Half-Heusler Structures with Full-Heusler Counterparts: Machine-Learning Predictions and Experimental Validation

Abstract

Heusler compounds form a diverse group of intermetallic materials encompassing many compositions and structures derived from cubic prototypes, and exhibiting complicated types of disorder phenomena. In particular, preparing solid solutions between half-Heusler ABC and full-Heusler compounds AB2C offers a means to control physical properties. However, as is typical in materials discovery, they represent only a small fraction of possible intermetallic compounds. To address this problem of unbalanced data sets, a machine-learning model was developed using an ensemble approach involving the synthetic minority oversampling technique to predict new compounds likely to adopt half-Heusler structures. The training set was based on experimental crystal structures, including those of nonstoichiometric compounds. The model achieved an accuracy of 98% on the validation set and gave excellent performance in terms of balanced statistical measures. A subset of compounds predicted to adopt half-Heusler structures having existing full-Heusler counterparts was then targeted for preparation. Six of seven of these candidates were successfully synthesized and confirmed to be half-Heusler compounds.

Superconductivity in LiGa2Ir Heusler type compound with VEC = 16

“…With more than a thousand members reported in the literature, the Heusler family remains one of the most interesting and intensively studied intermetallic systems in materials science 1 . Among this class of materials we can find catalysts 2 , ferromagnets 3 , 4 , thermoelectric 5 – 7 and magnetocaloric materials 8 .…”

Machine Learned Synthesizability Predictions Aided by Density Functional Theory

“…Our model reproduces an expected correlation between DFT stability and synthesizability, yet it also can identify metastable HHs and stable HHs that are unsynthesizable (compositions where a non-HH structure is metastable). We show prediction improvements for several HHs predicted by Gzyl et al [45], and our model achieves a precision of 0.83 and recall of 0.85. We conclude by applying our model on several thousand ABC compositions unreported in literature and identify promising candidates for further experimental study.…”

“…We begin by focusing on the synthesizability models by Gzyl et al [45] and Legrain et al [46], which predict the synthesizability of an ABC composition in the HH structure. Legrain et al used a random forest algorithm with a training dataset comprised of ABC compositions flagged as reported in the ICSD within the AFLOW [12] database.…”

“…Figure 2: Comparisons of E hull from the OQMD and synthesizability for synthesizable HHs as determined by Gzyl et al's [45] and Legrain et al's [46] models. Both models are not informed of DFT stability and predict many unstable HHs to be synthesizable.…”

“…Nevertheless, we are confident VRhSn and MnRuSb are not ordered HHs as originally claimed. For each composition in table 2, we also list the compounds that [45] for the seven compositions their model predicted to be HH. We simulate XRD patterns (see Supplemental fig.…”