Data fusion of middle-resolution NMR spectroscopy and low-field relaxometry using the Common Dimensions Analysis (ComDim) to monitor diesel fuel adulteration

Autor(es)

de Aguiar, Leticia Magalhaes ^[1] ; Galvan, Diego ^[1] ; Bona, Evandro ^[2] ; Colnago, Luiz Alberto ^[3] ; Killner, Mario Henrique M. ^[1]

Afiliação do(s) autor(es)

^[1] Univ Estadual Londrina, Dept Quim, POB 10-011, BR-86057970 Londrina, Parana - Brazil

^[2] Univ Tecnol Fed Parana, Programa Posgrad Tecnol Alimentos, Campus Campo Mourao, BR-87301899 Campo Mourao - Brazil

^[3] Embrapa Instrumentacao, Rua 15 Novembro 1452, BR-13560970 Sao Carlos, SP - Brazil

• Tipo de documento: Artigo Científico
• Fonte: Talanta; v. 236, JAN 1 2022.
• Citações Web of Science: 0

Resumo

Medium-resolution (MR-NMR) and time-domain NMR relaxometry (TD-NMR) using benchtop and low-field NMR instruments are powerful tools to tackle fuel adulteration issues. In this work, for the first time, we investigate the possibility of enhancing the low-field NMR capability on fuel analysis using data fusion of MR and TD-NMR. We used the ComDim (Common Dimensions Analysis) multi-block analysis to join the data, which allowed exploration, classification, and quantification of common adulterations of diesel fuel by vegetable oils, biodiesel, and diesel of different sources as well as the sulfur content. After data exploration using ComDim, classification (applying linear discriminant analysis, LDA), and regression (applying multiple linear regression, MLR), models were built using ComDim scores as input variables on the LDA and MLR analyses. This approach enabled 100% of accuracy in classifying diesel fuel source (refinery), sulfur content (S10 or S500), vegetable oil, and biodiesel source. Moreover, in the quantification step, all MLR models showed a root mean square error of prediction (RMSEP) and the residual prediction deviation (RPD) values comparable to the literature for determining diesel, vegetable oil, and biodiesel contents. (AU)

• Processo FAPESP: 13/25802-2 - Desenvolvimento e validação de métodos simples, rápidos e não destrutivos de RMN em média e baixa resolução para a caracterização e controle de qualidade de combustíveis derivados do petróleo e biocombustíveis.
• Beneficiário: Mario Henrique Montazzolli Killner
• Modalidade de apoio: Bolsas no Brasil - Pós-Doutorado