First-hand information for everyone
Researchers synthesized 34 candidate ternary alloys by arc melting in inert argon and, for most samples, annealing at 1000 °C for one week. Powder X‑ray diffraction and Rietveld refinement confirmed that three previously unreported compounds—GdPt₂Ge, TiPt₅Si, and ZrPt₅Si—formed in the predicted on‑hull crystal structures and that YCoSi₃ crystallized as a new orthorhombic Cmmm polymorph, 8 meV / atom above the expected I4mm structure.
The ternary targets were selected from three lists: (A) compositions within 100 meV / atom of the convex hull and lacking reported ternaries; (B) hull entries with all binary edges present and exactly one reported ternary; (C) Alexandria PBEsol hull entries with a predicted disorder probability below 0.2. Eight of the 34 targets (≈24 %) formed a single‑phase ordered compound or a high‑energy polymorph; the remaining 26 were multiphase, disordered, or dynamically unstable.
Figure 1 shows the XRD patterns for GdPt₂Ge, TiPt₅Si, ZrPt₅Si, and YCoSi₃. The refinements match the predicted structures: GdPt₂Ge adopts the Pnma lattice (lowest‑energy in Alexandria), TiPt₅Si and ZrPt₅Si adopt the P4/mmm lattice, and YCoSi₃ adopts the Cmmm lattice. The annealing conditions were identical for all samples.

These findings demonstrate that a narrow filtering of database predictions—based on hull proximity and low disorder probability—can raise the experimental success rate for new ternary compounds. However, the modest overall yield and frequent failures underline the limitations of current density‑functional theory predictions, including dynamical instability, compositional disorder, and incomplete database hulls.