Novel ground states in mixed bcc/fcc, high-/low-spin Fe-(Ni,Pd,Pt) from first-principles*

Among the observed ordered phases of Fe-X (X=Ni,Pd,Pt), some structures are curiously absent: L1$_2$ Fe$_3$Pt exists, but Fe$_3$Pd and Fe$_3$Ni do not; L1$_0$ FePt and FePd exist, but the stability of L1$_0$ FeNi is debated. Furthermore, the recently measured short range order (SRO) in Fe-rich Fe-Ni is at odds with L1$_2$-type SRO. Theory has been hindered by the appearance of geometric (bcc/fcc) and magnetic (high-spin [HS]/low-spin [LS]) bi-stabilities. We address such bi-stabilities by performing first-principles mixed-basis cluster expansion with added geometric and magnetic ``filters,'' separating structures according to HS/LS and ``degree'' of fcc-ness and bcc-ness. We performed separate fcc HS and (for Fe-Pd) bcc HS cluster expansions. We predict that {\em at low temperatures}: (i) New ordered structures exist, including fcc Pt$_8$Ti-type FeX$_8$ and the (100) superlattices Fe$_2$Pd and Fe$_2$Pt; (ii) {\em All } fcc Fe$_3$X compounds are unstable w.r.t. bcc Fe + L1$_0$ FeX (iii) Fcc FePd is unstable w.r.t. bcc Fe + (100) Fe$_2$Pd superlattice, while L1$_0$ FeNi is, in fact, stable w.r.t. bcc Fe + L1$_2$ FeNi$_3$. At high T, (iv) The SRO in Fe-rich Fe-Ni is governed by (100) superlattice ordering.