Multiwfn forum

Multiwfn official website: http://sobereva.com/multiwfn. Multiwfn forum in Chinese: http://bbs.keinsci.com/wfn. E-mail of admin: sobereva[at]sina.com

You are not logged in.

#1 2017-11-03 19:35:50

dronkit
Member
Registered: 2017-11-03
Posts: 3

Weird AIM charges in complexes with ghost atoms

I have calculated the DFT wavefunction of amine-metal complexes with g09 and outputted wfn files. The Cu and Fe atoms have ECPs.

Multiwfn recognizes the ECPs and handles the whole complex very well and calculates well the atomic charges by integrating electron density over the atomic basin+atomic centers.

Now I want to consider the fragments of the complex to see how the atomic charges change when in the complex. I calculated each fragment separately but keeping the rest of the complex as ghost atoms (basis sets centered on a point with no nucleus or electrons), so, i take it, I can avoid a kind of basis set superposition error, since the whole complex is calculated with all the atoms' basis sets.

When I input the fragments' WFN to Multiwfn, first, I get "unrecognized atom" errors for all the ghost atoms, which I ignore by hitting enter. Then I calculate the basins and get the first problem: some attractors are insignificant (<1E-5) and have three choices.

If I "do nothing", next when trying to integrate with mixed basin+atomic center I get an "unable to assign attractor to atom" error which I have the opion to ignore by hitting enter, but the process halts anyway. If I choose 2 "mark insignificant attractors as unassigned" or 3 "join insignificant basins to nearest significant ones", it correctly recognizes the basins belonging to the real atoms and calculates (basin integration+atomic center) charges. But the normalization factor in those cases is ridiculous (around 0,28) so I get ridiculously big charges.

Am I doing something wrong? Or is it Multiwfn getting confused by the ghost atoms?

How is the normalization factor calculated? Maybe I can calculate it correctly on my own?

Is a fix necessary?

Thanks!

ps: I pasted an output in https://drive.google.com/open?id=0BzbGy … ExVTklDQms

Offline

#2 2017-11-04 16:17:35

sobereva
Tian Lu (Multiwfn developer)
From: Beijing
Registered: 2017-09-11
Posts: 1,611
Website

Re: Weird AIM charges in complexes with ghost atoms

Unfortunately, the google drive is blocked by goverment of my country and thus I am unable to access it...
AIM is highly insensitive to basis set, so I don't think adding ghost atoms is needed to get correct result about change of AIM charges during formation of complex.
Your case is rather special and Multiwfn doesn't take this circumstance into account, it seems that the ghost atoms lead to artificial attractors. Using ghost atoms simply makes the thing much more complicated.

Offline

#3 2017-11-04 17:50:35

dronkit
Member
Registered: 2017-11-03
Posts: 3

Re: Weird AIM charges in complexes with ghost atoms

I'll follow your advice and calculate the fragments without ghost atoms.
I'm still curious about what is the normalization factor. I can't find it in the manual.

Thank you for your quick reply!

Offline

#4 2017-11-04 17:54:56

dronkit
Member
Registered: 2017-11-03
Posts: 3

Re: Weird AIM charges in complexes with ghost atoms

I re-uploaded the output to mega.nz. Hope you can see it there in case you want to take a look.

https://mega.nz/#!DZZRwYLB!-x1_OwgGbLD6 … zMHNzPhD8s

Offline

#5 2017-11-05 16:50:45

sobereva
Tian Lu (Multiwfn developer)
From: Beijing
Registered: 2017-09-11
Posts: 1,611
Website

Re: Weird AIM charges in complexes with ghost atoms

AIM population is calculated via numerical integral, therefore, there must be some numerical errors, and thus the sum of atomic population is not exactly equal to actual number of electrons in the system. To rectify this problem, Multiwfn automatically carry out normalization correction for AIM atomic populations, so that their sum is identical to actual number of electrons.
:-( Almost all netdisks outside China region was blocked by goverment, including megz (AFAIK, only filedropper, Sendit and sendspace are accessible).

Offline

Board footer

Powered by FluxBB