Corrosion Inhibition Study Of Carboxymethyl Celluloseionic Liquid Via Electrochemical And Machine Learning Technique

Corrosion is a natural phenomenon defined as the deterioration of a substance or its properties due to interactions between the substance and the environment. Prolonged exposure to corrosive environment had negative consequences, including increased repair and maintenance costs, decreased structural integrity, and fatalities. An approach to address the issue is to use a corrosion inhibitor. Numerous inhibitors have lately been developed or made accessible on the market. However, some could be dangerous or contain of volatile organic compounds (VOCs). Our study introduces carboxymethyl cellulose (CMC) mixed with 1-ethyl-3- methylimidazolium acetate ([EMIM][Ac]) ionic liquid, also known as CIL, as a corrosion inhibitor on mild steel in seawater. The functional group of combined CIL was examined using Fourier transform infrared spectroscopy (FTIR). The study employed mild steel specimens immersed in varying concentrations of CIL, subject to electrochemical impedance spectroscopy (EIS) measurements at different temperatures. The obtained result of EIS measurement were analyzed to calculate corrosion inhibition efficiency (IE) of CIL. 46 of electrochemical data were fed into a machine learning technique to forecast the effectiveness of the inhibition. Results indicate when inhibition concentration rises, so does inhibition efficiency (IE). The inhibitory efficiency of CIL decreased as the temperature of the test solution rose. At an ambient temperature of 950 ppm, an IE result of 83% was recorded. Levenberg-Marquardt (LM), Bayesian Regularization (BR), and Scale Conjugate Gradient (SCG) training algorithms were compared via Neural Network Fitting Tool (NNTool). LM was found to be the best backpropagation training algorithm, providing the highest regression value (R) of 0.907 and the lowest mean square error (MSE) of 0.006 when compared to BR and SCG. With R value closer to 1 and MSE close to 0, the use of Artificial Neural Networks (ANN) appears to offer a new insight in predicting methods with the goal of easing the hassle and time-consuming of experimental work.