The smartphone in your pocket, the solar cell on the roof, the battery in the electric car - behind all of these are years of attempts to find the right material. Researchers test, analyze and improve again and again. This costs time and money. In Dresden, two Leibniz Institutes are now working on significantly accelerating this process with artificial intelligence.
The Leibniz Institute for Solid State and Materials Research Dresden (IFW Dresden) and the Leibniz Institute of Polymer Research Dresden (IPF) have launched the "Leibniz-AI4MAT" project for this purpose. The name stands for "Artificial Intelligence for In Silico Materials Research" - in other words, AI-supported materials research on the computer. "In silico" means that experiments are first digitally simulated before they take place in the laboratory.
Less trial and error
Materials research has so far often been based on many successive experiments. AI aims to make this process more targeted. Computer models evaluate large amounts of research data and recognize correlations that are difficult to capture using traditional methods. This makes it possible to predict which materials could have certain properties - before complex laboratory experiments begin.
A powerful computing cluster, i.e. a network of many interconnected computers, is at the heart of the project. It will be jointly available to both institutes. With its help, researchers can calculate material properties, virtually test new material combinations and identify suitable production methods.
Research and save energy
The operation of such high-performance computers requires a lot of energy and generates heat. To ensure that this does not remain unused, a waste heat recovery system is planned that reuses the heat generated. The project is being financed by the European Regional Development Fund (ERDF) and the European Union's Just Transition Fund - programs that support regions in structural change and the transition to a climate-neutral economy. The Free State of Saxony is thus specifically promoting application-oriented research infrastructure.
In future, the materials of tomorrow - for example for energy, mobility or electronics - could first be developed and optimized on the computer before they are produced in the laboratory.
