Transition states.

A transition state is a first order saddle point on a potential energy surface (PES). The
vibrational spectrum of a transition state is characterized by one imaginary frequency
(implying a negative force constant), which means that in one direction in nuclear
configuration space the energy has a maximum, while in all other (orthogonal) directions
the energy is a minimum. In order to verify if a stationary point is a transition state you
have to do a vibrational frequency calculation at the same computational level as you did
the geometry optimization. The normal mode corresponding to the imaginary frequency
in the transition state usually reflects the change in geometry in going from reactants to
products. Let me mention here there are more advanced techniques to verify if the
transition state connects reactants and products.

Locating transition states and estimating reaction barriers can be a rather difficult task.
The first step is to obtain a reasonable guess for the structure of the transition state. In
general our intuition is not as well developed for transition states as it is for ground states.
In addition transition energies can be close to excited states and this often implies that the
usual MO based computational methods may not be as accurate for transition states than
they are for minima on a PES. A fairly high level of theory is often required to get a good
quantitative estimate of a reaction barrier. In this lab we will be mainly focused on
locating the transition state and we will use the AM1 and HF levels of theory. Once you
have found the transition state at the AM1 level you can refine your calculations by re-
optimizing the structure at the B3LYP/6-31G* level or by simply calculating the
energetics at this higher accuracy level. Let us for now simply use the HF / 6-31G(d,p)
level in this set of labs.

As a first simple example of a transition state search consider the process of Hydrogen
transfer in

H
H C N C N C N H

In this case the minima (HCN and CNH) can be obtained in the usual fashion. You can
verify they are minima by calculating vibrational frequencies. Next provide a reasonable
estimate of the transition state geometry. You can change the HCN angle from 180˚ to
about 60˚ but you should also elongate the HC and NH bond distances somewhat. The
bonds you see in gaussview are only there for the user. They do not influence the
outcome of calculations which depend purely on the positions of the atoms. To calculate
a transition state you select OPT+FREQ under the jobtype, and then select both
Calculate force constants (once) and Locate transition
state. If you calculate a transition state you always want to calculate vibrational
frequencies also because you have to verify that you have precisely one imaginary
frequency. After the calculation finishes successfully you can view the vibrational
frequencies. Please note that one frequency is imaginary and that the corresponding

1
normal mode represents the hydrogen migration. Also examine the other normal modes
in the transition state. Finally locate the energy of reactants, products and transition state
in the respective *.log files / summary.

Let us look at another example. Consider the transition state for hydrogen migration in
formaldehyde

H H
H
C O C O
C O
H
H H

First optimize the reactants and products. To make the product you might build methanol
CH3OH first and then delete two of the Hydrogens using delete atom. Next provide a
guess for the transition state and submit the proper calculation (force constants, locate
transition state). It is required that the force constant matrix has one negative eigenvalue
at your initially specified geometry for the transition state. If this is not the case the
program stops. If you open the log file and go to the end, you will see an error occurred.
By browsing back a little bit you will see (for example)

Wrong number of Negative eigenvalues: Desired = 1 Actual= 2

This indicates your initial geometry is too far from the transition state. You might adjust
your initial guess or you can use the option OPT=NoEigenTest in the additional
keyword section in Gaussview. This is often counterproductive, as it avoids the test that
indicates your guess is not good enough. It would allow you to find a better guess for the
transition state, that you can then optimize completely later on if need be.

We can make use of an advanced feature in Gaussian/Gaussview to search for the

transition state. In a single “molecule group” we can place both the H2CO and HCOH
structures. You can copy and paste the optimized geometries in the appropriate window.
Use add to molgroup when creating a new file. You can see both structures if you click
on the top right button. Under Edit find the connection editor. This editor allows you to
lign up the atoms, such that Gaussian can recognize the similarity of the structures. You
can simply click Enable Autofixing and it will renumber the atoms appropriately. This
will be apparent in the window. At this point we can submit a calculation using the QST2
option for a transition state search. Also specify you want to calculate the force constant
matrix once. This means the program will find a guess for the transition structure itself.
Unfortunately the QST2 option is often only moderately useful. This is also the case here,
even for this fairly simple case. The program stops before it is done, and it tells you that it
terminates with an error. It is better not to open the file, upon the query, but to open the
.log file in the editor to see what it did. To see this you have to open the output file
(file.log), and go to the end. There it prints the error message. You have to get into the
habit of checking the output file. It is not infrequent to see calculations of transition states
fail, and then you will have to find another path to get it going.

2
It is better if you add a third structure to the molgroup, which is your guess for the
transition state. Choose the OH to be about 1.18 and CH bonds to about 1.23 and take an
OCH angle of 60 degrees. The HCOH dihedral angle should be about 170. That looks
reasonable. Again line up all three structures using the connectivity editor, and now
submit as optimizing a transition state, with the QST3 option. Also specify you wish to
calculate the force constant matrix once, at the beginning of the optimization. Again
check the output file, opening the .log file in the editor. If you have succeeded to
calculate the transition state we can calculate vibrational frequencies for the transition
state structure. It has one imaginary frequency, listed as -2641 cm-1 at the HF / 6-
31G(d,p) level. This let’s you know if you are on the right track.

Let us examine one more feature that can be very useful when examining transition
states, the Internal Reaction Coordinate or IRC. The IRC is related to the minimum
energy path (MEP), which is defined as the steepest descent path starting from the
transition state, initially going in the direction of negative curvature in the Hessian. The
path can be followed backwards (to reactants) and forward (to products). Following this
path can indicate if you really found the transition state that connects the two minima of
interest. The IRC is defined in terms of mass-weighted Cartesian coordinates and it has a
well defined step length. Computationally, once you have a found a transition state and
have obtained the force constant matrix you can start from the CHK file and ask for IRC
under the method keyword. By default it makes 6 small steps in either direction. You can
ask for more steps or increase the stepsize (default =10 in units of 0.01 amu 1/2-Bohr).
The latter is done by specifying on the command line input: IRC=(StepSize=30). After
the calculation you can open a set of structures, opening the resulting .chk file. Under
results <- IRC you can see the energy profile.

Some transition state structures have a higher symmetry than the ground state structure.
These transition states are usually easy to locate because the energy is a minimum if the
molecule is constrained to have the higher symmetry. If you start the optimization from
the higher symmetry this is maintained during the optimization. As a very simple
example consider the inversion of ammonia, which proceeds through a planar D3h
transition state.

H
H H
H
N H N N
H
H H H

Optimize both the ground state and the transition state using simple optimization
(imposing the proper symmetry) and calculate vibrational frequencies. Alternatively you
can select calculate force constants and locate transition state.
This is safer if your symmetry is only approximate. Locate the single imaginary
frequency and examine the corresponding normal mode. Verify that it corresponds to the
inversion mode. You can define the geometry of planar NH3 by setting the appropriate
dihedral angle.

3
As another example of the use of symmetry in transition state searches calculate the
ground state and transition state for hydrogen transfer in the enol form of malonaldehyde

H
H H
O
O O O
O O

The ground state has Cs symmetry while the transition state has C2V symmetry. Please
not that the normal mode corresponding to the transition state breaks the symmetry (any
non-A1 normal mode will reduce the symmetry). You can start building this structure
using aromatic carbons, and adding a Hydrogen atom between the Oxygens in the end.
Under edit you can find the feature pointgroup. You will want to manipulate your
structure, such that it recognizes the C2v symmetry. You will want to impose this
symmetry. Playing with the tolerance helps, to make the program recognize this
symmetry. Searching for the minimum will work if you have initially specified a solid
C2V geometry. The symmetrize button is very useful to enforce the desired symmetry. In
C2v symmetry you can simply optimize for a minimum, and immediately calculate
frequencies. Verify through measurements that the structure is indeed symmetrical, and
that the negative (i.e. imaginary) frequency corresponds to Hydrogen migration.

Summarizing notes on finding transition states.

As you may have noticed finding transition states can be rather complicated even for
simple systems. Let me give you some further guidlines.

Characterization: A transition state is characterized by precisely one negative force

constant (to be precise: one negative eigenvalue of the Hessian or matrix of second
derivatives). If you have obtained a structure that has one negative force constant you are
in business. The program will usually locate the transition state.

1. Try to build a good guess for the transition state. In particular check the relevant bond
distances. You (often) are breaking one or more bonds so typical distances of the
bonds that are being broken are a little longer than usual.
2. You can start from either the reactants or product state to build the transition
structure. The transition state is usually somewhere in between and this may guide
your guessing of certain structural parameters.
3. Do not use the "clean" option in Gaussview because this will typically get you back
to reactants or products.
4. If possible use symmetry to define your structure. You may have to be creative to
tease the program into building the appropriate structure. The pointgroup option
under edit is very useful in this respect. We have had two examples (NH3 and
hydrogen transfer in malonaldhyde).
5. the QST2 and in particular QST3 options may be helpful to locate transition states.

4
6. When submitting the calculation under Jobtype select "Opt", or "Opt+Freq" (if
you’re confident!) and then select both "Calculate Force Constants
(once)" and "Locate Transition State".

It often happens that your calculation returns quickly with an error message. This means
that the initial guess did not satisfy the criterium of one negative eigenvalue. You can
view the Log file and see that at the end of the file you do not see
Normal termination of Gaussian 98 (preceded by a quote to keep up your
spirits). If you scroll up a little bit in the output file you will see the message:

Wrong number of Negative eigenvalues: Desired = 1 Actual= 0

This indicates your strucure is not close enough to the transition state. If Actual= 0
you are close to a local minimum still. You will have to adjust your structure for the
optimization to the transition state to proceed.

Excercises: additional examples of transition states.

1. In a later lab we will explore electrophilic aromatic substitution. The following

structure is an intermediate (hence a minimum) for the ortho addition of Cl+ to aniline

NH2

H
+
Cl

First find this structure (don't forget to make the charge equal to +1), which is a minimum
on the PES (no imaginary frequencies). Use the AM1 level of theory. There is a transition
state in which Cl+ approaches anniline and it is somewhere bridging carbons 2 and 3 (the
NH2-C is numbered carbon 1), and there is a transition state where H+ leaves the chlorine
substituted species and it is located on top of carbons 2 and 3. The transition state is more
like connecting the chlorine and hydrogen shift reaction products. Locate both transition
states, using what ever tools you need to use (QST3, calculating force constants in every
step). Save your results as they will be used later on. Use the AM1 level of accuracy
throughout.

2. Optimize cis- and trans-butadiene at the HF/6-31G(d,p) level. Make a guess for the
transition structure, taking a dihedral angle of about 90 degrees, and locate the transition
state, using the QST3 tool. Once you have obtained the transition state calculate
vibrational frequencies and then perform an IRC calculation that shows the path
connecting the Cis and trans structures. Use a stepsize of 30 and sufficient steps to
unambiguously establish the reaction path.

5
3. Find the transition state for the reaction H2CO → H2 + CO, by making a guess for the
transition state. Also calculate the energies of reactants and products, all at the HF/6-
31G(d,p) level. Combine this with our previous results for the transformation H2CO →
HCOH and its transition state. Make an energy profile, listing all minima and the
transition states that connect them. Make an IRC for both reactions and verify that you
have found the proper transition states. List the relevant geometrical features and total
energies in a table. To make this a more meaningful calculation we would need to include
zeropoint energy effects. We will save this for later.

4. Consider both [1,5] and [1,3] sigmatropic shifts in cyclopentadiene, again using AM1
level of theory.
Cl Cl
Cl H
H

H [1,5]
H [1,3]

Find the transition state for the [1,5] sigmatropic shift. Also plot the orbitals that are
involved in the hydrogen transfer reaction at all stages in the reaction (reactants, products
and transition state). You will see that the [1,5] shift is suprafacial. The [1,3] shift would
require an antarafacial hydrogen migration and this can be clearly seen by examining the
orbitals for the reactant. Try to find a transition state for the [1,3] shift and locate the
transition state. Explain what happened during your search. The [1,3] transition is quite a
tough problem. See how far you get.