ISSN print edition: 0366-6352 ISSN electronic edition: 1336-9075 Registr. No.: MK SR 9/7 Published monthly

AI-driven optimization of zinc removal using surface-modified biochar: a comparative study of ANN and DFNN

B. Kamala and B. Latha

Department of Information Technology, Sri Sairam Engineering College, Chennai, India

E-mail: kamalasundararaman81@gmail.com

Received: 7 March 2025  Accepted: 9 June 2025

Abstract:

Toxic heavy metal poisoning of water bodies, especially Zn (II) ions, is a serious risk to human health and the environment. Traditional treatment techniques may have low efficiency and significant operating expenses. In this study, alkaline-modified neem bark biochar (ANBB) and ultrasonicated neem bark biochar (UNBB) were evaluated as cost-effective and sustainable adsorbents for Zn (II) removal. To optimize adsorption conditions and enhance predictive accuracy, advanced artificial intelligence (AI) models such as artificial neural networks (ANN) and deep fuzzy neural networks (DFNN) were employed. The optimal conditions for zinc removal, as determined from batch experiments, were found to be at pH 4. The highest removal efficiency was achieved with a biosorbent dosage of 2.5 g/L for UNBB and 2 g/L for ANBB, a contact time of 40 min, and a temperature of 303 K. The kinetic tests verified pseudo-first-order kinetics as the best fit, while the isotherm analysis showed better fitting to the Freundlich model, suggesting adsorption on heterogeneous surfaces with maximal adsorption capacities of 153.07 mg/g for UNBB and 271.9 mg/g for ANBB. AI-based predictive modelling revealed the superiority of DFNN over ANN in adsorption performance prediction, with DFNN achieving an R² value of 0.9982, significantly outperforming ANN (R² = 0.5774). The integration of deep learning with fuzzy logic in DFNN provided enhanced interpretability and adaptability, enabling precise optimization of Zn (II) removal while capturing system uncertainties. These results demonstrate that AI-driven DFNN modelling is a reliable prediction tool for improving wastewater treatment decision-making, reducing experimental expenses, and optimizing adsorption processes. AI combined with biochar-based adsorbents offers an economically feasible approach for efficient and sustainable heavy metal removal.

Graphic abstract

Keywords: Adsorption; Artificial intelligence; Biochar; Deep fuzzy neural networks; Optimization

Full paper is available at www.springerlink.com.

DOI: 10.1007/s11696-025-04186-2

Chemical Papers 79 (9) 6247–6264 (2025)