A simple and green chemometrics-assisted spectrophotometric technique has beendeveloped and validated for the determination of antipyrine (ANT) and benzocaine HCl (BEN) along with the official impurity of ANT, antipyrine impurity A (ANT imp-A), and the degradation product of BEN, p-amino benzoic acid (PABA), in their quaternary mixture. Three models were developed and compared: partial least squares (PLS), artificial neural networks (ANN), and multivariate curve resolution-alternating least squares (MCR-ALS) where the four studied drugs were successfully quantified. The quantitative determination of the studied drugs was assessed using percentage recoveries, standard errors of prediction, and root mean square errors of prediction. The ANN model demonstrated the lowest error and the best correlation making it the most accurate method for analysis. The models were constructed in the ranges of 5.0-9.0 µg mL-1 for ANT, 1.0-5.0 µg mL-1 for BEN, 0.5-2.5 µg mL-1 for ANT imp-A, and 0.25-1.25 µg mL-1 for PABA. The established models successfully determined ANT, BEN, ANT imp-A, and PABA with detection limits of 0.312, 0.178, 0.093, and 0.042 µg mL-1 for PLS, 0.185, 0.085, 0.001, and 0.034 µg mL-1 for ANN; and 0.473, 0.240, 0.073, and 0.069 µg mL-1 for MCR-ALS, respectively. The greenness and the whiteness of the proposed method were assessed using two green evaluating approaches: analytical Eco-scale, and AGREE, along with one white analytical chemistry evaluating tool, RGB. The three proposed models were successfully applied for determination of ANT and BEN in their pharmaceutically co-formulated dosage forms. They are also recommended for stability assays and purity testing of these drugs in quality control laboratories.
Keywords: Antipyrine; Artificial neural network; Benzocaine HCl; Green analytical chemistry; Multivariate curve resolution-alternating least squares; Partial least squares.
© 2024. The Author(s).