Disprop ATG

[kspieks]

@am2145 and I refined reactions for an ammonia combustion mechanism in collaboration with @josemvdh. These 8 reactions were calculated at CCSD(T)-F12/cc-pVTZ-F12//B2PLYP-D3/def2-TZVP. Since "No coupled cluster methods based on UHF orbitals are implemented in MOLPRO" link, restricted open shell was used when calculating the single point energy for all species with spin > 0. All TSs have been verified with IRC.

To comment on some of the failing RMG-tests, it complains that the rate estimates have changed. In general, this is a bit curious since all new training reactions involve nitrogen yet all of the example reactions in RMG-tests are hydrocarbons. I examined some of these reactions and noticed that [CH]=C(7) + [c]1ccccc1(3) <=> ethyne(2) + benzene(1) and [c]1ccccc1(3) + [CH]=Cc1ccccc1(12) <=> benzene(1) + C#Cc1ccccc1(51) do not have any matching training reactions so the test just points out the new estimate differs from the old estimate but as of now, I don't think we have a true value to compare to. As another example, [H](4) + [c]1ccccc1(3) <=> [H][H](11) + C1#CC=CC=C1(10) matches index 697 from JetSurf2.0 kinetics library which shows that this reaction has a reasonably large barrier of 24.5 kcal/mol. The old rate rule said this was barrierless, which seems skeptical to me since the competing reaction is H addition to form benzene and I imagine that should actually be barrierless while this reaction should have some barrier. The new rate rule says this has a small barrier and a larger n so it has larger temperature dependence as shown by the comparison here so in general, I believe that RMG-tests just point out the estimates are changing but the ones I checked seem to be changing for the better. Some of these reactions also barely changed. For example, the estimated barrier for this reaction went from 0 to 0.3 kcal/mol and again, RMG has no matching training reactions so for now, we don't have a true value to compare to.

Looking forward to your review