Page 1 of 1

bulk Pd: magnetic?

Posted: Fri Nov 03, 2006 6:26 pm
by hong123ok
Dear all,

When I tried to test bulk properties of Pd, I found a very strange thing. I know Pd is a nonmagnetic material. However, in some cases the VASP calculations show it looks like magnetic. Here is a summary of the calculation results.

+++++++++++++++++++++++++++++++++++++++++++++++++++
1. Using GGA_PBE-PAW: lattice constant= 3.95323978 A

(1) bulk fcc (containing only one atom)
nonspin-polarized: E_tot= -5.178009
spin-polarized: E_tot= -5.178678, magnetization per atom= 0.272
=> E_diff= .000669 eV
(2) bulk 2x2x2 supercell (containing 32 atoms)
nonspin-polarized: E_tot= -165.576764
spin-polarized: E_tot= -166.012387, magnetization per atom= 0.360 (equal for all atoms)
==> E_diff= .435623 eV
+++++++++++++++++++++++++++++++++++++++++++++++++++
Here, Pd looks like magnetic at least for the 2x2x2 unitcell. What do you think? First, I thought that since the GGA calculation gave a larger lattice constant than the experimental one, it might give a nonzero magnetic moment on each atom. So I re-calculated the same thing using LDA.
+++++++++++++++++++++++++++++++++++++++++++++++++++
2. Using LDA-PAW: lattice constant= 3.85226884 A

(1) bulk fcc (containing only one atom)
nonspin-polarized: E_tot= -6.440954
spin-polarized: E_tot= -6.440954, magnetization per atom= 0.007 (Isn't it very small to neglect?)
=> E_diff= .0 eV
(2) bulk 2x2x2 supercell (containing 32 atoms)
nonspin-polarized: E_tot= -205.917187
spin-polarized: E_tot= -206.173514, magnetization per atom= 0.301 (equal for all atoms)
==> E_diff= .256327 eV
+++++++++++++++++++++++++++++++++++++++++++++++++++
Here, since the magnetic moment of 0.007 is quite small (?) to neglect, it looks OK for the bulk fcc structure. However, the bulk 2x2x2 structure has an equal magnetic moment on each atom, which implies that Pd is magnetic even in the LDA calculations.
+++++++++++++++++++++++++++++++++++++++++++++++++++

Please advise me.
Thanks.
hong

bulk Pd: magnetic?

Posted: Fri Nov 03, 2006 7:50 pm
by hong123ok
Hi,

Suddenly I realized I might have used a smaller number of k-points for the metal system. After the LDA calculations of 2x2x2 unitcell using a denser mesh, I found that the magnetic moment on each atom becomes close to zero. So, Pd is nonmagnetic within the LDA.

Sorry if I made you confused,
hong