Machine learning–driven optimization of lead-free Rb 2 GeSnCl 6 double perovskite solar cells for high-efficiency sustainable energy

Read full text
Authors: Md Hafizur Rahman, Md. Azizur Rahman, Noureddine Elboughdiri, S AlFaify, Karim Kriaa, Jehan Y. Al‐Humaidi, Ali Rayyes, Mohamed Benghanem
Cited by: 4
References: 67

Abstract

Perovskite solar cells (PSCs) have attracted significant attention as next-generation photovoltaic technologies due to their strong optoelectronic properties and potential for low-cost, high-efficiency energy conversion. In this work, the lead-free double perovskite Rb 2 GeSnCl 6 is systematically investigated as an environmentally benign absorber material for sustainable PSC applications. A combined device-simulation and machine-learning-assisted optimization framework is employed to analyze the effects of absorber thickness, bulk defect density, and electron transport layer (ETL) selection on photovoltaic performance. Among the investigated configurations, devices incorporating WS 2 as the ETL exhibit optimal performance, achieving a PCE exceeding 30% with an absorber thickness of 1000 nm and a low defect density of 10 12 cm -3 . Detailed current density–voltage (J–V) characteristics and external quantum efficiency (EQE) analysis confirm efficient charge extraction and strong photon-harvesting capability, with a stable EQE response spanning the 300–1100 nm wavelength range. The non-toxic and lead-free composition of Rb 2 GeSnCl 6 further enhances its suitability as a sustainable alternative to conventional Pb-based perovskites. In contrast to earlier studies that primarily rely on isolated parameter tuning or fixed transport-layer architectures, this work integrates defect-physics analysis with data-driven machine learning models, including XGBoost, CatBoost, LightGBM, and Gradient Boosting Regressor, to enable rapid and predictive device optimization with improved physical insight. This combined simulation–machine-learning strategy reflects an emerging trend in perovskite photovoltaics toward high-throughput screening and intelligent device design, particularly for lead-free materials. Beyond solar cells, the proposed methodology is readily extendable to other perovskite-based optoelectronic devices, such as photodetectors and light-emitting systems, broadening its technological relevance and impact.

Fields of Study & Subjects

Perovskite Materials and Applications Machine Learning in Materials Science Chemical and Physical Properties of Materials

Publication details

Journal / source
Optical Materials Express
Publisher
Optica Publishing Group
Volume
16
Issue
5
Pages
1182–1182
Publication date
23 February 2026
Language
EN