When you use UMP2, you should use "density" (equivalent to density=current). Do not use "density=MP2", in this case the density you obtained is not the MP2 density in usual sense.

The first option, namely using built-in density, is very safe and reasonable for any practical study. Only in theoretical investigation of wavefunction analysis methodology, using atomic UMP2 density may be indeed needed.

Thank you for your quick reply. If I understand it correctly, then I have three options:
- using the default, built in atomic densities
- using UMP2 with density=MP2
- getting the atomic wfn files from DFT, e.g. B3LYP.
Of course, the first option is the most convenient one. However, the charges are quite crucial for what I do. What would be 'special reasons' in that case?

Without special reasons, please simply use built-in atomic densities, in this case calculation of atomic .wfn file could be avoided.

Currently, Multiwfn is unable to automatically generate atomic .wfn file by invoking Gaussian. Because when Multiwfn asks you to input the calculation level, if you input "MP2/cc-pVTZ", the actual keywords in the generated .gjf file (which can be find in "wfntmp" subfolder in current folder) will be "out=wfn ROMP2/cc-pVTZ"; however, Gaussian doesn't support generating wavefunction at ROMP2 level. Therefore, your only choice is manually generating atomic .wfn files at UMP2 level and provide them to Multiwfn.

BTW: If you are using a Gaussian older than G09 C.01, "density" must be specified if you want to yield post-HF density. Since G09 C.01, when "out=wfn" is found, Gaussian automatically employs "density" keyword and thus in case you do not need to explicitly specify it.

In fact, even if your molecular wavefunction was produced at MP2 level, you can still safely use DFT atomic .wfn files for Hirshfeld/ADCH/CM5/Hirshfeld-I analysis in Multiwfn, the result is completely reasonable.

I am slowly approaching the production phase with what I am intending to do. Following the manual, I wanted to obtain the atomic .wfn files that are required by the ADCH process. My calculations will most likely be at the ROMP2/cc-pVTZ and RMP2/cc-pVTZ level of theory (I am comparing closed-shell molecules with their ionized counterparts). For the atomic wavefunctions, would you recommend ROMP2 or UMP2? And how do I make sure that the MP2 wavefunction is actually printed into the wfn file by gaussian - is Density=current sufficient?

Again, thank you very much. Your help is most appreciated.

ADCH can be normally applied to charged system, and the ADCH charges are independent of choice of origin. However, due to the fact that sum of all ADCH charges is not zero, the dipole moment evaluated by ADCH charges in this case is affected by origin.

I have a follow-up question. The dipole moment of an ion depends on the choice of a coordinate system - are ADCH charges reasonable for such a system? i.e. do they depend on the origin of the coordinate system as well?

Thank you very much for your helpful reply. And, of course, thank you for providing such a powerful software, with a manual that is more comprehensive that most lectures.

The easiest and fastest way of obtaining electric dipole moment in current version of Multiwfn is entering main function 200, and then choose subfunction 2, then dipole moment will be immediately printed on screen.

The electric dipole moment is contributed by both electron charges and nuclear charges, the formula used to evaluate it is clearly given in Section 3.200.2 of Multiwfn manual.

For a system carrying net charge, its electric dipole moment is dependent of the choice of original point. It is unable to say which choice is the best and most meaningful. The original point is commonly set to center of nuclear charges, this is what Gaussian do by default (that is, if you do not add "nosymm" keyword when performing single point calculation by Gaussian, Gaussian will put center of nuclear charges of the system to 0,0,0 position). Multiwfn calculates electric dipole moment based on the inputted wavefunction and standard formula.

I am interested in multipole analyses on ions, where I have a few questions regarding the use of Multiwfn.
The first question I have is how to access the Dipole moment in Multiwfn. Is there a way to just obtain the dipole moment?
Alternatively, which analysis that yields the dipole moment as byproduct is the fastest?
The second question is about how it is calculated. Does it include the nuclei or is it just <psi|r|psi>?
The third and last question is about the ions. Usually, the position of the ion (i.e. the choice of origin) influences the outcome of the multipole analysis. If I shift the molecule by a vector abc, then this vector adds to the dipole moment. In my point of view, it would be most sensible to subtract the center of charge, i.e. choose the latter as the origin. How is this handled in multiwfn? Does it yield sensible dipole moments for ions?

I would appreciate any help with this rather basic issue....