Toward Self-Driven Autonomous Material and Device Acceleration Platforms (AMADAP) for Emerging Photovoltaics Technologies

Jiyun Zhang, Jens A. Hauch, and Christoph J. Brabec

Accounts of Chemical Research 2024 57 (9), 1434-1445

DOI: 10.1021/acs.accounts.4c00095

This summary discusses recent progress made by the team in developing automated and autonomous labs to discover new materials and improve devices, especially for new types of solar cells like perovskite and organic photovoltaics.

Four key technologies were created:

  1. A microwave-assisted system for quickly creating new organic interface materials.
  2. A versatile robot-based system for making and testing new semiconductor materials.
  3. The SPINBOT system, which optimizes the layers in complex devices using a spin-coating process.
  4. AMANDA, a fully automated system that works on its own.

These systems use robots to handle the complicated task of finding the best compositions and parameters for organic and perovskite solar cell materials. Lastly, we propose a new idea: a fully autonomous lab called AMADAP that can discover and develop functional solar materials independently. We believe this concept can become even more powerful with future advancements in technology.

Equipment used:  Here our SpinBot functions as a device acceleration platform (DAP).  The SpinBot allows the processing paramenters to be precisely controlled, as often required in functional film coating.

Precise Control of Process Parameters for >23% Efficiency Perovskite Solar Cells in Ambient Air Using an Automated Device Acceleration Platform

Jiyun ZhangAnastasia BarabashTian DuJianchang WuVincent M. Le CorreYicheng ZhaoShudi QiuKaicheng ZhangFrederik SchmittZijian PengJingjing TianChaohui LiChao LiuThomas HeumuellerLarry LüerJens A. HauchChristoph J. Brabec

arXiv:2404.00106v1 []

DOI:  10.48550/arxiv.2404.00106

Here an automated system was used to improve the process of making perovskite solar devices in regular air. The team focused on eight key steps that can impact the device's performance. One important step is controlling the speed at which a specific chemical (organic ammonium halide) was added, which is hard to do by hand. The experiments showed that this speed affects how well the devices work, especially by changing the leftover PbI2 content in the films.  A moderate speed of 50 µl/s works best, while speeds that are too fast or too slow result in poorer performance and consistency. By optimizing these steps, a standard procedure was created for making perovskite devices without additives, achieving efficiencies over 23%, good consistency, and excellent stability.

Equipment used:  Our SpinBot enabled the researchers to investigate optimal deposition and dispersion levels to increase efficiency and stability in perovskite solar devices.  

Accelerating Photostability Evaluation of Perovskite Films through Intelligent Spectral Learning-Based Early Diagnosis

Ziyi Liu, Jiyun Zhang, Gaofeng Rao, Zijian Peng, Yixing Huang, Simon Arnold, Bowen Liu, Can Deng, Chen Li, Heng Li, Hanxiang Zhi, Zhi Zhang, Wenke Zhou, Jens Hauch, Chaoyi Yan, Christoph J. Brabec, and Yicheng Zhao

ACS Energy Lett. 2024, 9, 2, 662–670

Publication Date:  January 30, 2024

Copyright © 2024 American Chemical Society

This study introduces a spectral learning-based method to predict perovskite stability using photoluminescence and absorption spectra of new films. This approach avoids lengthy aging tests by using a custom spectral feature extraction algorithm and a machine learning model. Integrated with high-throughput experiments, this method achieves over 86% accuracy in predicting stable perovskites in 160 fresh samples.

Equipment used:  Our SpinBot provided uniform film fabrication that enabled researchers to investigate complex relationships in the production of stable perovskites.  

Optimizing Perovskite Thin-Film Parameter Spaces with Machine Learning-Guided Robotic Platform for High-Performance Perovskite Solar Cells (Adv. Energy Mater. 48/2023)

Jiyun Zhang, Bowen Liu, Ziyi Liu, Jianchang Wu, Simon Arnold, Hongyang Shi, Tobias Osterrieder, Jens A. Hauch, Zhenni Wu, Junsheng Luo, Jerrit Wagner, Christian G. Berger, Tobias Stubhan, Frederik Schmitt, Kaicheng Zhang, Mykhailo Sytnyk, Thomas Heumueller, Carolin M. Sutter-Fella, Ian Marius Peters, Yicheng Zhao, Christoph J. Brabec

First published: 22 December 2023

This study presents SPINBOT, a fully automated platform designed for engineering solution-processed thin films. SPINBOT can conduct unsupervised experiments on hundreds of substrates with precise control. This method accelerates the optimization of perovskite solar cells through simple photoluminescence characterization. The optimized films achieved a power conversion efficiency (PCE) of 21.6% and retained 90% efficiency after 1100 hours of continuous operation at 60–65 °C. Integrating robotic platforms with intelligent algorithms is expected to advance autonomous experimentation in materials science research.

Equipment used:  Using Bayesian optimization (BO) and machine learning (ML), our SPINBOT iteratively improves the quality and reproducibility of thin films.


+49 911 4505 6512