Ionic Polymer Metal Composite Based Smart Tongue Using Machine Learning

Abstract

Ionic polymer metal composite (IPMC)-based new type of taste sensing technology is focused on this study. The developed sensor can sense five taste senses (salty, sweetness, spicy, bitterness, and sourness) artificially. Development of electronic tongue using Nafion-based IPMC strip (Pt-electroded) is the background philosophy for the current finding. Due to the hydrophilic nature of the perfluorinated polymer membrane, IPMC strips while immersed into test solutions, cations like Li <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">+</sup> , Na <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">+</sup> , K <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">+</sup> , H <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">3</sub> O <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">+</sup> , etc., diffused into the polymer layer penetrating through the Pt-electrode surface of the IPMC and rearranged inside the polymer matrix. Bending of IPMC strip produces a small potential difference in millivolt (mV), which is developed across two Pt-electroded sides of IPMC. The output voltage across two electrode sides of IPMC for different taste samples of different concentration shows distinct voltage pattern and is measured in terms of potential difference across the Pt-electrode of IPMC. K-nearest neighbor classification approach of machine learning is useful for the identification of unknown taste agents. The average performance index of the developed machine learning model for the prediction of basic taste is 93.45%. Due to the nontoxic nature of IPMC, taste identification of different food contents would be feasible using this sensing technology.