Multiwfn forum

wham09

Member

Registered: 2025-09-19

Posts: 33

Frequency correction

Dear Prof. Lu,

The Shermo code utilizes frequency scale factors for ZPE, U(T)-U(0), S, and CV. I can find the scale factors for ZPE in the literature, but the other three factors are rarely documented (especially, I could not find the factors for CV). So I'd like to ask if the following routine would work.

1. Fit the fundamental frequency factor for my functional/basis set using a database (F38/10 for example).
2. Run a opt-freq calculation for my molecule, and tabulate the computed harmonic frequencies.
3. Multiply the computed frequencies by the fitted fundamental frequency factor.
4. Put the results as the custom frequencies into a new freq calculation input with freq=(ReadFC, ReadIsotopes) keyword. (Or use scale keyword)
5. Use the output of the new freq calculation as the input file for Shermo, with the following settings:
  5-1. sclZPE, sclheat, sclS, sclCV = 1.0
  5-2. ilowfreq = 2 or 3

Last edited by wham09 (2025-10-25 09:46:39)

wham09

Member

Registered: 2025-09-19

Posts: 33

Re: Frequency correction

Additional questions:

1. I understand that unlike sclZPE, other scale factors are normally close to unity. But I still want to understand how they're derived. From what I read in Moran/Radom paper (doi/10.1021/jp073974n), sclheat and sclS cannot be obtained by simply fitting U_calc and U_exp (or S_calc and S_exp) linearly. Am I correct?

2. Can I get any reference where the scale factors for CV are derived? I could never find it.

3. I tried fitting the frequency scale factor myself. For some of the molecules in F38/10 set, the degenerate frequencies appeared to be different from each other. For example, the frequencies corresponding to the 3rd mode (pi_u symmetry) of CO2 are 646.6657 and 646.6429 in my calculation output. Did I do something wrong? Or should I just average them?

4. If what I want at the end is G(sol), should I put into Shermo the E value from SCRF calculation, rather than gas-phase single-point?

Last edited by wham09 (2025-10-25 10:51:19)

sobereva

Tian Lu (Multiwfn developer)

From: Athens, Greece

Registered: 2017-09-11

Posts: 2,132

Website

Re: Frequency correction

Dear Prof. Lu,

The Shermo code utilizes frequency scale factors for ZPE, U(T)-U(0), S, and CV. I can find the scale factors for ZPE in the literature, but the other three factors are rarely documented (especially, I could not find the factors for CV). So I'd like to ask if the following routine would work.

1. Fit the fundamental frequency factor for my functional/basis set using a database (F38/10 for example).
2. Run a opt-freq calculation for my molecule, and tabulate the computed harmonic frequencies.
3. Multiply the computed frequencies by the fitted fundamental frequency factor.
4. Put the results as the custom frequencies into a new freq calculation input with freq=(ReadFC, ReadIsotopes) keyword. (Or use scale keyword)
5. Use the output of the new freq calculation as the input file for Shermo, with the following settings:
  5-1. sclZPE, sclheat, sclS, sclCV = 1.0
  5-2. ilowfreq = 2 or 3

Only ZPE scale factor is relatively important, you can simply set other factors to 1.

sobereva

Tian Lu (Multiwfn developer)

From: Athens, Greece

Registered: 2017-09-11

Posts: 2,132

Website

Re: Frequency correction

Additional questions:

1. I understand that unlike sclZPE, other scale factors are normally close to unity. But I still want to understand how they're derived. From what I read in Moran/Radom paper (doi/10.1021/jp073974n), sclheat and sclS cannot be obtained by simply fitting U_calc and U_exp (or S_calc and S_exp) linearly. Am I correct?

2. Can I get any reference where the scale factors for CV are derived? I could never find it.

3. I tried fitting the frequency scale factor myself. For some of the molecules in F38/10 set, the degenerate frequencies appeared to be different from each other. For example, the frequencies corresponding to the 3rd mode (pi_u symmetry) of CO2 are 646.6657 and 646.6429 in my calculation output. Did I do something wrong? Or should I just average them?

4. If what I want at the end is G(sol), should I put into Shermo the E value from SCRF calculation, rather than gas-phase single-point?

1 Yes. U and S are not determined linearly with respect to frequencies.
2 AFAIK, no known literature reported it
3 When using DFT, distribution of integration grids usually does not satisfy molecular symmetry. To make the degeneracy better, you can use better integration grid (e.g. int=superfine with CPHF=grid=fine)
4 Yes

wham09

Member

Registered: 2025-09-19

Posts: 33

Re: Frequency correction

Dear Prof. Lu,

The Shermo code utilizes frequency scale factors for ZPE, U(T)-U(0), S, and CV. I can find the scale factors for ZPE in the literature, but the other three factors are rarely documented (especially, I could not find the factors for CV). So I'd like to ask if the following routine would work.

1. Fit the fundamental frequency factor for my functional/basis set using a database (F38/10 for example).
2. Run a opt-freq calculation for my molecule, and tabulate the computed harmonic frequencies.
3. Multiply the computed frequencies by the fitted fundamental frequency factor.
4. Put the results as the custom frequencies into a new freq calculation input with freq=(ReadFC, ReadIsotopes) keyword. (Or use scale keyword)
5. Use the output of the new freq calculation as the input file for Shermo, with the following settings:
  5-1. sclZPE, sclheat, sclS, sclCV = 1.0
  5-2. ilowfreq = 2 or 3

Only ZPE scale factor is relatively important, you can simply set other factors to 1.

Thank you for the reply. As I said, I do understand that, but I just wanted to know if what I described would do the same thing as putting all scl values into Shermo?