It is really unprobable that you could extract the exact net charge from DFT based calculations, that is mostly because the theory implementation has its theoretical and numerical limitations.

-0.995 in the scope of the DFT can be considered (for interpretation purposes) the same thing as -1.0, It is far more realistic to report a value with some deviation than reporting an exact and perfect value.

Dear Lucas Gian Fachini, thank you very much for your response. your statement is completely reasonable and clear. However, what I am dealing with is a polymeric system that I need to replicate several of one chain in a system. However, with not having the exact value of charges for each chain, the additional charge would add up in my system and ultimately I cannot get a neutral system. That's why I need the exact value to handle the net charge in my simulations. Do you have any suggestions for this issue?

Thank you very much,

Mahsheed

Could you specify the idea of ​​your next step? What are you doing and how (software/theory level)? Because I'm assuming you are using the results of a monomer DFT calculation to conduct calculations on the polymer.

Firstly, Mulliken population analysis will not give you the exact charges of your system and that is not a problem(!), see, this analysis is just an integration scheme over the orbitals constructed for your system at a given theory level and, as an integration scheme, will not give you the exact value.

After finishing a DFT calculation you will have a wavefunction that describes your system, this wavefunction may be used to extract several information about your molecule, for example, applying the Mulliken scheme on this function will give you the Mulliken charges for it, but this does not mean that the wavefunction has this charge as an inherent property, it just means that integrating it using said boundary conditions will give rise to that value.

The important thing is having a good wavefunction, because it is the starting point for other calculations, it is not because the wavefunction for a monomer leads you to a net Mulliken charge of -0.995 (considering a particular integration scheme) that two monomers bonded together are going to give you -1,990. The new wavefunction considering two monomers is going to have its own net Mulliken charge and its own error associated with it.

*sorry for any mistakes, english is not my first language.

Dear Lucas Gian Fachini, many thanks for your thorough answer. Currently, I am trying to build a polymer system, based on the monomer unit that I built with Materials Studio. To get the charges for my system, I am trying to get Mulliken DMol3 charges for one monomer. After, I will use the Build Polymer module to construct the chain based on the assigned charges from Mulliken (i.e. I consider the end effect for charges, too). In the end, I need to use 10 chains of the built polymer to construct my whole structure. This information should be imported to the LAMMPS for further analysis.

However, as you truly understand my intention, I am getting an additional charge (mostly around 0.05 for each chain, and 0.5 for the whole system), which would be difficult to handle. Also, if I use the Mulliken charge to get charges for the whole chain, it would give a variety of partial charges for each atom, which would be so much more complicated to deal with. Not to mention that in the literature, it seems that they all have considered similar charges for all the monomer units in the same chain.

So, I need to figure out how to resolve this issue.

I really appreciate your time and consideration,

Cheers,

Mahsheed

I am not experienced with Material Studio, so I won't be able to help you further, but as it seems that you are taking the Mulliken charges just as a starting point for another calculation, have you thought about rounding the charges to -1.000? Considering that Mulliken charges are not very precise for most systems it wouldn't be a rude approximation. Most people would consider a net Mulliken charge of -0.995 actualy a -1.000 with some deviation due to the integration or the basis set.

Also, there are other types of charge analysis, as Atoms in Molecules (AIM). But I'm afraid any kind of analysis you carry on a DFT based wavefunction won't give you an exact value, as they are usualy integration schemes with finite grids, thus, they are just as precise as the grid employed.

Lucas Gian Fachini, Thank you very much for all the help and quick response.

No, I did not round up the charges, as I thought it might pay a toll on the accuracy of the simulations. However, knowing that this is a common approach, I would take it into consideration to see if it makes a difference.

Have a good one,

Mahsheed