Crossing Bariers in condensed phases: Rare but important events

Introduction

In many body systems stable states are separated by a region of low probability. We view this process as a
crossing of a free energy barrier and we refer to this process as an activated process. When this barrier
is much higher than kT this event is an infrequent or {\it rare event} which cannot be simulated by
conventional techniques.

However, often this is the interesting or determining event.

Examples: chemical reactions,

different phases of a condensed material
different structures of a polymer
isomerization of a polypeptide chain

Transition State Theory (TST)

If the activated process can be described by a phenomenological rate equation then the associated rate
constants can be calculated within the reactive flux correlation function formalism.

Theory
(i) Classically, we often choose a suitable function of configuration space
that can describe the transitions between the stable states, the
{\it reaction coordinate}, which is a function of the set of coordinates
of all the degrees of freedom of the system. It describes the evolution of
the system from one stable state to another.

If we choose the surface carefully, then the equilibrium flux through this surface may be a good
approximation to the observed non-equilibrium rate. TST is based on the calculation of just that
equilibrium flux. To understand the success of this theory we need to
consider its underlying assumptions carefully:

(1) The rate can be calculated for a system in thermal equilibrium

(2) The rate is associated with the equilibrium flux across the boundary
separating ``reactants'' and ``products''

In fact assumption 2 requires that all equilibrium trajectories crossing

the barrier are reactive, that is go to the well defined final state. For
this to be satisfied two further conditions must be met.

Example: TST rate of escape from a one dimensional well

Consider an activated rate process represented by the escape of a particle from a one dimensional potential
well. The Hamiltonian of the particle is

p2
H= + V ( x) (1)
2m
where V(x) is a potential barrier peaked at x=xB separating the reactants (x < xB) from the products
(x>xB). Under the above assumptions the rate coefficient for the escape of a particle out of the well via the
transition state is give by

kTST =< v f > P( xB ) (2)

where P(xB)dx is the equilibrium probability that the particle is within dx of xB,
exp( EB )
P(x B ) = xB
(3)
dx exp( V ( x))

and <vf> the average forward velocity is given by

2
dvv exp(1/ 2 mv ) 1
< v f >= 0

=
2 2 m
dv exp(1/ 2 mv )

For a high barrier, most of the contribution to the integral in the denominator in Eq. 2 comes
From regions of the coordinate x for which V(x) is well represented by an harmonic well,
V ( x ) = (1/ 2)mo2 x 2 so

1
mo2 x 2
dxe 2 and hence

o E
kTST = e B

Comments:
The transition state is a product of a frequency factor that may be interpreted as the number of attempts per
unit time that the particle makes to exit the well and an activation term associated with the height of the
barrier. (Arrenius type behaviour). Note particularly that that rate is independent of the environment.
In order to yield the correct result, the potential used in the one dimensional starting point (Eq. 1) should
be the effective potential of the reaction coordinate, that is the potential of mean force along the coordinate
where all other degrees of freedom are in thermal equilibrium at the given state of the coordinate.
This treatment assumes that such a coordinate can be identified and that a point along this coordinate
separates reactants from products.
It is clear from the formulation that the TST rate is an upper bound to the exact rate. The reason for this is
that essentially the probability that an escaping equilibrium trajectory is indeed a reactive trajectory is
smaller than unity. This observation allows for choosing the diving surface between reactants and products
and then choose the one that maximises the TST rate (Variational Transition state theory).
Corrections to the TST arise from dynamical effects to the rate: when they coupling to the thermal
environment is either too large or too small. In the former case, the total outgoing flux from the reactants is
not a good representative of the reactive flux because most of the trajectories cross the dividing surface
many times.
Given all the above, it may by a surprise that TST works well. The reason is that it accounts quantitatively
for the most important factor affecting the rate: the activation energy.

Transition rate for a multidimensional well

The Hamiltonian for an N+1 dimensional system is now

N
pi2
H = + V ( x0 ,..., xN ) ()
i =0 2mi
or in terms of mass weighted coordinates and momenta

xi = mxi ; pi / m = x&i
V ( x0,..., xN ) = V ( x0 ,..., xN )

N
pi2
H = + V ( x N +1 ) ()
i =0 2mi
where XN+1 denotes the set (x0,xN)

Following similar steps as above, we can derive the effective one dimensional potential for the reaction
Coordinate.

Notes: Low dimensional systems can only relax when interacting with their environment. Deviations from
TST are larger when this is either too large or too small. If it is too small, then the assumption of
maintaining thermal equilibrium in the reactants breaks down. In the extreme case, the rate is controlled not
by the time it takes a thermal particle to traverse the barrier but by the time it takes to accumulate enough
energy to reach it. (Lindemann mechanism). On the other regime, when the interaction is too strong most
trajectories cross the dividing surface many times . In the extreme case, the system cannot execute any large
amplitude motion and the actual rate vanishes. (Kramers)

Rate of escape: Kramers Theory: Overdamped limit

H.A. Kramers, Physica 7, 284 (1940).

P. Hanggi, P. Talkner, M. Borkovec, Rev. Mod. Phys. 62, 251 (1990).

We can place a particle at the bottom of the well and running a trajectory of the stochastic equation that
governs its motion. We can compute the average escape time <>. In the theory of stochastic processes this
is called the mean first passage time: the average over all realizations of the stochastic processes over all
sampled trajectories of the time it takes to exit the well. The rate is

k =< > 1

or k = A1 ( dA / dt ) where A dxP( x, t )
well
The rate coefficient is well defined as a time independent constant when a steady state characterized by a
constant flux out of the well has been reached. For this to happen in an experimentally meaningful time,

E >> 1 should be satisfied. We therefore limit the discussion here to the high barrier limit.
A maintains its identity while being depleted, and hence A1 ( dA / dt ) is constant even though A is not.
We may now evaluate the rate by considering an artificial situation in which A is maintained strictly
constant by imposing a source at the bottom of the well and a sink outside it. Mathematically, it is more
convenient to put the source at x = and a sink at x = + . In fact this source does not have to be
described in detail. We simply impose the condition that the population (or probability) inside the well, far
from the barrier region is fixed, while outside the well we impose the condition that it is zero. Under such
conditions the system will approach at long time a steady state in which

P
=0 but J 0
t

The rate is given by J/A.

We start by writing the equation on the form

P ( x ,t )
t J =
x
# V $
J = D & + ' P( x, t )
( x x )
At steady state
# V $
J = D & + ' Pss ( x, t ) where
( x x )
J is a constant that can be viewed as one of the integration constants to be determined by the boundary
conditions. The case D=D(x) can be considered right away.

A solution of the form

Pss ( x ) = f ( x )e V ( x )

gives

df J V ( x )
= e
dx D( x )

x ' '
e V ( x ) V ( x )

' e
f ( x ) = J dx ' = J dx
x D( x' )

D( x ' )

integrating both sides from to xB finally leads to

1
$ xB
'
e V ( x ) %

J
k = xB = & dxe V ( x ) dx ' '
( x
D( x ' ) )
dxPss ( x)

Because we are in the high barrier case,

 (V ( xB ) V (0)) >> 1 . In this case the largest contribution to the inner integral comes from the
neighborhood of the barrier, so
exp[ V ( x)] exp[ EB (1/ 2)mB2 ( x xB )2 ]
and for the outer integral from the bottom of the well at x=0 so
exp[ V ( x)] exp[(1/ 2)mB2 x 2 ]

We finally obtain

1 1
% [ EB m B2 ( x xB )2 &
2 2 e 2
k = ' dxe 1/ 2 mo x dx (
'
D ( x B ) (
') (*
using D=(m)-1

we obtain

o B E
k= e = kTST B
B

Since we have considered the case where damping is faster than any other characteristic time of the
System (ie overdamped limit) the correction term is smaller than unity as expected.

Application: Single molecule force spectroscopy or Dynamic force spectroscopy

E. Evans, Annu. Rev. Biophys. Biomol. Struct. 30, 10528 (2001).

Rate of Escape Over an Idealized Barrier Under Force:

However, application of an external field or force f acts to select a reaction path, which can then be
represented by a scalar coordinate x. Analyzed along this coordinate, the outcome is a generic expression for
escape or forward-transition rate k that depends on how the energy landscape E(x) is deformed under the
applied field
From E. Evans
From D. Chandler