Researches On Theory And Application Of Second-order Calibration Method And Its Development Of Application Software

The development of analytical instrument s can provide two-, three-orhigher-way response data array with more information for each sample,which inspires analysts and chemometricans to develop effective dataanalysis tools for the purpose of rapidly extracting the useful information.Due to the second-order advantage that quantifies analytes of interest evenin the presence of interferents, the second-order calibration has becom oneof the most popular activities in the world. In the present work, aiming toattack the drawbacks of the present second-order calibration in practicalapplications, the paper takes efforts to the development of newmethodologies and its applications in the filed with suitable degree ofcomplex. In addition, a new chemometric s oftware was developed. Thecontents of this parper can be summarized as:Chapter2: A comprehensive and sy stematic strategy for evaluating theperformances of several trilinear second-order calibration algorithms ispresented, in particular with a view of practical applications. Severaltrilinear second-order calibration methods such as parallel factor analysi s(PARAFAC), alternating trilinear decomposition (ATLD), self-weightedalternating trilinear decomposition (SWATLD) and alternating penalt ytrilinear decomposition (APTLD), which have the second-order advantageand are gaining widespread acceptance in the field of chemometrics, werecompared. Based on different input parameters including noise level, initialvalue, number of estimated components and collinearity in simulated andreal data, the performances of these methods were evaluated in terms ofpredict ing ability, consistency of resolved and real profiles, fitnessobtained by selected components and speed of convergence. The obtainedresults give a reevaluation of the position and role of these trilinearsecond-order calibration methods in chemometrics and provide guidance inpractical applications for solving complicated quantitative analysisproblems in analytical chemistry.Chapter3: A great number of trilinear algorithms which can obtain thesecond-order advantage (quantifies analytes of interest i n ther presence of interferents) have been proposed so far. However, each has darwbaks apartfrom its advantage. According to conclusion of the previous algorithmcomparison, there seems to be a complementary relationships among thesealgorithms. Therefore, a novel strategy for second-order calibration wasdeveloped on the basis of algorithms combination methodology in order toassemble the advantages of several of the most famous trilinearchemometric methodologies: ATLD, SWATLD and PARAFAC. Simulationand two real experimental data sets have demonstrated that the performanceof the novel method is better than that of any one of ATLD, SWATLD andPARAFAC in terms of either being sensitive to noise level or providingrobust results. Additionally, the new meth od is very stable with respect tonumber of components as long as the employed number is chosen to be equalor larger than the underlying one.Chapter4: In the case of complicated sample analysis, the presence ofmultiple interferences in background poses great challenge for analysts,which may lead the bias on the final solutions for analytes. In this chapter,we highlight a novel hybrid method on the basis of combining theaforementioned algorithm combination method with constraint backgroundbilineariza tion (ACM/CBBL) for the purpose of obtaining the well-knownsecond-order advantage. A significant advantage of the newly proposedmethod lies in the fact that1) the influences of nonlinear factors and noiseon the solutions of analytes can be partly if no t completely eliminated byintroducing excessive factors, and2) thus more reliable qualitative andquantitative results of analytes can be retrieved even multiple interferentswere presented in background. The theoretical application of theACM/CBBL has b een demonstrated with a simulated data and realfluorescence data sets. Finally, this novel method has been successfullyapplied for quantifying antibiotics in tap water using three-dimensionalfluorescence spectroscopy. Results indicate that the performan ce of the newmethod is comparable with that of bilinear least squares coupled to residualbilinearization (BLLS/RBL).Chapter5: Peak alignment and overlapping peaks deconvoluiton ofcomplex biological chromatographic data remain challenges in biologicalfluid studies. In this chapter, we highlight a novel peak alignment method,abstract subspace difference (ASSD), to improve the alignment quality for the complex samples containing overlapping interferences. The basics ofthe methodology are that time shif ts can be reflected by the difference inthe corresponding abstract chromatograms. We also combine ASSD wit hsuch chemometric trilinear decomposition methodology as ATLD to analyzesecond-order (three-way) chromatographic data having serious coelutionpeak s. The performance of the new method was evaluated on a simulateddata with chemically dependent time shifts as well as a complex biologicalsystem. The results indicated that ASSD can accurately align peak shiftsregardless of the presence of interference s. After alignment, ATLD cansuccessfully retrieve qualitative and quantitative results of analytes fromoverlapping peaks in the data. This approach is widely applicable forchromatographic studies and can be potentially suitable for comprehensivetwo-dimensional chromatographic data.Chapter6: The ubiquity of antibiotics in environment urges more rapidand sensitive analytical method for the purpose of assessing the risk ofthese compounds. Routine analysis of antibiotics has traditionally beenperformed using chromatographic methods. Prominent drawbacks of thes echromatographic methods are the lack of general selectivity as aconsequence of the occurrence of matrix effect that may change in variousanalytical systems. Aiming to address this problem, a ne w strategy based onthe combination of alternating trilinear decomposition with three-way datamodel (ATLD-TWDM) was developed in this chapter. The potential abilityof the proposed method was validated for the trace determination offourteen antibiotics i n two water matrices using a solid-phase extractioncoupled to high performance liquid chromatography with diode arraydetector. The results demonstrated that the combination of chromatographywith ATLD-TWDM has the advantages that1) quantifying analytes ofinterest regardless of the overlapped peaks, improving selectivity ofchromatography as well as significantly reducing the analytical time andcomplexity can be obtained;2) the generality of chromatographic separationcondition can be further enhanced, implying various analytical systems canbe analyzed by using the same chromatographic condition;3) the presenceof time-shifts and background drifts can be modeled by the new method.Moreover, ATLD-TWDM is very stable to component number provided thatthe number of components is chosen to be equal or greater than the real one. Chapter7: A strategy that combines trilinear decomposition methodswith high performance liquid chromatography with diode detector arra y(HPLC-DAD) has been proposed in chapter7to mathematically separate theoverlapped peaks and to quantify quinlones in honey samples. TheHPLC-DAD data were obtained within a short time in isocratic mode. Thedeveloped strategy could be applied to determine12quinolones even in thepresence of un calibrated interfering components in complex background.Besides, this work provided as an example and a guideline for analysts to bemuch more efficiency for using HPLC-DAD and to require less organicsolvents compared with the traditional methods. To acc ess the performanceof the proposed strategy for the determination of quinolones in honeysamples, the figures of merit were employed. The limits of detection for allanalytes were within the range1.2-56.7μg kg-1. The work presented in thispaper illustr ated the suitability and excellent confirmatory potential ofcombining second-order calibration method with second-order analyticalinstrument for multi-residues analysis in food samples.Chapter8: Three dimensional excitation emission matrix fluorescence(EEM) can provide more useful formation for single sample and has beenwidely used in various fields. However, the presence of nonlinear factorsand overlapped peaks pose great problem for users. In addition, there is stilla lack of the automated feature identification and resolution. In thischapter, the authors proposed a novel method for fluorescence data analysis.The method are capable of automatically identifying most relevantinformation of analytes of interest and can accurately estimated theunderlying chemical information involving excitation spectrum, emissionspectrum and relative intensities for each of underlying components. Thenew method has been used to analyze a fluorescence data that quantifies twoanalytes in human plasma. It successfu lly retrieves the qualitative andquantitative results for analytes and satisfactorily reconstructs the EEMs.