Tersoff Ge-O

Dear lammps users,

I am trying to reproduce the oxidation of a Ge surface adding oxygen atoms bonded to the surface. I manually add them at a reasonable distance and angle from the surface (little less than 2 A and 65° from the normal to the surface). Then i mixed the tersoff potential of Ge and O and created the potential. However i see that when i run the simulation the oxygen atoms get away very fast from the surface. I put in allied an image of the initial configuration and what happens when termalization begins.

Here are the my GeO tersoff file and the part of my input file related to the potentials:

INPUT

3D germanium block simulation

boundary p p p
units metal
#farci attenzione per tutti i parametri, anche di tempo

atom_style full
bond_style harmonic
angle_style harmonic

pair_style hybrid/overlay lj/cut/coul/long 10.0 10.0 lj/cut/soft 1 0.5 10.0 tersoff

kspace_style pppm 1.0e-4
#newton on

read_data 100boxossidatobondsnowater.txt

group germanium type 1 2
group geeOx type 1 2 3
group GeO type 2
group Oox type 3
group water type 4 5

pair_coeff * * tersoff GeO.tersoff Ge Ge O NULL NULL

pair_coeff 1 1 lj/cut/coul/long 0.0 0.0 #Ge-Ge
pair_coeff 1 2 lj/cut/coul/long 0.0 0.0 #Ge-GeO
pair_coeff 1 3 lj/cut/coul/long 0.0 0.0 #Ge-Oox
pair_coeff 1 4 lj/cut/coul/long 0.0 0.0 #Ge-H

pair_coeff 2 2 lj/cut/coul/long 0.0 0.0 #GeO-GeO
pair_coeff 2 3 lj/cut/coul/long 0.0 0.0 #GeO-Oox
pair_coeff 2 4 lj/cut/coul/long 0.0 0.0 #GeO-H

pair_coeff 3 4 lj/cut/coul/long 0.0 0.0 #Oox-H
pair_coeff 4 4 lj/cut/coul/long 0.0 0.0 #H-H
pair_coeff 4 5 lj/cut/coul/long 0.0 0.0 #H-O

pair_coeff 1 5 lj/cut/soft 0.0074437 3.723 1.0 #Ge-O
pair_coeff 2 5 lj/cut/soft 0.0074437 3.723 1.0 #GeO-O
#divido le interazioni tra O e O in coulomb e lj in modo da fare fep
pair_coeff 3 5 lj/cut/soft 0.006739886 3.1655 1.0 #Oox-O
#pair_coeff 3 5 coul/long
pair_coeff 5 5 lj/cut/coul/long 0.006739886 3.1655 #O-O

#energy minimization

dump TRAJ all custom 500 dump_ossidatonowater.lammpstrj id mol type element x y z ix iy iz
dump_modify TRAJ element Ge Ge O H O

thermo_style custom step temp &
ke vol lx ly lz press epair evdwl emol etotal
thermo 5000

minimize 1.0e-6 1.0e-6 1000 10000
reset_timestep 0

variable dt equal 0.001
variable TK equal 298.15
#variable PBAR equal 1.0

neighbor 2. bin #2 angstroms oltre cutoff
neigh_modify every 2 delay 0

timestep ${dt} #0.001

velocity geeOx create {TK} 12857 #dist gaussian rot yes velocity water create {TK} 12857 #dist gaussian rot yes

fix SHAKE water shake 0.0001 20 0 b 1 a 1 #con 1000 printa ogni 1000

#fix LANG geeOx langevin {TK} {TK} 0.025 264537
#fix fixgermanium geeOx nve
fix TEMPgeOx geeOx nvt temp {TK} {TK} 0.1

fix TEMPwater water nvt temp {TK} {TK} 0.1

thermo_style custom step time temp &
pe press epair evdwl etotal
thermo 5000

run 1000000

TERSOFF FILE

format of a single entry (one or more lines):

element 1, element 2, element 3,

m, gamma, lambda3, c, d, costheta0,

n, beta, lambda2, B, R, D, lambda1, A

#The n, beta, lambda2, B, lambda1, and A parameters are only used for two-body interactions. The m, gamma, lambda3, c, d, and costheta0 #parameters are only used for three-body interactions. The R and D parameters are used for both two-body and three-body interactions.
#The non-annotated parameters are unitless. The value of m must be 3 or 1

#element 1 (the center atom in a 3-body interaction)
#element 2 (the atom bonded to the center atom)
#element 3 (the atom influencing the 1-2 bond in a bond-order sense)

Ge Ge Ge 3.0 1.0 0.0 106430.0 15.652 -0.43884
0.75627 0.00000090166 1.7047 419.23 2.95 0.15 2.4451 1769.0

O O O 3.0 1.0 0.0 64692.1 4.11127 -.845922
0.0000 0.00000 0.0000 0.000000 1.85 0.15 0.0000 0.0000

Ge O Ge 3.0 1.0 0.0 106430.0 15.652 -0.43884
0.75627 0.00000090166 2.03081 285.763 2.95 0.15 3.3081 1824.89

Ge Ge O 3.0 1.0 0.0 106430.0 15.652 -0.43884
0.75627 0.00000090166 2.03081 285.763 2.336 0.15414 3.3081 1824.89

Ge O O 3.0 1.0 0.0 106430.0 15.652 -0.43884
0.00000 0.00000 0.0000 0.000000 2.336 0.15414 0.0000 0.0000

O Ge Ge 3.0 1.0 0.0 64692.1 4.11127 -.845922
0.00000 0.00000 0.0000 0.000000 2.336 0.15414 0.0000 0.0000

O O Ge 3.0 1.0 0.0 64692.1 4.11127 -.845922
0.00000 0.00000 0.0000 0.000000 2.336 0.15414 0.0000 0.0000

O Ge O 3.0 1.0 0.0 64692.1 4.11127 -.845922
0.00000 0.00000 0.0000 0.000000 1.85 0.15 0.0000 0.0000

I activate only the Ge-Ge-Ge, Ge-O-Ge and Ge-Ge-O interactions since I am putting oxygen on the surface only so if can’t be the center of the 3 body interaction (all O- x -x interactions null) and no bonds between two oxygen is possible (Ge-O-O).

Am i mistaken in this use of the GeO.tersoff?
Thank you for your advices

Kind regards,
Lorenzo

after few steps.PNG

initial.PNG

Greetings,
A few reasons might cause that; even assuming the potential and your model setup is correct a few oxygen atoms don’t reproduce a typical gas pressure that might be needed to see what you want. There’s also timescales associated with the process so those might be longer than you expect; should probably look into it deeply in the literature.

In reality i have been unclear in my previous explaination. I am studying the wetting of the Ge surface, so i try to reproduce an experimental % of Oxygen atoms on the surface and the slab is also in contact with water. I found a lot about amorphous GeO2, interface Ge/GeO2 and so on, too bad almost nothing on how to reproduce the external surface configuration. I will keep digging, so far my question is more “technical”, is my setup coherent with what I am trying? (even if in case it doesn’t reproduce exactly the real surface)

I think it’s not a good idea to modify the tersoff potential file to exclude something. The tersoff potential should not care where the O atoms are. The data format in the tersoff potential file is also not that simple.

Bruce

Lorenzo Chiavarini <lorenzo.chiavarini95@…92…> 于 2019年2月13日周三 18:44写道:

I followed the example of the original SiCGe.tersoff in lammps and another example of SiO.tersoff. In both only the possible interactions were activated,
for example C-Si(D)-Si(D) is activated, while C-Si(D)-C no.
I don’t get what you mean when you say “the tersoff potential should not care where O atoms are”. I want to use tersoff to model the interaction between Ge and O, so the position matters a lot i think. That’s way i also deactivate the O-O tersoff interactions, since i don’t want oxygen who get close to act like if they were bonded.
I also don’t get what you mean with “the tersoff potential file” is not that simple. One you have the parameters for the pure species, you can mix then to get the combinations, in addition you need Xij to modulate the entity of the interaction or am I wrong?

The later Tersoff potential for Carbon, Silicon, and Germanium

J. Tersoff, PRB, 39, 5566 (1989) + errata (PRB 41, 3248)

The Si and C parameters are very close to those in SiC.tersoff

C C C 3.0 1.0 0.0 3.8049e4 4.3484 -0.57058 0.72751 1.5724e-7 2.2119 346.74 1.95 0.15 3.4879 1393.6
Si(D) Si(D) Si(D) 3.0 1.0 0.0 1.0039e5 16.217 -0.59825 0.78734 1.1000e-6 1.7322 471.18 2.85 0.15 2.4799 1830.8
Ge Ge Ge 3.0 1.0 0.0 1.0643e5 15.652 -0.43884 0.75627 9.0166e-7 1.7047 419.23 2.95 0.15 2.4451 1769.0

C Si(D) Si(D) 3.0 1.0 0.0 3.8049e4 4.3484 -0.57058 0.72751 1.5724e-7 1.97205 395.1451 2.3573 0.1527 2.9839 1597.3111
C Si(D) C 3.0 1.0 0.0 3.8049e4 4.3484 -0.57058 0.72751 0.0 0.0 0.0 1.95 0.15 0.0 0.0
C C Si(D) 3.0 1.0 0.0 3.8049e4 4.3484 -0.57058 0.72751 0.0 0.0 0.0 2.3573 0.1527 0.0 0.0

Si(D) C C 3.0 1.0 0.0 1.0039e5 16.217 -0.59825 0.78734 1.1000e-6 1.97205 395.1451 2.3573 0.1527 2.9839 1597.3111
Si(D) Si(D) C 3.0 1.0 0.0 1.0039e5 16.217 -0.59825 0.78734 0.0 0.0 0.0 2.3573 0.1527 0.0 0.0
Si(D) C Si(D) 3.0 1.0 0.0 1.0039e5 16.217 -0.59825 0.78734 0.0 0.0 0.0 2.85 0.15 0.0 0.0

Si(D) Ge Ge 3.0 1.0 0.0 1.0039e5 16.217 -0.59825 0.78734 1.1000e-6 1.71845 444.7177 2.8996 0.1500 2.4625 1799.6347
Si(D) Si(D) Ge 3.0 1.0 0.0 1.0039e5 16.217 -0.59825 0.78734 0.0 0.0 0.0 2.8996 0.1500 0.0 0.0
Si(D) Ge Si(D) 3.0 1.0 0.0 1.0039e5 16.217 -0.59825 0.78734 0.0 0.0 0.0 2.85 0.15 0.0 0.0

Ge Si(D) Si(D) 3.0 1.0 0.0 1.0643e5 15.652 -0.43884 0.75627 9.0166e-7 1.71845 444.7177 2.8996 0.1500 2.4625 1799.6347
Ge Si(D) Ge 3.0 1.0 0.0 1.0643e5 15.652 -0.43884 0.75627 0.0 0.0 0.0 2.95 0.15 0.0 0.0
Ge Ge Si(D) 3.0 1.0 0.0 1.0643e5 15.652 -0.43884 0.75627 0.0 0.0 0.0 2.8996 0.1500 0.0 0.0

Kind regards,
Lorenzo

by saying that the tersoff potential file is not that simple, I mean

the C Si C entry contains some parameters that have not been used, which can be set to 0. This might let you think that you can set some parameters in the C Si Si entry to 0 to only turn off the the Si-Si interactions. This is unfortunately not the case, because this will turn off the C-Si interactions instead.

that is, in the CSiGe.tersoff file, nothing is turned off.

by saying that the tersoff potential does not care something, I mean it is designed to be applicable to any atom configuration (if you trust the potential; othetwise it should not be trusted as well when something is turned off by hand).

Bruce

Lorenzo Chiavarini <lorenzo.chiavarini95@…24…> 于 2019年2月13日周三 20:25写道:

In a certain way I get what you mean but not completely. If i set the C Si Si parameters for 2 bonds interactions to 0 I am of course turning off C Si interactions and not Si Si, that’s why i said that to turn off O-O bond interactions I put to zero the parameters of the entries O-O-X not the ones of X-O-X or X-X-O.

In the CSiGe.tersoff some interactions are indeed set to zero:

C Si(D) Si(D) 3.0 1.0 0.0 3.8049e4 4.3484 -0.57058 0.72751 1.5724e-7 1.97205 395.1451 2.3573 0.1527 2.9839 1597.3111
C Si(D) C 3.0 1.0 0.0 3.8049e4 4.3484 -0.57058 0.72751 0.0 0.0 0.0 1.95 0.15 0.0 0.0

The tersoff potential assumes that if an atom is in the cutoff and the entry is not null it is bonded, so if for some reason the atom is not bonded in that way you have to tell him I suppose. It is not a reactive force field which adjust the bond configuration.
This as far I understood.

Kind regards,
Lorenzo

ok, let me state two more points.

1.in your first email, you have turned off both O-O and Ge-O (due to some zeros in the Ge-O-O entry) interactions.
2. tetsoff IS reactive. REBO is just a more complicated tersoff. reaxff is just even more complicated. if REBO and reaxff are reactive, so does tersoff.

Zheong

Lorenzo Chiavarini <lorenzo.chiavarini95@…24…> 于 2019年2月13日周三 21:55写道:

Let me elaborate a bit more.

In Tersoff the “A B C” notation denotes that atom A sits in the middle, atom B is first neighbor of A making A-B bond the center bond, and atom C is another first neighbor of A affecting the center A-B bond. If you want some visualization, it is C—A---B.

Coming back to SiC potential, “C Si C” can be safely set to zero because in a SiC condensed phase (which Tersoff SiC potential is designed for) you wouldn’t find any C—C---Si bond because C wouldn’t see another C as its first neighbor. This is not the same for “C Si Si” because there are tons of Si—C---Si bonds/angles in a SiC phase.

Tersoff is not that simple as you can’t just take a set of O parameters, combine it with Ge parameter, and call it a Ge-O potential – not even with some adjustments to X_ij. Careful parameterizations are needed to fit/optimize Ge-O cross interactions as well as the angular terms.

It is also a very bad idea to use pair hybrid or hybrid/overlay to combine a bond order potential with a pair-wise force field. It is best you fit the Ge-O-H parameters carefully considering the key reactions between Ge, GeO2, and H2O. If you can’t do that, find a suitable potential and set of parameters in the literature. Perhaps it is already described by ReaxFF.

the reason for the permitted zeros in some entries of the tersoff potential file is not due to the absence of some bonds, but due to the fact that there is some redundancy (for two-body parametets such as A, B, lambda) when using M^3 entries for a system with M atom types.

Bruce

Ray Shan <rshan@…1795…> 于 2019年2月13日周三 22:10写道:

I took the info from a paper which used the GeO tersoff to describe the interactions of GeO2 oxide and fitted with dft and experimental results to get the entity of Xij. Regarding hybdrid pair style I am not mixing tersoff with pair wise interactions, I use tersoff to describe the interaction n the Ge crystal and i wanted to use it only fro some surface oxygen atoms which I manually added. The pairwise interactions lj and so on are between the surface and free water. In fact they are all set to zero for the combinations of 1,2,3 (Ge atoms and superficial oxygen).
I found also studies on Ge/GeO2 interactions parametrization with tersoff (less accurate) and reaxff, but no study the exact thing I need.
I thought of using the reax for this but I need to evaluate the adhesion between the germanium and the water to characterize some properties and that would not be possible if I use the reax.
I want also to apologize if I seemed rough, I am new of the field and I am following mostly the directives of the person who is following me on my thesis.

Kind regards,
Lorenzo

a simple suggestion:

if you think the O atoms are covalently bonded to Ge crystal, use the GeO tersoff without modification.

if you think they are weakly interacted, use LJ or similar between them and only use tersoff for Ge crystal.

Bruce

Lorenzo Chiavarini <lorenzo.chiavarini95@…24…> 于 2019年2月13日周三 22:25写道:

what i said can explain what you obser red. You observed that the O atoms fly away from the Ge crystal, because there is no Ge-O interactions at all (they are turned off in your tersoff file) in your simulation.

Bruce

Bruce Fan <brucenju@…24…> 于 2019年2月13日周三 22:34写道:

I am using this: https://doi.org/10.1016/j.susc.2012.12.013 , that is why I said starting from Ge and O I mixed.

You are right that I made the mystake of putting some O-GE-X parameters shutting so down Ge-O interactions. I corrected that.

Bruce,

That is not correct. The way Tersoff is implemented in LAMMPS it is considered as a 3-body potential and there are no redundancies in pair parameters. If A and B for a particular I-J-K line are zero then the 3-body has no interactions at all by the definition of Tersoff potential. It doesn’t matter what the angular parameters are.

If you look at the Tersoff source code in LAMMPS, you can see that for 3-body interactions it takes parameters from Itype-Jtype-Ktype – it doesn’t only depend on I and J, but also K. The reason that “C Si C” can have zeros for SiC is that one would never find C-C bonds in a perfect SiC phase. However, if one wants to simulate C atoms on top of a SiC surface (where one can find C-C bonds), one has to use a potential that explicitly defines the “C Si C” line without any zeros.

So for Lorenzo’s application, all 8 lines need to be explicitly defined (no zeros permitted) to have proper O/H2O interacting with GeO2.

Thanks a lot to both of you.

So you would use it like this:

Ge Ge Ge 3.0 1.0 0.0 106430.0 15.652 -0.43884
0.75627 0.00000090166 1.7047 419.23 2.95 0.15 2.4451 1769.0

O O O 3.0 1.0 0.0 64692.1 4.11127 -.845922
1.04968 0.00000011632 2.35692 218.787 1.85 0.15 4.17108 1882.55

Ge O Ge 3.0 1.0 0.0 106430.0 15.652 -0.43884
0.75627 0.00000090166 2.03081 285.763 2.95 0.15 3.3081 1824.89

Ge Ge O 3.0 1.0 0.0 106430.0 15.652 -0.43884
0.75627 0.00000090166 1.7047 419.23 2.336 0.15414 2.4451 1769.0

Ge O O 3.0 1.0 0.0 106430.0 15.652 -0.43884
0.75627 0.00000090166 2.03081 285.763 2.336 0.15414 3.3081 1824.89

O Ge Ge 3.0 1.0 0.0 64692.1 4.11127 -.845922
1.04968 0.00000011632 2.03081 285.763 2.336 0.15414 3.3081 1824.89

O O Ge 3.0 1.0 0.0 64692.1 4.11127 -.845922
1.04968 0.00000011632 2.35692 218.787 2.336 0.15414 4.17108 1882.55

O Ge O 3.0 1.0 0.0 64692.1 4.11127 -.845922
1.04968 0.00000011632 2.03081 285.763 1.85 0.15 3.3081 1824.89

I am thinking if it would be reasonable to use: (all 3 body active, 2 body only O-O to zero, however this is like me forcing the system to something i think to be right but may not be true?)

Ge Ge Ge 3.0 1.0 0.0 106430.0 15.652 -0.43884
0.75627 0.00000090166 1.7047 419.23 2.95 0.15 2.4451 1769.0

O O O 3.0 1.0 0.0 64692.1 4.11127 -.845922
0.0000 0.00000 0.0000 0.000000 1.85 0.15 0.0000 0.0000

Ge O Ge 3.0 1.0 0.0 106430.0 15.652 -0.43884
0.75627 0.00000090166 2.03081 285.763 2.95 0.15 3.3081 1824.89

Ge Ge O 3.0 1.0 0.0 106430.0 15.652 -0.43884
0.75627 0.00000090166 1.7047 419.23 2.336 0.15414 2.4451 1769.0

Ge O O 3.0 1.0 0.0 106430.0 15.652 -0.43884
0.75627 0.00000090166 2.03081 285.763 2.336 0.15414 3.3081 1824.89

O Ge Ge 3.0 1.0 0.0 64692.1 4.11127 -.845922
1.04968 0.00000011632 2.03081 285.763 2.336 0.15414 3.3081 1824.89

O O Ge 3.0 1.0 0.0 64692.1 4.11127 -.845922
0.00000 0.00000 0.0000 0.000000 2.336 0.15414 0.0000 0.0000

O Ge O 3.0 1.0 0.0 64692.1 4.11127 -.845922
1.04968 0.00000011632 2.03081 285.763 1.85 0.15 3.3081 1824.89

Kind regards,
Lorenzo

Ray,

thanks for your comments. do you agree with the following text from the manual?

The parameters used for a particular three-body interaction come from the entry with the corresponding three elements. The parameters used only for two-body interactions (n, beta, lambda2, B, lambda1, and A) in entries whose 2nd and 3rd element are different (e.g. SiCSi) are not used for anything and can be set to 0.0 if desired.

Bruce

Ray Shan <rshan@…1795…> 于 2019年2月14日周四 00:28写道: