Corrosion resistance enhancement of a sol-gel coating by incorporation of modified carbon nanotubes: Artificial neural network (ANN) modeling and experimental explorations

Abstract

A computational method of data analysis based on an artificial neural network (ANN) has been proposed to model the behavior of a sol-gel coating modified with different amounts of oxidized multi-walled carbon nanotubes (O-MWCNT). The constructed ANN model utilized a single hidden-layer perceptron. The Levenberg–Marquardt algorithm optimization procedure was applied as a learning algorithm. In this model, the input variables were the different concentrations of O-MWCNT, the immersion time, and the real part of the impedance, and consequently, the imaginary part of the impedance was considered as the output variable. Then, the accuracy of the optimized model was evaluated using the correlation coefficient and schematically comparing the simulated data with the experimental ones in the Nyquist diagrams. Furthermore, the protection performance of the sol-gel layer was enhanced by the incorporation of O-MWCNTs. To this end, the different concentrations of the O-MWCNTs up to 0.9 % wt./wt. have been added to a silane layer, and the performance was followed by electrochemical exploration using electrochemical impedance spectroscopy (EIS). The results revealed the improvement of the protective performance of the silane coating by increasing the content of the O-MWCNTs in the matrix, followed by the enhancement of barrier properties. Moreover, the polarization curves, in agreement with the AC impedance spectra, reflected the significant decrease in the corrosion current density by employing more content of O-MWCNTs in the silane-based coatings.