| Chemometdcs is a rising interdisciplinary subject. It has many advantages in.extracting and analyzing information, analyzing chemical test data, distinguishingcomplicated spectra and chemical classification etc. The combination ofchemometrics and modem analytical means would offer a new way to solve someproblems about food safety and food inspection.This paper was divided into six parts. The chemometrics combined withspectrophotometry and fluorescence method was applied in simultaneous analysis oforganic pesticide residues and food additives, which widened the application ofehemometric method. Satisfactory results demonstrated the feasibility of method.Chapter oneThe chapter briefly summarized the development of organic pesticide residuesand food additives, reviewed the analytical methods used in organic pesticide residuesand food additives especially the application of united use of chemometric methodsand modem spectrum technology. It included classical least squares (CLS), multiplelinear regression (MLR), principal component regression (PCR), partial least squares(PLS) and artificial neural network (ANN), which showed the foreground ofchemometrics in food analysis.Chapter twoThree dithiocarbamate fungicides, ferbam, ziram and maneb were simultaneouslydetermined in this chapter. Ferbam, ziram and maneb were three antiseptic whichwere highly similar in molecular structure. The procedure is based on the threedithiocarbamate fungicides individually reacts with phenylfluorone to form abluecolored complex in presence of CTMAB. These three compounds had maximumabsorption peaks at 571 nm, 551 nm and 554 nm respectively and the spectraoverlapped strongly. Kinds of chemometric methods were used to analyze the absorption data of 500～700 nm, it found that the results based on PLS when theoriginal spectra data attributed to first-derivation were better than that untreated byderivation. Thus, a simultaneous PLS-derivative speetrophotometric method wasfirstly developed for the determination of ferbam, ziram and maneb. The linear rangesof the calibration curves for Ziram, ferbam and maneb are 0.2～6.0μg-mL-1, 0.2～5.5μg·mL-1 and 0.2～3.8μg·mL-1. The limits of detection are 0.19μg·mL-1 for ziram, 0.15μg·mL-1 for ferbam, 0.19μg·mL-1 for maneb. The proposed method was simple,needed no separation of samples and the recovery rate were among 85.3％-109.7％when applied to samples such as rice, fruit and tap water.Chapter threeIn this chapter, the multivariate calibration and ANN of chemometrics combinedwith spectrophotometry were investigated for synchronous quantity determination ofsix food additives sodium benzoate, sorbic acid, saccharin sodium; sodium nitrate,sodium nitrite and vanillin which had seriously overlapped, spectra. The influence ofacid to absorption spectra has been probed. Various media including Sodiumcitrate-Chlorhydric acid, Britton-Robinson and potassium dihydrogenphosphate-sodium borate (PDP-SB) buffer solution were introduced to examine theireffects on the experiment. The results indicated that absorbance in B-R buffer (pH2.85) solution had best sensitivity and was selected as medium. The linear ranges ofthe calibration curves for sodium benzoate, sorbic acid, saccharin sodium, sodiumnitrate, sodium nitrite and vanillin are 0.2～7.7 mg·L-1, 0.1～3.85 mg·L-1, 0.2～6.1mg·L-1, 0.3～6.1 mg·L-1, 0.3～6.1 mg·L-1 and 0.2～7.7 mg·L-1, respectively. The limitsof detection are 0.13 mg·L-1, 0.067 mg·L-1, 0.15 mg·L-1, 0.21 mg·L-1, 0.21 mg·L(-1) and0.094 mg·L-1, respectively, and linear correlation coefficient at 0.9995～0.9999. It wasdifficult to determine the compounds individually from their mixtures by classicalspectrophotometric method for the absorption spectra of these six food additivesoverlapped strongly. Therefore in this research, multivariate calibration, ANN andother chemometrics approaches were applied to analyze the overlapped spectra of thesix compounds complex systems. Through the research to the synthetic samples ofthe six compounds, it was found that RBF-ANN is the best method to yield accurate analysis results with its relative prediction errors (RPET) at 5.1％and the recoveries at93.2％～99.5％; CLS, PCR and PLS models gave relatively higher RPET and％Recovery values as compared to ANN methods. RPET of CLS, PCR and PLSmethods are 6.3％, 6.2％and 6.1％, respectively and the recoveries at 88.8～103.6％.Satisfactory results were obtained when utilizing the set-up RBF-ANN assistingspectrophotometry to directly and simultaneously detect six.food additivesin realsamples.Chapter fourChinomethionat, Triazophos, Coumaphos and Carbendazim have the strongendogenesis fluorescence. And their spectra overlapped seriously, and it is difficult todetermine them individually from their mixtures without a pre-separation. A simplemethod was proposed for simultaneous fluoprescence spectrophotometrydetermination of chinomethionat, triazophos, coumaphos and carbendazim usingchemometrics. The experimental conditions were investigated and the spectra from260 to 440 nm were scanned. The obtained data matrices were. processed bychemometrics. The linear ranges of the calibration curves for chinomethionat,triazophos, coumaphos are 0.024～0.432μg·ml-1,and for carbendazim is 0.048～0.768μg·ml-1. The limits of detection for chinomethionat, triazophos, coumaphos andcarbendazim are 0.020μg·ml-1, 0.017μg·ml-1, 0.018μg·ml-1 and 0.036μg·ml-1,respectively, and linear correlation coefficient at 0.9990～0.9993. It is difficult todetermine the compounds individually from their mixtures. Henceforth in thisresearch, multivariate calibration, ANN and other chemometrics approaches wereapplied to analyse the overlapped spectra of the four compounds complex systems.Through the research to the synthetic samples of the four compounds, it was foundtha PCR is the best method to yield accurate analysis results with its relativeprediction errors (RPET) at 7.5％. RPET of RBF-ANN and PLS methods are 8.2％and9.5％, respectively. The proposed method was applied to the analysis of water, rice,cabbage and apples with satisfied results.Chapter five A simple method was proposed for simultaneous spectrophotometric determinationof two herbicides, atrazine and cyanazine using principal component regression andpartial least square.They are the triazine herbicide of the similar molecular structure.Atrazine or Cyanazine reacts with pydidine and coupled with p-aminoacetophenone(PAAP) to form a yellow orange polymethine dye in certain condition.Theexperimental conditions were investigated and the spectrum from 400 to 600 nm werescanned. The absorption spectrums of these compounds, are overlapped seriously, andquantitative estimations cannot be carried out successfully without a pre-separation.In this paper, the conventional and first-derivative spectrograms of the mixtures wereused to perform the optimization of the calibration procedure by chemometric models.Several chemometrics methods are applied viz., CLS, PCR, PLS and RBF-ANN, andtheir performance of prediction are compared. The linear ranges of the calibrationcurves for atrazine and cyanazine are 0.2～3.5μg·ml-1 and 0.3～5.0μg·ml-1. The limitsof detection are 0.099μg·ml-1 for atrazine, 0.16μg·ml-1 for eyanazine. The proposedmethod was applied to the analysis of food samples with satisfied results.Chapter sixThe comprehensive summary and precis for our research was mainly noticed.
|
PDF Full Text Request