This article is compiled by artificial intelligence observation
Translator: Sandy
Scientists have teamed up to use artificial intelligence to discover a new alternative to steel in a record time. In this experiment, they found a total of three mixtures for synthesizing metallic glasses 200 times faster than before.
Metallic glass is essentially an alloy of the future. Often, mixing small amounts of metals together means that the desired properties of each metal can be "added" together to form a "super metal". On the surface, alloys appear to be indistinguishable from metals, and their atomic structures exhibit rigid geometric shapes.
Although metallic glass does not have rigid geometry, it has a disordered atomic structure similar to glass. Therefore, metallic glass is stronger and lighter than today's steel, and it is also more resistant to corrosion and wear, and is considered an ideal replacement for steel.
Speaking of which, metallic glass is relatively new, but not all combinations of metallic glass components have passed the test. Over the past 50 years, of the millions of possible combinations of ingredients, thousands have been evaluated, but only a few are likely to be useful. The way to get a prediction or model the best combination is to look in the world of metallic glass.
So, a scientific team led by scientists from the U.S. Department of Energy's SLAC National Accelerator Laboratory, the National Institute of Standards and Technology (NIST), and Northwestern University conducted experiments and reports that they have found a shortcut to discovering and improving metallic glasses by artificially Smart technology that can discover new materials with less time and cost.
Using SLAC's Stanford Synchrotron Radiation Light Source (SSRL), the team discovered three new mixtures of raw materials through a machine learning system to form metallic glasses. The findings are reported in Science Advances, a sub -journal of Science .
Northwestern University professor Chris Wolverton, a pioneer in the use of computers and artificial intelligence to predict new materials, is one of the paper's collaborators. He said that it usually takes ten or twenty years for new materials to complete the process from discovery to commercialization. "This achievement has greatly shortened the time it takes to discover new materials."
This work can be used not only for metallic glass but also for other materials, which means it is a very valuable technology for industry. The ultimate goal, according to Professor Wolfton, is to allow scientists to get direct feedback from machine learning models and have another set of samples ready to be tested the next day or even the next hour.
In the past half century, scientists have studied the composition of about 6,000 metallic glasses, and the new system can make and screen 20,000. While other teams are also using machine learning predictions to find different kinds of metallic glasses, this time scientists' rapid validation and prediction through experiments, and then looping the results into the next round of machine learning and experimentation, is unique to this advance. place.
In fact, this method can be used in various experiments, especially in finding materials, such as metallic glasses and catalysts, to great advantage. NIST materials research engineers have said that artificial intelligence will change the landscape of materials science.