COSMO input and output

[Mario Schleep]

Hi all

I try to use COSMO with NWChem dft calculations. However, unfortunately I have some trouble with input and output.

I have attachd two examples of output to explain my problems. In the first example, I wanted to do a calculation of TFA in acetonitrile (37.5). But in the output file it says dielec 37.5 as I wanted it to be, but also "solvname_short: h2o" and "solvname_long:  water". How do I have to change this? And where (I am using NWChem 7.0.1-1 on debian testing) can I find the default COSMO parameter files?

Apart from this, the calculation seemed to have worked, but what COSMO output do I have to use? There are several "COSMO solvation results", is just the last one the one to use?

In the secend example, I tried to do a calculation of Ac-Gly-OH in acetonitrile. Apart from the above described problem with water, the calculation stops with "dft optimize failed" although the output contains "The DFT is already converged" before. Could someone tell me what is gone wrong here?

I would really appreciate your help!

Thanks,

Mario

[Edoardo Aprà]

1.  You can ignore the "solvname_short: h2o" and "solvname_long:  water" bits in the output since, as you already pointed out, the dielectric constant you provide by input is used.
   
2. With the input "print low" on, it's hard to tell what happened. Event though you can find the string "converged" that refers to reading of the molecular orbitals prior to the fatal error. Therefore it could well be a SCF convergence problem. Increasing the DFT maxiter option to 90 https://nwchemgit.github.io/Density-Functional-Theory-for-Molecules.html#iterations-or-maxiter-number-of-scf-iterations might fix the problem

PS The solvname warnings are can be fixed by using the undocumented solvent input option

cosmo
 dielec 37.5
 solvent acetntrl
end

[Mario Schleep]

Thanks Edoardo, I rerun the calculation with medium print level and chech again. However, I understood that iterations and maxiter is the same, right? And in my input I set iterations to 1000.

Could you tell me in the TFA example which COSMO results I have to use?

[Edoardo Aprà]

On Monday, October 26, 2020 at 11:05:30 PM UTC-7, Mario Schleep wrote:

Could you tell me in the TFA example which COSMO results I have to use?

I am assuming you are try to compute the frequencies of this TFA example, right?

The frequency output is a the bottom of output file.

The frequency output report one negative frequency. This is an indication that the geometry optimization did not converge into a minimum.

You can use the displacement analysis to redo the geometry optimization. For example, you can copy the geometry that occurs after the string

 Geometry after -100.0% step for mode  1; Step length =  0.154 angstroms

[Mario Schleep]

Alright, sorry for the misunderstanding. Actually, I have just quickly calculated TFA to have an example with a smaller molecule as mine didn't converge (as Ac-Gly-OH). I didn't check the vibrations...

I want to calculate the solvation enthalpy of the molecules and normally I would use the gas phase calucation "Total DFT energy" + "Zero-Point correction to Energy" and calculate H and G using "Thermal correction to Energy" and Thermal correction to Enthalpy". Then I would do the same with the results of the calculation with COSMO and finally calculate DeltaH and DeltaG for solvation.

However, I do not know, if it should be done differently, as the output file contains various "COSMO energy" and (in Ha?) and various "COSMO solvation results" and I do not know which one is the one to use (the last one?). The "COSMO solvation results" contain the "gas phase energy", "sol phase energy" and "(electrostatic) solvation energy" what seems good, but the "gas phase energy" differs slightly from the "Total DFT energy" I obtained from the optimization without COSMO. And using the way to calculate DeltaH and DeltaG, I do not get the same result as the number in any of these COSMO results.

So, sorry, I hope this is not written to chaotic and thanks a lot for your time and patience!

Mario