Ferromagnetic bulk nickel supercell optimization - multiplicity?

[mdsimula...@gmail.com]

Hello,

I am attempting an optimization of bulk nickel in a 5x5x5 super cell.  Nickel is supposed to be ferromagnetic so I'm wondering how I set the appropriate magnetization for each Ni?

Based on literature, the magnetization should be ~0.64. 

Do I need to set some crazy high multiplicity and the magnetization in the kind section?

My initial input is attached.

Thanks,

Frank

[uca...@ucl.ac.uk]

Hi Frank,

I attach files where I have converged a spin moment of 0.72 in 36 SCF steps for your 5x5x5 supercell.

Indeed, Ni should be ferromagnetic and we can confirm this by converging a solution with a multiplicity of 1 and comparing this to the ferromagnetic solution with a multiplicity of 500*2+1=1001. The latter is -84660.541395560532692--84661.817516270704800=1.3 Ha or 35 eV lower in energy.

Hope this helps,  
Chris

[mdsimula...@gmail.com]

Ah ok...so I do need to set the multiplicity to some high number! 

Thanks for the help Chris! 

[GENG YUAN]

Hi Chris,

I am also testing the optimization of bulk Ni using your files, which are very helpful materials for me. I found that the energy converged when I directly used your input file without any changes (a 5x5x5 supercell with a multiplicity of 1001), but when I changed the supercell from 5x5x5 to others (e.g. 4x4x4, 256 atoms total) with a modified multiplicity (256*2 +1 = 513), the energy couldn't converge within 500 SCF steps. I am wondering what may cause this issue, and if you could share some thoughts.

The only changes I made are the multiplicity and the multiple_unit_cell settings, I attached my input and output files for your reference. Please let me know if any concerns.

Many thanks in advance,

Sincerely,

Geng

[uca...@ucl.ac.uk]

Hi Geng,

It looks like you have already converged to around EPS_SCF 1E-7 in 32 steps, see line: 

32 Broy./Diag. 0.10E+00   38.7     0.00000010    -43346.6332044618 -4.17E-05

Therefore I would recommend changing EPS_SCF 1E-8 -> 1E-7, and if you must converge to 1E-8 (generally unnecessary) then either use SCF_GUESS RESTART using the wavefunction file from EPS_SCF 1E-7 or try to optimise the BROYDEN_MIXING ALPHA and BETA values. Increasing the ELECTRONIC_TEMPERATURE may also help.

Best wishes,

Chris