Modelagem quimiométrica para análise de óleos essenciais de lúpulo: quantificação de β-mirceno e análise exploratória de perfis químicos

• Ano de defesa: 2024
• Autor(a) principal: Marden Claret Fontoura Teixeira
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Minas Gerais; Brasil; ICX - DEPARTAMENTO DE QUÍMICA; Programa de Pós-Graduação em Química; UFMG
• Palavras-chave em Português: Cerveja; Mirceno; Calibração multivariada; Infravermelho médio; Cromatografia gasosa; Química analítica; Quimiometria; Terpenos; Essências e óleos essenciais; Análise cromatográfica; Espectroscopia de infravermelho
• Link de acesso: http://hdl.handle.net/1843/66312; https://orcid.org/0000-0002-8602-8164

This study aimed to develop a chemometric PLS (Partial Least Squares) model for quantifying β-myrcene in hop essential oils using ATR-FTIR (Attenuated Total Reflectance Fourier Transform Infrared) spectroscopy. Additionally, an exploratory analysis of the data was conducted through PCA (Principal Component Analysis) and HCA (Hierarchical Cluster Analysis) models, using gas chromatography for obtaining the reference values. The extraction of essential oils was performed according to ASBC (American Society of Brewing Chemists) standards using a Clevenger system. The developed chromatographic analysis methodology required a small sample volume of only 20 µL, with a runtime of 18 minutes. The linear regression of the chromatographic method was statistically validated within a range of 0.75-1.75% v/v. Intermediate precision and accuracy were assessed according to ISO 17025 validation guidelines. PCA and HCA models proved highly useful in identifying patterns of oil samples clusters, not only concerning β-myrcene content but also related to the contents of other compounds. The PLS model demonstrated good performance according to the estimate of proper figures of merit, yielding calibration and validation RPD (Ratio of Performance to Deviation) values of 2.60 and 2.59, respectively, with an RMSEP/RMSEC (Root Mean Squares Errors of Prediction and Calibration) ratio below 2.5, an indication of absence of overfitting. Consequently, the PLS model can be utilized as a screening method, leading to substantial savings in sample preparation time, analysis costs, and solvent consumption, aligned with green chemistry concepts.