Efficient and stable mesoscopic perovskite solar cell in high humidity by localized Dion-Jacobson 2D‐3D heterostructures

Mesoscopic perovskite solar cells (PSCs) suffer from poor stability despite high efficiency, due to irreversible decomposition of perovskite in moisture. Herein, we demonstrate localized Dion−Jacobson (DJ) 2D–3D heterostructure (L2D–3DH), in which DJ 2D perovskite is formed on the localized surface of 3D perovskite without fully covering it, to improve the efficiency and stability of PSCs simultaneously. Different from conventional 2D–3D composite, L2D–3DH does not hinder the charge transfer between perovskite and hole transport layer due to partial coverage of DJ 2D yet effectively improve the moisture stability. Moreover, control of crystal orientation is not critical for efficient charge transport. As a result, L2D–3DH yields a power conversion efficiency of 20.1%, which is higher than that obtained from pure 3D perovskite or conventional 2D–3D composite. The improved PCE can be attributed to the passivation of 3D perovskite with DJ 2D plates, which reduces the trap density, suppresses non-radiative recombination, and enhances quasi-Fermi level splitting. The L2D–3DH-based device retains 86% of the initial efficiency after 1300 h under high humidity (70% RH). This work offers a simple and effective way to improve the efficiency and stability of PSCs by incorporating the DJ 2D in 3D perovskite. • Localized DJ 2D–3D heterostructure is proposed by forming DJ 2D perovskite on the localized surface of 3D perovskite. • This structure does not hinder the interfacial charge transfer due to partial coverage yet improve the moisture stability. • The control of crystal orientation is not critical for efficient charge transport in the localized 2D–3D heterostructure. • The unencapsulated device yields PCE of 20.1% and retains 86% of the initial PCE after 1300 h under 70% RH high humidity. • The passivation of 3D perovskite with DJ 2D plates reduces the trap density and enhances quasi-Fermi level splitting.