In everyday life, we utilize many high-tech materials: touch screen displays in smartphones, durable but lightweight composites in cars and airplanes, or innovative batteries for electromobility. However, scarce resources and the quest for energy saving and environmentally friendly production and disposal put enormous pressure on the materials industry to stay at the forefront of innovation. New paradigms are required for developing multifunctional and multiresponsive materials in the future, including new types of catalysts, self-healing polymers, or programmable matter. A holistic approach to materials design is indispensable to accelerate progress in this area beyond incremental improvements based on trial and error.
For decades, the development cycle in materials science relied on natural scientists to develop microscopic models either from first principles or from experimental observations. Similar reliance was placed on engineers to construct continuum theories for explaining macroscopic behavior. The gap between these scales and approaches still hampers the development of virtual test laboratories. These laboratories allow predicting the properties even of composite materials at all scales, ranging from micro- and mesostructures to composites and structured metamaterials. Said gap can be closed by combining the labs with a systematic exploitation of the vast amount of available data via machine learning methods. This will permit tackling the reverse problem of finding the ideal material to fit a set of required properties in silico.
To implement this vision in a virtual materials lab, data-based methods will be combined with traditional simulation strategies, and machine learning approaches will be evolved from generic predictors into models that obey fundamental physical principles. Data gathered from simulations, experiments, and sensors during the real-world use of current products will enhance the development of custom functional material systems of the future. A new generation of simulation environments will evolve to allow systematic model adaption based on required accuracy, time-to-solution, and available computational resources. Integrated and pervasive visualization of the results, their causes, and their uncertainties on different length scales will provide insights that cannot be achieved experimentally.
In cooperation with partners worldwide, basic academic research in such a virtual materials lab will generate knowledge to boost technological development of industrial engineering products. It will also lead to development cycles that are dramatically more efficient, flexible, reliable, and predictive. This will result in improved robots, batteries, and turbines all the way to cars or airplanes, and possibly to new products that we cannot even conceive as of today. Progress in this field will play a significant part in keeping the Stuttgart region a vibrant, world-leading technology center.