Identifier: TDX:787
Authors: Garrido, Mariano Enrique
Abstract:
This Thesis, entitled: 'Quantitative monitoring of epoxy resins reactions by near infrared spectroscopy and curve resolution methods' has two main objectives. Firstly, to evaluate the capability of near infrared spectroscopy, assisted by Multivariate Curve Resolution-Alternating Least Squares meted, to quantitative monitor curing reactions of epoxy resins. This objective also involves verifying if the combination of these techniques (both, instrumental and chemometric) is able to detect all the sources of variability present in the system under study. Likewise, this objective involves the quantitative determination of the concentration of the species that take place in the reaction, as well as their corresponding spectral profiles. As a consequence, a study related to the uncertainty of the results and to the appropriate validation strategies must be designed. The second principal objective is to estimate the kinetic rate constants liked to the reactions of epoxy resins. This objective involves the study of the curing mechanism and the relationship between this mechanism and the final properties of the polymerization product.Epoxy resins are among the most important polymer materials in use because of their chemical resistance, notable adhesive characteristics, and good mechanical and physical properties. The final properties of the polymer depend on the chemical structure of the epoxy resin and the curing agent (in this case, the amines), but also on the curing process. Therefore there is a need for new analytical methods that are able to monitor the curing reactions.Near infrared spectroscopy is an attractive tool for monitoring the curing reactions and has certain advantages: is a non-destructive technique and has a wavenumber domain in which the absorption bands are overtones or combinations bands with lower absorptivities. This means that a relatively large quantity of sample can be used and that the kinetics of the curing process can be studied in situ. These characteristics make it possible to control the quality of the polymerization product in real time and to extract information about the curing reaction.Using curve resolution methods (called also 'soft-modelling methods') makes it possible to analyse the spectral data in a multivariate way. In this way, more information about the system is achieved. One condition, necessary for the correct application of this methodology, is that the data must have a bilinear structure, i.e. the experimental data matrix should be expressed as the product of two matrices; one matrix of concentrations and one matrix of pure spectral responses of the species present in the system. Thus, by this matrix decomposition, it is possible to know the evolution of the species that take part in the process, as well as their corresponding pure spectra.In this Doctoral Thesis, these techniques are applied to a set data obtained from the near infrared monitoring of the epoxy resins reactions, in order to obtain useful information (both qualitative and quantitative) about the species involved in the reaction process as well as about the reaction mechanism and the kinetic parameters linked to it. Despite the advantages of the multivariate data analysis, there are some problems related to the structure of the spectroscopic data such as the rank deficiency. This problem is common when the spectroscopic data come from evolving systems. Rank deficiency happens when the number of factors found in the set of data is lower than the number of absorbing species. In this Thesis, we have applied two different strategies to solve the rank deficiency problem.Also, when decomposition of bilinear data matrices is carried out, there is a ubiquitous problem named ambiguity. There is to types of ambiguities: rotational and intensity ambiguities. This means that, instead of unique solutions, a set of feasible solutions is obtained, that fit the experimental data equally well. Part of the work carried out in this Thesis is focused to the estimation of these ambiguities and minimizing them. The results obtained by the combination of near infrared spectroscopy and the soft modelling methods have been contrasted with other technique, already well established for the monitoring of this kind of reactions, such as high performance liquid chromatography (HPLC) and other techniques, such as 13C Nuclear Magnetic Resonance, which have not been used until now to quantitative monitor reactions of epoxy resins.Also, the kinetic study of model reactions of epoxy resins (i.e. reactions that do not polymerize) has been carried out. This study involves the identification of the correct reaction mechanism and obtaining the corresponding kinetic parameters, i.e. the kinetic rate constants. The kinetic study has been performed by combining soft and hard modelling techniques. Also, a brief discussion about extrapolating these techniques to study epoxy-amine systems that evolve toward crosslinked products is included.
Repositori URV