Artificial neural networks applied to fluorescence studies for accurate determination of N-butylpyridinium chloride concentration in aqueous solution

N-butylpyridinium chloride ([bpy][Cl]) is an ionic liquid (IL) extensively employed as an effective catalyst for many chemical reactions. Therefore, precise monitoring of its concentration can ensure the desired high yields in these catalyzed chemical processes. In this work, a fluorescence study has been carried out to determine the concentration of [bpy][Cl] in aqueous solution. A light emitting diode (LED), a continuous wave laser diode (CWLD), and a femtosecond pulsed laser (FPL) have been employed as light sources to electronically excite IL samples at a central wavelength of 400 nm. The measured fluorescence spectra obtained at different concentrations have been used to design three mathematical models for each one of the light sources. These models rely on artificial neural networks (ANNs) to assess the concentration of IL aqueous solutions in a wide range of concentrations. ANNs have been selected thanks to their ability to discover and adequately interpret nonlinear relationships among datasets. ANNs have been successful due to the existence of a nonlinear dependence between the IL fluorescence signal and its concentration, most likely due to the inner filter effect. The three light sources employed were suitable to fulfill the goal (mean prediction errors were 4.9%, 2.5%, and 1.7% for the LED, CWLD, and FPL models, respectively). These results suggest the existence of a potential source of reliable sensors based on the combination of fluorescence and ANNs. Furthermore, the accurate concentration estimation of many fluorescent compounds in aqueous solution appears achievable and, therefore, applicable to multiple chemical processes.