Changwon Suh

The beginning of the new year is a great time to look ahead at what new technologies are on the horizon and think about how these new or improved tools will affect daily life. This eye-catching infographic by Michell Zappa visualizes some of what we can expect in the coming decades. Yet despite our boundless excitement for futuristic technologies like hover cars, we often overlook the value of retrospection; by looking back at how technologies have evolved, we can better understand current technology trends and project future technologies.

One good opportunity for retrospection is in data-driven materials research and development (R&D). The materials community has come a long way with data science, yet has put forth little effort to reflect on what and how pioneers were able to accomplish the milestones driving materials R&D forward today. As a result, some materials scientists believe the concept of “data-driven materials R&D” is relatively new. Moreover, some scientists may think that the materials community lags far behind other industries in applying emerging data technologies such as big data, artificial intelligence, machine learning, or deep learning. While this is partly true, data-driven materials technology is pervasive nowadays and evolving fast.

This blog post intends to provide insight into how data-driven technology has contributed to materials R&D. Continued strengthening of innovative ways to gather, glean, and share meaningful materials data will help efforts to deliver innovative products and promote U.S. materials and manufacturing competitiveness. Read on to find out more about how a historical understanding of applied science and engineering data has helped scientists effectively discover, develop, and deploy new materials.

Current Trends in Data-Oriented Materials R&D
By improving data-driven technology, the materials community can reduce R&D costs and accelerate time-to-market. In this case, data is defined as scientific or technical measurements, values, or facts in the materials community, and data manipulation can be regarded as a process to change data into information (i.e., collected, organized, and transformed data) and knowledge (i.e., collected information) to make a rational designs for materials or process/property optimization. As the titles of the recent workshops on Materials Research and Data Science Conference and AI applied to Materials Discovery and Design imply, the current focus of such data-driven materials R&D is on implementing AI-encompassing machine learning and deep learning to analyze materials data in a more automated way.

The History of Data-Oriented Materials R&D
Below is a brief chronology of several major publications that exemplify how the materials community has poured its energy into data-intensive R&D for the last three decades. This timeline emphasizes the enormous efforts undertaken to utilize materials data for materials discovery and design. You will be amazed to see that several of the keywords used actively in today’s materials community originated decades ago!