| With the development of modern scientific instruments and chemometrics,molecular spectral analysis technology(including infrared spectroscopy,near infrared spectroscopy and Raman spectroscopy)can quickly and simultaneously determine various of physical and chemical properties of substance,and is widely used in many fields as a process analysis technology.At present,near infrared spectroscopy is widely used in process analysis,which is mainly used for homogeneous liquid with good fluidity and for solid powder,but can not be used for rapid analysis of porridge and sticky materials.The reason is that this kind of material not only has the problems of sampling and cleaning,but also has uneven distribution of ingredients.When the spectrum is collected repeatedly,the measured optical path is difficult to keep consistent.The obvious optical path error causes the spectrum to deviate from Beer’s law,which seriously affects the prediction ability of quantitative analysis model.The common spectral pretreatment methods can not eliminate this influence.Therefore,it is one of the research contents in this paper to solve the technical problems of rapid analysis of porridge and sticky materials.Raman spectroscopy is also commonly used in process analysis.However,when Raman signal is excited by laser,fluorescence is also excited,and fluorescence background affects the accuracy of analysis results.Therefore,a method of eliminating the influence of fluorescence background is one of the other research contents in this paper.The main research contents and results of this paper are as follows:1.Study on the method for the instant determination of cellulose slurry/solution by attenuated total reflection infrared spectroscopyIn this paper,the spectrum collection and quantitative analysis methods of lyocell fiber were studied in order to solve the problem of rapid quantitative analysis of cellulose slurry/solution.Attenuated Total Reflection Infrared Spectroscopy(ATR-IR)is selected as the analysis signal,which solves the problem of spectrum collection when using near infrared spectroscopy to analyze porridge and glue solution.However,it is found that there are obvious optical path errors in repeatedly collecting ATR-IR of cellulose slurry(uneven distribution of internal components)and viscous solution(easy to entrap bubbles),which seriously affects the prediction performance of PLS model.Therefore,a processing method of converting absorption spectrum into angle spectrum is proposed,in which the angle spectrum is not affected by optical path error.Several PLS quantitative models of cellulose,water and NMMO content in porridge and glue solution were established by using raw spectrum,common calibrated spectrum and angle spectrum,respectively.The results show that the performance of the model established by the raw spectrum and the common calibrated spectrum can not meet the control accuracy of lyocell fiber production process,and the influence of optical path error on the model performance can be effectively eliminated based on angular spectrum modeling.Among them,the SEC and SEP values of each component content PLS model established by using first principal component spectrum PC1 as the angular spectrum calculated by reference spectrum are obviously lower than the reproducibility error specified by reference method,which is 0.5,successfully solving the technical problem of quickly determining the component content of congee and viscous substances.2.Study on Raman fluorescence background removal methodRaman spectroscopy has the advantages of convenience,rapidity and nondamage,and has great development potential in the field of process analysis.However,for substances with high fluorescence yield,intense fluorescence is generated with Raman signal at the same time,which brings adverse effects on quantitative and qualitative analysis.The commonly used spectral preprocessing methods can not eliminate the fluorescence background reasonably and effectively,and it is still an urgent problem to remove the fluorescence background in Raman signal.Therefore,an iterative moving window projection of segmentation right triangle(IMWPSRT)method is proposed to determine gasoline octane number by Raman spectroscopy.The fluorescence background of gasoline Raman spectrum is corrected by common fluorescence pretreatment method and IMWPSRT method,respectively and the quantitative models of gasoline octane number(MON and RON)is established by combining with PLS method.The results show that after the Raman spectrum is preprocessed by IMWPSRT method,the corrected spectrum has a characteristic peak corresponding to the raw spectrum,and the baseline is 0.It is verified that IMWPSRT method can accurately separate fluorescence from Raman signal without losing important Raman information.SEP of RON/MON quantitative model established by IMWPSRT method combined with PLS method is 0.37 and 0.56,respectively and with a repeatability of 0.1 and 0.2 respectively,and meets the repeatability requirements of GB/T 5487-2015 method,which further verifies the accuracy and effectiveness of removing fluorescence background by this method,significantly improves the prediction accuracy of gasoline octane number quantitative model,and has important practical significance for gasoline quality process control.3.Apply IMWPSRT method to surface enhanced Raman fluorescence background correctionSurface enhancement technology can enhance Raman scattering signals by several orders of magnitude,which significantly improves the detection sensitivity of Raman spectrum.However,like conventional Raman,the fluorescence intensity decays with time,which makes the reproducibility of SERS spectrum extremely poor,which is an important reason for limiting the application of SERS spectrum technology in quantitative analysis.At present,the commonly used spectral preprocessing methods can not accurately and reasonably eliminate the fluorescence background in SERS spectrum.Therefore,this section focuses on rhodamine B based on silver sol substrate.Adaptive iterative least squares(air-PLS)and IMWPSRT methods were used to correct 10 consecutive rhodamine B SERS spectra,and the relative standard deviations(RSD)of the intensities of three characteristic peaks(622,1363 and 1649cm-1)in the repeated spectra were calculated.The research shows that the baseline of SERS signal corrected by IMWPSRT method is obviously 0,and all main characteristic peaks are not lost.The RSD values at 622,1363 and 1649cm-1 characteristic peaks in repeated SERS spectrum are reduced from 33.4%,38.8%and 21.9%after original smoothing to 11.2%,18.6%and 13.9%,respectively,which are significantly lower than that of original smoothing.Compared with common methods,IMWPSRT method can effectively remove the fluorescence background in SERS spectrum and improve the reproducibility of its detection,which has a good significance for the application of surface enhanced Raman spectroscopy in quantitative analysis. |
PDF Full Text Request