Univariate vs. Multivariate Calibration in the Quantification of Carbamazepine in Tablets by Raman Spectroscopy using PCA as Spectral Selection Tool

Author(s)

de Menezes, Daniela R. ^[1] ; Carneiro, Renato L. ^[2]

Affiliation

^[1] Univ Fed Sao Carlos, UFSCar, CCET, Dept Chem Engn, BR-13565905 Sao Carlos, SP - Brazil

^[2] Univ Fed Sao Carlos, UFSCar, CCET, Dept Chem, BR-13565905 Sao Carlos, SP - Brazil

• Document type: Journal article
• Source: Latin American Journal of Pharmacy; v. 31, n. 9, p. 1341-1347, 2012.
• Web of Science Citations: 1

Abstract

When Raman spectroscopy is employed to quantify pharmaceutical compounds, one typically resorts to the multivariate calibration, in order to minimize errors arising from the scattered signal fluctuation. However, molecules of traditional excipients used in tablet formulation emit low Raman scattering, allowing the use of univariate calibration, which is simpler to achieve. In this study, the quantification of carbamazepine in tablets by Raman spectroscopy was performed using univariate and multivariate calibration. To minimize the problem of signal intensity variation, the most representative replicates were selected by Principal Component Analysis (PCA). Univariate and multivariate calibration curves were obtained by simple linear regression and Partial Least Squares (PLS), respectively, resulting in Root Mean Square Errors of Cross Validation (RMSECV) of 4.24 and 3.42 %. The results of this procedure was satisfactory and indicates that with the aid of appropriate numerical treatment, it is possible to perform a simple and reliable method of quantification even when using the univariate approach. (AU)

• FAPESP's process: 10/16520-5 - Application of calibration chemometrics methods and multivariate curve resolution in Raman spectroscopy for qualitative, quantitative and polymorphism analyses in medicines
• Grantee: Renato Lajarim Carneiro
• Support Opportunities: Regular Research Grants