Park, HyunjungMorisset, AudreyKim, MunhoLee, Hae-SeokHessler-Wyser, AichaHaug, Franz-JosefBallif, Christophe2023-10-232023-10-232023-10-232023-09-1110.1016/j.egyai.2023.100299https://infoscience.epfl.ch/handle/20.500.14299/201778WOS:001077893500001Tandem solar cells are a key technology for exceeding the theoretical efficiency limit of single-junction cells. One of the most promising combinations is the silicon-perovskite tandem cells, considering their potential for high efficiency, fabrication on a large scale, and low cost. While most research focuses on improving each subcell, another key challenge lies in the tunnel junction that connects these subcells, significantly impacting the overall cell characteristics. Here, we demonstrate the first use of tunnel junctions using a stack of p+/n+ polysilicon passivating contacts deposited directly on the tunnel oxide to overcome the drawbacks of conventional metal oxide-based tunnel junctions, including low tunneling efficiency and sputter damage. Using Random Forest analysis, we achieved high implied open circuit voltages over 700 mV and low contact resistivities of 500 m omega cm2, suggesting fill factor losses of less than 1% abs for the operating conditions of a tandem cell.Computer Science, Artificial IntelligenceEnergy & FuelsComputer Sciencetunnel junctiontandempassivating contactsolar cellmachine learningdetailed balance limitcrystalline siliconefficiencyreassessmentPassivating contact-based tunnel junction Si solar cells using machine learning for tandem cell applicationstext::journal::journal article::research article