中红外光谱结合向量夹角直接定量三氯蔗糖

多组分复杂体系的红外光谱由于成分多样、谱带严重重叠、背景干扰等导致谱图特征减弱，且体系中的红外光谱往往不是各组分光谱的简单叠加，需要借助于化学计量学技术以实现解析。基于朗伯-比尔定律（Lambert-Beer law）并且随着衰减全反射（ATR）技术取样普及，结合偏最小二乘回归法（PLS）、人工神经网络（artificial neural network, ANN）等化学计量法将数据进行累积、特征提取以及校正建模等工作解决红外光谱中严重重叠和非线性等问题，再采用二次测量定量分析已初步具有应用成效，而有效增强信息量表达，降低背景、噪声干扰，减少复杂的计算过程，进而保证定量分析结果的可靠性，则成为研究的主要方向。通过讨论光通道和背景对光谱强度变化的影响，结合向量角度转换提出一种新的近似线性定量方法。经初步理论计算，发现多组分混合物与该混合物中待测组分的相对含量二者间的向量夹角值存在一定的关系，而该关系并不受批次制样的影响，进一步通过高斯曲线模拟混合信号，充分说明选择合适的波长范围，二者在一定范围内存在线性相关性，相关性不受测量条件变化的影响。采用KBr压片法，向待测样品中逐步加入待测组分得到混合样本，分别得到待测组分光谱与混合样本光谱信号，先将其进行一阶求导，消除加性误差后，再将其转化为空间向量角度，消除批次制样中光程差异，再以百分比定义待测物样品的含量，所得到混合样的夹角值与其中所含参照品含量成简单线性关系，即以计算结果夹角值与含量作标准曲线，实现定量分析。本测量方法用于三氯蔗糖（Sucralose, TGS）与三氯蔗糖-6-乙酸酯（4,1,6-trichloro sucrose-6-acetate, TGS-6-A）二者进行一定比例混合（双组分体系），以及TGS-6-A的脱酰产物（多组分体系）中的TGS的含量定量分析，在两种体系中定量均得到令人满意的结果，所建立标准曲线的相关系数r均达到0.995 0以上，相对误差均低于8.0%，所建立的分析方法对利用中红外透射光谱分析研究多组分样品具有重要的参考价值。

Mid-Infrared Spectral Combined Vectorial Angle Method for Direct Quantifying the Content of Sucralose

The diversity of components and interferences of spectral overlapping highly and background reduce the features of the mid-infrared of the complex mixtures. Furthermore, the mid-infrared spectrum of a multi-component complex sample is generally not a simple sum of spectra of the components. Chemometrics techniques are needed for quantitative interpretation of the mid-infrared spectra. The problems of serious overlap and nonlinearity in infrared spectra were solved by data accumulation, feature extraction and correction modeling from chemometrics methods such as PLS and ANN, etc. And bi-measurement quantitative analysis was applied preliminarily based on Lambert Bill’s law and the popularization of the sampling technology of ATR. Then how to reduce these disturbances and ensure the accuracy of the results effectively was the main direction of research. Based on evaluating the spectral intensity of optical channel and background, this study proposed a new approximate linear quantitation method combined with vector angle transformation. Through the preliminary theoretical calculation, it was found that there was a certain relationship between the multi-component mixture and the vector angle value of the relative content in the mixture, and the relationship was not influenced by the batch sampling. Further, the Gauss curve was used to simulate the mixed signal, which fully illustrated a linear correlation within a certain range if suitable wavelength range was selected, and the correlation was not affected by the change of the measurement conditions. The KBr compression method was used to get the mixed samples, into which the components were added step by step, and the spectral and mixed-sample spectral signals were obtained. The first order derivation was carried out to eliminate the additive error signals which were converted into the space vector angle to the elimination of light path difference in batch sampling, and then they converted the content into a percentage. The angle values of the mixed samples have simple linear relationships with the contents of the reference materials. Then the quantitative analysis was made by using the standard curve of the calculated result angle values and the contents. This method was utilized to determine the contents in samples of the mixing of TGS and TGS-6-A (two-component systems), and the deacylated products of TGS-6-A (multi-component systems) with satisfactory results. The established correlation coefficients (r) of the standard curves were all above 0.995 0 and the relative errors were less than 8%. The result shows the established analytical method could meet the content analysis of multi-component samples. It has great reference value for the study of multi-component samples by mid-infrared transmission spectroscopy.