MIT Researchers Use AI to Identify Atomic Defects in Materials

In biology, defects are generally seen as negative; however, in materials science, they can be intentionally tuned to endow materials with useful new properties. Today, atomic-scale defects are carefully introduced during the manufacturing process of products such as steel, semiconductors, and solar cells to enhance strength, control electrical conductivity, and optimize performance. Despite the powerful role of defects, accurately measuring different types of defects and their concentrations in finished products remains a challenge, especially without damaging the final material. Without knowledge of the defects present, engineers risk producing items that perform poorly or exhibit unintended properties.

Now, researchers at MIT have developed an AI model capable of classifying and quantifying specific defects using data from a non-invasive neutron scattering technique. The model, trained on 2,000 different semiconductor materials, can detect up to six types of point defects in a material simultaneously, which would be impossible with conventional methods alone. “Existing techniques can’t accurately characterize defects in a universal and quantitative way without destroying the material,” says lead author Mouyang Cheng, a PhD candidate in the Department of Materials Science and Engineering. “For conventional techniques without machine learning, detecting six different defects is unthinkable. It’s something you can’t do any other way.”

The researchers claim that the model is a significant step towards harnessing defects more precisely in products like semiconductors, microelectronics, solar cells, and battery materials. “Right now, detecting defects is like the saying about seeing an elephant: Each technique can only see part of it,” says senior author and associate professor of nuclear science and engineering Mingda Li. “Some see the nose, others the trunk or ears. But it is extremely hard to see the full elephant. We need better ways of getting the full picture of defects because we have to understand them to make materials more useful.”

Manufacturers have become adept at tuning defects in their materials, but measuring precise quantities of defects in finished products is still largely a guessing game. “Engineers have many ways to introduce defects, like through doping, but they still struggle with basic questions like what kind of defect they’ve created and in what concentration,” says postdoc Chu-Liang Fu. “Sometimes they also have unwanted defects, like oxidation. They don’t always know if they introduced some unwanted defects or impurities during synthesis. It’s a longstanding challenge.”

The result is that there are often multiple defects in each material. Unfortunately, each method for understanding defects has its limitations. Techniques like X-ray diffraction and positron annihilation characterize only certain types of defects. Raman spectroscopy can discern the type of defect but can’t directly infer the concentration. Another technique known as transmission electron microscopy requires cutting thin slices of samples for scanning.

In several previous papers, Li and collaborators applied machine learning to experimental spectroscopy data to characterize crystalline materials. For the new paper, they sought to apply that technique to defects. For their experiment, the researchers built a computational database of 2,000 semiconductor materials. They created sample pairs of each material, one doped for defects and the other left without defects, and then used a neutron scattering technique that measures the different vibrational frequencies of atoms in solid materials. They trained a machine learning model on the results.

The model covers 56 elements in the periodic table and leverages a multi-head attention mechanism, similar to what ChatGPT uses. It extracts the difference in data between materials with and without defects and outputs a prediction of what dopants were used and in what concentrations. The researchers fine-tuned their model, validated it on experimental data, and demonstrated that it could measure defect concentrations in an alloy commonly used in electronics and in a separate superconducting material. They also doped the materials multiple times to introduce multiple point defects and test the limits of the model, ultimately finding it can make predictions about up to six defects in materials simultaneously, with defect concentrations as low as 0.2 percent.

The researchers were pleasantly surprised by how well it worked. “It’s very challenging to decode the mixed signals from two different types of defects — let alone six,” Cheng says. Typically, manufacturers of items like semiconductors run invasive tests on a small percentage of products as they come off the production line, a slow process that limits their ability to detect every defect. “Right now, people largely estimate the quantities of defects in their materials,” says Bowen Yu. “It is a painstaking experience to check the estimates by using each individual technique, which only offers local information in a single grain anyway. It creates misunderstandings about what defects people think they have in their material.” The results were exciting for the researchers, but they note that their technique of measuring vibrational frequencies with neutrons would be difficult for companies to quickly implement in their quality control processes.