At a kick-off meeting held in May, researchers from the three NWO institutes gathered to align their roadmap and research strategy. HELIOS is coordinated by Amsterdam based institute AMOLF, that will contribute its expertise in fundamental research on functional materials. DIFFER, located in Eindhoven, brings in strong systems engineering, machine learning, and data management capabilities, while CWI will lead the AI efforts.
The five-year project, with a total budget of €3.71 million, focuses on developing an advanced, multilayered halide perovskite solar cell. This solar cell is made up of many ultra-thin layers – potentially up to twenty – each with a specific function to improve performance. A regular solar cell has five layers.
Halide perovskite solar cells
Unlike traditional silicon-based solar panels, halide perovskites are highly efficient at capturing sunlight, can be produced more cheaply, and allow for flexible applications. Stacking multiple layers could significantly boost efficiency and stability, but it also introduces complex challenges: changes in one layer can affect others. This is why AI plays a key role in the HELIOS project, helping researchers navigate the vast number of possible material combinations and layer configurations.
CWI leads AI development
CWI will lead the AI component of HELIOS, designing and implementing novel approaches to guide the complex choices involved in the solar cell’s development. These approaches will need to extract patterns from vast quantities of experimental data and help researchers optimize the material layers for both efficiency and integration.
The CWI team will combine deep learning, explainable AI, and evolutionary algorithms to model key physical processes, extract knowledge, and optimize solar cell design for multiple objectives. These computational tasks require powerful hardware infrastructure – including a large number of GPUs for training machine learning models, significant CPU capacity for running simulations and performing optimisation, and ample storage for the terabytes of experimental and optimisation data generated during the project.
Peter Bosman, group leader Evolutionary Intelligence at CWI explains: “The combination of different expertise, acknowledging the complexity of all aspects that underlie this project – from physics to AI to automated experiments for state-of-the-art materials discovery – is crucial if you are aiming to achieve true breakthroughs in science with societal impact. HELIOS lays exactly such a multi-disciplinary foundation for the automated design of ultra-efficient Perovskite solar cells. It is not something that AMOLF, DIFFER, or CWI could do alone.”
About the NWO-I funding
HELIOS is one of the first projects funded through NWO-I’s new Strategic Renewal Fund. This fund enables NWO institutes to invest in innovative, experimental research directions that align with their long-term mission. Projects like HELIOS are intended to stimulate collaboration between institutes and to strengthen the national role of each research centre in tackling future scientific and societal challenges.
Header photo: Shutterstock
