https://doi.org/10.1140/epjp/s13360-023-04646-2
Regular Article
Prediction and tuning of the optical energy gap and refractive index of amorphous titania-alumina thin films prepared by atomic layer deposition using adaptive neuro-fuzzy inference system model
1
Physics Department, Faculty of Education, Ain Shams University, Roxy, Cairo, Egypt
2
Department of Basic Sciences, Obour High Institute for Engineering and Technology, Cairo, Egypt
3
Department of Solid State Physics, Faculty of Science and Technology, University of Debrecen, PO Box 400, 4002, Debrecen, Hungary
4
Department of Environmental Physics, Faculty of Science and Technology, University of Debrecen, Debrecen, Hungary
Received:
5
June
2023
Accepted:
31
October
2023
Published online:
16
November
2023
Titania was mixed with different percentages of alumina during the preparation of an 8 nm film by atomic layer deposition. The effect of mixing on the transmittance and absorption spectra was investigated. Increasing the alumina percentage increased the transmittance and decreased the absorption of light. The films showed an indirect allowed transition with an energy gap that slightly increased with increasing alumina percentage. The refractive index was calculated using four different methods. The average value of the refractive index decreases from 2.350 for pure titania to 2.314 for titania (70%) and alumina (30%). The optical properties were estimated by using ANFIS model. To achieve the best fit, different configurations were trained using MATLAB-R2021a, and the ANFIS optimal network of training the transmittance of titania-alumina thin films was determined. Experimental measurements were compared with the ANFIS simulated outputs and it is clear that the experimental data and the ANFIS simulated results almost coincide. The major goal is to employ the ANFIS model to predict the optical properties of the underlying titania thin film at alumina concentrations, which are experimentally unmeasured. Thus, the net is reduced in the number of samples, which is important in saving effort, time, and costs, which is an urgent requirement today. The method based on machine learning presented here also works when few measurement data are available and the relationship between the parameters is nonlinear, making the use of other numerical methods, e.g., inter/extrapolation, questionable. Compatibility between practical and theoretical results provides availability for more applications in the field of material science.
Copyright comment Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
© The Author(s), under exclusive licence to Società Italiana di Fisica and Springer-Verlag GmbH Germany, part of Springer Nature 2023. Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
