Trying to get pressure dependence from experimental data for radical recombination

[CCM]

Hi!

I'm trying to get pressure dependent Arrhenius parameters for some radical recombination reactions, starting with CF3 recombination. I'm using rovibrational data from electronic structure calculations I've done in Gaussian 16. There are several papers with experimental data that I would like to incorporate in addition to getting the Plog rate constants. I've tried to use the Marquardt calculation method with the rate data from a paper by Rossi & Golden (1979), but it seems like the chi squared values blow up and the calculation terminates due to error. I've pasted the error message below.

WARNING: Chemically significant eigenvalues (CSE) not well separated from internal energy relaxation eigenvals (IEREs).

The last CSE = -2.71155e+07 and the first IERE = -4.35597e+07
(last CSE)/(first IERE) ratio = 0.622491, which is less than an order of magnitude

Results obtained from Bartis Widom eigenvalue-vector analysis may be unreliable

Chi^2 = 5.7519e+22 Lambda = 1
R1:preExponential=6.49e+10  R1:nInfinity=1
Singular Matrix in routine ludcmp

Marquardt Hessian Matrix:
{
   0.000000e+00,   0.000000e+00,
   0.000000e+00,   0.000000e+00,
}
Error encountered in LU decompostion invoked from solveLinearEquationSet.

I've read about the CSE issue in the manual, and am not concerned with that part just yet (unless someone knows something I don't, and the CSE/IERE problem is the cause of these issues). I've attached my input .xml file for review. Any help would be greatly appreciated!

[struanr]

Hello,

Sorry for my slow response.

I have looked at your system and found a few issues. One issue was the control section, which needed to be separated out into individual tasks, which I have done and added a third so that you can see the change in the calculated rate coefficients. I found a bug MESMER concerning the units that you were using, cm3mol-1s-1, which are clearly correct but were not being processed by MESMER properly. I will fix this for the next release of MESMER (scheduled for autumn of this year), but in the meantime I have altered the units to be cm3molecule-1s-1. The last problem is that the data were presented without experimental errors, and these are needed to calculate the figure of merit correctly. Arguably MESMER should provide a default for this but for now I have assumed a value of 10% of the experimental value. I attached a .zip file with the revised input and the results I obtained. According to the Chi_2 statistic the fit is not great but with so few points this might not be a good measure.

Looking through your input I noticed a few couple of other things that you might want to consider: I suspect that one of the vibrational modes of the C2F6 might be an internal rotor and it may be batter to treat it as such. Also, you might want to look at the symmetry number of the CF3 radicals as there may be an inversion mode there which might mean the symmetry number is 6.

I hope this helps a little.

Thank-you for using MESMER.

Regards, Struan

[CCM]

Hi Struan,

Thank you for taking the time to look at my input file and making the necessary corrections. There are other experimental studies I can compare against, so I'm not too concerned with the Chi_2 statistic for this case, but I appreciate the note.

Regarding the internal rotor for C2F6, the vibrational mode that corresponds to the rotor is the lowest value vibrational frequency in this case (confirmed in GaussView). I have been reading through the manual and I am a little confused on how to go about accounting for this in the Mesmer calculation, as well as accounting for the inversion mode of CF3 radicals. I checked the Gaussian log file for CF3 and it indeed has an inversion and should have a symmetry number of 6, so I'm not sure how it was recorded with a symmetry number of 3 in my original input. Based on what I've read in the manual, I think I need to account for these things in the density of states calculations, but I'm unsure how to proceed. If you have time and could provide some insight, I would greatly appreciate it.

Thank you again for taking the time to help me with my calculations!

Best,
CCM

[struanr]

Hello,

I have revised the input to include the hindered rotor. The revised input and the results obtained are in the attached .zip file. I made a guess for the barrier height of the potential of 20 kJ/mol but obviously you should check this against ab initio. I also note that you were using a <delta e="">down of 300 cm-1 which, in my experience, is a little high, so I floated this parameter as well. This does result in a marginally better Chi^2 but as you will see the errors on this parameter are huge so the value should be treated with caution. </delta>

I hope this helps.

Regards, Struan