A team reporting in Nature Synthesis has now used artificial intelligence to search the enormous chemical space of self-assembled monolayers, the single-molecule-thick films that anchor charge extraction in these devices, and has landed on a candidate that is both easy to make and remarkably durable in operation.
The framework, described by researchers including Fengchun Cai, Geping Qu and Jixian Xu, combines three ingredients: a graph neural network trained to predict retrosynthetic routes, high-throughput screening based on density functional theory, and expert chemical judgement. Starting from 308 molecular fragments, the pipeline generated more than 90,000 candidate structures and then narrowed them down to 300 finalists.
Ranking those finalists required more than the usual electronic yardsticks. The team built a composite score from eight descriptors that extended beyond intrinsic electronic properties to include synthesizability, the hole reorganization energy λh, and structural features such as heteroatom content and aromaticity. This deliberately balanced scoring was designed to avoid molecules that look perfect on a computer but cannot be made in a laboratory or survive at a real device interface.
The top candidate, named AI-1, emerged as readily synthesizable and incorporates favourable heteroatom functionalities. According to the study, AI-1 achieves optimal energetic alignment with the perovskite and forms robust S–Pb coordination bonds at the buried interface, the hidden boundary where the monolayer meets the light-absorbing perovskite crystal.
Those bonds matter for two reasons. First, by strengthening the connection at the interface, AI-1 effectively suppresses non-radiative recombination, the wasteful process in which excited charge carriers annihilate each other instead of contributing to current.
The performance figures are striking. Devices incorporating AI-1 achieved a power-conversion efficiency of 27.2%, and retained more than 95% of their initial efficiency after 2,000 hours of maximum-power-point tracking at 65 °C. Sustained operation at elevated temperature under continuous load is one of the most demanding stability tests a solar cell can face, making the retention figure a meaningful benchmark rather than a shelf-life claim.
Mechanistic studies traced the success to a specific structural feature: the multifunctional methylthio side chain. The researchers found this side chain is vital because it reinforces the interfacial bonding and balances passivation with chemical robustness. In other words, the same molecular appendage that helps neutralize defects at the interface also resists being torn apart by the harsh chemistry of the perovskite surface.
The workflow presented here, illustrated across figures covering retrosynthetic prediction with a message-passing neural network, descriptor distributions and correlations, device photovoltaic characteristics, and mechanistic characterization, offers a systematic alternative in which synthesizability and stability are scored alongside electronic performance from the outset.
The authors argue that their findings demonstrate artificial intelligence-guided design can systematically deliver self-assembled monolayers uniting high efficiency with long-term operational stability, and that the approach provides general principles for interface engineering.
Subject of Research: AI-guided discovery of synthesizable, stable self-assembled monolayers for perovskite solar cells
Article Title: AI-guided discovery of self-assembled monolayers combining synthetic accessibility and interfacial stability for perovskite solar cells
Article References: Cai, F., Qu, G., Zhang, B., Zhang, L., Chen, L., Mao, K., Tang, L., Wang, Y., Li, M., Yuan, S., Guo, H., Liu, F., Ma, Y., Xiao, C., Jen, A. K. Y., Jiang, J., Xu, Z.-X., & Xu, J. (2026). AI-guided discovery of self-assembled monolayers combining synthetic accessibility and interfacial stability for perovskite solar cells. Nature Synthesis. https://doi.org/10.1038/s44160-026-01203-4
Image Credits: AI Generated
DOI: 10.1038/s44160-026-01203-4
Keywords: perovskite solar cells, self-assembled monolayers, artificial intelligence, retrosynthetic prediction, density functional theory, interfacial stability, power-conversion efficiency, non-radiative recombination, methylthio side chain, interface engineering, AI-guided, discovery
News Source: Bethany Barker. (October 28, 2026). AI-designed molecule lifts perovskite solar cells to 27.2% efficiency with lasting stability. Scienmag.
