What is Zr₆Fe₁₆Si₇C, the more stable of our two new carbides, actually like, and what could it be used for?
No new candidate: this loop characterises one we already had. Zr₆Fe₁₆Si₇C is a dense, stiff, cheap-apart-from-zirconium metal with a weak ferrimagnetic moment. It is not a magnet. Its most plausible use is as a strengthening precipitate in ferritic steel: its cubic cell is four iron cells on a side (to 0.1 % in our DFT; 1 to 2 % if the measured size of the unfilled phase is the guide), and with carbon in it the phase is 18 meV/atom more stable than the known G phase plus zirconium carbide.
Everything below except the stability, the moment and the phonons is new this loop. The elastic numbers come from a machine-learned model and are estimates; the density of states is DFT. Nobody has made this compound, and none of the uses has been tested.
The cubic cell, then its energy against the four phases it would otherwise break into:
Seven-second video of the 120-atom cubic cell of Zr₆Fe₁₆Si₇C (carbon-filled G phase, Fm-3m) beside its PBE formation energy against Zr₆Fe₁₆Si₇ + ZrC + Zr₂Fe₂Si₂C + Fe₃Si (18 meV/atom below). Energies from ABACUS, DFT-relaxed.
One cell throughout: the PBE-relaxed 30-atom primitive cell (ABACUS, TZDP, k-spacing 0.2 Å⁻¹). The cubic cell has 120 atoms and a = 11.459 Å.
Property | Value | How |
|---|---|---|
Stability (meV/atom below the hull) | PBE, relaxed, against Zr₆Fe₁₆Si₇ + ZrC + Zr₂Fe₂Si₂C + Fe₃Si (loop 5) | |
Phonons, lowest mode (THz) |
From the two moduli, and so no better than they are: Young's modulus about 194 GPa, Poisson ratio 0.33, bulk-to-shear ratio 2.6 (the ductile side of the usual 1.75 line), and a Vickers hardness near 5 GPa from Chen's formula.
The carbon is the single atom at the centre of an octahedron of six Zr atoms, 2.31 Å from each:
The magnetism is ferrimagnetic and weak. The sixteen Fe atoms sit on two sites of eight: one carries +1.36 μB per atom, the other −0.28 μB, and Zr and Si are polarised slightly against the first. That is why 54 at.% iron gives only 0.21 T.
It is a metal. The density of states at the Fermi level is 25 states/eV per formula unit, 86 % of it on Fe (almost all Fe d), with 7 in the majority spin and 18 in the minority. The majority spin sits in a shallow dip at the Fermi level and the minority on a shoulder; carbon contributes under 1 %. The same run gives 6.9 μB per formula unit, the value from the relaxation in the larger basis. As a bare band value that is an electronic specific-heat coefficient of about 59 mJ/(mol K²) per formula unit, 2.0 per mole of atoms.
Total density of states, both spins, zero at the Fermi level:
Density of states per spin, resolved by element and orbital
The simulated pattern is here. Its strongest line, (440) at d = 2.026 Å, falls on the (110) line of bcc iron at 2.027 Å, so in a steel the phase would have to be identified from the (511)/(333) line at 40.9° or by electron diffraction.
These are conjectures from the numbers above and from what the structure family is known for. None is measured.
A strengthening precipitate in ferritic steels. G-phase silicides (Ti₆Ni₁₆Si₇, Mn₆Ni₁₆Si₇) form as nanometre precipitates in ferrite and are now used on purpose to strengthen ferritic stainless and maraging steels (review, Yang et al. 2023). Our cell edge is 11.459 Å against 4 × 2.8665 = 11.466 Å for iron at its measured lattice constant, a mismatch of 0.06 % (about 1 % if iron's PBE lattice constant is used instead), so a coherent cube-on-cube precipitate is geometrically easy. The new part is the carbon: in an alloy that holds Zr, Si and C, PBE puts the filled phase 18 meV/atom below the empty G phase plus ZrC. If that holds, carbon would go into the precipitate, not into separate coarse ZrC.
A nickel-free G phase. The G phases used in steels are nickel silicides. This one is built from Fe, Si, Zr and a little carbon. The unfilled Zr₆Fe₁₆Si₇ is a known ternary phase of the Fe–Si–Zr system (τ₅, a = 11.53–11.68 Å, Du et al. 2009), which was studied for silicon steels (Wang et al. 2013).
An interstitial host. The parent structure type (Th₆Mn₂₃) takes up hydrogen: Y₆Fe₂₃ and Ho₆Fe₂₃ form hydrides up to H₁₂ per formula unit (review table
Not made. "Stable" means below every competitor we computed, in PBE at zero temperature. The margin against the phases a real melt would form first (ZrC is very stable and forms early) is a statement about equilibrium, not about kinetics.
The elastic constants are ML estimates with no calibration on a G phase. A DFT elastic tensor would replace them.
The lattice match uses a PBE lattice constant for the precipitate. The measured a of the unfilled phase is 0.6–1.9 % larger than ours, so the real mismatch to iron is probably 0.6–1.9 %, not 0.06 %.
No finite-temperature free energy, no solubility of carbon between the filled and empty phase, no Fe/Si site disorder.
Novelty: the composition is absent from Materials Project, Alexandria, GNoME and COD, and two literature searches found no carbon-filled Zr G phase. License: ours (Materials Project parent, CC-BY-4.0).
Earlier: loop 5
ML potential, no imaginary modes; Debye temperature 484 K (loop 5) |
Formation energy (eV/atom) | −0.563 | PBE, from bcc Fe, hcp Zr, diamond Si and diamond C |
Density (g/cm³) | 7.28 | from the relaxed cell |
Saturation polarisation (T) | PBE; 6.9 μB per formula unit, 23.5 A m²/kg |
Density of states at the Fermi level (states/eV per formula unit) | PBE, DZP basis, k-spacing 0.3 Å⁻¹ (0.2 for the DOS mesh) |
Bulk modulus (GPa) | ALIGNN (ML estimate) |
Shear modulus (GPa) | ALIGNN (ML estimate) |
Curie temperature (K) | ML estimate that assumes a ferromagnet; treat as "below room temperature" |
Raw-material cost ($/kg) | element prices; 97 % of it is zirconium |
Strongest X-ray lines, Cu Kα (°2θ) | simulated; cubic (440), (511)/(333), (600)/(442) |
New carbon-filled G phase, 30-atom primitive cell relaxed in PBE (ABACUS, TZDP, k-spacing 0.2, forces < 0.04 eV/Å). 18.2 meV/atom below Zr₆Fe₁₆Si₇ + ZrC + Zr₂Fe₂Si₂C + Fe₃Si; weakly magnetic; no imaginary phonon modes. Relaxed from the Orb geometry in this file's family. Sun-tetragonal campaign, loop 5.
Not a magnet, hard or soft. 0.21 T and an ordering temperature that the one estimate we have puts below room temperature. In an electrical steel it would be a nearly non-magnetic inclusion.
Properties of the carbon-filled G phase (density of states, moduli, cost, diffraction) and a conjecture: a nickel-free strengthening precipitate for ferritic steel. No new candidate.