LEAP: A closed-loop framework for perovskite precursor additive discovery
Abstract
Efficient discovery of precursor additives is essential for improving the performance of perovskite solar cells, yet the large chemical space makes conventional trial-and-error screening inefficient. We develop LEAP(LLM-driven Exploration via Active Learning for Perovskites), an expert-in-the-loop closed framework that couples a domain-specialized large language model(LLM) with active learning for iterative additive prioritization. The LLM is trained to extract mechanism-relevant knowledge from the perovskite additive literature and to represent candidate molecules through interpretable descriptors, which are further integrated into a Bayesian optimization workflow for uncertainty-aware prioritization under low-data conditions. Benchmark results on unseen literature show that the domain-specialized model outperforms general-purpose models in mechanism-consistent reasoning. Experimental validation in an expert-in-the-loop proof-of-concept study suggests improved additive prioritization across three screening rounds, leading to average device PCEs of 20.13% and 20.87% for the later-round 6-CDQ- and 2-CNA-treated devices, respectively, compared with 19.25% for the control, with a champion PCE of 21.32%. These results provide preliminary evidence that literature-grounded mechanistic descriptors, when coupled with Bayesian optimization and expert feasibility review, can support mechanism-aware additive prioritization in perovskite photovoltaics.
Versions
Related
- Extensive A-Site Magnesium Solubility in Perovskite Oxides via Precursor-mediated Synthesis Pathways
- Robot-Accelerated Perovskite Investigation and Discovery: Copper Halide Perovskitoids
- Prototype-guided transfer of sparse literature knowledge for electrolyte additive discovery
- A physics-driven machine learning framework for predicting XRD patterns, magnetization, and magnetocaloric response in rare-earth-doped double perovskite oxides
- LLM-AutoSciLab: Closed-Loop Scientific Discovery via Active Experimentation with LLMs
BibTeX
@misc{wang2026leap,
title = {{LEAP: A closed-loop framework for perovskite precursor additive discovery}},
author = {Xin-De Wang and Zhi-Rui Chen and Ze-Feng Gao and Peng-Jie Guo and Cheng Mu and Zhong-Yi Lu},
howpublished = {arXiv},
year = {2026},
month = may,
doi = {10.48550/arxiv.2605.20242},
eprint = {2605.20242},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2605.20242},
}
RIS
TY - UNPB
TI - LEAP: A closed-loop framework for perovskite precursor additive discovery
AU - Xin-De Wang
AU - Zhi-Rui Chen
AU - Ze-Feng Gao
AU - Peng-Jie Guo
AU - Cheng Mu
AU - Zhong-Yi Lu
PY - 2026
DA - 2026/05/18
PB - arXiv
DO - 10.48550/arxiv.2605.20242
UR - https://arxiv.org/abs/2605.20242
AB - Efficient discovery of precursor additives is essential for improving the performance of perovskite solar cells, yet the large chemical space makes conventional trial-and-error screening inefficient. We develop LEAP(LLM-driven Exploration via Active Learning for Perovskites), an expert-in-the-loop closed framework that couples a domain-specialized large language model(LLM) with active learning for iterative additive prioritization. The LLM is trained to extract mechanism-relevant knowledge from the perovskite additive literature and to represent candidate molecules through interpretable descriptors, which are further integrated into a Bayesian optimization workflow for uncertainty-aware prioritization under low-data conditions. Benchmark results on unseen literature show that the domain-specialized model outperforms general-purpose models in mechanism-consistent reasoning. Experimental validation in an expert-in-the-loop proof-of-concept study suggests improved additive prioritization across three screening rounds, leading to average device PCEs of 20.13% and 20.87% for the later-round 6-CDQ- and 2-CNA-treated devices, respectively, compared with 19.25% for the control, with a champion PCE of 21.32%. These results provide preliminary evidence that literature-grounded mechanistic descriptors, when coupled with Bayesian optimization and expert feasibility review, can support mechanism-aware additive prioritization in perovskite photovoltaics.
ER -