QTAIM atomic basin energy

vinimart · 2019-10-22 20:37:19

Hello.
I'm studying rotational barrier of ethane when it goes from staggered to eclipsed with a combination of techniques in which the QTAIM is one of the main ones. So, one of the main methods to do this type of study is to optimize the staggered conformer, then rotate the C-C bond to eclipsed conformer but do not optimize any other geometric parameter, so you are left with dihedral angle of eclipsed conformer but bond length and 3 atom angles of staggered conformer, something like a "staggered-eclipsed" conformation. Finally you optimize the eclipsed conformer. This leaves you with 3 conformers, the staggered being the lower in energy, the "staggered-eclipsed" being the higher in energy and the eclipsed being the middle term in energy. This is what i get from the single point calculation performed with MP2/ma-def2-QZVPP level of theory. When i use QTAIM to separate these 3 conformers into atomic basins, the summation of the electronic energy of all basins of a given conformer should give me the molecular electronic energy of that conformer just like the single point energy. So i sum up all the basins electronic energies and i have the molecular electronic energy just like the one from the single point except for the "staggered-eclipsed" conformation. This one is lower in energy than the other two conformers even though it should be the higher. Does anyone have any idea why QTAIM reproduces the molecular energy for the staggered and for the eclipsed conformers but not for the "staggered-eclipsed" conformer?

Thank you.
Att. Vinícius Martinelli

sobereva · 2019-10-23 03:47:37

I don't know what is the "basins electronic energies" you exactly referred to. If it denotes integral of energy density G(r) over the AIM basin produced by basin analysis module of Multiwfn, it seems that currently the sum of this value of all basins is not quite close to the MP2 single point, leading to incorrect order of energy. There are several possible reasons:

(1) The quality of grid is too poor. You should select a better grid setting (i.e. smaller grid spacing) prior to generation of basins and retry.
(2) You didn't select mixed type of integration grid for calculating integral of G(r). Note that if evenly spaced grid is used for calculating the integral, the numerical error may be significant.
(3) The wavefunction recorded in your input file is HF wavefunction rather than MP2 wavefunction. Please check beginning of Chapter 4 of Multiwfn manual on how to correctly yield .wfn/wfx or .fch file containing post-HF wavefuntion.
(4) The Virial ratio (V/T) of current system is not quite close to -2. This condition is exactly satisfied when the wavefunction if fully variational (e.g. HF, CASSCF wavefunction) and force of all atoms are vanished. If this condition is evidently violated, one cannot expect that sum of energy of all basins is quite close to single point energy. Commonly the ratio can be found from output file of quantum chemistry code.

vinimart · 2019-10-23 19:16:10

So, if i use wfn files from gamess dat files or molden files from ORCA i should have the MP2 wavefunction instead of the HF wavefunction, right?

sobereva · 2019-10-23 23:40:55

Yes.

After loading the input file into Multiwfn, select option 3 in main function 6, if you find occupation of all orbitals are integer, that means the current wavefunction is HF. If some of values are non-integer, that means the orbitals loaded from the input file are MP2 natural orbitals, and thus the subsequent analyses will be at MP2 level.

vinimart · 2019-10-31 23:35:41

Thank you very much for your support.
Let me make another question.

"Note: Although GTFs with angular moment of g and h are not formally supported by original .wfn format, if g
and h-type GTFs are recorded in following manner, then Multiwfn is able to recognize them: 21~35 in "TYPE
ASSIGNMENT" correspond to YZZZ, XYYY, XXYY, XYZZ, YZZZ, XYYZ, XXXX, XXXY, XZZZ, XXYZ,
XXXZ, XXZZ, YYYY, YYYZ, ZZZZ, respectively. 36~56 correspond to ZZZZZ, YZZZZ, YYZZZ, YYYZZ,
YYYYZ, YYYYY, XZZZZ, XYZZZ, XYYZZ, XYYYZ, XYYYY, XXZZZ, XXYZZ, XXYYZ, XXYYY,
XXXZZ, XXXYZ, XXXYY, XXXXZ, XXXXY, XXXXX, respectively."

Does this mean that i need to change the numbers 21, 22,... for these respective letters YZZZ, XYYY,...? If yes, do i need to take care with the length of the lines? I mean, too long lines are a problem like in a gamess input?

sobereva · 2019-11-01 01:02:08

If you are a GAMESS-US user and you directly use GAMESS-US output file as Multiwfn input file (of course, the suffix should be changed to .gms to make Multiwfn recognize its format), you do not need to concern this problem. If due to special reason you have to use .wfn file produced by GAMESS-US as input file, then:

(a) If the basis set you used doesn't contain g or h angular moment Gaussian functions, you do not need to care about this problem.

(b) If g or h functions indeed present in your wavefunction (for example, cc-pVQZ is employed for organic system), you may need to write a code to reorganize the indices in the "TYPE ASSIGNMENT" field of the.wfn file. If you open the .wfn file by text editor, you can find "TYPE ASSIGNMENT" field, different indices correspond to different types of Gaussian function. If the definition of the index for g or h Gaussian functions is not in line with Multiwfn, the loaded wavefunction will be wrong and the result will be meaningless.

Multiwfn forum

#1 2019-10-22 20:37:19

QTAIM atomic basin energy

#2 2019-10-23 03:47:37

Re: QTAIM atomic basin energy

#3 2019-10-23 19:16:10

Re: QTAIM atomic basin energy

#4 2019-10-23 23:40:55

Re: QTAIM atomic basin energy

#5 2019-10-31 23:35:41

Re: QTAIM atomic basin energy

#6 2019-11-01 01:02:08

Re: QTAIM atomic basin energy

Board footer