Dear LAMMPS users,
My system consists of clustered silver particles. I would like to change the cluster charge by modifying the charge of the consisting atoms. However, since the number of atom within each cluster is not constant I used dynamic group to classified each cluster. By using dynamic group, I cannot use fix adapt or fix adapt/fep command to modify the charge of the cluster. Is there anyway to modify atom properties with dynamic group? or is it possible to use chunk-ID to do this?
I used the following lines in my input script:
compute cluster 1 cluster/atom 2.9
compute cc2 1 chunk/atom c_cluster compress yes
compute size 1 property/chunk cc2 count
variable agg1 atom “c_cc2==1”
variable agg2 atom “c_cc2==2”
group c1 dynamic 1 var agg1 every 2000
group c2 dynamic 1 var agg2 every 2000
variable lambda equal 0.2
fix afep c1 adapt 0 atom charge v_lambda
fix afep1 c2 adapt 0 atom charge v_lambda
Thank you.
i don’t follow what the model here is that you are trying to implement.
what pair style are you using?
how can this work for clusters? if you charge all the constituent particles of the cluster, it should fall apart due to the repulsion of equal charges. is this what you want to achieve?
are you sure your system can be treated by a classical model with “effective” charges at all? i would expect that you’d need to treat the electrons separate from the atom cores. but then again, this is pure speculation, i have not done a simulation of such a system myself, so i have no authority (just suspicions)
axel.
Dear Axel,
Thank you for your reply and comments.
Sorry if my description was not clear enough. Here are some more detail explanations.
The simulation model consists of charged Ag ions which interacting with EAM potential, based on my preliminary test and also from the previous thread in the mailing list, EAM potential will always consider neutral charge, and thus the clusters can be formed.
It is also true that this kind of system cannot be modeled purely with classical model charge, so EAM potential here is combined with charge equilibration (QEq) which has been used in (dx.doi.org/10.1007/s00894-014-2111-1) and (dx.doi.org/10.1021/nl502503t).
However, apart from Ag-Ag interaction, the interaction between Ag and the solvent (ethylene glycol) in this case is significantly affected by the charge of Ag atoms. Hence, my idea is to reduce the cluster charge based on free energy calculation on each cluster. It is why if it is possible, I want to modify the charge of the cluster on the fly.
Thank you.
Dear Axel,
Thank you for your reply and comments.
Sorry if my description was not clear enough. Here are some more detail
explanations.
The simulation model consists of charged Ag ions which interacting with
EAM potential, based on my preliminary test and also from the previous
thread in the mailing list, EAM potential will always consider neutral
charge, and thus the clusters can be formed.
It is also true that this kind of system cannot be modeled purely with
classical model charge, so EAM potential here is combined with charge
equilibration (QEq) which has been used in (dx.doi.org/10.1007/s00894-
014-2111-1) and (dx.doi.org/10.1021/nl502503t).
However, apart from Ag-Ag interaction, the interaction between Ag and the
solvent (ethylene glycol) in this case is significantly affected by the
charge of Ag atoms. Hence, my idea is to reduce the cluster charge based on
free energy calculation on each cluster. It is why if it is possible, I
want to modify the charge of the cluster on the fly.
i don't understand this either. if you are running charge equilibration,
you cannot also modify the charges with fix adapt at the same time. ...and
if you want polarization of the clusters by the solvent, it would be
handled by the charge equilibration as well.
axel.
Dear Axel,
Thank you once again for your insightful comments.
If you don’t mind, I would like to add further explanation.
Yes it is true that charge distribution should be handled by charge equilibration, however since the simulation system started with all positively charged Ag ions, I intend to apply sort of “reduction process” during MD run which followed by charge equilibration.
Previous study, which I mentioned earlier was done using GCMC and MD. Thus, the reduction occurs on every growth step of GCMC, and MD is just used to relax the cluster in the presence of solvent.
Therefore, in order to achieve similar reduction process, I would like to reduce the cluster charge based on free energy calculation on every certain time steps using fix adapt.
Hope this clarify your concern. I understand that this approach may not be without flaws, since I’m still in the process of learning. Please feel free to give your opinions and suggestions.
Thank you for your kind attention.
Best Regards,
Kristanto
Dear Axel,
Thank you once again for your insightful comments.
If you don't mind, I would like to add further explanation.
Yes it is true that charge distribution should be handled by charge
equilibration, however since the simulation system started with all
positively charged Ag ions, I intend to apply sort of "reduction process"
during MD run which followed by charge equilibration.
Previous study, which I mentioned earlier was done using GCMC and MD.
Thus, the reduction occurs on every growth step of GCMC, and MD is just
used to relax the cluster in the presence of solvent.
Therefore, in order to achieve similar reduction process, I would like to
reduce the cluster charge based on free energy calculation on every certain
time steps using fix adapt.
Hope this clarify your concern. I understand that this approach may not be
without flaws, since I'm still in the process of learning. Please feel free
to give your opinions and suggestions.
now here is another thing that i don't understand: how can you try
something as elaborate and complex as this without some form of validation?
why not start with a single cluster where you don't have to worry about
variable cluster sizes?
axel.
Dear Axel,
Thank you again for your suggestion. I guess I did overlook the validation part just because I thought it is similar with the references, I just realized it actually differs significantly in the approach.
I will start with the single cluster as per your suggestion to test my procedure whether it behaves as I expected or not.
Many thanks for your attention.
Best Regards,
Kristanto