Research And Industrial Application Of Raman Spectra Transfer And Quantitative Analysis

Raman spectroscopy is an important tool for chemical and physical analysis, and has been widely applied. The aim of this thesis is to research on Raman spectra standardisation or transfer between different instruments, and to make improvement on traditional quantitative analysis methods. Some achievements have been implemented in the software for our online Raman analyzer. The main contents of this thesis are:1. Dispersive Raman spectrometer is sensitive to the temperature of environment, which affects the consistency of measured spectra. To solve the problem, a novel method is proposed for correcting the temperature effect on instruments by applying convolution with a Gaussian function. A standard sample is used to represent the instrumental response functions under various temperature conditions, and a Gaussian function for convolution is built, then wave number drift and resolution variation are corrected. Based on spectral analysis and curve fitting, this method does not need to measure a set of well-prepared samples. Results show that it is an effective method to reduce temperature influence on dispersive Raman spectrometer, and the consistency of Raman spectra measured at different temperatures is greatly improved after correction.2. This thesis proposes a calibration transfer method based on the instrument response function, allowing transfer of Raman spectra from higher to lower resolution. No need to measure a set of well-prepared samples to construct a statistical transfer model, the method can correct the instrumental differences from source to target by convolution with a Gaussian function. Combined with exsiting methods for wavenumber calibration and relative intensity correction, an integrated procedure for Raman spectra transfer is presented. Experiment results show that, with a calibration source and a standard luminescence glass, this procedure can be applicable for all kinds of Raman spectra transfer.3. To transfer well-established Raman database, a spectra transfer method based on piecewise direct standardization (PDS) is proposed. This method needs no analysis of instrumental response function but a few transfer samples, to build a transfer model linking two Raman spectrometers. A Mahalanobis distance based Kennard-Stone method is used for transfer samples selection. Gasoline sample spectra acquired by a laboratory spectrometer are transferred to an online spectrometer, and a quantitative model predicting RON number is built with transferred sample. The experiment results show that, the transfer model performs as well as the rebuilt model, which means this transfer method for Raman spectra has practical value.4. A quantitative analysis method based on Raman spectra combination is proposed to avoid the limitation in commonly-used soft modeling and hard modeling methods. In this method, the mixture spectrum is represented by a linearly combination of each component spectrum, and each component content is determined by the weighting coefficient. Only a few real samples are needed for the calibration method, which is suitable to the analysis of sample which varies little. This method is used to predict methanol content in methanol gasoline, and the results are compared with a PLS model. It shows that, when the number of calibration samples is limited, the general performance is similar with PLS model, but the extrapolation risk is lower.5. An online Raman analyzer is introduced for process monitoring of methanol gasoline blending. The Raman spectrometer is placed in an operating room far from the production unit, and a couple of long optic Fibers are used to connect with the probe on the sample line. With separated structure, there are no electrical signal in production unit and no explosion-proof equipment for spectrometer. The above spectral data processing and quantitative analysis algorithms are implemented in the analysis software. Application results show that, the online analyzer is capable of continuously monitoring the methanol content in the production oil. With the advantages of rapid analysis, high accuracy and less maintenance, this analyzer can provide an efficient tool to the enterprise for stable and optimal production.