My Community

Hi,

I am trying to study the evolution of magnetization during the chemisorption of an hydrogen atom on a surface. In a calculation with H far from the surface we have only one spin unpaired and then during the adsorption this value may change. I also need the DOS projected on each atom.

Using LORBIT =5 and setting RWIGS to the covalent radius of the species I obtain when H is far from surface a magnetization of only 0.15 as shown in the last OUTCAR section: a panel deidcated to charge and magnetization on each atom. To obtain the right magnetization I should use a WS radius much higher for H than for the atoms of the surface! Of course I cannot use the
"volume per type =" as a reference beacuse of I have a huge vacuum in the supercell.

Is this due to the problem in counting the electrons that sometimes happens using PAW?

As the total magnetization of my system have I to take into account the last panel:

magnetization (s):
# of ion s p d tot
------------------------------

...or the value given in each step?

number of electron 123.456678 magnetization 1.00000

In my case these two values are respectively 0.15 and 1.00009
and 0.45, 0.9999999 when H adsorbs

Thanks in advance for your suggestions.
Regards

local magnetic moments and charges indeed cannot be regarded as absolute numbers in all PW and mixed basis set calculations, because the numbers depend on the choice of RWIGS (and therefore the volume which is taken into account for the integration).
Hence these numbers should be used for taking differences only (with RWIGS fixed of course). Nevertheless they give a good qualitative estimate.

The magnetization written after each ionic step in OSZICAR (mag=) and in OUTCAR (the line you show) are on the other hand the magnetic moments for the whole unit cell, (obtained by taking the difference between the total spin up and down charges), thus they sum up all local and interstitial contributions.