For greater than 100 years, scientists have been using X-ray crystallography to find out the structure of crystalline materials resembling metals, rocks, and ceramics.
This method works best when the crystal is unbroken, but in lots of cases, scientists have only a powdered version of the fabric, which incorporates random fragments of the crystal. This makes it more difficult to piece together the general structure.
MIT chemists have now give you a brand new generative AI model that could make it much easier to find out the structures of those powdered crystals. The prediction model could help researchers characterize materials to be used in batteries, magnets, and plenty of other applications.
“Structure is the very first thing that you should know for any material. It’s vital for superconductivity, it’s vital for magnets, it’s vital for knowing what photovoltaic you created. It’s vital for any application you can consider which is materials-centric,” says Danna Freedman, the Frederick George Keyes Professor of Chemistry at MIT.
Freedman and Jure Leskovec, a professor of computer science at Stanford University, are the senior authors of the brand new study, which appears today within the Journal of the American Chemical Society. MIT graduate student Eric Riesel and Yale University undergraduate Tsach Mackey are the lead authors of the paper.
Distinctive patterns
Crystalline materials, which include metals and most other inorganic solid materials, are made from lattices that consist of many similar, repeating units. These units could be considered “boxes” with a particular shape and size, with atoms arranged precisely inside them.
When X-rays are beamed at these lattices, they diffract off atoms with different angles and intensities, revealing information concerning the positions of the atoms and the bonds between them. Since the early 1900s, this method has been used to investigate materials, including biological molecules which have a crystalline structure, resembling DNA and a few proteins.
For materials that exist only as a powdered crystal, solving these structures becomes way more difficult since the fragments don’t carry the total 3D structure of the unique crystal.
“The precise lattice still exists, because what we call a powder is actually a set of microcrystals. So, you will have the identical lattice as a big crystal, but they’re in a completely randomized orientation,” Freedman says.
For hundreds of those materials, X-ray diffraction patterns exist but remain unsolved. To attempt to crack the structures of those materials, Freedman and her colleagues trained a machine-learning model on data from a database called the Materials Project, which incorporates greater than 150,000 materials. First, they fed tens of hundreds of those materials into an existing model that may simulate what the X-ray diffraction patterns would seem like. Then, they used those patterns to coach their AI model, which they call Crystalyze, to predict structures based on the X-ray patterns.
The model breaks the means of predicting structures into several subtasks. First, it determines the scale and shape of the lattice “box” and which atoms will go into it. Then, it predicts the arrangement of atoms throughout the box. For every diffraction pattern, the model generates several possible structures, which could be tested by feeding the structures right into a model that determines diffraction patterns for a given structure.
“Our model is generative AI, meaning that it generates something that it hasn’t seen before, and that permits us to generate several different guesses,” Riesel says. “We are able to make 100 guesses, after which we are able to predict what the powder pattern should seem like for our guesses. After which if the input looks exactly just like the output, then we all know we got it right.”
Solving unknown structures
The researchers tested the model on several thousand simulated diffraction patterns from the Materials Project. In addition they tested it on greater than 100 experimental diffraction patterns from the RRUFF database, which incorporates powdered X-ray diffraction data for nearly 14,000 natural crystalline minerals, that that they had held out of the training data. On these data, the model was accurate about 67 percent of the time. Then, they began testing the model on diffraction patterns that hadn’t been solved before. These data got here from the Powder Diffraction File, which incorporates diffraction data for greater than 400,000 solved and unsolved materials.
Using their model, the researchers got here up with structures for greater than 100 of those previously unsolved patterns. In addition they used their model to find structures for 3 materials that Freedman’s lab created by forcing elements that don’t react at atmospheric pressure to form compounds under high pressure. This approach could be used to generate latest materials which have radically different crystal structures and physical properties, despite the fact that their chemical composition is identical.
Graphite and diamond — each made from pure carbon — are examples of such materials. The materials that Freedman has developed, which each contain bismuth and one other element, could possibly be useful within the design of latest materials for everlasting magnets.
“We found numerous latest materials from existing data, and most significantly, solved three unknown structures from our lab that comprise the primary latest binary phases of those mixtures of elements,” Freedman says.
With the ability to determine the structures of powdered crystalline materials could help researchers working in nearly any materials-related field, based on the MIT team, which has posted an internet interface for the model at crystalyze.org.
The research was funded by the U.S. Department of Energy and the National Science Foundation.
