Newly discovered connection could help design of nextgen alloys

A study conducted by Assistant Professor Liang Qi's group provides a robust example and a key step to construct advanced theories to describe the quantitative connections between the chemical bonding characteristics at the electronic level and the macroscopic materials’ properties.
Newly discovered connection could help design of nextgen alloys

Ceramics are brittle but hard even at high temperatures; metals and metallic alloys are ductile but relatively soft. These differences originate from characteristics of chemical bonds determined by electronic structures. Can we manipulate electronic structures to design alloys that can be as hard as ceramics but still ductile? This dream, if achieved, can help us to produce more efficient turbine engines and safer nuclear reactors.

Many engineering applications, like those mentioned above, require the usage of alloys that are strong, tough, and stable in extreme environments, like high temperatures and high irradiation. Alloys based on refractory metal elements that have high melting temperatures (> 2000 oC), like tungsten, tantalum, molybdenum, and niobium, are potential candidates. Recently, the research team of MSE Professor Liang Qi and his collaborators discovered a quantitative model to describe the electronic contributions to alloy properties. This result will significantly speed up the design and development of these advanced alloys.

Metals are crystals with almost all atoms in perfectly ordered positions, but there are always small exceptions. Those exceptional atoms and the corresponding atomistic structures are called “defects.” The properties of defects decide mechanical, thermal, and irradiation performances of metals because atoms at defects usually have fewer constraints to move around compared with those at perfect positions. Alloying elements so-called “solutes” are intentionally added into metals, or inevitably included as impurities, to change material performances. It is because solute atoms can go to these defects, interact with them, and change their behavior. 

Since the interactions between solute atoms and defects are essential for predicting alloy properties, many experimental and computational studies were conducted to understand them. Based on quantum mechanics, first-principles calculations can accurately predict the defect-solute interaction energy, a key parameter to evaluate the solute effects on defects. However, these calculations still have too high computational costs. For advanced alloys that serve in extreme environments, there are many types of sophisticated defect structures and various choices of solute atoms. It is impractical to use direct first-principles calculations to map out all critical interaction energies.

In the paper recently published in Nature Communications with Dr. Yong-Jie Hu, a postdoc researcher in Prof. Qi’s group, as the first author, the research team presents an efficient and general way to predict the solute-defect interaction energies in binary alloys of refractory metals. The critical idea is that the different electronic structures at defect sites compared with those at perfect lattice sites are the fundamental driving forces for the defect-solute interactions. The key problems are how to describe such differences and connect them to defect-solute interactions quantitatively.

Inspired by electronic structure models of perfect crystals, the team constructed two quantitative parameters (so-called “descriptors”) to describe electronic structure variations for each atom in pure metals. These descriptors can be obtained by electronic structures generated from first-principles calculations of pure metals containing defects. Then they found a quantitative linear correlation between these descriptors and the solute–defect interaction energies obtained from many expensive first-principles calculations.

Amazingly, this linear correlation is independent of defect types or sites for a given pair solute atom and metal crystal. The team has confirmed that such a linear correlation can be used to predict solute concentrations for multiple types of solutes at complex defects, like high-angle grain boundaries. These grain boundaries are common weak points in crystals and responsible for material performance degradations and failures. This linear model can also be used to predict the solute interactions with dislocations, which are defects to help crystals deform without catastrophic fractures. Controlling these defects by solutes are vital steps to achieve our dream alloys mentioned at the beginning of this article.

This study builds a scientific link between electronic structures and the defect behaviors in alloys. More studies are still needed to achieve efficient alloy design based on first-principles calculations. The fast-developing machine-learning techniques are promising tools to accelerate these processes. It is usually hard to choose representative descriptors with physical meaning for material properties as machine learning inputs. Fortunately, this study found potential electronic structure descriptors mentioned above. The discovery process was achieved based on “human learning” from classical electronic models. It indicates that, in the age of big data and artificial intelligence, human intelligence still provides reliable resources for scientific discoveries.

This study was achieved by collaborations between Prof. Qi’s group and several collaborators, including Prof. Xiaofeng Qian’s group from Texas A&M University, and Dr. Ge Zhao and Prof. Zi-Kui Liu from Penn State University. They provide critical contributions to electronic structure models, statistical analyses, and defect calculations. This study was supported by the startup fund from U-M and National Science Foundation awards. All data and codes have been uploaded to an open-access database Materials Commons, which is funded by Department of Energy and located at U-M.

Read the full article.