Optimization of Metal-Oxide Spectral Filters for Reduced Photovoltaic Heating via Machine-Learning Guided Thickness
Ahasanur Rahman1, Kevin Thomas1, Amith Khandakar1
1Department of Electrical Engineering, College of Engineering, Qatar University, Doha 2713, Qatar.
ACS Omega
|August 8, 2026
Summary
A novel multilayer thin-film infrared (IR) filter reduces photovoltaic (PV) system heating by reflecting IR light. This AI-optimized filter enhances solar cell efficiency and lifespan by lowering operating temperatures.
Area of Science:
- Materials Science and Engineering
- Photovoltaics and Renewable Energy
- Optical Coatings and Thin Films
Background:
- Infrared (IR)-induced heating in photovoltaic (PV) systems critically lowers efficiency and accelerates module degradation.
- Existing PV technologies face challenges in managing thermal load without compromising energy conversion.
- Development of advanced optical filters is essential for enhancing PV performance and durability.
Purpose of the Study:
- To design and optimize a multilayer thin-film IR filter (TiO2/NiOx/Ag) for integration with PV modules.
- To reduce operating temperatures in PV systems by reflecting IR radiation while transmitting visible light.
- To leverage machine learning, specifically Gaussian Process Regression (GPR), for efficient filter optimization.
Main Methods:
- Fabrication of a multilayer thin-film filter comprising TiO2 (50 nm)/NiOx (100 nm)/Ag (variable thickness).
- Utilized experimental optical data to train Gaussian Process Regression (GPR) models for optimizing Ag layer thickness.
- Integrated the optimized filter with silicon solar cells and evaluated performance under simulated conditions.
Main Results:
- Optimized Ag thickness (approx. 10 nm) achieved >50% IR reflectance (750-1200 nm) and >50% visible transmittance.
- Silicon solar cells with the filter exhibited increased short-circuit current and power output due to reduced thermalization losses.
- A significant operating temperature drop of ~2-3 °C was observed, leading to corresponding PV efficiency gains.
Conclusions:
- The TiO2/NiOx/Ag thin-film filter effectively mitigates IR-induced heating in PV systems, enhancing performance and longevity.
- AI-driven optimization using GPR accelerates the design of high-performance photonic coatings with minimal experimental data.
- The scalable e-beam evaporation manufacturing process and demonstrated performance improvements indicate strong commercial potential.
