ELF basin populations and grid quality, use of solvent

monthem · 2025-03-29 16:04:48

Hello!

I am trying to analyze ELF on a system [M2X6]2- M = Be, X = F, Cl, Br in gas phase and water.

Structures have been optimized in gas phase and water, respectively. WFN file was generated along with the SP calculation of the minima structures (Gaussian16) with route section "M062X/aug-cc-pvdz SCF(VeryTight) Integral(SuperFine) output=wfn" + "SCRF(SMD, solvent=water)" if water in the title name.

Multiwfn 3.8 dev binary build for Linux. The command sequence i am using is: 17 1 9 (options 1-4) 12

Problem 1 (gas phase, no solvent!): For X = Br, low and medium quality grids produce too low total populations, high quality results in hundreds of electrons more than it should. Lunatic quality results in segmentation fault during search for attractors. Note, for X = F, Cl, high and lunatic quality produce very good result.

Problem 2 (solvent present): This i am not sure if it has to do with grid quality or rather artefacts of the SCRF(SMD) generated WFN. Be2F6 in water integrates to 80 electrons.

The wawefunction files are too large to upload to the forum. I am sharing a google drive link with relevant .wfn files, .xyz structures of the minima and .log files of mentioned multiwfn runs (multiwfn title.wfn | tee title.log).

If i may use the chance to ask how is ELF gradient calculated, are the derivatives analytical?

https://drive.google.com/file/d/1PI1xnQ … sp=sharing

Thank you very much for your time!

monthem · 2025-03-30 08:00:08

An update: Problem 1 was solely due to an issue in my laptop configuration. I am using a zsh shell in a Linux based OS and it appears that "ulimit -s unlimited" has to be the last line in the .zshrc file. Otherwise it won't get executed when opening a new shell session. I don't know why in my above reported high quality Be2Br6 log file calculation found 800+ electrons, but that is not an issue anymore, neither is segmentation fault. I have successfully ran some even heavier variations of the compounds im studying overnight and they all terminate gracefully even with the lunatic grid. Although, precision in basin populations drop as the size of the system increases, regardless of the lunatic quality grid.

May I still ask is it feasible to do this analysis (17 1 9 4 12) when implicit solvent was used via SCRF(SMD) command in Gaussian16 when generating the wfn file?

Last edited by monthem (2025-03-30 08:01:16)

sobereva · 2025-03-30 09:08:51

Integration of ELF basins in Multiwfn is carried out based on uniform grids. However, for very heavy atoms, because electron density around their nuclei varies very sharply, it is impossible to accurate integrate core ELF basins based on the uniform grids, even if lunatic quality grid is used. Since core basin is usually not of chemical interest (and if they are accurately integrated, then their populations must be very close to integer, that means you can easily predict their populations without any calculation), you can only focues on valence basins. Alternatively, using pseudopotential for these heavy atoms, then at least innermost ELF basins will not be presented.

It is fully possible to use implicit solvation model when generating wavefunction files.

monthem · 2025-03-30 12:50:28

Yes, it is entirely possible to generate a wfn file using implicit solvent. However, is it reasonable to use ELF analysis on a solvated system wawefunction?

Running 17 1 9 4 12 on a [Be2F6]2- wfn file obtained with SCRF(SMD, solvent=water) in Gaussian16, results in a population of ~81 electrons. Be2F6 in vacuum results in 64.0 electrons as it should.

Here is part of the output:

The following information is printed according to basin indices
Basin indices, populations (e), volumes (Angstrom^3) and assigned labels:
Basin 1 Pop.: 3.5680 Vol.: 139.506 Label: V(F7)
Basin 2 Pop.: 3.6193 Vol.: 133.044 Label: V(F8)
Basin 3 Pop.: 2.7588 Vol.: 42.311 Label: V(F2)
Basin 4 Pop.: 2.7588 Vol.: 41.842 Label: V(F6)
Basin 5 Pop.: 2.6866 Vol.: 40.412 Label: V(F2)
Basin 6 Pop.: 2.6868 Vol.: 39.994 Label: V(F6)
Basin 7 Pop.: 1.1917 Vol.: 1.520 Label: V(F2,Be5)
Basin 8 Pop.: 1.1916 Vol.: 1.520 Label: V(Be5,F6)
Basin 9 Pop.: 2.1478 Vol.: 0.025 Label: C(F7)
Basin 10 Pop.: 2.1436 Vol.: 0.025 Label: C(F8)
Basin 11 Pop.: 0.1401 Vol.: 5.491 Label: V(F8)
Basin 12 Pop.: 2.6661 Vol.: 61.656 Label: V(F8)
Basin 13 Pop.: 0.1405 Vol.: 5.506 Label: V(F8)
Basin 14 Pop.: 1.2711 Vol.: 1.617 Label: V(Be5,F8)
Basin 15 Pop.: 0.0638 Vol.: 3.190 Label: V(F7)
Basin 16 Pop.: 0.0638 Vol.: 3.171 Label: V(F7)
Basin 17 Pop.: 2.8711 Vol.: 64.571 Label: V(F7)
Basin 18 Pop.: 1.2669 Vol.: 1.601 Label: V(Be5,F7)
Basin 19 Pop.: 2.0565 Vol.: 0.651 Label: C(Be5)
Basin 20 Pop.: 2.8704 Vol.: 64.522 Label: V(F3)
Basin 21 Pop.: 2.6648 Vol.: 61.330 Label: V(F4)
Basin 22 Pop.: 1.1918 Vol.: 1.520 Label: V(Be1,F2)
Basin 23 Pop.: 1.1917 Vol.: 1.520 Label: V(Be1,F6)
Basin 24 Pop.: 1.2711 Vol.: 1.616 Label: V(Be1,F4)
Basin 25 Pop.: 1.2669 Vol.: 1.601 Label: V(Be1,F3)
Basin 26 Pop.: 2.1513 Vol.: 0.025 Label: C(F2)
Basin 27 Pop.: 2.1513 Vol.: 0.025 Label: C(F6)
Basin 28 Pop.: 0.1406 Vol.: 5.561 Label: V(F4)
Basin 29 Pop.: 0.1406 Vol.: 5.509 Label: V(F4)
Basin 30 Pop.: 0.0638 Vol.: 3.175 Label: V(F3)
Basin 31 Pop.: 0.0640 Vol.: 3.168 Label: V(F3)
Basin 32 Pop.: 3.5685 Vol.: 138.659 Label: V(F3)
Basin 33 Pop.: 3.6200 Vol.: 132.383 Label: V(F4)
Basin 34 Pop.: 2.1478 Vol.: 0.025 Label: C(F3)
Basin 35 Pop.: 19.0373 Vol.: 0.651 Label: C(Be1)
Basin 36 Pop.: 2.1436 Vol.: 0.025 Label: C(F4)

Sum of core basin populations: 33.9792
Sum of valence basin populations: 46.9992
Sum of all basin populations: 80.9785

Note that the sum of valence basin populations is nearly identical to that found in Be2F6 vacuum wfn file on the same quality grid (lunatic), which is 46.9920. Sum of core basin populations on the vacuum wfn file is 17.0051. The error appears to come entirely from C(Be1) with estimated 19.0373 electrons.

The .wfn file for this calculation is Be2F6_dianion_water.wfn, posted in the google drive link above.

What could be causing the issue here?

Last edited by monthem (2025-03-30 12:57:15)

sobereva · 2025-03-30 12:58:39

You Google driver only contains Be2Br6_vacuum_high.log, I don't find .wfn file.

.wfn file produced under implicit solvation model can be normally analyzed by Multiwfn.

ELF gradient is calculated analytically in Multiwfn.

monthem · 2025-03-30 15:29:05

I apologize, please find here the problematic .wfn file:

Be2F6_dianion_water.wfn

Some testing though seems that Be2Cl6 and Be2Br6 in water are just fine. And is some issue with this particular wnf file perhaps.

An update: The error seems to be only with the Be2F6 in water and only with the option "Assign ELF basin labels" (input sequence 17 1 9 (1-4) 12). Using a sequence "2 Integrate real space functions in the basins 9 ELF" during the same job yields valid results. Somehow, option 12 adds ~17 electrons to one of the basins.

Here is copy of the terminal output ([Be2F6]2- (64 e), in water) from a low quality grid just for showing quickly the results, basin 21 is the problematic one: (first tests show the problematic basin changes each run)

12
Assigning labels for core basins...
Assigning labels for valence basins...
Progress: [##################################################] 100.0 % /
Evaluating basin populations and volumes, please wait...
Progress: [##################################################] 100.0 % |

The following information is printed according to basin indices
Basin indices, populations (e), volumes (Angstrom^3) and assigned labels:
Basin 1 Pop.: 6.2712 Vol.: 198.400 Label: V(F8)
Basin 2 Pop.: 4.9320 Vol.: 80.681 Label: V(F2)
Basin 3 Pop.: 4.9320 Vol.: 76.561 Label: V(F6)
Basin 4 Pop.: 6.2712 Vol.: 202.918 Label: V(F4)
Basin 5 Pop.: 2.2368 Vol.: 0.040 Label: C(F8)
Basin 6 Pop.: 2.2368 Vol.: 0.040 Label: C(F4)
Basin 7 Pop.: 1.3316 Vol.: 1.693 Label: V(Be5,F8)
Basin 8 Pop.: 1.3445 Vol.: 1.923 Label: V(F2,Be5)
Basin 9 Pop.: 1.3316 Vol.: 1.693 Label: V(Be1,F4)
Basin 10 Pop.: 1.3445 Vol.: 1.920 Label: V(Be5,F6)
Basin 11 Pop.: 1.3445 Vol.: 1.924 Label: V(Be1,F2)
Basin 12 Pop.: 1.3445 Vol.: 1.920 Label: V(Be1,F6)
Basin 13 Pop.: 6.3454 Vol.: 202.133 Label: V(F7)
Basin 14 Pop.: 2.0434 Vol.: 0.650 Label: C(Be5)
Basin 15 Pop.: 1.9907 Vol.: 0.040 Label: C(F2)
Basin 16 Pop.: 1.9907 Vol.: 0.040 Label: C(F6)
Basin 17 Pop.: 2.0434 Vol.: 0.650 Label: C(Be1)
Basin 18 Pop.: 6.3454 Vol.: 206.630 Label: V(F3)
Basin 19 Pop.: 1.2627 Vol.: 1.581 Label: V(Be5,F7)
Basin 20 Pop.: 1.2627 Vol.: 1.581 Label: V(Be1,F3)
Basin 21 Pop.: 17.9989 Vol.: 0.040 Label: C(F7)
Basin 22 Pop.: 2.0149 Vol.: 0.040 Label: C(F3)

Sum of core basin populations: 32.5556
Sum of valence basin populations: 45.6637
Sum of all basin populations: 78.2193

Sorting basins according to labels...
The following information is printed according to order of basin labels
Basin indices, populations (e), volumes (Angstrom^3) and assigned labels
# 1 Basin 22 Pop.: 2.0149 Vol.: 0.040 Label: C(F3)
# 2 Basin 21 Pop.: 17.9989 Vol.: 0.040 Label: C(F7)
# 3 Basin 17 Pop.: 2.0434 Vol.: 0.650 Label: C(Be1)
# 4 Basin 16 Pop.: 1.9907 Vol.: 0.040 Label: C(F6)
# 5 Basin 15 Pop.: 1.9907 Vol.: 0.040 Label: C(F2)
# 6 Basin 14 Pop.: 2.0434 Vol.: 0.650 Label: C(Be5)
# 7 Basin 6 Pop.: 2.2368 Vol.: 0.040 Label: C(F4)
# 8 Basin 5 Pop.: 2.2368 Vol.: 0.040 Label: C(F8)
# 9 Basin 18 Pop.: 6.3454 Vol.: 206.630 Label: V(F3)
# 10 Basin 3 Pop.: 4.9320 Vol.: 76.561 Label: V(F6)
# 11 Basin 13 Pop.: 6.3454 Vol.: 202.133 Label: V(F7)
# 12 Basin 2 Pop.: 4.9320 Vol.: 80.681 Label: V(F2)
# 13 Basin 4 Pop.: 6.2712 Vol.: 202.918 Label: V(F4)
# 14 Basin 1 Pop.: 6.2712 Vol.: 198.400 Label: V(F8)
# 15 Basin 11 Pop.: 1.3445 Vol.: 1.924 Label: V(Be1,F2)
# 16 Basin 20 Pop.: 1.2627 Vol.: 1.581 Label: V(Be1,F3)
# 17 Basin 9 Pop.: 1.3316 Vol.: 1.693 Label: V(Be1,F4)
# 18 Basin 12 Pop.: 1.3445 Vol.: 1.920 Label: V(Be1,F6)
# 19 Basin 8 Pop.: 1.3445 Vol.: 1.923 Label: V(F2,Be5)
# 20 Basin 10 Pop.: 1.3445 Vol.: 1.920 Label: V(Be5,F6)
# 21 Basin 19 Pop.: 1.2627 Vol.: 1.581 Label: V(Be5,F7)
# 22 Basin 7 Pop.: 1.3316 Vol.: 1.693 Label: V(Be5,F8)

Number of core basins is 8, their indices:
5,6,14-17,21,22
Number of 1-synaptic basins is 6, their indices:
1-4,13,18
Number of 2-synaptic basins is 8, their indices:
7-12,19,20

============= Basin analysis =============
-10 Return to main menu
-6 Set parameter for attractor clustering or manually perform clustering
-45 Export attractor information and cube file of present grid data
-5 Export basins as cube file
-4 Export attractors as pdb/pqr/txt/gjf file
-3 Show information of attractors
-2 Measure distances, angles and dihedral angles between attractors or atoms
-1 Select the method for generating basins
0 Visualize attractors and basins
1 Regenerate basins and relocate attractors
2 Integrate real space functions in the basins
3 Calculate electric multipole moments and <r^2> for basins
4 Calculate localization index (LI) and delocalization index (DI) for basins
5 Output orbital overlap matrix in basins to BOM.txt in current folder
10 Calculate high ELF localization domain population and volume (HELP, HELV)
11 Calculate orbital compositions contributed by various basins
12 Assign ELF basin labels
2

Please select the integrand:
-2 Return
-1 The values of the grid data stored in an external file (.cub/.grd)
0 The values of the grid data stored in memory
----------- Available real space functions -----------
1 Electron density (rho) 2 Gradient norm of rho 3 Laplacian of rho
4 Value of orbital wavefunction 44 Orbital probability density
5 Electron spin density
6 Hamiltonian kinetic energy density K(r)
7 Lagrangian kinetic energy density G(r)
8 Electrostatic potential from nuclear charges
9 Electron localization function (ELF)
10 Localized orbital locator (LOL)
11 Local information entropy
12 Total electrostatic potential (ESP)
13 Reduced density gradient (RDG) 14 RDG with promolecular approximation
15 Sign(lambda2)*rho 16 Sign(lambda2)*rho with promolecular approximation
17 Correlation hole for alpha, ref. point: 0.00000 0.00000 0.00000
18 Average local ionization energy (ALIE)
19 Source function, mode: 1, ref. point: 0.00000 0.00000 0.00000
20 Electron delocal. range func. EDR(r;d) 21 Orbital overlap dist. func. D(r)
22 Delta-g (promolecular approximation) 23 Delta-g (Hirshfeld partition)
24 Interaction region indicator (IRI) 25 van der Waals potential (probe=C )
100 User-defined function (iuserfunc= 0), see Section 2.7 of manual
1
Integrating, please wait...
Progress: [##################################################] 100.0 % \
#Basin Integral(a.u.) Volume(a.u.^3)
1 6.2711728011 1338.87200000
2 4.9320407832 544.46400000
3 4.9320349937 516.65600000
4 6.2711766577 1369.36000000
5 2.2368212255 0.27200000
6 2.2368212255 0.27200000
7 1.3316213589 11.42400000
8 1.3444968056 12.97600000
9 1.3316213589 11.42400000
10 1.3444968055 12.96000000
11 1.3444968057 12.98400000
12 1.3444968055 12.96000000
13 6.3453661586 1364.06400000
14 2.0433783222 4.38400000
15 1.9906591894 0.27200000
16 1.9906591894 0.27200000
17 2.0433783222 4.38400000
18 6.3453687602 1394.40800000
19 1.2626604030 10.67200000
20 1.2626604030 10.67200000
21 2.0149239202 0.27200000
22 2.0149239202 0.27200000
Sum of above values: 62.23527622
Integral of the grids travelled to box boundary: 0.00000009
Integrating basins took up wall clock time 1 s

Last edited by monthem (2025-03-30 17:58:34)

sobereva · Yesterday 09:17:28

It is found to be a bug, I have fixed it, please download the latest version from Multiwfn website. Thank you for bringing this bug to my attention.

monthem · Yesterday 09:24:26

Dear professor, thank you very much for attending to the bug so quickly.

I was about to post wfn_results.xlsx this spreadsheet file with the results of the calculations I've done overnight. Column A is the compound name (all are dianions), columns B to M are number of electrons expected to be found. Columns O to U are what 17 1 9 4 12 finds (blue colored refers to water solvent systems). It seems that the error is always in one basin (core or valence). I will now download the latest version and re run the calculations.

May I ask though, how are core electrons treated when ECP is present? The program detects "Total/Alpha/Beta electrons correctly, but in the 17 19 4 12 integration it seems to find roughly total core populatons without ECP - ~ 5 electrons.

sobereva · Yesterday 10:52:17

If you use a large-core ECP, then there will only be valence ELF basins. If a small-core ECP is used, core basins will also be present, which correspond to subvalence shell electrons. You can use option 0 in basin analysis module to visualize the basins to better understand their characters.

monthem · Yesterday 10:54:36

Dear professor, thank you for your time and help. The latest version program 17 1 9 4 12 works as intended indeed.

Multiwfn forum

#1 2025-03-29 16:04:48

ELF basin populations and grid quality, use of solvent

#2 2025-03-30 08:00:08

Re: ELF basin populations and grid quality, use of solvent

#3 2025-03-30 09:08:51

Re: ELF basin populations and grid quality, use of solvent

#4 2025-03-30 12:50:28

Re: ELF basin populations and grid quality, use of solvent

#5 2025-03-30 12:58:39

Re: ELF basin populations and grid quality, use of solvent

#6 2025-03-30 15:29:05

Re: ELF basin populations and grid quality, use of solvent

#7 Yesterday 09:17:28

Re: ELF basin populations and grid quality, use of solvent

#8 Yesterday 09:24:26

Re: ELF basin populations and grid quality, use of solvent

#9 Yesterday 10:52:17

Re: ELF basin populations and grid quality, use of solvent

#10 Yesterday 10:54:36

Re: ELF basin populations and grid quality, use of solvent

Board footer