article · International Journal of Numerical Modelling Electronic Networks Devices and Fields
A random-forest machine learning algorithm was applied to evaluate how nine design parameters influence the power conversion efficiency of cesium lead halide perovskite solar cells. The analysis examined the effects of layer thicknesses, defect densities, and doping levels across the perovskite absorber layer, the hole transport layer, and the electron transport layer. To train the predictive model, a comprehensive dataset exceeding 1.5 million points was constructed by combining experimental records and experimentally validated numerical simulations. The model analysed efficiency variations across three distinct metal halide compositions: cesium lead iodide, cesium lead bromide, and cesium lead chloride. The findings identified distinct efficiency ceilings for each material formulation, achieving a peak power conversion efficiency of 17.8 percent for cesium lead iodide, 14.6 percent for cesium lead bromide, and 6.5 percent for cesium lead chloride.
Designing efficient perovskite solar cells involves balancing multiple physical factors, such as layer dimensions and defect levels. Using machine learning to simulate and predict performance across vast parameter combinations allows researchers to pinpoint optimal configurations much faster than relying solely on repetitive laboratory trials, thereby accelerating the development of higher-performing solar technologies.
This predictive modelling approach could assist solar cell designers and photovoltaic device developers in optimising device architectures prior to physical fabrication. By pinpointing ideal layer thicknesses, doping, and defect tolerances, the method can help reduce experimental prototyping costs. At present, the work represents early-stage computational and numerical research, requiring further targeted physical fabrication and real-world testing to establish commercial viability.
AI-generated from the published abstract. Always read the original work before citing.
Abstract The integration of machine learning (ML) models in studying, investigating, and optimizing various electronic devices and materials has significantly glow up. With the aid of ML algorithms and input datasets, data regression and prediction can show the output characteristic performance under a wide range of input combinations. Herein, we utilize a random‐forest ML algorithm to study the influence of nine input design parameters on the overall power conversion efficiency (PCE) of cesium lead halides perovskites cells. The doping levels, the defects densities, and the thicknesses among the perovskite thin film, as well as the hole and electron transport layers, are studied against the cell PCE. The seeded dataset is managed using experimental data and experimentally validated numerical simulations. Datasets of more than 1 512 000 points were generated and seeded to the ML model. The PCE variation against the inputs for the three metal halides was investigated. A 17.8% PCE for CsPbI 3 was reached, while PCE of 14.6% and 6.5% were recorded for CsPbBr 3 , and CsPbCl 3 , respectively.
This page summarises published work. The authoritative version sits with the publisher.
DOI: 10.1002/jnm.3164
Is something wrong with this record? Report it or request removal.
Discussion
Have you built on this work, tried to replicate it, or seen it applied in practice? Share what you know. Verified researchers and MARATTO™ domain experts can open a discussion, and any member can reply. Contributions are reviewed before they appear.
No discussion yet. Open the first thread.
New to MARATTO™? Create a free account.
