Our unique deep learning approach has been developed to model, optimize, and discover new materials that simultaneously satisfy multiple physical criteria. An artificial neural network is trained from sparse historical data to enable the prediction of all material properties both as a function of composition and the processes applied. This allows the identification of optimized material formulations and discover new materials with properties most likely to exceed target criteria.
- Maximize material performance for multiple target properties
- Reduce prototype costs
- Reduce number of experiments by 90%
- Minimize expensive properties and reduce environmental impact
- Standardize the design process across the company
The current approach to developing new materials is experiment-driven trial-and-improvement. This approach may take up to twenty years to design and verify a new material. The long lead time rules out designing new materials alongside products, forcing engineers to compromise products around the shortcomings of pre-existing materials. The capability to discover materials computationally has the potential to empower engineers to utilize materials optimized for their application at the same time as new products, bringing materials into the heart of the design process. Previous approaches to design new materials on a computer have not been capable of simultaneously optimizing the compromise between material properties and capture the deep correlations between composition and final properties. The development of new algorithms and increase in computing power has paved the way to screening large numbers of prospective compositions.