4.2 Finding the ammonia inversion transition state
These are unexecuted teaching inputs and starting models. Original diagrams are schematics, not calculated results. Validate version-specific syntax, licensed or authorized data, numerical convergence and the scientific model before using this workflow.
4.2.1 Model, units and provenance
Geometry in Å; electronic energy in hartree; vibrational wavenumbers in cm⁻¹. Check each printed field and keep thermal and standard-state terms distinct.
Shared inputs, conventions and evidence
4.2.2 Unexecuted inputs and explicit deltas
Use the accompanying instructions to identify the parent calculation and placement of every delta; a snippet is not automatically a standalone input. Preserve all blank-line and file-provenance requirements.
4.2.2.1 Input block 1
%chk=case14_nh3_ts.chk
%mem=4GB
%nprocshared=2
#p B3LYP/6-31G(d) Opt=(TS,CalcFC,Tight,Cartesian,MaxCycle=100)
Freq SCF=Tight Integral=UltraFine NoSymm
Temperature=298.15 Pressure=1.0
Ammonia inversion saddle from original near planar coordinates
0 1
N 0.000000 0.000000 0.040000
H 1.010000 0.000000 0.000000
H -0.505000 0.874686 0.000000
H -0.505000 -0.874686 0.000000
4.2.2.2 Input block 2
%chk=case14_nh3_min_control.chk
%mem=4GB
%nprocshared=2
#p B3LYP/6-31G(d) Opt=(Tight,CalcFC) Freq
SCF=Tight Integral=UltraFine NoSymm
Temperature=298.15 Pressure=1.0
Ammonia pyramidal minimum control
0 1
N 0.000000 0.000000 0.360000
H 0.940000 0.000000 0.000000
H -0.470000 0.814064 0.000000
H -0.470000 -0.814064 0.000000
4.2.3 Worked investigation
4.2.3.1 Question and intuition
How can an optimizer locate a mountain pass instead of falling into a valley? A transition-state search follows a saddle-point objective: it seeks a stationary structure with one downhill vibrational direction and upward curvature in the others. Ammonia inversion is an accessible example. The nitrogen moves through the plane of the three hydrogens while the molecular “umbrella” turns inside out. The pyramidal minima and nearly planar transition structure have the same connectivity and electronic state, so the exercise focuses on geometry and curvature rather than bond breaking or spin changes.
Prerequisites are a successful minimum optimization and the ability to animate a normal mode. NH3 contains ten electrons and is modeled as a neutral singlet. The original near-planar coordinates below intentionally begin close to the expected saddle region, with nitrogen slightly displaced from the hydrogen plane. They are guesses, not a claim that the exact transition geometry or barrier has already been computed. The hydrogen order must remain fixed in Cases 15–16 so that a signed inversion coordinate can be followed consistently.
4.2.3.2 Complete transition state input
See input block 1 above.
TS selects the saddle search, CalcFC computes the initial Hessian, Tight tightens convergence, and Cartesian avoids relying on an internal-coordinate representation near a planar umbrella. NoSymm prevents symmetry constraints from blocking the inversion motion. Freq is still necessary after optimization: an optimizer's evolving approximate Hessian does not replace a final frequency calculation. NoEigenTest is deliberately omitted; suppressing the curvature test is not a beginner's default remedy for an unsuitable guess.
4.2.3.3 Independent minimum control
Run this separate control to establish a pyramidal minimum at the identical model chemistry. It is useful for the provisional electronic barrier, but the stronger connectivity evidence will come from both IRC endpoints in Case 15.
See input block 2 above.
4.2.3.4 Workflow
- View both initial structures side by side and confirm atom ordering and N–H connectivity. Measure the nitrogen height above the hydrogen plane rather than only the absolute Cartesian z coordinate, which can change under reorientation.
- Run the minimum and TS jobs independently. In each log, verify SCF convergence and geometric convergence; a large maximum cycle count only permits more iterations, not success.
- At the TS, count six internal vibrational modes for this nonlinear four-atom system. Exactly one should be imaginary and the remaining five should be real. At the minimum, all six should be real. Count internal modes rather than translation/rotation diagnostics.
- Animate the imaginary mode. It must correspond to the umbrella inversion, with nitrogen and the hydrogen framework moving relative to one another. A mode describing an unrelated distortion would validate a different saddle, not this proposed pathway.
- Compare ETS with Emin in hartree, then convert their difference. A positive electronic barrier is the qualitative expectation. Do not substitute this difference for ΔG‡ or quote it as an observed result until both outputs are validated.
- Preserve the completed TS frequency checkpoint untouched. If the search finds a minimum, an incorrect mode, or multiple imaginary modes, inspect the mode vectors and revise the near-planar guess or numerical settings, then rerun. Launch the IRC only after the one-mode and convergence tests pass.
4.2.3.5 Checks pitfalls and exercise
A planar-looking image is not proof of a transition state. The evidence is a small gradient, one negative Hessian eigenvalue in the vibrational subspace, the right displacement pattern, and subsequently the correct connections. Likewise, “one imaginary frequency” does not prove that every conceivable reaction mechanism was considered. This model's simplicity makes diagnosis easier; a difficult multidimensional reaction may require scanning, QST2/QST3, or other methods for obtaining a good saddle guess.
Ordinary unlabeled ammonia does not become a different isolable chemical compound upon inversion, and its pyramidal equilibrium structure is not a chiral molecule with three different substituents. The two wells are equivalent configurations along a signed coordinate. That equivalence becomes a stringent endpoint check rather than a reason to skip the reverse path. Record the printed rotational symmetry numbers for later entropy analysis; using NoSymm makes it especially important to verify rather than assume those statistical factors.
EN exercise: Change only the initial nitrogen height from +0.040 Å to −0.040 Å and repeat the TS search with a new checkpoint. Explain why the validated saddle energies should agree within numerical tolerance. Then start an ordinary minimum optimization from each side and compare the resulting signed heights.
4.2.4 Related calculations
- 4.1 Computing a balanced reaction energy at a consistent level
- 4.3 Following both IRC directions and validating endpoints