Determinação de cloridrato de propranolol em medicamentos por espectroscopia no infravermelho com calibração multivariada (PLS)

• Ano de defesa: 2005
• Autor(a) principal: Zeni, Diego
• Orientador(a): Flores, Érico Marlon de Moraes
• Banca de defesa: Bittencourt, Celso Figueiredo
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Santa Maria
• Programa de Pós-Graduação: Programa de Pós-Graduação em Química
• Departamento: Química
• País: BR
• Palavras-chave em Português: Química; Química analítica; Espectrometria; Quimioterapia; Medicamentos; Qualidade; Calibração multivariada
• Área do conhecimento CNPq: CNPQ::CIENCIAS EXATAS E DA TERRA::QUIMICA
• Link de acesso: http://repositorio.ufsm.br/handle/1/10478

In this work calibration multivariate models were developed with the use of the technique of partial least scores (PLS) for the dosage of tablets of propranolol hydrochloride using infrared spectroscopy. Samples were ground and homogenized in cryogenic mill in order to have all the samples in the same conditions in relation to the particle size and to avoid problems related to the samples heterogeneity. Sample masses of 45.0 ± 2.0 mg of tablets were used and the concentration of reference tablets were between 0.1 and 0.45 mg of propranolol hydrochloride for mg of tablet. Eighteen calibration samples and 8 validation samples were used, for which were obtained spectra in 5 replicatas. The 3 more similar replicates were chosen by the analysis of HCA. The pre-processing and used pre-treatments were: autoscaling of the data, multiplicative scatter correction and first and second derivatives. Five models were obtained with good capacity of prediction of the concentrations of propranolol hydrochloride in the tablets. These models presented the coefficient of linear correlation higher than 0.92 and the root mean standard error of cross validation (RMSECV) smaller than 0.020. By the ANOVA test was verified that there was not significant difference between the models with a confidence level of 95%. The proposed procedure was fast, cheap, allowing good accuracy and it can be easily adapted to the quality control of the pharmaceutical industry. In addition, the proposed methodology doesn't generate dangerous chemical residues, because it is a technique non destructive and it doesn't use solvents for your application.