LSR doesn't reproduce Pt(111) thermo

[cfgoldsmith]

We set the LSR parameters to correct values for H, C, N, O on Pt(111). We generate a mechanism. We then compare the thermo in output.html with values in the thermo library. They should be the same, but they are not. Fortunately, they are close (e.g. I think the reference enthalpy for CH3* differed by ~3 kcal/mol), so it is not catastrophic, but we need to fix it.

I think switching the way we do LSR (by specifying the intercept) should fix this problem (see Issue 40 #40)

[kblondal]

For CH3*, the reference enthalpy in output.html is -13.17 kcal/mol, whereas in the thermo library it is -10.94 kcal/mol. The C adatom value for Pt I used is -6.750 eV.

[rwest]

We need these values to match your "correct values":
13d0780

Feel free to fix them

[cfgoldsmith]

Perfect, thanks.

[kblondal]

Thanks, I did as you said and pushed to my cat branch:
https://github.com/kblondal/RMG-Py/blob/5ab526122262566968f466e053bb2ccc028a580d/rmgpy/data/thermo.py#L1267-L1273

Now the reference enthalpy and entropy in the output.html file is consistent with the Pt thermo library. However the heat capacity is still off. It's consistently a bit lower in the output.html file than if calculated based on the thermo library. Taking CH3* as an example again, Cp(300K) = 8.86 cal/molK in output.html, but 10.95 cal/molK based on Pt thermo.

[rwest]

That's weird. It grabs thermo for an adsorbate from a library, and outputs something different?
Could it be extrapolation to zero kelvin or infinite temperature extremes are shifting things?
Does the detailed annotated chemkin file (or the HTML) include comments about where the thermo for that species came from? what do they say?

[cfgoldsmith]

My first thought is that the extrapolation/Wilhoit/etc is responsible, but it seems odd that 300K would be off, so I'm not sure either.

[kblondal]

I just calculated Cp(300) for CH3* by hand based on the NASA polynomials given in the .cti file and the Pt thermo file and now I get 8.86 cal/molK for both af them. So RMG-Cat is actually probably grabbing and outputting all the right values based on the Pt thermo library.
The code I am using to print heat capacity values is for the thermo before being fitted to a NASA polynomial. Maybe the discrepancy I get is just due to the fitting. I'll double check that it is done correctly. I should have just compared the polynomials to begin with, sorry about that. But it seems that this issue is solved!

(To answer the question: The annotated chemkin file says "surfaceThermoPt Binding energy corrected by LSR", but there is no such note anymore in the html file.)