Transition States and

Reaction Paths
Computational Chemistry lab
2013

Potential Energy Surface (PES)

A 3N-6 -dimensional hypersurface in, where N is the
number of atoms.

Saddle points

Minima equilibrium
structures of one or
several molecules

Saddle point - Transition state

Minimal maximum
maximum in one direction, minimum in other directions

Transition State
Arrhenius equation

k = Ae

E A / kbT

k - rate constant
EA activation energy
T temperature
kb Boltzmann constant

Multidimensional Optimization

- gradient
The coordinates can be
Cartesian (x,y,z..),
or internal, such as
bondlength and angle
displacements)
Force: F=- E (Potential energy)
f=0 - Stationary Point (minimum, maximum, or saddle point)

Hessian matrix
A matrix of second-order derivatives
of the energy with respect to atomic
coordinates (e.g., Cartesian or internal
coordinates)
Sometimes called force matrix
matrix size of (3N-6)x(3N-6)

H ij ( f ) =

2 f
xi x j

Quadratic approximation
Approximate the complex energy landscape by harmonic potentials
r
r( st)
r( st) r r( st) 1 n n
r( st)
E( x) = E( x ) + E( x )(x x ) + 2 Hij ( x )(xi xi( st) )(x j x(jst) )
i =1 j =1

around a stationary point xr ( st ) = ( x1( st ) ,... xn( st ) )

r ( st )
[ E ( x ) = 0]

Quadratic approximation
Approximate the complex energy landscape by harmonic potentials
r ( st )
( st )
( st )
around a stationary point x = ( x1 ,... xn )
n

( st )
1

E ( x1 ,... xn ) = E ( x

,... x

( st )
n

r ( st )
[ E ( x ) = 0]

r ( st )
) + H ij ( x )( xi xi( st ) )( x j x (j st ) )
1
2

i =1 j =1
n
(k )
k j

Hessian matrix diagonalization (eigenproblem) l

= H ij li( k )
i =1

2
Eigenvalues k = mk

k - vibrational frequencies
n

Eigenvectors l (kj ) give normal coordinates qk = li( k ) ( xi xi( 0 ) )

n

E ( x1 ,... xn ) = E ( x1( st ) ,... xn( st ) ) + 12 k qk2

k =1

i =1

2
k - vibrational frequencies
Eigenvalues k = mk
n
(k )
Eigenvectors l j give normal coordinates qk = li( k ) ( xi xi( 0 ) )
i =1

E ( x1 ,... xn ) = E ( x1( st ) ,... xn( st ) ) + 12 k qk2

k =1

All k > 0(k real )

minimum

q1

All k < 0(k imaginary) k <> 0(k imaginary/real )

maximum

q2

saddle point

Transition state 1<0, k>0 (k>1)

Eigenvector l1- the path direction

Algorithms for Finding Transition

States
No general methods which are guarantied to find!
Global methods interpolation between reactant and product
Linear Synchronous Transit (LST)
Quadratic Synchronous Transit (QST)
Local methods augmented Newton-Raphson
Eigenvector following
Berny algorithm
Synchronous Transit-Guided Quasi-Newton (STQN)
method QST+quasi-Newton
Force-field parameters in molecular mechanics are defined for
equilibrium structures and can be inapplicable to transition structures.

Quantum calculations only

(recent semi-empirical potentials are applicable too)

Linear synchronous transit (LST)

-search for a maximum
along a linear path between
reactants and products

Reactants

LST
Saddle point

Products

Linear synchronous transit (LST)

- search for a maximum along a
linear path between reactants
and products

Reactants

Quadratic synchronous transit

(QST) - search for a maximum
along an arc connecting reactants
and products, and for a minimum
in all directions perpendicular to
the arc
Best case - the transition state is
found.
General case search is finished in
a wrong saddle point or in a point
with wrong number of negative
eigenvalues (>1).

LST
Saddle point
QST
Products

Newton(-Raphson) method:
Quadratic approximation : E ( R ) = E ( R 0 )
R=(x1,...xn) - (3N-6) -dimensional vectors

n
2
1
(
F
q

q
k k 2 k k)
k =1

qk- normal coordinates, R R 0 =

q l , E (R
k k

n
0

) = Fk l k

k =1

k =1

k and lk - eigenvalues and eigenvectors of the Hessian matrix at R0

Quasi-Newton no explicit computation

of the Hessian matrix.
One-step search for
qk = Fk / k
quadratic functions

k =1

Finds the closest stationary point (either

minimum, maximum, or saddle point).
Transition state 1(R)<0, k(R)>0 (k>1)

For arbitraryn functions

R i +1 = R i + l k (R i ) Fk ( R i ) / k ( R i )

R ts

Rational function optimization

(RFO)
n

Fk (R i )
R i +1 = R i + l k ( R i )
k ( R i ) k
k =1
Berny algorithm (Gaussian, Opt=TS)

1=min(k) < k<2/2

Eigenvector Following Method
(Gaussian, Opt=EF; Hyperchem)

1 = (1 + 12 + 4 F12 ) / 2
k > 1 : k = Fi 2 /( i )
i >1

Clamber up valley

Clamber up valley

Synchronous Transit-Guided Quasi-Newton (STQN)

approach to the quadratic region quadratic synchronous transit
complete the optimizationeigenvector-following algorithm

Reactants

Gaussian, Opt=QST3: input

reactants, products, and estimate
of transition state
Gaussian, Opt=QST2;
Hyperchem: input reactants and
products only, automatic
estimate of transition state

TS estimate
Eigenvector following
Saddle point
QST
Products

Is the correct TS found?

Look at the transition state geometry to make
sure its the right one.
Use several algorithms.
Try several estimates of transition state.
Follow reaction path to be sure that the
transition state connects the correct
reactants and products.

Reaction Paths
Steepest descent path from transition state to reactants
and products
IRC-internal reaction coordinate
Gaussian - IRC
New point xk+1
minimization on
the hypersphere
surface