Big data in material science

From the emergence of various inventions can be seen a law, new materials nurture new inventions. Gorilla Glass has been heavily used in smartphones; and Kevlar has been saving lives since entering consumer-grade products; lithium-ion battery technology has provided a source of power for high-energy facilities. In spite of this, developing new materials is an extremely time-consuming task.

  And those who depend on these materials in the product development cycle, compared to create a new breakthrough material is extremely time-consuming process. Boeing 787 Dreamliner from concept to commercial voyages only took less than 9 years; Apple iPhone design since 2005, 2007 can be officially listed. By contrast, the birth of new materials can take up to 20 years of research and experimentation.

Two years ago, the U.S. government set up the Materials Genome Initiative (MGI) project with the goal of breaking through this technological bottleneck. MGI's goal is to drastically reduce the time and money involved in the development of new materials. Similar to the Human Genome Project's task of mapping our genes, scientists want to find out through MGI that the interaction between elements has a wide range of effects on the type and nature of the material. Based on this knowledge, scientists and engineers will have the hope of "customizing" the corresponding materials for different applications in shorter cycles.

The number of elements arranged between the array of numerous, most of them have no meaning for us. Trying to exhaust these permutations and combinations in the lab is totally impractical. So, some of MGI's project teams have begun to leverage the power of big data to model all possibilities and then dive into the potential directions by analyzing the data.

For many years since its establishment, MGI has led to the cooperation of some third-party projects. Among them are the Materials Project from the Massachusetts Institute of Technology and the Harvard Clean Energy Project. The two plans seek different answers on a similar theoretical basis. The former focuses on inorganic solids, especially battery materials, while the latter's clean energy program centers around molecular materials that can be used in solar cells. Both use the huge database collected by Density Functional Theory to predict the actual properties of the simulated matter model.

MIT's Material Project was founded about eight years ago with the help of Professor Gerbrand Ceder. As a consultant for many companies, Ceder has accumulated a great deal of results. But working with a handful of companies keeps these valuable data closed. "If we provide that data to everyone, people will create a lot of amazing results, which is the Materials Project," he said. Currently, MIT's database holds about 100,000 kinds of known or theoretical material information. To take full advantage of these data in the development of new materials, MIT scholars use artificial screening and machine learning to explore various chemical laws.

Similarly, Harvard's Clean Energy Program also explores its database with a combination of manual and machine tools. The program started with a validation experiment on the concept of organic solar cell materials. In the completely virtual case, scholars have calculated the performance of about 15 new compounds in the real world. The end result of the simulation is a new material with super-electrical properties. This is just the result of a few experiments conducted by a graduate student. Imagine if you could borrow their computational power from a large army of volunteers and how much would the result be magnified?

That's exactly the strategy Clean Energy plans nowadays: Anyone can do the job by downloading a degree on the computer and return the result. With this huge resource at your disposal, academics have calculated millions of potential combinations - it's just the beginning. "The project has entered a very interesting phase," said Dr. Hachmann. "It is almost time for us to collect the fruits of hard work." Currently, Harvard publishes 2.3 million mixture combinations for research by all. Although the original intention of these data is to help the development of solar cells, scientists can also use any valuable information to help other aspects of research. MIT also has a web portal for people to read Data Project data.

Ceder hopes the great MGI program will accomplish its mission. In fact, Ceder has seen some results, and he is patenting a new battery material, which is a good result for the growing MGI and battery technology development. Ceder believes the addition of the Internet and big data can bring unpredictable improvements and discoveries to this program. "When you have this combination, you can not predict what people are going to bring to you."

Original link: http://www.36kr.com/p/206581.html

【Editor's Choice】

Oracle big data machine to protect enterprise big data security flu prevention and big data Big Data boom in Informatica new opportunities big data analysis US Open women's singles semi-finals: why Li Na lost to Serena store five years: the sudden emergence of flash memory big data cloud computing quickly 【Editor: Xiao Yun TEL: (010) 68476606】