Expanding quasiperiodicity in soft matter: Supramolecular decagonal quasicrystals by binary giant molecule blends

The quasiperiodic structures in metal alloys have been known to depend on the existence of icosahedral order in the melt. Among different phases observed in intermetallics, decagonal quasicrystal (DQC) structures have been identified in many glass-forming alloys yet remain inaccessible in bulk-state condensed soft matters. Via annealing the mixture of two giant molecules, the binary system assemblies into an axial DQC superlattice, which is identified comprehensively with meso-atomic accuracy. Analysis indicates that the DQC superlattice is composed of mesoatoms with an unusually broad volume distribution. The interplays of submesoatomic (molecular) and mesoatomic (supramolecular) local packings are found to play a crucial role in not only the formation of the metastable DQC superlattice but also its transition to dodecagonal quasicrystal and Frank–Kasper σ superlattices.

Machine learning prediction of the electronic property of binary transition metal alloys

Machine learning methods have garnered much attention and use in computational catalysis. Previous studies have demonstrated rapid and accurate prediction of a variety of catalytic properties as well as the underlying potential energy landscapes. In particular, d-band center, defined as the first moment of the d-projected density of states, has been widely used as the key descriptor of activity trends for reactions catalyzed by metal surfaces. In this work, we construct a gradient boosting regression (GBR) model for prediction of the d-band center of bulk binary transition metal alloys. An accurate model is obtained using a dataset of over 1200 alloys from the Materials Project database spanning the entire d-block of the periodic table. The d-band centers, periodic groups, and relative compositions of the constituent metals are determined to have the highest feature importance scores, consistent with the underlying physics of the alloy. The regression model presented here offer a promising strategy of rapid property prediction with physical interpretability to aid the optimization and discovery of efficient heterogeneous catalysts.