High-entropy alloys challenge traditional paradigms of alloy composition. Historically, alloys have been optimized around one dominant element with a secondary component enhancing properties, such as strength or corrosion resistance. By contrast, HEAs leverage a complex cocktail of multiple elements, which intermix in a solid solution phase, generating materials with remarkable mechanical strength, corrosion resistance, and thermal stability. Yet, the atomic-scale arrangement of these elements, especially on surfaces where catalytic and mechanical properties are often governed, has remained an intractable puzzle.

The key to untangling this puzzle lies in understanding the concept of local chemical ordering—or the subtle, non-random arrangement of atoms within the otherwise disordered crystalline lattice. Although prior assumptions suggested some degree of local ordering, direct experimental evidence, especially at surfaces, has been conspicuously absent. This knowledge gap has thwarted efforts to precisely tailor surface properties, hampering innovations in sectors requiring materials that endure extreme environments, such as jet engines, nuclear reactors, and energy storage devices.

Kim’s research, conducted under the guidance of Professor TeYu Chien and in collaboration with a multidisciplinary team spanning several universities, sets a new standard for probing the atomic-scale surface chemistry of HEAs. Their focus centered on the well-studied CoCrFeMnNi system—a canonical high-entropy alloy known for its mechanical robustness and stability. The team employed an integrative approach combining surface-sensitive scanning tunneling microscopy (STM) with advanced computational density functional theory (DFT) simulations.

Scanning tunneling microscopy, renowned for its exceptional resolution, enabled the visualization of atomic arrangements on the alloy’s surface with quasi-long-range ordering. This means that while perfect periodicity was absent, discernible patterns in atomic distribution could be detected, challenging the prevailing notion of wholly random element placement. To refine these observations, Kim and colleagues applied DFT calculations, a quantum mechanical modeling method, which provided insights into the energetics and stability of specific atomic configurations within these quasi-ordered domains.

This dual-experimental and theoretical framework culminated in the first unequivocal observation and characterization of surface local chemical ordering in a high-entropy alloy. The implications are multifold: by correlating surface atomic organization with physical and chemical properties, scientists can now envisage engineering HEAs with tailor-made functionalities, be it enhancing catalytic activity for chemical processing or improving corrosion resistance for harsh operating environments.

“The revelation that surface local chemical ordering exists fundamentally shifts our understanding of HEAs,” explains Professor Chien. “It means that by manipulating this order, we gain a powerful lever to control surface properties—a breakthrough that had previously been out of reach due to limitations in detection technologies.”

This advancement is not only a triumph of instrumental innovation but also of international collaboration. The project synergized expertise from the University of Wyoming, University of New Haven, University of Tennessee-Knoxville, and Taiwanese institutions National Yang Ming Chiao Tung University and National Tsing Hua University. Among the notable contributors is Jien-Wei Yeh, a pioneer who first demonstrated the stability of high-entropy alloys over two decades ago, signifying the lineage and evolution of HEA research.

Beyond academic merit, the practical applications of this insight could be transformative. With precise control over atomic-scale surface arrangements, materials scientists can devise alloys that do not just meet but exceed the rigors of future technologies. Imagine turbine blades in jet engines that maintain integrity at higher temperatures, or battery components with enhanced durability and efficiency due to optimized surface catalytic reactions.

This research was facilitated by funding from the U.S. National Science Foundation and the Air Force Office of Scientific Research, underscoring the strategic importance of materials innovation in national scientific agendas. The findings were recently published in the prestigious journal Nature Communications, signaling high recognition by the broader scientific community.

Kim’s methodology includes mapping surface atoms, detecting disparities in elemental distribution, and modeling their energetic preferences using DFT calculations. This hybrid approach overcomes prior technical hurdles, such as distinguishing neighboring atoms of similar atomic numbers, and surpasses earlier indirect inference techniques that could not definitively prove local chemical order.

Looking ahead, this breakthrough opens avenues for systematic exploration of surface phenomena across other HEAs, potentially unraveling new physical principles governing alloy behavior at the nanoscale. Moreover, it suggests that entropy, traditionally viewed as a measure of disorder, might be harnessed in a nuanced manner to design materials that balance order and randomness for exceptional performance.

In summary, the direct visualization of surface local chemical ordering in HEAs marks a pivotal moment in materials science. It bridges a critical knowledge gap, delivers a versatile analytical toolkit, and lays the groundwork for designing next-generation alloys tailored at the atomic level. As industries push the boundaries of performance and durability, these advances will doubtlessly play a central role in shaping the materials of the future.

Subject of Research: Not applicable

Article Title: Direct visualization of the existence of surface local chemical order in a high-entropy CoCrFeMnNi alloy

News Publication Date: 28-Mar-2026

Web References: https://www.nature.com/articles/s41467-026-71170-z

References: 10.1038/s41467-026-71170-z

Keywords

Physical sciences, Materials science, Physics, Chemistry, High-entropy alloys, Surface local chemical ordering, Scanning tunneling microscopy, Density functional theory, CoCrFeMnNi alloy, Alloy design, Atomic-scale visualization, Advanced materials

MPEAs, also known as high-entropy alloys, differentiate themselves by being composed of three or more principal metallic elements, a composition strategy that delivers formidable strength, corrosion resistance, and thermal stability. These attributes make MPEAs vital in critical technologies such as knee replacements, bone plates, catalytic converters, and high-performance aircraft parts. Despite their immense potential, the traditional approach to crafting such alloys has been hindered by extensive trial-and-error experimentation, often resulting in protracted and costly development cycles.

Deshmukh and his multidisciplinary team sought to overcome these limitations by integrating a data-driven methodology with state-of-the-art explainable AI techniques. Explainable AI diverges from conventional “black box” models by illuminating the rationale behind its predictions, providing tangible insights into the interplay of elemental compositions and their influence on alloy characteristics. Employing SHAP (SHapley Additive exPlanations) analysis, the researchers unpacked the AI’s decision-making process, revealing elemental synergies and local atomic environments critical to achieving superior mechanical performance.

This innovative AI framework enabled rapid screening and optimization of countless hypothetical MPEA formulations, narrowing down options that would otherwise take years of experimental work to evaluate. By leveraging expansive datasets incorporating both experimental results and high-fidelity simulations, the model could predict with remarkable accuracy how various combinations and concentrations of elements affect key properties such as yield strength, ductility, and fatigue resistance. This predictive capability transforms materials discovery from a largely empirical endeavor into an informed, highly focused exploration.

The team went beyond pure computational exploration by integrating evolutionary algorithms into the design pipeline. These algorithms mimic natural selection mechanisms, iteratively refining alloy compositions toward optimal performance. Through this synergistic pairing of explainable AI and evolutionary computation, the researchers accelerated the identification of promising MPEAs that not only outperform traditional alloys but also offer greater resistance to wear and corrosion under extreme conditions.

A pivotal aspect of this research is the tangible interpretability that explainable AI provides, enabling scientists to extract fundamental materials science insights rather than merely relying on opaque predictions. Understanding which elements and atomic-scale factors drive mechanical behavior allows for more rational, hypothesis-driven alloy engineering. Such insights pave the way for expanding this methodology to design other complex materials beyond metals, including polymeric glycomaterials, which hold promise for applications in personal care products, food additives, and biomedical devices.

Fangxi "Toby" Wang, a postdoctoral associate on the project, emphasized that this approach is not just about faster discovery but creating versatile, transferable design tools. The workflow’s interpretability and adaptability provide a blueprint for the broader materials community to tackle various complex systems where traditional trial-and-error design approaches fall short. It embodies a future where materials scientists can systematically tailor material properties with unprecedented precision and understanding.

Collaboration across disciplines and institutions was instrumental in reaching these milestones. Partnerships with experts like Tyrel McQueen, a professor of materials science at Johns Hopkins University, and Maren Roman, a sustainable biomaterials professor at Virginia Tech and director of the NSF-supported GlycoMIP platform, enriched the research’s depth and scope. This interdisciplinary synergy bridged computational modeling, experimental synthesis, and characterization, enabling a closed-loop process that validated novel MPEA predictions and accelerated the transition from design to application.

Graduate researchers, including Allana Iwanicki at Johns Hopkins, contributed by synthesizing and testing the new alloy compositions in laboratory settings. Their experimental work confirmed the enhanced mechanical attributes predicted by the AI model, validating the novel computational methodologies. This comprehensive approach from theoretical design to empirical validation underscores the robustness and real-world relevance of the new materials discovery framework.

While initial efforts concentrated on solvent-free metallic systems, Deshmukh’s team has expanded their AI-driven computational design to more intricate biomaterials like glycomaterials—polymers containing carbohydrate structures. These materials possess broad applicability in biotechnology and consumer sectors, highlighting the translational potential of combining explainable AI with materials innovation platforms. The ability to rationally engineer such complex systems could catalyze breakthroughs in health, sustainability, and manufacturing technologies.

Deshmukh remarked that this research exemplifies the transformative power of integrating AI-driven predictive modeling with experimental science. It not only accelerates the discovery of superior metallic alloys for industrial applications but also establishes a versatile discoverability framework that can cross traditional disciplinary boundaries. The union of machine learning, evolutionary computation, and interpretability marks a significant stride towards a new era of scientific materials design, where predictive understanding replaces empirical guessing.

As AI continues to evolve, its impact on materials science is expected to deepen, shifting paradigms of how alloys and advanced materials are conceptualized, optimized, and commercialized. This research highlights the profound value of transparency in AI-driven decisions—ensuring that scientific innovation remains explainable, trustworthy, and actionable. With such advancements, industries will more swiftly develop materials that meet bespoke performance specifications while reducing costs and environmental impact.

The synthesis of explainable AI with materials engineering heralds a new frontier wherein human insight and computational power operate in concert, unraveling the complexities of atomic interactions at unprecedented speed and clarity. This not only accelerates innovation but empowers researchers to innovate with a purpose, creating next-generation materials that advance technology and improve lives on a global scale.

Subject of Research: Development of advanced multiple principal element alloys (MPEAs) using explainable artificial intelligence for superior mechanical properties.

Article Title: Not explicitly provided.

News Publication Date: 15-May-2025

Web References:

• Sanket Deshmukh faculty profile: https://che.vt.edu/People/faculty/Deshmukh.html
• Chemical engineering department at Virginia Tech: http://che.vt.edu/
• GlycoMIP NSF Materials Innovation Platform: https://glycomip.org/
• Article DOI: http://dx.doi.org/10.1038/s41524-025-01600-x

References: Published in npj Computational Materials (Nature), DOI: 10.1038/s41524-025-01600-x

Image Credits: Photo by Hailey Wade for Virginia Tech

Keywords

Chemical engineering, Engineering, Materials science, Metals, Alloys, Artificial intelligence, Computer science, Machine learning, Biomaterials