Modeling and Maximum Power Extraction from Perovskite Solar Cell

Abstract

This study depicted maximum power extraction from perovskite solar cells under the noticeable current density – voltage (J-V) hysteresis. The main objectives of this study were to analyze the impact of scan rate on the J-V characteristics of PSC, and then to develop a maximum power point tracking algorithm that could be efficient enough to predict the most suitable maximum power point, even under the high levels of J-V hysteresis. IonMonger, a MATLAB-based environment developed by Courtier et.al, was used to study the impact of scan rate and its direction on the J-V characteristics of a threelayered PSC (Electron transport layer, Perovskite absorber layer, Hole transport layer). It was observed that by increasing the scan rate from 50mV/s to 200mV/s, the losses in the current density, at the interfaces, due to interfacial recombination increase, and ultimately the hysteresis level increases. The hysteresis index is used as a parameter to evaluate the hysteresis level present in the J-V characteristics of PSC. When the interfacial recombination rates and effective doping densities of ETL and HTL were tuned, the hysteresis was reduced to the minimum and the impact of scan rate and direction on the JV characteristics became negligible. A Random Forest Regression model is used to develop an MPPT that could track the most suitable MPP even under a high level of hysteresis. It was depicted in this study that even at high values of J-V hysteresis, the RFR – MPPT predicted the MPP characteristics of PSC efficiently and did not overestimate and underestimate the performance of PSC. 0.42mW/cm2 or 1.9% of MPP-power difference was noticed in the prediction of MPP when there was a high level of hysteresis content present and when there was negligible hysteresis present. The credibility of RFR – MPPT was also investigated by comparing it with the two highly performed MPPTs in the solar industry, and the RFR – MPPT performed very accurately and efficiently as compared to Perturb & Observe and Incremental Conductance algorithms.