Full text
| |
| Author(s): |
Marcos Viero Guterres
Total Authors: 1
|
| Document type: | Doctoral Thesis |
| Press: | Campinas, SP. |
| Institution: | Universidade Estadual de Campinas (UNICAMP). Instituto de Química |
| Defense date: | 2006-08-23 |
| Examining board members: |
Marcia Miguel Castro Ferreira;
Paulo Roberto Gardel Kurka;
Roberto Kawakami Harrop Galvão;
Celio Pasquini;
Francisco Benedito Teixeira Pessine
|
| Advisor: | Marcia Miguel Castro Ferreira |
| Abstract | |
On this work, multivariate analyses were used to develop methodologies for monitoring chemical reactions. Two techniques were used: Multivariate Curve Resolution (MCR) for analysis of spectroscopic measures and the Multivariate Statistical Process Control (MSPC) for the analysis of the physical-chemical variables. The first work developed through this dissertation used the MCR methods to build empirical models of chemical reactions from reflectance measures of infrared spectra. This model was used to predict the behavior of new batches of the same reaction. The obtained results showed that the predictions are more exact when the model is adapted according to the (PACLS) Predicted Augmented Classical Least Squares to include a component in the model which model the baseline changes of the infrared spectrum. The second work presents a new methodology developed for determining the creatinine in human serum, based on the monitoring of Jaffé reaction with visible spectroscopy and analysis of the data with the treeway MCR methods. According to the obtained results, creatinine can be determined in presence of others substances which interferes on its spectroscopic measure. The Standard Error of Performance (SEP) on this essay was 2,22mg L with a respective coefficient of variation around 6%. The third work presented on this dissertation describes a methodology developed for investigating the fluorescent and intracellular metabolite produced by Chromobacterium Violaceum during fermentation conducted in bioreactor under aerobics conditions. The fluorescence spectra collected from biomass samples were analyzed through the PARAFAC method (Parallel Factor Analysis). The tryptophan and tyrosine aminoacids beyond NADH (nicotinamide adenine dinucleotide) and violacein were identified through the analysis of the parameters from trilinear model obtained by analysis of fluorescence spectra. The estimations obtained for the concentration profiles of the chemical species, showed that this bacterium consumes the tryptophan before starting the progressive production of violacein. This result is according to the mechanism proposed on the literature for the biosynthesis of this pigment, which is a candidate to tumors and tropical disease treatment. The forth work on this dissertation presented a methodology of MSPC development for the analysis of autocorrelated data. This methodology was used to monitor the digestion of bauxite, used to extract de aluminum of this ore. Four Physical-chemical variables from this process were analyzed by time series models to model the autocorrelation among measures collected from this process. In a subsequent step of analyses, the residuals of time series models were analysed by Principal Component Analysis (PCA) to obtain a bilinear model that was used to construct multivariate control charts. The obtained results showed that the industrial plant operated out of statistical control (AU) | |