meam/c and tersoff

Dear experts:

I want to make a simulation for SiO2 and Si and Cu.
but the paper says that the potential should be set for tersoff and meam/c and morse.
I try to set them, but it makes a mistake.

Si 1
O 2
Cu 3

Si-O,Si-Si,O-O >>>tersoff
Si-Cu,Cu-Cu >>>meam/c
Cu-O >>>morse

pair_style hybrid tersoff morse 4.5 meam/c

pair_coeff * * tersoff SiO.tersoff Si O NULL
pair_coeff 2 3 morse 1.174711 2.058100 1.885238 4.5
pair_coeff * * meam/c library.meam AlS SiS MgS CuS FeS AlSiMgCuFe.meam SiS NULL CuS

I have a question for that:
I set the meam/c for si and cu,it is that of true?
I want the Si is tersoff with O,and Si is meam/c with Cu,and Cu is meam/c with Cu.
But I think if i use this meam/c,it will be Si-Si ,Si-Cu,Cu-Cu all be meam/c
It is have a problem for this set?

Dear experts:

I want to make a simulation for SiO2 and Si and Cu.
but the paper says that the potential should be set for tersoff and meam/c and morse.
I try to set them, but it makes a mistake.

Si 1
O 2
Cu 3

this is not correct. please keep in mind, that you are using a classical model and not quantum chemistry, thus the Si in Si is (chemically) different from the Si in SiO2 (different oxidation state) and thus has to be assigned a different atom type.

Si-O,Si-Si,O-O >>>tersoff
Si-Cu,Cu-Cu >>>meam/c
Cu-O >>>morse

this also cannot work. when using manybody potentials like tersoff or meam, you have to choose interactions inclusively by atom type. those are not pairwise additive.

let’s call the Si in SiO2 Si1 and the other Si atoms Si2. then you can do:

Si1 and O → tersoff
Si2 and Cu → meam/c
Si1-Si2, Si1-Cu, O-Si2, O-Cu → morse

you will still have a systematic error at the interface, because 1) the SiO2 tersoff potential seems to be parameterized to only work well for bulk systems and 2) the SiO2 atoms next to the Si and Cu atoms will not contribute to the embedding energy and their polarity is also not considered.

in short, unless you are ok with that (not insignificant) level of errors or find some consistent parameterization that can handle the whole system with a single pair style, you may be better off considering using the semi-empirical or full quantum method for this kind of system.

axel.

Dear expers:

oh ,I see.Maybe I should find a potential file (reaxff) with Cu Si O .
Maybe this problem can do it.

Thanks a lot for help !!
.

Axel Kohlmeyer <akohlmey@…24…> 於 2020年7月22日 週三 下午9:15寫道: