SOLAR-SKIES: AI-supported optimization of CIGS thin-film PV technology

Challenge: Increasing efficiency and reducing costs through data-driven approaches

Photovoltaic (PV) technologies are a key component of the energy transition but face challenges in terms of efficiency, cost, and environmental compatibility. Flexible CIGS thin-film solar cells offer tremendous potential but require continuous optimization of materials and processes. SOLAR-SKIES addresses these challenges through data-driven approaches and AI-based modeling.

The SOLAR-SKIES project focuses on three main objectives:

• Increase PV efficiency: AI-based modeling and data analysis aim to significantly improve the conversion efficiency of flexible CIGS PV cells. The goal is 18% efficiency for 1 cm² solar cells and 14% for 10×10 cm² modules.
• Reduce production costs: By analyzing extensive process data and applying automated optimization via AI, the production cost of flexible CIGS PV cells is to be reduced to under €0.90/Wp.
• Sustainability and stability: Data-driven material analysis enables the selection of environmentally friendly, easily recyclable materials. The goal is to ensure performance stability with less than 10% power loss after 1,000 hours at 85% relative humidity and 85 °C.

Technological solution: AI-based analysis and high-throughput screening

SOLAR-SKIES combines high-throughput screening with AI-based analysis to accelerate and optimize the development process of thin-film solar cells. Data from combinatorial thin-film deposition and automated measurement methods are collected and analyzed, with AI models identifying optimal material combinations and process parameters.

Using machine learning and Bayesian networks, correlations between process parameters, material properties, and solar cell performance are determined. These data-driven insights enable precise control of manufacturing processes and prediction of new material combinations and process parameters. This accelerates development and helps evaluate and scale innovative materials faster.

Identified optimizations are scaled from lab to pilot scale without complex production line modifications. This approach enables cost-efficient and sustainable production scaling.

RISC Software GmbH: Expertise in AI-based methods

In the context of the project, RISC Software GmbH will contribute by applying AI-based methods to identify causal dependencies between generation process parameters, material properties, and the efficiency of the resulting solar cells.

The underlying measurement data for these analyses are obtained through high-throughput screening. In this context, RISC Software GmbH also supports the collection and efficient management of data, which ultimately enables the upscaling of research results into industrial processes.

This project is funded by the FFG.