葡萄糖是生命活动中重要的有机分子, 研究葡萄糖在太赫兹波段的吸收特性并开展定性定量分析研究具有重要意义。 太赫兹光谱对大分子物质具有特异性, 当光谱穿透生物大分子时, 由于分子间作用力和分子转动振动等因素影响, 会产生太赫兹光谱下的特征指纹峰, 不同大分子物质的指纹峰具有特异性, 利用这一特征可以对不同大分子物质进行鉴别。 本研究选取D-无水葡萄糖为研究对象, 首先运用太赫兹时域光谱技术测量得到葡萄糖样品的时域光谱, 测得的时域光谱经过傅里叶变换得到频域光谱, 再采用Dorney和Duvillaret方法计算得到吸收系数, 研究样品的吸收特性, 然后开展了数学建模研究并建立了D-无水葡萄糖含量与太赫兹光谱的定量预测数学模型。 研究结果表明, D-无水葡萄糖在太赫兹波段呈现明显的吸收特性, 比较多元线性回归与偏最小二乘方法建立的回归模型, 通过多元线性回归对样品的特征指纹峰建立的回归模型, 得到的预测模型精度较高, 稳定性较好, 模型的校正集相关系数达0977 2, 均方根误差为0061 6, 预测集相关系数为0992 7, 预测均方根误差为0055 2, 从而太赫兹时域光谱技术可用于有效开展D-无水葡萄糖定性定量研究, 该研究为运用太赫兹光谱技术手段开展果蔬、 食品、 药品中葡萄糖含量快速检测提供方法参考。

Glucose is one of the most important organic molecules in life activities. It is significant to study the absorptive fingerprints of glucose in terahertz band and carry out quantitative and qualitative analysis research. Terahertz is specific for macromolecules. As for the rotational and vibrational modes within a molecule, when terahertz spectra penetrate the macromolecules, they have fingerprints in terahertz wavelength. This feature can be used to identify macromolecules. In this paper, D-glucose anhydrous was selected as the research object. Terahertz time-domain spectrum of D-glucose anhydrous was measured using Terahertz time-domain spectroscopy technique first and then the frequency domain spectrum was calculated with Fast Fourier Transform (FFT). The method of the Dorney and Duvillaret were used to process the frequency spectra to get the absorption coefficient. The absorption features of D-glucose anhydrous samples were studied and then mathematical models of quantitative relations between D-glucose levels and its absorptive spectroscopy were estimated. The results showed that, D-glucose anhydrous has remarkable absorptive fingerprints in terahertz band, and the regression model based on sample fingerprints using multivariate linear regression method performed well when compared with the partial least squares method. Its correlation coefficient and the error of mean square root of the calibration set model was 0977 2 and 0061 6 respectively, and the correlation coefficient and the error of mean square root the of the prediction set model was 0992 7 and 0055 2 respectively, which showed that terahertz time-domain spectroscopy technique can be used for quantitative and qualitative analysis of D-glucose anhydrous. It provides a reference for applications of rapid detection of the glucose content in fruits, vegetables, food and medicine in future using terahertz spectroscopy.