IRC calculations in Quantum Chemistry

[brydustin]

I was wondering if I do an IRC calculation and the energy range of the IRC does not quite reach the energy of the reactant or product of the system, do I need to make the IRC calculation "wider" (i.e. go more forward and reverse) so that I can interpolate the reaction coordinate of the system given the known energies of the reactants and products?

 

[SpectraCat]

I was wondering if I do an IRC calculation and the energy range of the IRC does not quite reach the energy of the reactant or product of the system, do I need to make the IRC calculation "wider" (i.e. go more forward and reverse) so that I can interpolate the reaction coordinate of the system given the known energies of the reactants and products?

[@Moderator: This should be moved to computational physics, or perhaps chemistry.]

What are you using for your basis of comparison to decide that the IRC "does not quite reach the energy of the reactant or product of the system"?

Is the value from the IRC higher or lower?

Are you sure the end points of your IRC calculations correspond to stable minima (i.e. no imaginary frequencies)?

 

[brydustin]

Energy (Relative to Transition as computed from IRC in Hartrees) = -0.02306;
with Rx Coordinate = -4.29864
Compared to:
Reactant with -0.028400 energy (Relative to Transition as computed in IRC in Hartrees)

I didn't do a frequency calculation because this requires more time; but I had the idea that I should probably want to find what reaction coordinate the reactant would coorespond to (however most of the literature doesn't actually concern itself with this); and I believe that the reaction coordinate is a bit of an abstraction anyway... i.e. what would it mean to interpolate a reaction coordinate for the reactant anyway?
I just figure it would be nice to label the IRC graph so that one could identify the products on a certain part of the graph and the reactants on the other.

I can't be sure that there would be no imaginary frequencies because I suppose I could have set the number of calculations ( # of reaction coordinates) to be higher (but that's precisely my question, I don't know if I should or not). I set it to do 70 (35 forward, 35 back; and that seems like quite a bit when compared to the default).
I agree this should be in computational physics, I click on the wrong link (sorry about that).

My basis for comparison was the energy which was calculated separately for a reactant energy/geometry optimization.

 

[SpectraCat]

Energy (Relative to Transition as computed from IRC in Hartrees) = -0.02306;
with Rx Coordinate = -4.29864
Compared to:
Reactant with -0.028400 energy (Relative to Transition as computed in IRC in Hartrees)

I didn't do a frequency calculation because this requires more time; but I had the idea that I should probably want to find what reaction coordinate the reactant would coorespond to (however most of the literature doesn't actually concern itself with this); and I believe that the reaction coordinate is a bit of an abstraction anyway... i.e. what would it mean to interpolate a reaction coordinate for the reactant anyway?
I just figure it would be nice to label the IRC graph so that one could identify the products on a certain part of the graph and the reactants on the other.

I can't be sure that there would be no imaginary frequencies because I suppose I could have set the number of calculations ( # of reaction coordinates) to be higher (but that's precisely my question, I don't know if I should or not). I set it to do 70 (35 forward, 35 back; and that seems like quite a bit when compared to the default).
I agree this should be in computational physics, I click on the wrong link (sorry about that).

My basis for comparison was the energy which was calculated separately for a reactant energy/geometry optimization.

Well, if you can give me some more information, perhaps I can help you more. I do agree that 35 points in each direction sounds like a lot. The different of 5 millihartree between your IRC calculation and the separate reactant calculation (I am assuming you did this using the same method and basis set) is also too large to ignore. Without knowing more, I suspect that you probably need to go further along the reaction coordinate in each direction in order to reach the minima. Another possibility might be that your IRC reaction coordinate does not actually intersect with the minimum on the potential energy surface representing your reactant. Have you looked at the differences in the geometries between your IRC endpoint and the reference calculation? If the molecule is distorted along a normal coordinate that is orthogonal to your IRC coordinate, then your IRC trajectory will never reach that point.

One final point is that you might consider doing a geometry optimization starting from the end-point of your IRC calculation. If that converges to the same energy and geometry as your reference calculation, then you can safely conclude that there are no significant barriers to rearrangement on the potential energy surface between your IRC geometry and the "real" stable minimum found by your reference calculation.

With regard to the interpretation of the IRC trajectory back toward the reactants, the physical significance of that is that it represents the reaction coordinate for the reverse reaction.

 

[brydustin]

Cool... thanks. I'm not sure what you mean by distorted in a orthogonal coordinate though (do you mean that a bond length, angle is off)? Its the "distorted" part I don't understand.
To answer your question, the geometry calculation and irc were both done using b3lpy/6-311(d,p), no frequency calcualation was used in either. This is why I feel that perhaps I should calculate more.
I'm really interested though about what you mean by normal/orthogonal distortion (I'm aware of what orthogonal means) and why this might cause the IRC pathway to not intersect the local minimum (reactants). [it might be note worthy now for me to point out that the reactants' energy (-0.028400 H) is the sum of two reactants, E(A) + E(B); whereas the IRC calculation will just calculate E(A-----B),-0.02306 H, where it pulls them far apart, right?] This might cause the issue, no?!
i.e. When I sum the energy of the reactants its as if they were calculated at infinite distances apart, right? but for practical purposes the numbers should be sufficiently close, right?