A self-calibrating porphyrin-based optofluidic system with edge-deployable chemometrics for online fluoride...

In this work, we report a self-calibrating two-channel flow system for a porphyrin-molecule-based optical sensor for the quantification of fluoride (F^-) in drinking water, with online monitoring capabilities. The mechanism employs absorbance-based detection of fluoride ion (F^-) using 5, 10, 15, 20-Tetraphenyl-21H, 23H-porphine iron(III) chloride (FeTPP) as the sensing probe. The absorbance spectrum of FeTPP was recorded for fluoride ions (F^-) in the presence of various interferents including chloride (Cl^-), bromide (Br^-), iodide (I^-), nitrate (NO₃^-), sulfate (SO₄^2-), phosphate (PO₄^3-), sulfite (SO₃^2-), bisulfide (HS^-), azide (N₃^-) and acetate (CH₃COO^-) to assess its applicability for real-time detection. Statistical analysis, in conjunction with a multivariate prediction model, has been implemented to estimate the sensitivity and predictive efficiency of the porphyrin sensor. The prediction model showed promising results with a coefficient of determination (R²) of 0.9983 ± 0.0009 and mean squared error (MSE) of 0.0018 ± 0.0068 in the absence of interferences, and R² of 0.9928 ± 0.0018 and MSE of 0.0472 ± 0.0422 in the presence of interferences. The effect of environmental parameters such as pH and temperature was also studied to develop a compensated prediction model that yielded a coefficient of determination (R²) of 0.9867 ± 0.0039 and MSE 0.0590 ± 0.0718. Furthermore, an online fluoride monitoring system has been designed, which includes a flow system with two parallel channels for the chemosensor and the water sample, a compact optical detector unit with an integrated LED and lens assembly, a microcontroller, and a display unit for data acquisition and user interface. The edge-deployed chemometrics, which included a multivariate prediction model, was configured in the microcontroller to predict the fluoride concentration in water. The proposed system showed congruence with recorded absorbance data for the fluoride concentration in a range from 0.1 to 10 ppm. The efficiency of the sensor was also effectively demonstrated for real water samples spiked with fluoride ions. These encouraging findings provide a firm foundation for the implementation of practical field applications in water quality sensing.