Analysis of adsorption isotherms of Ag+, Co+2, and Cu+2 onto zeolites using computational intelligence models

This manuscript discusses the application of computational intelligence models, such as Artificial Neural Network (ANN) and Adaptive Neuro-fuzzy Inference Systems (ANFIS), as alternatives for predicting the adsorption isotherms of metal ions on different zeolites. In particular, the adsorption of silver (Ag+), cobalt (Co+2), and copper (Cu+2) ions onto zeolites ZSM-5, ZHY, and Z4A was evaluated. The results showed that Z4A has better adsorption capacity for the three ions, followed by zeolite ZHY and ZSM-5. Furthermore, it was found that the Ag+ ion was more adsorbed compared to the Co+2 and Cu+2 ions; this can be related to the fact that Ag is a monovalent ion. In ANN and ANFIS, the equilibrium adsorption capacity for silver (Ag+), cobalt (Co+2), and copper (Cu+2) ions (target variables) were correlated with input variables, including temperature, Si/Al ratios of zeolites, molecular weights of metal ions and solution initial concentration. The 10-neuron hidden layer with Bayesian regularization backpropagation as a training function has shown better results for updating ANN weight and bias values. In contrast, the 16 fuzzy rule layer with the hybrid membership function has performed better for parameter training in ANFIS. ANN and ANFIS were able to predict all the adsorption systems with the advantage that they can take into consideration the adsorbent and adsorbate characteristics.