The design of advanced multi-junction solar cells using genetic algorithm for the optimization of a SILVACO® novel cell model.

MULTIJUNCTION solar cells consisting of series-stacked p-n junction layers offer a significant improvement in efficiency over conventional solar cells by generating power over a larger spectrum of sunlight. The design of multijunction solar cells is complicated by both the desire to have maximally efficient junction layers and the need to match the current produced in each junction layer under optimal load conditions. The ATLAS device simulator from Silvaco (R) International has been shown in exclusive research at the Naval Postgraduate School to have the capability to simulate multijunction solar cells. This simulation tool has the ability to extract electrical characteristics from a solar cell based on virtual fabrication of its physical structure and bypass the costly "build-and-test" design cycle. A paper introducing this modeling technique has been previously presented at the PVSC conference [2]. The current-matching problem is especially challenging for cells containing four or more junction layers. This paper proposes a method for using ATLAS data to optimize the power output of individual junction layers of an InGaP/GaAs/InGaNAs/Ge four junction solar cell and to construct these junction layers into a current-matched, optimum power multijunction solar cell. Individual junction layer optimization was accomplished through the use of a genetic search algorithm implemented in Matlab. The final multijunction cell current matching was performed using an iterative optimization routine also implemented in Matlab.