Can Public DFT Databases Guide Composition Selection for Metastable Magnets? A Validation of Energy-Above-Hull Screening Against Rapidly Solidified and Milled TbCu₇-Type Sm–Fe–M Alloys

Computational screening against open density-functional-theory (DFT) databases is now routinely used to choose compositions for new permanent magnets. Its reliability for metastable phases made by non-equilibrium routes has not been tested directly. Here we validate six public databases (Materials Project, AFLOW, JARVIS-DFT, NOMAD/Alexandria, OQMD and GNoME; 628 Sm–Fe–M entries, M = Co, Ti, V, Zr, Nb, Cu, Si, Ga, Al, Mo, Cr, Mn, Hf) against 38 literature observations of phases formed in melt-spun or mechanically milled Sm–Fe and Sm–Fe–M alloys, 30 of which are scorable across 11 alloy systems. Because no database contains the disordered TbCu₇ structure, we define a stability proxy as the lower convex envelope of each database's ordered CaCu₅, 2:17, 3:29 and ThMn₁₂ relatives, referenced to its own convex hull. Materials Project, OQMD and Alexandria place no Sm–Fe binary compound on the hull. As a consequence, equilibrium Sm₂Fe₁₇ (19–23 meV/atom), TbCu₇ SmFe₉ (16–19 meV/atom) and SmFe₁₂, which melt spinning does not produce (19–24 meV/atom), are separated by at most 4.4 meV/atom, an order of magnitude below typical DFT formation-energy errors. JARVIS-DFT orders the three phases correctly but places TbCu₇ SmFe₉ at 89 meV/atom and misses 19 of 20 covered observations at a 50 meV/atom window. AFLOW and GNoME contain no comparable entries. Computed Fe moments for TbCu₇ SmFe₉ bracket the measured value scaled to 0 K (1.88–1.96 μB/Fe): Alexandria overestimates it by 16% and JARVIS underestimates it by 22%. Energy above hull from current public databases therefore cannot, on its own, discriminate formable from non-formable metastable phases in Sm–Fe–M systems. We release the stability proxy table, validation matrix and pipeline as an open dataset. Preprint; not peer reviewed. Data, code and the stability proxy table: 10.5281/zenodo.23456632.