AMP_06_September_2021

A D V A N C E D M A T E R I A L S & P R O C E S S E S | S E P T E M B E R 2 0 2 1 2 3 processing. The model was used to de- termine if recipes with U.S.-based ingre- dients still met the OEM requirements. The model was then also used to optimize cost. Ingredients from a glob- al catalog were filtered by price, and recipes only using cheaper ingredients were evaluated for performance. PERFORMANCE, COST AND SUSTAINABILITY OPTIMIZATION If companies want to succeed in the market, performance and value are no longer enough—sustainability is also key. Multidimensional, multi-objective problems are perfectly suited for AI. The availability of a bank of AI-gen- erated recipes makes it more likely that target properties can be met by more than one material candidate. When that is the case, scientists can optimize cost and carbon footprint by systematically estimating them based on the ingredi- ents’ costs and life cycle assessments (Fig. 4). Costs can also be reduced when scientists are able to choose recipes that use ingredients that are already in stock. INSIGHT CREATION AND KNOWLEDGE CAPTURE Digitalizing R&D in materials can also benefit companies by making sure that the experience of their top employ- ees is not lost. Domain knowledge is captured and codified in AI models and design spaces. New scientists at a com- pany will be able to review graphical models and see which underlyingmate- rials mechanisms were thought import- ant by previous colleagues, as they will be represented. They will also be able to see which input data were import- ant to the final predictions and which were not. TRANSFORMING MATERIALS DEVELOPMENT AI for materials has the potential to transform materials development both in business and academia. It is now be- ing used by multinational materials and chemicals companies and will only be- come more prevalent over time as other companies catch on to the accelerated product development timelines and in- creased agility that AI unlocks in R&D. Materials developments are key to creating a sustainable future. Ac- cording to the United Nations, there is only an 11-year window to work on this [3] . On average, it has taken 20 years to commercialize a new material [4] . It is essential that the whole materials com- munity embraces materials informatics, so as to speed up development by two to five times. Education is key, both in univer- sity curricula and as part of continuing professional development. All materials scientists should feel comfortable part- nering with data scientists. In summary, materials informat- ics is proven to accelerate materials research and development. The tech- nology should be embraced, just as previous generations of scientists em- braced the scanning electron micro- scope. Materials informatics isn’t magic. But it is a powerful tool that can aid ma- terials scientists in achieving quicker and more reliable results. ~AM&P For more information: Hannah Me- lia, product management consultant, Citrine Informatics, 2629 Broadway St., Redwood City, CA 94063, 650.276.7318, hmelia@citrine.io , https://citrine.io/. References 1. J. Ling, et al., High-Dimensional Materials and Process Optimization Using Data-Driven Experimental Design with Well-Calibrated Uncertainty Esti- mates, Integrating Materials and Manu- facturing Innovation, 6 (3), p 207–21, 2017, https://link.springer.com/article/ 10.1007/s40192-017-0098-z. 2. NASA Sale Launches HRL Labora- tories’ Commercial 3D-Printed Alumi- num Effort, https://www.hrl.com/news/ 2019/09/30/nasa-sale-launches-hrl- laboratories-commercial-3d-printed- aluminum-effort. 3. Only 11 Years Left to Prevent Irreversible Damage from Climate Change, Speakers Warn during General Assembly High-Level Meeting, March 28, 2019, https://www.un.org/press/ en/2019/ga12131.doc.htm. 4. About the Materials Genome Ini- tiative, https://www.mgi.gov/about. Fig. 4 — AI can help optimize cost and carbon footprint by systematically estimat- ing them based on the costs and life cycle assessments of the ingredients. Fig. 3 — An AI model can predict the properties of plastics based on ingredients and processing.