Inverse design of selective multilayer emitters for thermophotovoltaic systems via Bayesian optimization
SOLAR ENERGY, cilt.304, 2026 (SCI-Expanded, Scopus)
- Yayın Türü: Makale / Tam Makale
- Cilt numarası: 304
- Basım Tarihi: 2026
- Doi Numarası: 10.1016/j.solener.2025.114210
- Dergi Adı: SOLAR ENERGY
- Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Applied Science & Technology Source, Compendex, Environment Index, Geobase, INSPEC, Academic Search Ultimate (EBSCO), Natural Science Collection (ProQuest), Engineering Source (EBSCO)
- Orta Doğu Teknik Üniversitesi Adresli: Hayır
Özet
Inverse design of the emitter for a thermophotovoltaic system is introduced. For this, the spectral emittance yielding the ideal performance is identified by considering the efficiency and power output through a parametric analysis. 1-D multilayer structures stand out among several alternative conceptual designs due to their ease of fabrication to achieve selective emission. The inverse problem is then formulated as identifying the design pa rameters that define the multilayer emitter that matches the ideal emission behavior. Such an inverse problem is highly nonlinear and yields a non-convex topology in the parameter domain necessitating the use of computation ally expensive global search algorithms to achieve an optimal solution. Bayesian optimization relies on machine learning, utilizing a trained surrogate model to search the global extremum throughout the topology with a lim ited number of function evaluations. This study introduces a novel systematic approach for material selection based on a preliminary analysis that investigates the classical models of optical properties to achieve the ideal spectral emission behavior. Three multilayer structures, dielectric-dielectric, metal-dielectric, and metal-metal emitters, are considered and their performances are compared. Because of its ability to suppress sub-bandgap emission, the metal-metal emitter exhibited superior performance. The tunability of doped silicon is then utilized to match the desired material property identified. Targeting an emittance profile that represents an optimal com bination of the two performance metrics, an estimated efficiency of 17.5 % and a power density of 4.88 kW/m2 are achieved for a 1100 K emitter.