Artificial intelligence models for methylene blue removal using functionalized carbon nanotubes

This study aims to assess the practicality of utilizing artificial intelligence (AI) to replicate the adsorption capability of functionalized carbon nanotubes (CNTs) in the context of methylene blue (MB) removal. The process of generating the carbon nanotubes involved the pyrolysis of acetylene under conditions that were determined to be optimal. These conditions included a reaction temperature of 550 °C, a reaction time of 37.3 min, and a gas ratio (H₂/C₂H₂) of 1.0. The experimental data pertaining to MB adsorption on CNTs was found to be extremely well-suited to the Pseudo-second-order model, as evidenced by an R2 value of 0.998, an X2 value of 5.75, a q_e value of 163.93 (mg/g), and a K2 value of 6.34 × 10–4 (g/mg min).The MB adsorption system exhibited the best agreement with the Langmuir model, yielding an R² of 0.989, R_L value of 0.031, q_m value of 250.0 mg/g. The results of AI modelling demonstrated a remarkable performance using a recurrent neural network, achieving with the highest correlation coefficient of R² = 0.9471. Additionally, the feed-forward neural network yielded a correlation coefficient of R2 = 0.9658. The modeling results hold promise for accurately predicting the adsorption capacity of CNTs, which can potentially enhance their efficiency in removing methylene blue from wastewater.