A bit of a layman question. If I have a reaction mechanism which I suspect has about 18 intermediate steps, is it wise to take each step one at a time? i.e., I optimise my reactant and product, then perform a TS search (using scan geometry for a possible TS then I use QST3 with these three structures). I then link up all three structures with an IRC calculations. I then use the product of my last step as my reactant of my next step. From this new reactant I can take a guess of the geometry of my new product and perform opt. Then I scan geometry, take a guess at a TS, run a QST3, run IRC.. then repeat.
I am taking all of this one at a time, from left to right as the mechanism goes. Is there anything I can do to start work on the other stages of the mechanism as I am simply waiting for some very long calculations (6-311+g(p,d))? Or must it be one at a time, so theoretically, each intermediate and TS runs smoothly across the potential energy surface?
Or rather, I have a low level basis set optimised structured of the product from my first transition state. Whilst this is now being reoptimised using a higher basis set, would it make sense to use the low level basis set as a starting point for generating my new product and TS? Or should I wait until I have the higher basis set versions complete? Thanks
You are doing it the best way, I'd say. The problem with using small basis set approximations to move in ahead is that small basis sets often give incorrect characterization of stationary points. A minimum at a small basis set level of theory might, in fact, be a TS with your best basis set, or vice versa, especially if there are any very low frequency vibrations of a structure.
Hi Allen, Thanks for the answer. I am quite fortunate that the TS found with a small basis set have quite large frequency vibrations and most of them were further refined with a higher basis set. Unfortunately a couple of TS states did not converge to a saddle point so I am probably going to have to work on finding a TS just using a high basis set with no previously calculated force constants.
OPT=QST2,CALCFC usually does a pretty good job of finding TS's, especially ones that are well-defined (not too floppy) as you say yours seem to be, in my experience. And that may save you a step compared to doing a scan then QST3 -
Thanks for the information - so far I've been lucky with a by-eye guess of a TS in a QST3 with calcfc, and a little bit of bond scanning if a particular TS is being rather elusive.
So far, all the TS have either been a dehydration reaction or tautomerisation. These have been quite easy. The next TS in this pathway will be working a cyclization reaction - not sure how to go about this given that the reactant of this TS was found to be linear. I need to bring the carbonyl group at the end of the molecule right back up near to the top. I am guessing I need to make an additional intermediate step before attempting to find a TS.
- Badges
- Science topic