Ag/TiO2-functionalized photocatalytic concrete for visible-light photocatalytic degradation of reactive black 5: Experimental optimization, mechanistic insights and machine learning modeling

Read full text
Authors: Nfor Elvis Nfor, Muhammad Saqib Khan, Rizwana Sarwar, Muhammad Bilal, Ajmal Khan, Syed Tasqeeruddin, Muhammad Faisal Javaid, Nadia Riaz
Cited by: 1
References: 66

Abstract

Photocatalytic concrete is a key advancement in eco-friendly materials that helps eliminate contaminants, improving the water and air quality in modern infrastructure. This work aimed to develop silver-loaded titanium dioxide (AgT) photocatalysts and AgT-photocatalytic concrete (AgTC) for visible light-driven degradation of Reactive Black 5 (RB5). AgT was synthesized employing a modified sol-gel method with calcination temperatures of 300, 400 and 500 °C and incorporated into concrete at concentrations ranging from 0 to 10 wt%. The goal was to optimize photocatalytic parameters for RB5 degradation and evaluate the predictive accuracy of machine learning (ML) models compared to manual optimization. Physicochemical characterizations, including structural and morphological analyses, including HRTEM, SEM, EDS, XRD, XPS, DRS, DLS, and PZC investigations, confirmed effective Ag integration, establishment of a crystalline Ag/TiO 2 heterojunction structure and retention of the anatase phase in the concrete matrix. Under optimized conditions (AgTC dose, initial dye concentration (IDC) and pH), complete RB5 photodegradation was achieved within 80 minutes of irradiation, with AgTC dose having the most significant impact compared to IDC and pH. Ten supervised ML models were trained on 1440 experiments to predict the photodegradation efficiency of AgTC using six regression metrics: R 2 , MAE, RMSE, MAPE, RMSLE, and NSE. The CatBoost (CB) model showed exceptional performance with R 2 and NSE values of 1.00, and MAE, RMSE, and MAPE values of 0.05, 0.10, 0.07, and 0.00, respectively. CB outperformed manual optimization techniques in predictive accuracy. These models effectively capture nonlinear parameter interactions and generalize well to new data. Combining AgTC photocatalysis with ML-driven optimization provides a scalable, intelligent approach for sustainable water treatment and multifunctional construction material.

Fields of Study & Subjects

TiO2 Photocatalysis and Solar Cells Advanced Photocatalysis Techniques Pigment Synthesis and Properties

Publication details

Journal / source
Journal of Science Advanced Materials and Devices
Publisher
Elsevier BV
Volume
11
Issue
2
Pages
101170–101170
Publication date
24 April 2026
Language
EN