Full text
| |
| Author(s): |
Douglas Rozendo da Silva
Total Authors: 1
|
| Document type: | Master's Dissertation |
| Press: | São Paulo. |
| Institution: | Universidade de São Paulo (USP). Escola Politécnica (EP/BC) |
| Defense date: | 2023-10-25 |
| Examining board members: |
Reinaldo Giudici;
Dennis Chicoma Lara;
Maria Giuliana Torraga Uchiyama
|
| Advisor: | Reinaldo Giudici |
| Abstract | |
Emulsion polymerization is a largely and increasingly used technique in industrial production of paints, resins, among others, due to its versatility and capacity of producing products with different properties. For improving emulsion polymerization processes, it is necessary the development of techniques for in-line monitoring, including the monitoring of particle size of the polymer produced during the reaction. NIR spectroscopy, combined with fiber optic sensors, has shown potential for this application; however, the data obtained by this technique is not a direct measurement of the desired variables: it is necessary to develop a calibration model to correlate spectra data with the desired properties to be monitored. The fitting of a good calibration model is crucial for the success of this method and, although robust models are generally fitted, a model that could be applied for different formulations and process conditions has not been found yet. In this work, 11 homopolymers and 3 copolymers were synthesized using the monomers Methyl methacrylate (MMA) and styrene (St), and then their spectra were analyzed in the NIR. The spectral region located between 9000 and 14000 cm-1 showed high sensitivity to the evolution of the particle size of the emulsion polymerization tests performed. With the creation of calibration models using principal component regression (PCR), partial least squares regression (PLS) and artificial neural networks (ANN) developed with two databases of different spectral bands (14000 4000 cm-1 e 14000 9000 cm-1), satisfactory results were obtained in terms of coefficient of determination (R2) and mean square error of the cross-validation (RMSECV), indicating good precision and performance. Among the models obtained, it is worth highlighting that the diameter model obtained from ANN and the spectral database from 14000 to 4000 cm-1 obtained the best performance. Thus, the results showed that the development of a robust calibration model is possible, which is valid for monitoring particle size in different formulations (monomers) and process conditions. (AU) | |
| FAPESP's process: | 20/00332-7 - Systematic study of the relation between NIR spectrum and particle size of polymer for application in the monitoring of emulsion polymerization processes |
| Grantee: | Douglas Rozendo da Silva |
| Support Opportunities: | Scholarships in Brazil - Master |