No abstract

Energy-efficient, healthy lighting is vital for human beings. Incandescent lighting provides high-fidelity color rendering and ergonomic visual comfort yet is phased out owing to low luminous efficacy (15 lumens per watt) and poor lifetime (2000 hours). Here, we propose and experimentally realize a photon-recycling incandescent lighting device (PRILD) with a luminous efficacy of 173.6 lumens per watt (efficiency of 25.4%) at a power density of 277 watts per square centimeter, a color rendering index (CRI) of 96, and a LT70-rated lifetime of >60,000 hours. The PRILD uses a machine learning–designed 637-nm-thick visible-transparent infrared-reflective filter and a Janus carbon nanotube/hexagonal boron nitride filament to recycle 92% of the infrared radiation. The PRILD has higher luminous efficacy, CRI, and lifetime compared with solid-state lighting and thus is promising for high–power density lighting.

Thermophotovoltaic (TPV) generators provide continuous and high-efficiency power output by utilizing local thermal emitters to convert energy from various sources to thermal radiation matching the bandgaps of photovoltaic cells. Lack of effective guidelines for thermal emission control at high temperatures, poor thermal stability, and limited fabrication scalability are the three key challenges for the practical deployment of TPV devices. Here we develop a hierarchical sequential-learning optimization framework and experimentally realize a 6″ module-scale polaritonic thermal emitter with bandwidth-controlled thermal emission as well as excellent thermal stability at 1473 K. The 300 nm bandwidth thermal emission is realized by a complex photon polariton based on the superposition of Tamm plasmon polariton and surface plasmon polariton. We experimentally achieve a spectral efficiency of 65.6% (wavelength range of 0.4–8 μm) with statistical deviation less than 4% over the 6″ emitter, demonstrating industrial-level reliability for module-scale TPV applications.

Extracting the equivalent parameters of the weak-coupling and strong-coupling fishnet structure metamaterial based on the traditional retrieval algorithm and the improved algorithm of Kramers-Kronig relations are proposed, respectively. A comparative analysis of the effectiveness and applicability of the two algorithms are also included. The theoretical analysis and numerical results show that the traditional retrieval algorithm can retrieve the equivalent parameters of the weak-coupling and strong-coupling cases of electromagnetic metamaterials accurately, but with high computational complexity. While the improved algorithm based on the Kramers-Kronig relations can reduce the computational complexity and extract the equivalent parameters only for the weak-coupling case of electromagnetic metamaterials. However, it is not suitable for the strong-coupling case which may disobey the continuity requirement of the Kramers-Kronig relations. The presented results may extend the equivalent medium theory and provide a theoretical reference for the design of new metamaterials.

No abstract