Loading...

Development and Application of Multivariate Curve Resolution Techniques for the Analysis of Complex Chromatographic Data

Parastar Shahri, Hadi | 2011

684 Viewed
  1. Type of Document: Ph.D. Dissertation
  2. Language: Farsi
  3. Document No: 42223 (03)
  4. University: Sharif University of Technology
  5. Department: Chemistry
  6. Advisor(s): Jalali Heravi, Mahdi
  7. Abstract:
  8. In recent years, chromatographic techniques as one of the most important analytical techniques have been greatly developed from instrumentation and data volume point of views. On the other side, multivariate curve resolution (MCR) techniques as a branch of chemometrics have potential for the analysis of multivariate and huge data sets. In addition, MCR methods have been developed very fast in these years and their potential have been shown in the analysis of different types of chemical data. Thus, combination of sophisticated chromatographic techniques (e.g. gas chromatography-mass spectrometry (GC-MS) or comprehensive two-dimensional gas chromatography (GC×GC)) with MCR techniques can be proposed as a powerful tool for the analysis of complex mixtures. In this thesis, development and application of MCR techniques have studied for the analysis of complex chromatographic data. In the first part, we have developed integrated software named MCRC for chemometric analysis of two-way chromatographic data. In this software, we have tried to include the most important chemometric methods developed in recent decade for the analysis of chromatographic data. This software gives an opportunity to non-expert users especially chromatographers to use chemometric methods in a user-friendly environment. From the MCR point of view, two common methods of heuristic evolving latent projection (HELP) and multivariate curve resolution-alternating least squares (MCR-ALS) are included in this software. In the second part, the potential of multivariate curve resolution-objective function minimization (MCR-FMIN) have been studied. A new procedure to calculate the initial estimate of rotation matrix is proposed and its effect along with other constraints on the extent of rotational ambiguity are studied using MCR-BANDS method. This MCR technique is proposed to assess co-elution problem in GC-MS analysis. In the third part, MCR techniques (MCR-FMIN and MCR-ALS) are used for the GC-MS analysis of the volatile components of saffron as the most expensive spice in the world. First the effective parameters of the ultrasonic-assisted solvent extraction (USE) are optimized using factorial-based response surface methodology (RSM). Then, in the optimum conditions, different samples from different regions of Iran are analyzed by GC-MS. MCR is proposed to solve different problems associated to their GC-MS analysis. Finally, an index has been obtained for the composition of volatile components in Iranian saffron. Parts fifth and sixth are dedicated to study the potential of independent component analysis (ICA)-based methods for multivariate resolution purposes in analytical chemistry. First, mean-filed ICA (MFICA) is used for the extraction of more information from the GC-MS analysis of essential oils. A new strategy based on design of experiment is proposed for the optimization of MFICA algorithm. Then, in a theoretical study, potential of ICA methods and the meaning of independence concept are studied from an analytical chemistry point of view. In this way, the statistical meaning of independence measured by mutual information (MI) and probability density function (PDF) are related to its chemical meaning. Finally, the ICA and MCR results are compared using MCR-BANDS by considering the presence of rotational ambiguity. Two final parts of this thesis are devoted to analysis of three- and four-way GC×GC data by using MCR methods. There are two different situations to handle in GC×GC analysis. The first situation is in the case of GC×GC coupled to multivariate detector response like time-of-flight mass spectrometry (TOFMS) which is used for the qualitative and quantitative analysis of polycyclic aromatic hydrocarbons (PAHs) in aromatic fraction of heavy fuel oil (HFO). In this case, first four-way data (two time axes, one spectral axis and one sample axis) is column-wise super-augmented and then analyzed by MCR-ALS under proper constraints. In this way, the retention time shifts within and between runs can be handled during MCR analysis without need to correct them before analysis. In addition, baseline can be modeled instead of correction in this strategy. The second situation is in the case of GC×GC coupled to univariate detector response such as flame ionization detector (FID) or MS in TIC mode. In this case, the retention time shifts is more severe than previous case due to the presence of two irreproducible chromatographic directions. First, a new method called bilinear peak alignment is proposed for correction of retention time shifts within run. In addition, retention time shifts between runs is corrected using correlation optimized shifting (COShift) in one chromatographic direction. Finally, the three-way corrected data is column-wise augmented and then analyzed by MCR-ALS to obtain resolution and quantitative results. The potential of this method is tested for the analysis of GC×GC-TIC data of PAHs and GC×GC-FID analysis of gasoline.

  9. Keywords:
  10. Chromatography ; Gas Chromatography/Mass Spectrometry (GC/MS) ; Multivariate Curve Resolution-Alternative Least Square (MCR-ALS) ; Independent Component Analysis (ICA) ; Multivariate Curve Resolution

 Digital Object List

 Bookmark

No TOC