Background: In the current study, nanocrystalline thin films of TiO2:xSiO2 (x:
mole percentage) with high photocatalytic activity were grown on glass surfaces using sol-gel
method. The films were then treated under high temperature of 500°C for the crystal growth.
The synthesized films presented high photocatalytic activity when they were in contact with
methyl orange (MO) solution and UV irradiation. Due to complexity and nonlinearity photocatalytic
features of TiO2 films doped by SiO2, Artificial Neural Network (ANN) and Fuzzy
Logic (FL) models have been applied to predict and calculate the MO concentration variations
with SiO2 concentration and MO degradation time.
Results: The simulations have resulted in accurate and reliable predictive models since the
squared correlation coefficient (R2) and the standard squared error (SSE) have been R2>0.99,
0.004, R2 >0.96, and 0.14 for ANN and FL models, respectively. The reported figures have
shown that the independent predicted values of MO concentration are extremely close to their
corresponding experimental data.
Conclusion: The results of simulations have shown that ANN and Fuzzy models are reliable
predictive models to study the photocatalytic activity of TiO2 thin film doped by SiO2.