Development of AI algorithms for the detection of defects and inhomogeneities as well as film quality correlations in in-situ image data of solution-processed perovskite thin films for solar cells.
How can AI methods revolutionise combinatorial materials science and processing? Find out in this week’s Helmholtz AI project showcase, where scientists at the Karlsruhe Institute of Technology (KIT), the German Cancer Research Center (DKFZ) and the Forschungszentrum Jülich (FZJ) are working on the transfer of existing and the development of new AI methods for the analysis of perovskite thin films in solar cells.
Could you introduce yourself, giving your affiliation, area of work, and of course, the project title?
The Helmholtz AI project “AI-InSu-Pero” is led by Tenure-Track Prof. Dr. Ulrich W. Paetzold from the Institute of Microstructure Technology (IMT) at KIT. The group’s research is directed towards unfolding the potential of next-generation photovoltaic technologies and focuses on the interaction between light and structured matter for the purpose of engineering novel optical concepts, materials, and device architectures for low-cost solar energy harvesting. We are strongly engaged in the fabrication, characterization, simulation, and understanding of the device physics of perovskite photovoltaics and perovskite-based tandem photovoltaics. The project is a cooperation with Helmholtz AI consultants @ KIT, in particular Dr. Charlotte Debus & Dr. Markus Götz, and the Helmholtz Imaging (HI) service unit “Applied Computer Vision Lab” at the DKFZ, represented by Dr. Fabian Isensee and Sebastian Ziegler.
In simple words, what specifically is your project about? And, how and why do you think it is a high risk, high gain endeavour?
In AI-InSu-Pero, we will transfer existing and develop novel AI methods for the analysis of perovskite thin films. High performance metal halide perovskite-based absorbers are a promising candidate for future commercial solar cells. The key challenge of upscaling fabrication to larger areas must be addressed to realise high-throughput industrial production. To achieve large-area fabrication, this project seeks to enhance the understanding of the underlying process steps and dynamics during the formation of the highly complex material system. To implement a high risk, high gain endeavour we seek to realise an explainable AI model. Making use of AI methods to enhance the understanding of process steps during the perovskite thin-film formation will serve as a first step towards the development of AI for real-time deposition monitoring of highly complex materials.
How important has the Helmholtz AI funding and platform been to carry out this project?
The funding and platform provided by Helmholtz AI has been crucial for the implementation of this project. The joining of forces of the different disciplines was only made possible through Helmholtz AI. The synergies created through the multidisciplinary collaboration in the form of high risk, high gain projects will help accelerate scientific progress in perovskite photovoltaic and many other research areas.
Figure: Schematic illustration of the experimental imaging setup utilised for generation of in situ data during the formation of the perovskite thin film. The data is then used as input for machine learning algorithms to enhance understanding of the underlying process steps during the formation of the highly complex material.