Mathematical modeling and simulation enable on-screen experiments in order to obtain insights into a real system or to make predictions. This accelerates and improves the material development process, aimed at shorter development times and innovation cycles. SCHOTT uses these methods in the optimization and development of almost all of its technological processes and products.
“Thanks to mathematical simulation, in modern glass-ceramic fabrication, crucial properties such as zero thermal expansion or color impression, can be precisely adjusted to the desired values.” Dr. Fotheringham explains.
The development of innovative materials with surprising qualities is also promoted. A ‘hot’ example is a new borosilicate glass, which is ideal as a fire protection safety glass. However, to achieve this quality, it must first be thermally toughened, a complex physicochemical process, the result of which largely hinges on the ’hidden parameters’ of the glass. “For the first time, we were able to determine these parameters through simulation, rather than painstaking experimentation, significantly reducing development time and cost,” continues Dr. Fotheringham.
In addition, the development of specialty glass ceramic powders for the next generation of battery technologies offers impressive potential for success in this attractive future market. The material is key to the production of innovative solid- state batteries, which are aimed at increasing the range of electric vehicles.
As these examples underline, the development of new products often hinges on new materials with improved properties. It is estimated that new materials are currently pivotal to around two out of three innovations. Materials informatics can quite literally act as a turbo effect. This emerging discipline has recently begun to draw on artificial intelligence, machine learning and automated analysis processes in data processing. Such concepts already support successful applications in other areas such as autonomous driving, speech recognition on mobile phones and the utilization of Internet data for user profiles and customized marketing. This involves powerful computers processing reams of data based on self-learning algorithms.
“With the help of machine learning, patterns and relationships can be extracted and predictions made – which is exactly what we are aiming for. However, such data volumes are not available to us in glass development,” says Dr. Benedikt Ziebarth, Principal Scientist Materials Informatics. While the Internet basically generates billions of data for free, a single datapoint in the field of materials science may well be the result of a measurement that has cost thousands of euros. “We therefore develop hybrid models that combine the small amount of material data with domain knowledge. This domain knowledge is present in the extensive toolbox of mathematical simulations and modeling already available at SCHOTT. We are also involved in relevant consortium projects,” adds Ziebarth. SCHOTT thus maintains a lively exchange with institutes and companies on the digitalization of material development. This shows that besides artificial intelligence, human intelligence is still a necessary part of the equation. SCHOTT is thus set to remain a prime address for upcoming generations of ‘glass doctors,’ true to the spirit of its founder, Otto Schott.