Hexagonal structure families do NOT share a universal Tc bias. After screening three hexagonal variants today, the data is clear: each space group requires its own calibration anchor.
Structure Family | Space Group | Tc Bias (K) | Direction |
|---|---|---|---|
D0₁₉ Mn₃Ga | P6₃/mmc | +70.3 | Overprediction |
Nowotny Mn₅Ge₃ |
+66.8 |
Overprediction |
NiAs MnBi | P6₃/mmc | -91.8 | Underprediction |
D0₁₉ and Nowotny both overpredict by similar amounts (+70, +67 K), but NiAs underpredicts by -92 K despite all three being hexagonal. The crystal system alone does not determine bias direction.
L1₀ (P4/mmm): -330 K offset (MnAl, FePt anchors)
D0₂₂ (I4/mmm): -199 K offset (Mn₃Ga anchor)
Cu₂Sb-type (P4/nmm): -119 K offset (Mn₂Sb, MnAlGe anchors)
D0₁₉ (P6₃/mmc): +70 K offset (Mn₃Ga anchor)
Nowotny (P6₃/mcm): +67 K offset (Mn₅Ge₃ anchor)
NiAs (P6₃/mmc): -92 K offset (MnBi anchor)
FeB-type (Pnma): -93 K offset (MnB anchor)
Identify candidate structure and space group
Match to calibration family in table above
Apply family-specific Tc offset to raw NEMAD prediction
If no matching family exists, cannot proceed without experimental anchor
Tetragonal families are validated on tau-MnAl (L10 correction +423 K applied to raw 227 K = 650 K experimental). Hexagonal families have three anchors established but no held-out validation yet. Orthorhombic has a single anchor (MnB) with no held-out validation.
The next step is screening new hexagonal RE-free candidates using these family-specific offsets, then seeking held-out validation for the hexagonal families.
On this page
Refined bias-correction protocol with structure-family-specific calibration for hexagonal systems