Digital-image photometry with multi-energy calibration

Abstract

Simple analytical procedures that are able to circumvent matrix effects are always in high demand given the wide variety of analytical applications and the need to deal with complex-matrix samples. In this context, digital-image photometry (DIP) outstands because of its simplicity, cost-effectiveness, and portability. Typically, it involves external standard calibration (EC) based on single-channel measurements and is often prone to matrix effects. Alternatively, multi-energy calibration (MEC) has been successfully exploited as a calibration strategy to minimize matrix effects in spectrochemical analysis, as it relies on measurements at different wavelengths using a pair of calibration solutions containing the same amount of sample. Aiming at a simpler and more practical approach, the present work pioneeringly proposes multi-channel smartphone-based digital-image photometry to minimize interferences and matrix effects using the general concept of MEC. Analytical signals were based on the responses obtained at different color channels, RGB (red, green, and blue), CMY (cyano and magenta channels) and HSV (saturation values). The strategy was evaluated with model colored solutions (methyl red, brilliant blue dye, and nickel(II) sulfate) to establish boundary working conditions. For calibration solutions with a sample/standard analyte concentration ratio between 0.2 and 1.0, analyte recoveries in the 85–103 %, 92–104 %, and 87–113 % ranges were found for methyl red, brilliant blue, and nickel(II) sulfate, respectively. The method’s feasibility was then demonstrated by determination of methanol in solutions containing different amounts of glycerol, and hydrogen peroxide in milk with different fat contents. Relative errors (Er) for determination of methanol were within 2–16 % and 17–48 % using the MEC strategy and EC, respectively. For determination of hydrogen peroxide in milk, Er achieved by the proposed approach were within 6–10 %, whereas values as high as 50 % were estimated with EC. Coefficients of variation (n = 10) and limits of detection (99.7 % confidence level) for methanol and hydrogen peroxide determination were 3.5 % and 7 mg/L, and 6.6 % and 1 mg/L, respectively. These results demonstrate the proposed approach is a simple, practical, and environment friendly strategy to minimize matrix and other interfering effects.