Slater-Pauling or volcano plot for 2D magnets. (a) High-throughput
screening undertook over several crystal structures and elements of
the periodic table including formulas MX′2, MX,
MX′3, MPX3, MX2, and CrFTe3 with M = Sc–Zn, La, Y; X′=Cl, Br, I;
X=O, S, Se, Te; F=Si, Ge. The simulations included mainly
transition metals with 3d electrons, but some with 4d and 5d were
included for comparison. (b) Variation of the local magnetic moment M(μB) at the metal atom as function of
its valence Z(e–). Bader charge
analysis was used to extract Z for each metal atom
at the compound. The solid lines show a fit to the data set on two
different regimes according to the filling of the valence. The positive
slope (weak magnets) can be fairly well fitted using M+ = 0.84Z – 1.15 (with a linear
regression coefficient R2 = 0.96) and
the negative (strong magnets) with M– = −0.87Z + 9.27 (R2 = 0.90). An electron counting argument can be used to explain
both regimes as discussed in the text. (c) Spin resolved density of
states (DOS) for monolayer MPTe3 (M = V, Cr, Mn, Fe, Co,
Ni) as function of the energy ε displaying the spin up density nup (faint gray) and spin down ndown (faint brown) at opposite
sides. The energy is shifted to the Fermi energy εF at zero. (d) Variation of the model predicted magnetization
versus DFT + U calculated magnetization for the compounds
showed in (a). Calculations were performed using the VASP code884 using a 21 × 21 × 1 k-sampling grid, the Dudarev (GGA+U) scheme885 with Hubbard U values following
those in ref (341).
The energy cutoff is set to 600 eV, the convergence criteria for energy
to 10–7 eV and for the forces to 0.01 eV/Å.
In order to avoid interactions between the layers, we applied periodic
boundary conditions with a vacuum space of 25 Å. We used the
projector augmented wave (PAW)886 methods
with a plane wave basis. The Vosko–Wilk–Nusair modification
scheme887 is applied for the spin-polarized
calculations. All images in this figure are original, and no permissions
are required.