I am investigating the mechanism of hydrogen transfer between a hydroperoxyl radical and an aromatic hydroxyl group donating it. The problem is that I cannot find transition state structure. So far, I've tried with QST3, QST2 and TS with initial geometries either based on similar reactants I'm studying or taken from the literature. In general - the angle between the hydroperoxyl radical and the hydroxyl group of the donor is about 165°, and the distances from the hydrogen atom are 1.358A and 1.076A, respectively. Initial structure

These failed despite changing the optimization algorithm, by:

  1. lowering maxstep (even to 1);
  2. using notrust;
  3. calculating force constants using calcfc or only derivatives for the atoms involved (available in modred).
  4. calculation of calcfc coupled with an imposed hessian of -1 for both bonds

The other way I tried to achieve my goal was to freeze the two bond and the angle, optimize the system, and only then reactivate them and run TS (with combinations of the above changes). No satisfactory results.

The last thing I did was to perform the TS task at a lower level of theory. By default I use M052X/6-311+G(d,p), although I have checked 6-31+G(d), 6-311G(d,p), 6-31+G(d,p), 6-311+G(d), 6-31G(d,p), but still without success.

And now, after a long introduction, I come to the substance: I ran scan at the desired level of theory. I set up the geometry similarly to the one shown above, although I reduced the bond length between the hydrogen atom and the hydroxyl group to 0.94 (typical for my systems in their ground state) simultaneously increasing the distance between the hydroperoxy oxygen and this hydrogen atom to about 2.0Å.


I ran the simulation for 50 steps every -0.05Å and.... here a certain problem arises. Following the literature and previous works, in such transition states the hydrogen atom is perpendicular to the oxygen of the aromatic hydroxyl group, while the hydroperoxy radical is parallel to the system in sandwich form. To be precise, it lies above/below it. Therefore, the initial structure was constructed this way. However, as a result of the scanning, the hydroperoxy radical shifts to completely the same plane, influencing the hydrogen to follow the same geometry.

incorrect scan

The resulting energy scan suggests a barrier-free reaction, but due to the mentioned glitch I can't believe the outcomes. I tried lowering the maxstep for the scan, and also doing an calcfc (although I'm not sure if either of those two affect the scan algorithm) though with the effect.

I appreciate a solution to this issue. Or what are other options I can take to find a TS or energy scan that really reflects the barrierless path of the process (which I think is more true in this particular case).

Thanks in advance.

  • 1
    $\begingroup$ Did you check the frequencies if there are no imaginary frequencies ? The functional is good as it includes non covalent interactions, you can also test a similar functional, the basis-set should absolutely include diffuse functions (++) even at a low level (6-31G). You should not expect to have exactly an experimental result, but this result is surprising. $\endgroup$
    – M06-2x
    Jun 19 at 19:28
  • $\begingroup$ @M06-2x Not at all. I presume, the potential energy surface is very flat. Checked M062X and the same ! I do not expect, though I'be been working a while with such systems thereby know what should the outcome be. $\endgroup$
    – farmaceut
    Jun 20 at 12:56
  • 1
    $\begingroup$ @farmaceut M052X and M062X are both highly parametrized Minnesota functionals. Try wB97XD which may be more accurate as it also includes dispersion corrections; it appears to be used a lot to model peroxyl radicals. $\endgroup$ Jun 20 at 17:09
  • 1
    $\begingroup$ Haven't checked the output in detail to see if there are any generally applicable lessons that can be gleaned, but I ran your reactants/products through Schrodinger's AutoTS and it produced a TS with the peroxyl perpendicular to the plane. $\endgroup$
    – Tyberius
    Jun 20 at 22:54
  • $\begingroup$ @SusiLehtola Need to keep with M052X. Have already tried to run wB97XD previously, alongside with B3LYP, and while the TSs have been found, upon reoptimization with the desired functional, they are no more!. $\endgroup$
    – farmaceut
    Jun 21 at 18:09

1 Answer 1


I solved the problem by reducing maxstep to 1 and using notrust and calcfc. The proposed transition states were simply very close to the real one and, as I assumed, the potential energy surface was very shallow hence any step greater than 0.01Bohr caused the basin to exit. However glum it worked.... Instead of 6 months, in less than 1 day I found what I was looking for.


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .