An overview of the strategies adopted in flow analysis towards cleaner analytical methods is presented. The discussion deals with reagentless procedures, replacement of hazardous chemicals, strategies for waste minimization as well as on-line waste treatment or recycling. The potential of flow approaches such as sequential injection, multicommutation and monosegmented flow is emphasized. Automation, employment of solid-phase reagents and miniaturization are highlighted as alternatives for waste minimization.

A new approach for photometric measurements in procedures involving liquid–liquid extraction is proposed. It is based on the direct acquisition of the digital image of the sample extract in a conical tube by using a smartphone and a free application to convert the image to RGB values. The approach avoids the drawbacks of UV-vis spectrophotometric measurements caused by the restricted volume of the sample extract, such as reduction of the optical path or the need for dilution. Moreover, the need for phase isolation and thus losses of the extract before measurements is avoided. The feasibility of the procedure was demonstrated by (i) dispersive liquid–liquid microextraction aimed at the determination of anionic surfactants in freshwaters by the methylene blue (MB) method and (ii) cloud point extraction for the determination of free chlorine in tap waters by the N,N-diethyl-p-phenylenediamine (DPD) method. The values of the RGB channels were taken as the analytical signal. For the MB method, a linear response was achieved within 10.0 and 70.0 μg L⁻¹, with a detection limit of 2.6 μg L⁻¹. With the proposed strategy, the consumption of chloroform was 1000-fold reduced (only 100 μL is required per determination) in comparison to the conventional procedure. Good recoveries (within 87.8 and 99.7%) and agreement with the reference method were observed. The DPD method yielded a linear analytical response within 0.1 and 2.0 mg L⁻¹, with a detection limit of 70 μg L⁻¹, and results in agreement with the reference procedure. For both procedures, the coefficient of variation was estimated at 4.0% (n = 10).