A Practical TS Optimization Workflow in Gaussian

February 2026 · Computational Chemistry

When I optimize a transition state (TS) in Gaussian, I rarely jump directly into a full TS search. I usually run constrained optimization first, then TS optimization, but in one continuous Gaussian job.

Workflow summary: constrain key coordinates in Link 1, then read the checkpoint and run TS optimization in Link 2.

One continuous job: constrained optimization + TS search

In the first link, I keep critical internal coordinates fixed while relaxing the rest of the structure with Opt=ModRedundant.

In the second link, I read the optimized structure from the checkpoint file and launch Opt=(TS,CalcFC,NoEigenTest,NoFreeze) for TS optimization.

Gaussian input example

%chk=ts.chk
%nprocshared=4
%mem=8GB
#p B3LYP/6-31G(d) Opt=ModRedundant

Title Card Required

0 1
... geometry ...

B i j F

--Link1--
%chk=ts.chk
%nprocshared=4
%mem=8GB
#p B3LYP/6-31G(d) Opt=(TS,CalcFC,NoEigenTest,NoFreeze) Freq Geom=AllCheck Guess=Read

Diels–Alder reaction transition state example

%chk=DA.chk
%nprocshared=4
%mem=8GB
#p B3LYP/6-31G(d) Opt=ModRedundant

DA

0 1
C           0.44704         1.43712         0.48957
H           0.12659         1.06846         1.45731
H           0.39419         2.51748         0.37881
C           1.32291         0.70351        -0.28987
H           1.87072         1.21327        -1.08117
C           1.32281        -0.70367        -0.28984
H           1.87053        -1.21353        -1.08115
C           0.44686        -1.43713         0.48965
H           0.12647        -1.06837         1.45738
H           0.39385        -2.51748         0.37893
C          -1.57675         0.69306        -0.22727
H          -2.07546         1.23594         0.57059
H          -1.47383         1.23605        -1.16002
C          -1.57680        -0.69292        -0.22738
H          -1.47383        -1.23577        -1.16021
H          -2.07561        -1.23590         0.57035

B 1 11 F
B 8 14 F

--Link1--
%chk=DA.chk
%nprocshared=4
%mem=8GB
#p B3LYP/6-31G(d) Opt=(TS,CalcFC,NoEigenTest,NoFreeze) Freq Geom=AllCheck Guess=Read

Why these TS options

TS: tells Gaussian to locate a first-order saddle point.
CalcFC: builds an initial force constant matrix for a better search direction.
NoEigenTest: avoids early failure when the initial Hessian is not ideal.
NoFreeze: ensures previously constrained coordinates are fully released in TS optimization.

Always verify the result with frequency analysis and IRC when needed: the TS should have exactly one imaginary frequency, and the IRC should connect to the expected reactant and product minima.

This linked-job workflow is not the only method, but it is often more controllable than a single-shot TS search, especially for medium-sized organic systems.