demonstrated that this method is simple and rapid, and can be used to replace the traditional methods in this field. The other analytical results of fermented grains and soybeans with NIR spectroscopic analysis showed that this method was more convenient for quality control of fermented grains and soybeans compared with traditional methods. The analytical model can be set up by collecting enough representative samples at first and analyzing all kinds of compounds to get chemical analytical data secondly. Then we can predict the chemical analytical data of unknown sample with its spectra from the built model. Compared with traditional analysis methods, the methods of molecular spectroscopic analysis are more simple, rapid and effective