Artificial Intelligence-Guided Cosolvent Design for High-Performance Perovskite/Silicon Tandem Solar Cells

Lu Liu; Xinying Cai; Bita Farhadi; Xinrui Dong; Kai Wang; Yufei Shao; Shulin Wang; Jiaxue You; Wanyi Li; Hao-Chung Kuo; Hanying Wang; Dong Yang; Alex K.-Y. Jen; Shengzhong (Frank) Liu

doi:10.1007/s40820-026-02291-9

Artificial Intelligence-Guided Cosolvent Design for High-Performance Perovskite/Silicon Tandem Solar Cells

Lu Liu (Co-first Author)
, Xinying Cai (Co-first Author)
, Bita Farhadi
, Xinrui Dong
, Kai Wang^*
, Yufei Shao
, Shulin Wang
, Jiaxue You
, Wanyi Li
, Hao-Chung Kuo
, Hanying Wang
, Dong Yang^*
, Alex K.-Y. Jen^*
, Shengzhong (Frank) Liu^*

^*Corresponding author for this work

Research output: Journal Publications and Reviews › RGC 21 - Publication in refereed journal › peer-review

1 Citation (Scopus)

4 Downloads (CityUHK Scholars)

Abstract

Realizing high-performance perovskite/silicon tandem solar cells requires precise control of wide-bandgap perovskite crystallization. Solvent engineering is the most direct lever for this task; yet, its intricate, multi-variable mechanisms defy intuition-driven design. Herein, we overcome this bottleneck by pioneering a retrieval-augmented large language model to screen > 8000 solvents, identifying γ-valerolactone (GVL) as a non-toxic, high-performance cosolvent. It is found that the GVL strongly coordinates FA⁺, thus precisely modulating crystallization kinetics, retarding nucleation, and promoting oriented, micrometer-scale grain growth. The resulting films exhibit not only superior crystallinity, reduced non-radiative recombination, but also improved scalability to large area and the tolerance to increased film thickness. Consequently, both the single-junction and tandem devices achieve efficiencies of 23.3% and 32.5%, respectively, along with excellent stability under moisture and illumination. This study establishes the first artificial intelligence (AI)-guided cosolvent strategy for 1-μm-thick perovskite layers in perovskite/silicon tandem architectures, underscoring the transformative role of generative AI in advancing high-performance photovoltaics. (Figure presented.) © The Author(s) 2026.

Original language	English
Article number	446
Number of pages	18
Journal	Nano-Micro Letters
Volume	18
Issue number	1
Online published	21 Jul 2026
DOIs	https://doi.org/10.1007/s40820-026-02291-9
Publication status	Online published - 21 Jul 2026

Funding

This work was supported by the\u00A0Strategic Priority Research Program (Category B) of the Chinese Academy of Sciences (Grant No. XDB1140000), Xingliao Talent Program (XLYC2503007), the Shanxi Provincial Key Research and Development Program (202402060301005), the Hon Hai (AC-202403-02-3), the National Nature Science Foundation of China (22279140, U21A20102, 62174103, 52350710208), the innovation fund of Dalian institute of chemical physics (DICP I202431), Dalian Excellent Young Scientists Project (Grant No. 2024RY023), the Natural Science Foundation of Liaoning province (Grant No. 2024-MSBA-62), the 111 Project (B1404), the Project of Knowledge Innovation Engineering (Y261261606), the Cooperation Foundation of Yulin University and Dalian National Laboratory for Clean Energy (YLU-DNL fund 2022011), the Fundamental Research Funds for the Central Universities (GK202103106), and the Shanxi Science and Technology Department (20201101012). Special thanks are due to Dr. Cai Rui in Instrumental Analysis Center of Dalian University of Technology for the assistance with AFM analysis. A.K.Y.J. thanks the sponsorship of the Lee Shau-Kee Chair Professor (Materials Science), and the support from the APRC Grants (9380086, 9610419, 9610440, 9610492, 9610508) of the City University of Hong Kong, the TCFS grant (GHP/121/22SZ) from the Innovation and Technology Commission of Hong Kong, and the GRF grants (11304424, 11316422, 11308625) and CRS grants (CRS_CityU104/23, CRS_HKUST203/23) from the Research Grants Council of Hong Kong. This work was partially financially supported by City University of Hong Kong (9610739) for the project \u201CFostering Innovation for Resilience and Sustainable Transformation,\u201D officially endorsed by the United Nations Educational, Scientific, and Cultural Organization under the International Decade of Sciences for Sustainable Development (2024\u20132033).\u00A0The work was financially supported by City University of Hong Kong for the project 'Fostering Innovation for Resilience and Sustainable Transformation' (FIRST), officially endorsed by the United Nations Educational, Scientific and Cultural Organization (UNESCO) under the International Decade of Sciences for Sustainable Development (IDSSD) (2024-2033) via the internal City University of Hong Kong account of 9610739.

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

SDG 7 Affordable and Clean Energy

Research Keywords

Green solvent
High crystallinity
Machine learning
Tandem solar cells
Wide-bandgap perovskite

Publisher's Copyright Statement

This full text is made available under CC-BY 4.0. https://creativecommons.org/licenses/by/4.0/

RGC Funding Information

RGC-funded

Access Output

10.1007/s40820-026-02291-9

509207032.pdfFinal Published version, 698 KBLicence: CC BY

Other Access Options

Check CityUHK Library

Permanent Link

Right click to copy link

Cite this

Liu, L., Cai, X., Farhadi, B., Dong, X., Wang, K., Shao, Y., Wang, S., You, J., Li, W., Kuo, H.-C., Wang, H., Yang, D., Jen, A. K.-Y., & Liu, S. (2026). Artificial Intelligence-Guided Cosolvent Design for High-Performance Perovskite/Silicon Tandem Solar Cells. Nano-Micro Letters, 18(1), Article 446. Advance online publication. https://doi.org/10.1007/s40820-026-02291-9

@article{b32bd23c4a774200af39194bba128b4a,

title = "Artificial Intelligence-Guided Cosolvent Design for High-Performance Perovskite/Silicon Tandem Solar Cells",

abstract = "Realizing high-performance perovskite/silicon tandem solar cells requires precise control of wide-bandgap perovskite crystallization. Solvent engineering is the most direct lever for this task; yet, its intricate, multi-variable mechanisms defy intuition-driven design. Herein, we overcome this bottleneck by pioneering a retrieval-augmented large language model to screen > 8000 solvents, identifying γ-valerolactone (GVL) as a non-toxic, high-performance cosolvent. It is found that the GVL strongly coordinates FA+, thus precisely modulating crystallization kinetics, retarding nucleation, and promoting oriented, micrometer-scale grain growth. The resulting films exhibit not only superior crystallinity, reduced non-radiative recombination, but also improved scalability to large area and the tolerance to increased film thickness. Consequently, both the single-junction and tandem devices achieve efficiencies of 23.3\% and 32.5\%, respectively, along with excellent stability under moisture and illumination. This study establishes the first artificial intelligence (AI)-guided cosolvent strategy for 1-μm-thick perovskite layers in perovskite/silicon tandem architectures, underscoring the transformative role of generative AI in advancing high-performance photovoltaics. (Figure presented.) {\textcopyright} The Author(s) 2026.",

keywords = "Green solvent, High crystallinity, Machine learning, Tandem solar cells, Wide-bandgap perovskite",

author = "Lu Liu and Xinying Cai and Bita Farhadi and Xinrui Dong and Kai Wang and Yufei Shao and Shulin Wang and Jiaxue You and Wanyi Li and Hao-Chung Kuo and Hanying Wang and Dong Yang and Jen, \{Alex K.-Y.\} and Liu, \{Shengzhong (Frank)\}",

year = "2026",

month = jul,

day = "21",

doi = "10.1007/s40820-026-02291-9",

language = "English",

volume = "18",

journal = "Nano-Micro Letters",

issn = "2311-6706",

publisher = "SpringerOpen",

number = "1",

}

TY - JOUR

T1 - Artificial Intelligence-Guided Cosolvent Design for High-Performance Perovskite/Silicon Tandem Solar Cells

AU - Liu, Lu

AU - Cai, Xinying

AU - Farhadi, Bita

AU - Dong, Xinrui

AU - Wang, Kai

AU - Shao, Yufei

AU - Wang, Shulin

AU - You, Jiaxue

AU - Li, Wanyi

AU - Kuo, Hao-Chung

AU - Wang, Hanying

AU - Yang, Dong

AU - Jen, Alex K.-Y.

AU - Liu, Shengzhong (Frank)

PY - 2026/7/21

Y1 - 2026/7/21

N2 - Realizing high-performance perovskite/silicon tandem solar cells requires precise control of wide-bandgap perovskite crystallization. Solvent engineering is the most direct lever for this task; yet, its intricate, multi-variable mechanisms defy intuition-driven design. Herein, we overcome this bottleneck by pioneering a retrieval-augmented large language model to screen > 8000 solvents, identifying γ-valerolactone (GVL) as a non-toxic, high-performance cosolvent. It is found that the GVL strongly coordinates FA+, thus precisely modulating crystallization kinetics, retarding nucleation, and promoting oriented, micrometer-scale grain growth. The resulting films exhibit not only superior crystallinity, reduced non-radiative recombination, but also improved scalability to large area and the tolerance to increased film thickness. Consequently, both the single-junction and tandem devices achieve efficiencies of 23.3% and 32.5%, respectively, along with excellent stability under moisture and illumination. This study establishes the first artificial intelligence (AI)-guided cosolvent strategy for 1-μm-thick perovskite layers in perovskite/silicon tandem architectures, underscoring the transformative role of generative AI in advancing high-performance photovoltaics. (Figure presented.) © The Author(s) 2026.

AB - Realizing high-performance perovskite/silicon tandem solar cells requires precise control of wide-bandgap perovskite crystallization. Solvent engineering is the most direct lever for this task; yet, its intricate, multi-variable mechanisms defy intuition-driven design. Herein, we overcome this bottleneck by pioneering a retrieval-augmented large language model to screen > 8000 solvents, identifying γ-valerolactone (GVL) as a non-toxic, high-performance cosolvent. It is found that the GVL strongly coordinates FA+, thus precisely modulating crystallization kinetics, retarding nucleation, and promoting oriented, micrometer-scale grain growth. The resulting films exhibit not only superior crystallinity, reduced non-radiative recombination, but also improved scalability to large area and the tolerance to increased film thickness. Consequently, both the single-junction and tandem devices achieve efficiencies of 23.3% and 32.5%, respectively, along with excellent stability under moisture and illumination. This study establishes the first artificial intelligence (AI)-guided cosolvent strategy for 1-μm-thick perovskite layers in perovskite/silicon tandem architectures, underscoring the transformative role of generative AI in advancing high-performance photovoltaics. (Figure presented.) © The Author(s) 2026.

KW - Green solvent

KW - High crystallinity

KW - Machine learning

KW - Tandem solar cells

KW - Wide-bandgap perovskite

UR - https://www.scopus.com/pages/publications/105045256895

UR - https://www.scopus.com/pages/publications/105045256895#tab=impact

U2 - 10.1007/s40820-026-02291-9

DO - 10.1007/s40820-026-02291-9

M3 - RGC 21 - Publication in refereed journal

SN - 2311-6706

VL - 18

JO - Nano-Micro Letters

JF - Nano-Micro Letters

IS - 1

M1 - 446

ER -

Artificial Intelligence-Guided Cosolvent Design for High-Performance Perovskite/Silicon Tandem Solar Cells

Abstract

Funding

UN SDGs

Research Keywords

Publisher's Copyright Statement

RGC Funding Information

Access Output

Other Access Options

Permanent Link

Other Files and Links

Fingerprint

GRF: Molecular Engineering of Polyoxometalate (POM) and Self-Assembled Monolayer (SAM) Composites for Improving Efficiency and Stability of Organic and Perovskite Solar Cells

GRF: Development of Multifunctional Redox Mediators to Mitigate Halide Segregation in Wide-Bandgap Perovskites for Perovskite Based Tandem Solar Cells

ITF: Development of Commercializable Perovskite-Organic Tandem Solar Cells with High Efficiency and Stability

Cite this

Artificial Intelligence-Guided Cosolvent Design for High-Performance Perovskite/Silicon Tandem Solar Cells

Abstract

Funding

UN SDGs

Research Keywords

Publisher's Copyright Statement

RGC Funding Information

Access Output

Other Access Options

Permanent Link

Other Files and Links

Fingerprint

Projects

GRF: Molecular Engineering of Polyoxometalate (POM) and Self-Assembled Monolayer (SAM) Composites for Improving Efficiency and Stability of Organic and Perovskite Solar Cells

GRF: Development of Multifunctional Redox Mediators to Mitigate Halide Segregation in Wide-Bandgap Perovskites for Perovskite Based Tandem Solar Cells

ITF: Development of Commercializable Perovskite-Organic Tandem Solar Cells with High Efficiency and Stability

Cite this