Most materials, especially metals and ceramics, are crystals. Their atoms are arranged in three-dimensional lattices that repeat the same exact pattern, over and over again. But there's a well-known saying in materials science: "Crystals are like people. It is the defects that tend to make them interesting."
In a new study, published in Physical Review Letters, researchers from Lawrence Livermore National Laboratory (LLNL) created a new model for crystal defects at realistic temperatures. The simulation technique overcomes long-standing challenges in the field to calculate material structure and properties that were previously impossible to obtain. The result points the way toward improved production and performance of materials.
The work focused on two types of defects: point defects and grain boundaries. Point defects arise when atoms are missing in the lattice or when extra atoms are wedged in between the regular structure. Grain boundaries occur where two crystals with different orientations meet. Imagine the latter defect like a patchwork quilt where multiple pieces of fabric are stitched together at the seams.
"Cracks often find it easier to grow along grain boundaries, which can cause materials to fracture," said author and LLNL postdoctoral researcher Flynn Walsh. "This is just one example of how defects affect the properties of materials ranging from protective walls in fusion energy plants to the magnets that power most electric motors."
To improve technology based on these materials, researchers need to understand what's happening to the crystal structure in complex defects like grain boundaries. While it is technically possible to image these defects, the associated experiments are very difficult. Modeling, therefore, is critical.
The new simulation technique advances the field with a simple but powerful idea: It allows atoms to come and go from the simulation. In a real-world defect, nature adjusts by moving atoms around until it finds a stable state. The team wanted to replicate that phenomenon.
"The conventional way to perform these simulations is to directly add and remove atoms, but this doesn't work in solid crystals because the energy barriers are too high," said Walsh. "Our approach is instead based on gradually adding and removing atoms. The basic idea is simple but doing it efficiently and correctly was surprisingly difficult."
Instead of abruptly shoving an atom through a packed crowd of its fellows, the model softly pushes or pulls it into place.
"For the first time, this new technique opens the door to predicting grain boundary structures and phase transitions at finite temperatures," said Timofey Frolov, LLNL scientist and principal investigator on the project. "This enables more accurate modeling of materials used in extreme environments such as fusion reactors."
The method is more computationally demanding than traditional approaches and greatly benefited from LLNL's supercomputing resources. But Walsh emphasized that the most important factor in its success was the research environment at the Laboratory. Much like defects make crystals interesting, the people involved (and their unique quirks and expertise) made this project possible.
"I was able to think deeply about this problem for a year and half with the guidance of experts in different areas of physics and materials science," Walsh said.
Other LLNL authors included Babak Sadigh and Joseph McKeown. The work was funded by Frolov's Department of Energy early career project and McKeown's Laboratory Directed Research and Development Strategic Initiative. The LLNL Institutional Computing Grand Challenge provided computational resources.
