Application And Study Of (Near-) Infrared Spectroscopy On Pharmaceutical Analysis

Near-infrared spectroscopy (NIRS) would be occurred when there were absorption due to overtone and combination bands originating from fundamental molecules vibrations correspond mainly to C-H,N-H and O-H groups in compound. The use of NIRS by the agriculture, chemical food and so on has grown steadily as the result of its advantages over conventional alternative, it provides a non-destructive, non-invasive, reagent-free, no sample preparation, on-line method, which was developed greatly accompany with the development of chemometrics. General and main mode using NIRS is model-setting, model-verification, model-application and model-maintaining. Many conditions are needed for Model-setting, which is the classic method in using NIRS and has a lot of advantages, such as many calibration set are necessary and the results are gotten by other standard methods. Model can not be used if the conditions are not enough.PartⅠTo develop a new quantitative method without calibration set for NIR spectroscopy, based solely on the spectrum of the sample and the component. The new method, termed as minimal similarity method, depends on the searching of the second-derivative differential NIR spectral region with minimal similarity before statistical quantitative analysis is performed. The samples are from two sources: some samples are from W.Winig workgroup. Three components (methlylene chloride, dichloropropane and acetone) are selected. Simulation sample containing the three components were made by OPUS software. Other samples were self-made, a system contained acetone, methyl cyanide and ether, another system contained cyclohexane, petroleum benzine and chloroform. The contents of three components in three real samples from W.Winig workgroup are determined with average recovery of 100.9% and the relative standard deviation of 6.1% (n=9). Correspondingly the results from self-made samples are 93.5% and RSD=5.6%(n=5) except ether. Quite satisfactory results can be obtained by the most similar spectral region method without traditional calibration set. The new method provides an alternative for the quantitative analysis in NIRs.PartⅡAs a continue work of the method above, the correlation idea was applied on the modification of Local Straight-Line Screening (LSLS). LSLS algorithm is a newly designed method to detect synthetic drug(s) adulterated in herbal medicines based on infrared spectroscopy. Some modifications are made in this paper to improve the existing LSLS algorithm, including interpolation which can make the IR spectrum more concentrated without the noise enlarged, second derivation which can eliminate blank noises and baseline shift, change of calculation regions from 3 to 7 data points which represent the screened-out peaks character better. The algorithm was exemplified by five kinds of synthetic drugs (sibutramine, fenfluramine, lovastatin, sildenafil and methyldopa) in 40 herbal medicine samples and verified by LC-MS results. The concentration of the synthetic drug(s) predicted by the modified LSLS algorithm was closer to those determined by LC-MS. Consequently, the correct results rise from 30 by the original LSLS to 36 by the modified LSLS, the false negative responses drop from 5 to 1 in negative samples, the false positive responses from 5 to 3 in positive samples. These allows the method to be more appropriate for the primary screening of synthetic drugs suspiciously adulterated in herbal medicines, with high rapidity, accuracy and cost-effectiveness.The feasibility of fast and correctly detecting the quantity in FT-207 micro emulsion by NIR was studied. The blank micro emulsion contains water, lecithin, ethanol and isopropyl myristate (IPM). The real micro emulsion contains FT-207 besides blank micro emulsion. R2 (%) of correction equation on blank micro emulsion are 99.59 (water), 99.26 (lecithin), 98.83 (ethanol) and 99.91 (IPM). RMSECV (P) of that are 0.424 (water), 0.794 (lecithin), 0.820 (ethanol) and 0.490 (IPM). Correspondingly R2 (%) of correction equation on real micro emulsion are 99.75 (water), 98.37 (lecithin), 98.97 (ethanol), 99.93 (IPM) and 75.74 (FT-207). RMSECV (P) of that are 0.359 (water), 1.107 (lecithin), 0.767 (ethanol), 0.426 (IPM) and 2.43E-5 (FT-207). Accuracies in a day and between days are both between 95%～105%. Model-setting is the classic mode in using NIRS. The results proved again that NIR spectroscopy analysis technology has superiority in pharmaceutical analysis,Three part works were introduced in the paper: a new method (minimal similarity) has been developed; LSLS algorithm was modified and complex system was analyzed. Predicted results were gained.