Multiwfn forum

pt007

Member

Registered: 2024-08-10

Posts: 5

Quantitative Molecular Surface Analysis recommended input file?

Hi,

I am interested in extracting the ESP minima, maxima, and average value using the Quantitative Molecular Surface Analysis (option 12).

The molecules I work with are isolated ions. For example, 1-butyl-3-methylimidazolium, 1-butylpyridinium, etc. Each ion is a different file. I used DFT methods M06-2X/AUG-cc-pVTZ with SMD water solvent and GD3 correction (using Gaussian 16) to get the optimized structure of the ion. I can then output .log file (from which I can get optimized coordinated and make .xyz file) and .wfn file.

I have tried using the two different input files (.log/.xyz and .wfn) for Quantitative Molecular Surface Analysis and to no surprise, get different results for ESP descriptors.

What is the recommended input file for accurate calculations?
Some of my data correlates well with ESP average value using .log as input... When the same correlation is weak when .wfn input is used. I want to better understand why is that the case and which one should be considered more accurate.

Regards,
Pranav

sobereva

Tian Lu (Multiwfn developer)

From: Beijing

Registered: 2017-09-11

Posts: 1,773

Website

Re: Quantitative Molecular Surface Analysis recommended input file?

Dear Pranav,

You need to use a file containing wavefunction information for the analysis, please check Section 2.5 of Multiwfn manual on which file formats contain this kind of information, you will find you should use e.g. wfn/mwfn/fch/molden... as input file.

Using .xyz/log file for this analysis is meaningless, which only provide atom information to Multiwfn, the ESP produced in this way is only contributed by nuclei, while contribution from electrons is fully missing.

Best,

Tian

pt007

Member

Registered: 2024-08-10

Posts: 5

Re: Quantitative Molecular Surface Analysis recommended input file?

Tian,

Thank you for your response. Makes sense.

I have two follow up questions:

1) So does it matter when optimizing geometry and generating .wfn on Guassian16, I do it in gas-phase (vacuum) or solvent-phase (SMD with water as the solvent)? I do these as two steps... First optimize and get .log file. Then I convert .log to .xyz, and feed that as an input to make .wfn. I keep DFT methods same for both steps.

I tried to compare my DFT method (gas- and solvent-phase) for 4 chemicals with published MPI from your work Carbon, 171, 514 (2021).
The values here are in kcal/mol

Chemical / Published (gas-phase) / MPI from Gas-phase / MPI from SMD water
cyclo[18]carbon / 2.6 / 2.45 / 2.46
ethane / 2.6 / 2.66 / 2.86
ethene / 6.7 / 6.86 / 7.55
benzene / 8.4 / 8.2 / 9.94

As observed, when optimizing and generating .wfn in a solvent-phase, MPI value increases. And my gas-phase numbers are comparable to your published values (good validation!)

2) I also ran the same process for a non-polar gas CO2. I am surprised as to why the MPI for CO2 is 11.19 kcal/mol... Because CO2 is regarded as a non-polar molecule. I checked my structure as it is linear as expected.

Regards,
Pranav

sobereva

Tian Lu (Multiwfn developer)

From: Beijing

Registered: 2017-09-11

Posts: 1,773

Website

Re: Quantitative Molecular Surface Analysis recommended input file?

Dear Pranav,

1 It is well known that solvent environment polarizes charge distribution of solute, and the larger the dielectric constant of the solvent, the stronger the polarization effect. So, it is fully expected that employing solvation model will increase the MPI index.

2 CO2 is not so nonpolar. You can plot electrostatic potential map in the molecular plane via main function 4 of Multiwfn (see Section 4.4.4 of Multiwfn manual for example), you will find anisotropic character of CO2 charge distribution is quite evident. The MPI index just well represents the "effective polarity" of CO2, which cannot be exhibited by dipole moment. MPI provides important complementary information on judging molecular polarity.

It is strongly suggest also checking my previous work about interaction between H2 and N2, namely J Mol Model (2013) 19:5387–5395 DOI: 10.1007/s00894-013-2034-2. Though H2 and N2 are commonly (somewhat naively) believed to be fully nonpolar, in this work I deeply discussed how electrostatic interaction (in a sense, local polar-polar interaction) greatly influences dimer configuration.

Best regards,

Tian

pt007

Member

Registered: 2024-08-10

Posts: 5

Re: Quantitative Molecular Surface Analysis recommended input file?

Tian,

This is really interesting. I also checked a citation in your paper wherein many molecules were screened for their charge separation or local polarity (PI-index) from Tore Brinck, Jane S. Murray & Peter Politzer.
They had CO2 in their dataset and actually got a large value for PI = 16.2 kcal/mol (my value from a very different DFT theory is close to 11.2 kcal/mol but close enough). They suspect this is because of high degree of charge separation in the molecule, which cannot be captured from dipole moment, just like you mentioned.

I look into what other descriptors can I extract from Multiwfn to see what correlates best with my experimental data.

Thank you once again

Best regards,
Pranav