Understanding how elements mix is one of the grand challenges in materials science. With 118 known elements and millions of possible combinations, identifying promising alloys has often relied on heuristics, trial-and-error, or computationally expensive models.
In a newly published paper in Advanced Science, COMPASS-supported researchers demonstrate a powerful new approach: applying graph theory to map elemental miscibility across the periodic table.
By representing elements as nodes in a network and their thermodynamic relationships as edges, the team shows how graph metrics—such as closeness centrality—can reveal which elements are broadly miscible, which resist mixing, and which are best suited for interfacial or passivating roles. This network-based method not only aligns well with established models like CALPHAD and Miedema’s model, but also opens the door to machine learning–enabled predictions under extreme or previously unexplored conditions.
Thanks to COMPASS collaboration and support, this work goes beyond explaining known behavior. It points toward the next frontier: identifying alloy systems that have never been made before and, in future work, predicting their properties in collaboration with partners like Xiaoming.
This research highlights the role COMPASS plays in advancing data-driven, interdisciplinary tools for complex materials discovery—where physics, chemistry, and network science intersect.