Experimental modeling study demonstrates 99.65% Rhodamine B degradation using BiOCl/Bi2Mo3O12 nanoparticles, highlighting machine learning-guided optimization for water treatment.
Key Points
To optimize process conditions and model the photocatalytic degradation of Rhodamine B dye in water using BiOCl/Bi2Mo3O12 heterogeneous nanoparticles.
Evaluated Rhodamine B photodegradation using BiOCl/Bi2Mo3O12 nanoparticles under irradiation across varying dye concentrations, pH levels, and polyvinylpyrrolidone (PVP) amounts.
Integrated catalyst surface roughness characterization with response surface methodology (RSM) and statistical correlation analyses (Pearson, Kendall, and Spearman).
Trained and evaluated machine learning models—including Random Forest, Gradient Boosting, XGBoost, Autoencoders, and Deep Neural Networks—on a dataset of 27,681 observations.
Response surface methodology identified optimum degradation conditions at 10 mg/L Rhodamine B, 0.1 g PVP, and pH 3, achieving a maximum removal efficiency of 99.65%.
Machine learning models predicted dye removal efficiency with 95.4% to 99.7% accuracy, with a feedforward neural network establishing Rhodamine B concentration as a key factor with a sensitivity of -0.9243.
Catalyst preparation with 0.1 g PVP generated an arithmetic average surface roughness of 35.2 nm (Ra/Rq of 0.75, Rz/Ra of 2.98, and Rv/Rp of 1.41), identifying surface topology as a controlling factor in degradation.