Caffeine Determination in Coffee Samples by Smartphone Application-Based Thin-Layer Chromatography Combined with Multivariate Image Analysis

Abstract The consumption of caffeine-containing products such as energy drinks, coffee, and tears needs to be regulated due to health risks. Caffeine content is standardized in decaffeinated coffee products, which must comply with a maximum of 0.1 wt%. A thin-layer chromatography (TLC) method and partial least squares regression (PLS) were developed for the quantification of caffeine in decaffeinated coffee and arabica coffee beans. Initially, caffeine standard was evaluated on three elution solvents on TLC analysis. The TLC plates (5 × 20 cm) were placed on UV viewing cabinet at 254 nm and recorded using smartphone camera. TLC plates were eluted using ethyl acetate with 0.1% acetic acid, which presents a retention factor ( R f ) of 0.33. Caffeine contents were predicted based on sample spots and standard analytical curve from 0.8 to 7.2 μg band –1 . The region of interest (ROI) from caffeine spots was selected by a free smartphone app, which employs PLS and red-green-blue (RGB) extraction from image for data processing and caffeine content prediction. For caffeine determination in decaffeinated coffee, an extraction with methanol was carried out with spiked samples, and the conformity was evaluated based on a default value. Green and roasted arabica coffee beans had a high value with 2.86 and 2.84 μg of caffeine per band, respectively. The smartphone application-thin layer chromatography (app–TLC) method showed precision and accuracy (coefficient of variation < 3.5%) with an average recuperation of 104%. The app–TLC was useful, fast and precise for the determination of caffeine in coffee products, certification of conformity, and comparable to GC–MS.