I'm studying a mechanism of aromatic substitution of fluorine which proceeds via an intermediate.

I'm searching for a transition state using a relaxed surface scan, then I optimize the TS and perform an IRC (intrinsic reaction coordinate) calculation to verify the products. The problem is that the intermediate (that I get from the IRC) has nearly the same energy as the TS (about 1 kcal/mol lower), which doesn't seem right.

I assume that the intermediate geometry is wrong because the IRC makes 2-3 steps towards it and stops, saying the minimum was reached. How can I verify that an IRC gives correct results?

  • 3
    $\begingroup$ I recommend that you take the result of the IRC calculation and perform a "standard" geometry optimization. Does the energy of the resulting structure match your reactant? Does the TS connect both the reactant and product? Is the barrier you get from your surface scan relatively close to that of the TS search? If that all checks out, then perhaps you just have a very early TS. There's nothing inherently wrong with that. $\endgroup$ Feb 12 at 20:09
  • $\begingroup$ @AndrewRosen I think it's a very flat PES near the TS, I've tried tight IRC (I'm using Gaussian) and it reached minima on both sides with only one step, and obviously energy difference is very small. Actually, I'm comparing different attack sites in my molecule, and some of them has reasonable energy difference between TS and intermediate and other doesn't. I have very little experience in such calculations, so I'm very confused. $\endgroup$ Feb 13 at 6:17
  • $\begingroup$ I've tried calculating energy at CCSD(T) level of theory, but at that level my intermediate lies higher than both TS, which confuses me even more. Maybe something is fundamentally wrong in my calculations. $\endgroup$ Feb 13 at 14:30
  • 1
    $\begingroup$ Roma, would you mind giving us the relevant XYZ files and letting us try? This seems very system specific and it's hard to troubleshoot your problem without being able to do the calculations ourselves. We also don't know if CCSD(T) is valid or not for the molecules you're dealing with (if you use CCSD(T), which is a single-reference method, on a system that requires a multi-reference method, then there will be something "fundamentally wrong in your calculations" as you feared). Furthermore, it could be basis set incompleteness causing you to see strange results, but we don't know your basis! $\endgroup$ Sep 5 at 23:38
  • $\begingroup$ I agree with Nike's comment, without the geometry we can really only guess or give general advice. For now, I have closed the question to clear it out of the unanswered queue, but I can reopen it if you include a bit more detail about the calculation in the question (basis, geometry, ideally the whole input/output files). $\endgroup$
    – Tyberius
    Oct 29 at 17:50