A solar thermophotovoltaic (STPV) system can transform incident concentrated solar energy into electrical energy with an efficiency that could be higher than the Shockley–Queisser limit. Near-field thermophotovoltaic (NF-TPV) devices can generate larger electrical power output than traditional far-field TPV devices with the aid of photon tunneling. Moreover, multi-junction PV cells can boost the performance of TPV devices by effectively distributing the absorbed photon energy inside the PV cell. In this work, we design a multi-junction-based near-field STPV system with a practical and high-temperature stable graphite intermediate structure. To optimize the system configuration, we employ a genetic algorithm and a surrogate model based on an artificial neural network, which enables us to suggest a better design approach for the multi-junction-based NF-STPV system between the power output density and power conversion efficiency maximization scenarios. When the concentration factor of the incident solar energy is 5000 and the absorber-to-emitter area ratio is 3, we can achieve a system efficiency of 23%. By introducing a material whose emissivity is as high as a blackbody on the solar absorber, the system efficiency can be further enhanced up to 35%.

You do not currently have access to this content.