Microanalysis Methods By Near Infrared Spectroscopy

Near infrared (NIR) spectroscopy analytical method is efficient, rapid, noninvasive, environmental friendly, and it can be run at low costs. It is not only suitable for laboratory analysis, but also in-field fast measurement and real-time on-line analysis. However, NIR spectral bands are relatively weak and highly overlapping. Therefore, chemometrical methods are commonly used for building calibration model for qualitative and quantitative analysis. However, in order to constructing and maintaining models, large numbers of representative samples are collected for wet chemical analysis, and providing data information of components and characters. NIR spectrum is based on measurement of the overtone and combination frequencies of the vibrations of chemical bonds, and mainly records the absorption information of C-H, O-H, N-H and S-H, which make its relatively weak spectral bands and high detection limit. Up to now, these problems have not been solved thoroughly, which obstruct further development and application of NIR spectroscopy analytical method. In the dissertation, the design of calibration samples and improvement of quantitative determination ability for low concentration analytes were investigated for the quantitative analysis of complex samples.The main research contents of the dissertation involve:1. The design of calibration samples is often used to improve the cost-effectiveness of near-infrared (NIR) spectral analysis. A feasibility of constructing a parsimonious multivariate calibration model for NIR spectral analysis was demonstrated by constructing partial least squares (PLS) models with both standard and complex samples. Waveband selection algorithm and other parameters were also be discussed for improve predictive ability of models. The results indicate when constructing quantitative analysis PLS model, the calibration samples should be as similar as possible to the prediction samples. When there is a big difference in the properties of samples between calibration and prediction sets, it can improve the predictive ability of the models by extending the difference of calibration set samples. At the same time, it is very important to adopt waveband selection algorithms before modeling for improving the predictive ability of models.2. A method for quantitative analysis of chlorogenic acid in tobacco extracts was developed by using near infrared spectroscopy (NIRS) and PLS regression. Effects of the coexistence solutes on PLS model were discussed by constructing different models with standard solutions and real tobacco extracts. Results show that standard samples can be used as a part of the calibration samples when there are not enough real samples, and the interference of the coexistence solutes can be eliminated by selection of suitable wavelength regions. The proposed approach appears to be a valid alternative for fast analysis of chlorogenic acid in tobacco industry.3. Near-infrared diffuse reflectance spectroscopy (NIRDRS) has been proved to be a convenient and fast quantitative method for complex samples. The high detection limit or the low sensitivity of the method, however, is a big problem obstructing its application in the analysis of low concentration samples. A strategy for quantitative determination of low concentration samples was developed by using NIRDRS. Taking benzoic and sorbic acids as the analyzing targets and the alumina as the adsorbent, PLS model is built from the NIRDRS of the adsorbates. The results show that the detection limit can be improved by using adsorption preconcentration, and the interferences of co-adsorbates can be eliminated by using multivariate calibration method.4. Organic pollutants in water are one of the major water quality pollution. A method for quantitative determination of low concentration phenol and p-nitrophenol from wastewater was developed by using NIRDRS and a preconcentration procedure of resin adsorption. In the method, the analytes were firstly adsorbed onto NKA-II resin for preconcentration, and then NIRDRS of the resin is measured for quantitative analysis. The results show that both the phenolic compounds can be immobilized onto the adsorbent and directly measured by NIRDRS. The method provides new opportunities for analyzing low concentration components in aqueous solutions. At the same time, the proposed method may be an effective way for improving the detection ability of the NIRDRS for quantitative analysis of low concentration analytes. 5. The feasibility of NIRS and adsorption preconcentration procedure for determination of low concentration free amino acids including threonine (Thr), methionine (Met), lysine (Lys) and leucine (Leu) in aqueous solutions and amino nitrogen was investigated. In the method, two ion exchange resins were used for adsorption preconcentration of amino acids, and NIRDRS was used for collecting the spectra of resin after adsorption. The effect of the different kinds of resins on the determination results was investigated. The results obtained by using least-squares support vector regression (LS-SVR) and partial least squares regression (PLSR), which were used to eliminate the interferences resulting from near infrared diffuse reflectance spectra of resin and other coexistence components, were compared. The results show that preconcentration procedure of resin adsorption and NIRDRS were successfully used in determination of low concentration amino nitrogen, while the quantitative result of each amino acid is unsatisfactory. The resins with different colours, apertures and types have no effect on quantitative analysis. The performance of LS-SVR was slightly better than that of PLSR. Complex matrix interferences of co-existence in amino acid beverage can be eliminated by using multivariate calibration method.