Kinetics, central composite design and artificial neural network modelling of ciprofloxacin antibiotic photodegradation using fabricated cobalt-doped zinc oxide nanoparticles

Cobalt-doped zinc oxide nanoparticles were fabricated and examined in this study as a potential photocatalyst for the antibiotic ciprofloxacin (CIPF) degradation when exposed to visible LED light. The Co-precipitation technique created Cobalt-doped zinc oxide nanoparticles that were 5, 10, and 15% Co-loaded. Different known techniques have been used to characterize the synthesized ZnO and cobalt-doped ZnO nanoparticles. Compared to ZnO and other Cobalt-doped ZnO nanoparticles, the experiments showed that 10% Cobalt-doped ZnO nanoparticles were a very effective catalyst for CIPF photodegradation. According to XRD, these NPs have a hexagonal Wurtzite structure with an average size of between 38.47 and 48.06 nm. Tauc plot displayed that the optical energy band-gap of ZnO NPs (3.21) slowly declines with Co doping (2.75 eV). The enhanced photocatalytic activity of Cobalt-doped ZnO nanoparticles, which avoids electron-hole recombination, is brought on by the implantation of Co. Within 90 min, a 30 mg/L solution of ciprofloxacin was destroyed (> 99%). The kinetics studies demonstrated that the first-order model, with R² = 0.9703, is appropriate for illuminating the pace of reaction and quantity of CIPF elimination. The recycled Cobalt-doped zinc oxide nanoparticles enhanced photocatalytic performance toward CIPF for 3 cycles with the same efficiency. Furthermore, optimization of the 10% Cobalt-doped zinc oxide nanoparticles using a Central composite design (CCD) was also studied. The optimal parameters of pH 6.486, 134.39 rpm shaking speed, 54.071 mg catalyst dose, and 31.04 ppm CIPF initial concentration resulted in the highest CIPF degradation efficiency (93.99%). Artificial neural networks (ANN) were used to simulate the experimental data. The backpropagation technique was used to train the networks with 152 input-output patterns. After experimenting with various configurations, the best results with a correlation value (R²) of 0.9780 for data validation were obtained using a three-hidden layered network that included five, five, and eight neurons, respectively.