Distinction des mécanismes de dégradation des cellules solaires à pérovskites par des approches statistiques sur de grands ensembles de données simulées
- Authors
- Publication Date
- Oct 24, 2023
- Source
- HAL-Descartes
- Keywords
- Language
- English
- License
- Unknown
- External links
Abstract
Perovskite solar cells have attracted a lot of attention in the past years, due to high power conversion efficiencies and low cost of fabrication. Material and interface properties in these devices have been intensely studied, allowing to significantly improve performances. However, the expected lifetime remain short, because of numerous potential degradation mechanisms, triggered by various environmental factors.This work aims at helping the understanding of these degradation processes and supporting the development of stable perovskite solar cells. Precisely, modelling methods have been developed to distinguish and identify the mechanisms responsible for performance degradation under given aging experiments.Current voltage curves and photoluminescence spectra measurements, performed periodically over the course of aging experiments have been investigated. The associated evolutions of optoelectrical parameters over time were at the core of the approach developed here.In order to investigate the photovoltaic behavior of perovskite solar cells, optical (transfer matrices) and electrical (drift diffusion) modeling have been coupled. Furthermore, a statistical approach has been developed, because of some unwell known input parameters. A genetic algorithm has been designed, providing numerous sets of inputs that reproduce the initial performances of a given sample. These sets were the basis to simulate various hypothetical unitary degradation mechanisms.Importantly, pathways are obtained by considering the correlated evolution of optoelectrical parameters. They constitute characteristic footprints of the processes responsible for the performance degradation, and simulated and experimental pathways can be directly compared. As a result, compatible mechanisms can be proposed, and others excluded when pathways differ. The causality between performance losses and degradation mechanisms is here directly tackled.After applying this approach to experimental measurements reported in literature, making possible to compare results to authors analyses and demonstrate the validity of the approach, aging experiments performed at IPVF were investigated. A first set of samples prepared with four variations in the deposition method of the perovskite layer. Results showed that the perovskite could be excluded as a cause for degradation in most cases, except for a specific method, also having the least stable samples.A second set, containing devices having different hole and electron transport layers was investigated through coupled current-voltage and photoluminescence measurements. Interestingly, hole transport layer degradation could be attributed to several samples, and a protective role of the electron transport could be envisaged. Also, coupling characterization techniques helped to distinguish pathways through new complementary planes.Finally, the last part of this work took advantage of the numerous simulations performed to investigate degradation. It aimed at simplifying the design of drift diffusion simulations by reducing the number of necessary inputs and identifying the most important ones. Meta-parameters candidates have been proposed by considering relevant quantities in an analytical model. Moreover, their validity to define a solar cell performance was assessed through their correlation with its optoelectrical outputs. Finally, principal components analyses were also employed on subsets selected according to solar cells performances, to point out the most important parameters or provide new simple phenomenological models.This work shows how modelling can support experimental development of stable perovskite solar cells. Notably, insights on the causes of degradation of various samples have been proposed. Finally, this also demonstrates that statistical approaches can support the solar cell modeling research field, by being less dependent on the knowledge of given parameters.