Eletroextração multifásica e ferramentas quimiométricas aplicadas a análise de corantes antimicrobianos em água superficial por espectrofotometria UV-VIS e imagem digital

Abstract

Antimicrobial dyes such as malachite green, gentian violet, and methylene blue are environmental micropollutants whose analysis presents significant analytical challenges due to low monitored concentrations and matrix complexity. Despite prohibitions, these dyes continue to be irregularly used in aquaculture for infection prevention. The present study aimed to develop efficient analytical methods, proposing the use of simple, rapid, accessible techniques with high analytical frequency. To achieve this, multiphase electroextraction was employed in sample preparation, while for analytical determination was utilized digital image analysis and spectrophotometry associated with chemometric tools. In the first application, magnesium silicate was employed in multiphase electroextraction, proving pivotal for its integration with digital image analysis, therefore being a suitable alternative for the analysis of cationic compounds as it is cost-effective, easy to handle, and known to have a negative residual charge. Furthermore, a discrimination method using PLS-DA was proposed, effectively distinguishing between contaminated and uncontaminated samples, even at extremely low levels (0.2 μg L–1) within complex surface water matrices, obtaining a remarkable accuracy rate exceeding 95%. In the second application, dental paper points and an automated electroextraction system facilitated increased practicality and reproducibility of the technique. Additionally, a one-class classification method was proposed using a DD-SIMCA model, utilizing both spectrophotometric and digital image data. Spectrophotometric data exhibited great performance, obtaining a sensitivity of 0.96 (training) and 1 (test), in addition to a specificity of 0.94, where the classification errors was confined to the lowest tested concentration (2 μg L–1). Conversely, the model incorporating digital image data outperformed, achieving sensitivity of 0.96 (training) and 1 (testing), along with a specificity of 1. In conclusion, the obtained models are simple, rapid, and promising for screening antimicrobial dye contamination in surface water matrices.