Chemical Kinetics

and Transport

Chemical Kinetics

and Transport
Peter
C. Jordan
Brandeis University
Waltham, MasSllchusetts

Plenum Press . New York and London

Library of Congress Cataloging in Publication Data

Jordan, Peter C
Chemical kinetics and transport.
Includes bibliographical references and index.
1. Chemical reaction, Rate of. 2. Transport theory. I. Title.
QD501.J75731979
541'.39
ISBN-13: 978-1-4615-9100-9
e-ISBN-13: 978-1-4615-9098-9
DOT: 10.1007/978-1-4615-9098-9

78-20999

First Printing - March 1979

Second Printing - January 1980
Third Printing - September 1981

1979 Plenum Press, New York

Softcover reprint of the hardcover 1st edition 1979
A Division of Plenum Publishing Corporation
233 Spring Street, New York, N.Y. 10013

All righ ts reserved

No part of this book may be reproduced, stored in a retrieval system, or transmitted,
in any form or by any means, electronic, mechanical, photocopying, microfilming,
recording, or otherwise, withou t written permission from the Publisher

Preface
This book began as a program of self-education. While teaching undergraduate physical chemistry, I became progressively more dissatisfied with
my approach to chemical kinetics. The solution to my problem was to
write a detailed set of lecture notes which covered more material, in greater
depth, than could be presented in undergraduate physical chemistry. These
notes are the foundation upon which this book is built.
My background led me to view chemical kinetics as closely related to
transport phenomena. While the relationship of these topics is well known,
it is often ignored, except for brief discussions of irreversible thermodynamics. In fact, the physics underlying such apparently dissimilar processes
as reaction and energy transfer is not so very different. The intermolecular
potential is to transport what the potential-energy surface is to reactivity.
Instead of beginning the sections devoted to chemical kinetics with a
discussion of various theories, I have chosen to treat phenomenology and
mechanism first. In this way the essential unity of kinetic arguments, whether
applied to gas-phase or solution-phase reaction, can be emphasized. Theories
of rate constants and of chemical dynamics are treated last, so that their
strengths and weaknesses may be more clearly highlighted.
The book is designed for students in their senior year or first year of
graduate school. A year of undergraduate physical chemistry is essential
preparation. While further exposure to chemical thermodynamics, statistical
thermodynamics, or molecular spectroscopy is an asset, it is not necessary.
As a result of my perspective, the choice of topics and organization of
the material differs from that of other kinetics texts. The introductory
chapter, on equilibrium kinetic theory, is mainly review. The next two
chapters, treating transport in both gas and solution, introduce many
v

vi

Preface

chemically useful concepts. Chapter 4 focuses on phenomenology-the

experimental methods used to determine a rate law, the meaning of rate
laws and rate constants. The next three chapters are devoted to mechanisms of thermal and photochemical reaction, first in systems that are
stable to perturbation and then in those for which explosion or oscillation
is possible. Much of this material is in the standard repertory. However,
the sections devoted to heterogeneous catalysis, infrared laser photochemistry, and oscillating chemical reactions incorporate the results of
recent, occasionally speculative, research. Chapter 8 treats single-collision
chemistry-first, to relate the reaction cross section and the rate constant)
and then, to demonstrate how mechanistic information can be deduced from
scattering data.
The first of the theoretical chapters (Chapter 9) treats approaches to
the calculation of thermal rate constants. The material is familiar-activated
complex theory, RRKM theory of unimolecular reaction, Debye theory of
diffusion-limited reaction-and emphasizes how much information can be
correlated on the basis of quite limited models. In the final chaptpr, the
dynamics of single-collision chemistry is analyzed within a highly simplified
framework; the model, based on classical mechanics, collinear collision
geometries, and naive potential-energy surfaces, illuminates many of the
features that account for chemical reactivity.
I have tried to present both the arguments and the derivatioflS."so that
they can be readily followed. However, some steps have been intentionally
omitted. I believe that unless a student works through the material individually, the subject matter cannot be mastered. Examples have been chosen
to illustrate general principles, not to be exhaustive. The problems, which
are an indispensable adjunct to each chapter, provide further illustrations.
While the arrangement of the material is not traditional, each chapter,
and many of the sections, are, to a great degree, self-contained. Topics
may, if the instructor so desires, be presented in any order.
I am indebted to many individuals for their assistance. Foremost
among these are my colleagues, Michael Henchman and Kenneth Kustin,
who clarified my ideas and suggested many of their own; without their
help this book would never have been written. The reviewers, Lawrence
J. Parkhurst and Robert E. Wyatt, made many helpful and perceptive suggestions. Students at Brandeis University have rendered considerable assistance, especially in pointing out ambiguous or unclear passages. Finally I
am grateful to Evangeline P. Goodwin for her rapid and accurate typing and
to Virginia G. Steel for her excellent draftsmanship.
Peter C. Jordan

Contents
Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . ..
1.

Kinetic Theory of Gases-Equilibrium

1.1. Introduction
1.2. The Maxwell-Boltzmann Distribution of Velocities
1.3. Determination of Pressure
1.4. Properties of the Maxwell Distribution
1.5. The Molecular Flux and Effusion of Gases .
Problems
General References

2.

1
2
6
9
13
14
15

Kinetic Theory of Gases-Transport

2.1. Introduction . . . .
2.2. Mean Free Path . . . . . . .
2.3. Collision Frequency. . . . . .
2.4. Macroscopic Equations of Transport
2.5. Solution of the Transport Equations
2.6. Kinetic Theory of Transport and Postulate of Local Equilibrium
2.7. Simplified Transport Theory: Dilute Hard-Sphere Gas.
2.8. Comparison with Experiment . . . . . . . . . . .
Appendix. Coordinate Transformation by Jacobian Method.
Problems . . . . .
General References

3.

xi

17
18
20
24
26
29
30
35
44
45
47

Electrolytic Conduction and Diffusion

3.1.
3.2.

Introduction . . . . . . . . . .
Ionic Conduction in Solution-a Review.
vii

49
50

viii

Contents
3.3. Nernst-Einstein Relation
3.4. Electrolyte Diffusion
3.5. Stokes' Law and the Microscopic Interpretation of AD
3.6. The Mobility of H+ and OH- in Water
3.7. The Concentration Dependence of A
3.8. The Wien Effects
Problems . . . . .
General References

4.

69

Determination of Rate Laws

4.1. Introduction . . . . . .
4.2. Stoichiometry, Rate Law, and Mechanism . . . . . . .
4.3. Elementary Rate Laws and the Principle of Mass Action.
4.4. Reaction Order and Molecularity . . . . . . . .
4.5. The Principle of Detailed Balance . . . . . . .
4.6. Experimental Methods for Determining Rate Laws
4.7. Integrated Rate Laws-First Order
4.8. Isolation Methods . . . . . . .
4.9. Relaxation Methods . . . . . .
4.1 o. Determination of Rate Law from Data-Two Examples
4.11. Temperature Dependence of Rate Constants
Appendix. Linear Least-Squares Analysis
Problems . . . . .
General References

5.

53
56
57
61
62
65
67

71
72
73
77
78
80
85
87
89
93
98
101
103
11 0

Stationary State Mechanisms

5.1. Introduction . . . . . .
5.2. Consecutive Reactions-First Order. . .
5.3. Consecutive Reactions-Arbitrary Order.
5.4. Unimolecular Decomposition-Gases . .
5.5. Radical Recombination-Gases
5.6. Complex Mechanisms-Gaseous Chain Reactions.
5.7. Simple Mechanisms-Solution .
5.8. Complex Mechanisms-Solution
5.9. Homogeneous Catalysis
5.10. Enzyme Catalysis
5.11. Heterogeneous Catalysis
Appendix A. Matrix Solution to the Coupled Rate Equation (5.2)
Appendix B. Approximations to "+ and ,,_
Problems . . . . .
General References

113
114
120
124
127
131
135
139
142
145
147
153
156
157
163

ix

Contents
6.

Photochemistry
6.1.

Introduction

6.2. Measurement of Intensity

6.3. Spectroscopic Review
6.4. Primary Processes
6.5. Secondary Processes
6.6. Chemiluminescence.
6.7. Orbital Symmetry Correlations: Woodward-Hoffman Rules
6.8. Laser Activation . . . . . . . . .
6.9. Laser-Controlled Isotope Separation
Problems . . . . .
General References

7.

N onstationary State Mechanisms

7.1.

Introduction . . . . . . . .

7.2. Thermal Explosion . . . . .

7.3. Population Explosion-Autocatalysis
7.4. The Lotka Problem: Chemical Oscillations.
7.5. The Belousov-Zhabotinskii Reaction . . .
7.6. Other Oscillating Systems . . . . . . . .
7.7. Multiple Stationary States: The NO.-N.0 4 Reaction
Problems . . . . .
General References

8.

165
166
167
169
178
184
185
188
192
195
199

201
202
205
209
214
220
226
229
232

Single-Collision Chemistry
8.1.

Introduction . . . . .

8.2. Reaction Cross Section: Relation to the Rate Constant

8.3. Scattering Measurements-Molecular Beams .
8.4. Reaction Cross Section: Hard-Sphere Model . .
8.5. Reaction Cross Section: Ion-Molecule Systems
8.6. Reaction Cross Section: Atom-Molecule Systems
8.7. Product Distribution Analysis . . . . . . . .
8.8. Direct, Forward-Peaked-The Ar+ + D. System
8.9. Recoil-The D + Cl. System . . . . . .
8.10. Collision Complexes-The 0 + Br. System
8.11. State Selection Experiments
Problems . . . . .
General References

233
234
237
240
244

248
252
254
257
259
261
263
267

Contents

x
9.

Theories of Reaction Rates

9.1. Introduction . . . . .
269
270
9.2. Potential-Energy Surfaces .
276
9.3. Activated Complex Theory.
9.4. Model Calculations of k.(T)
281
9.5. The Kinetic Isotope Effect .
287
291
9.6. Unimolecular Reaction-General Features
9.7. Unimolecular Reaction-RRKM Theory
293
9.8. Critique of RRKM Theory-Isomerization of Methyl Isocyanide 299
302
9.9. Thermodynamic Analogy
304
9.10. Gas-Phase Reactions
306
9.11. Solution Reactions . . .
307
9.12. Primary Salt Effect . . .
309
9.13. Diffusion-Limited Kinetics-Debye Theory
Appendix A. Statistical Thermodynamics . . . .
316
Appen4ix B. Evaluation of koo in RRKM Theory
318
Problems . . . . .
319
General References
323

10. A Dynamical Model for Chemical Reaction

10.1. Introduction . . . . . . . . . . .
10.2. Kinetic Energy of a Triatomic System . .
10.3. Model for Inelastic Collision . . . . . .
lOA. Effects Due to More Complex Potential-Energy Surfaces .
10.5. Model for Thermoneutral Chemical Reaction
10.6. Exoergic Reactions . . .
10.7. Endoergic Reactions
10.8. Noncollinear Geometries
10.9. Realistic Scattering Calculations
Problems . . . . .
General References

Physical Constants
Index . . . . . .

325
326
328
333
335
339
342
345
348
351
353

355
357

Notation
Those symbols which are only used once in the text do not appear in this
listing. Familiar mathematical functions are also not included, nor are
quantities that are simply intermediates in a derivation. Capital letters are
listed before lowercase letters. Greek letters are listed at the end.
A

ABC

Ci
Cv
D
D

Di

FA(C)

E
Ea
E
F
FA(u, v, w)

FA(C)
L1Go+
G{J,(u)

Arrhenius A-factor, reaction affinity, energy acceptor in excitation transfer reaction

Product of principal moments of inertia of a
nonlinear molecule
Normality of the ith component
Constant volume heat capacity
Energy donor in excitation transfer reaction
Dissociation energy for dynamical model of
Chapter 10
Diffusion coefficient of the ith component
Energy
Arrhenius activation energy
Electric field
Faraday's number
Three-dimensional velocity distribution function
for molecules of species A
Speed distribution function for molecules of
species A
Gibbs free energy of activation
x component of molecular flux as a function of
x component of velocity
xi

Notation

xii

iJHo+

Enthalpy of activation

J Ionic strength, moment of inertia of a linear

h
J(v)
J(J.., x)
JA(x)
J
Ji
Jq
K(T), K(T, p), K(T, p, I)

K+
Km
M
MA
No
NA
Nt
N( E,,)

Px
P( E)
Q

QA

Q+
Q(E, i,j), Q(E), Q(E)

82 Q( E, e, 1> )/8e 81>

R
Rj (Rr)

Si
iJ So t
T

Ti
VA

molecule
Intensity of a beam of molecules of species A
Normalized fluorescence intensity at frequency v
Position-dependent light intensity at wavelength J..
Position-dependent intensity of a beam of molecules of species A
Rotational quantum number
Diffusive flux of the ith component
Heat flux
Equilibrium constants
Equilibrium constant for activation
Michaelis constant
Total mass
Molecular weight of species A
Avogadro's number
Number of particles of species .A
Hypothetical number of activated complexes
Density of states for the ath degree(s) of freedom
of an activated molecule with energy E~ in the
specified degree(s) of freedom
Momentum flux in x direction
Probability density for states of energy E
Enthalpy release per mole of reaction
Molar partition function for species A
Reduced molar partition function for the activated complex
Reaction cross sections as functions of molecular
states i, j, collision energy E, or molar collision
energy E
Differential cross section for molecules with
collision energy E scattering into solid angle 1>
Gas constant, total reaction rate
Rate of forward (reverse) reaction
The ith singlet state of an electronic manifold
Entropy of activation
Absolute temperature
The ith triplet state of an electronic manifold
Potential energy of a molecule of species A

e,

Notation

xiii

V
Vcoll

L1 vot
X, Y

Yi
[Yd
ZA
ai

at
b
C
Co

Ci
Ci

CM

tv
d (dAB)

e
erf(x)
fA (c)
f
g~

h
ji
k
ken), kn
kd
kdiff

(k diss )

koo (k o)
kef:)
me
mi

Volume, limiting velocity of an enzyme-catalyzed

reaction
Collision volume of crossed molecular beams
Volume of activation
Relative coordinates for collinear motion in the
dynamical model of Chapter 10
Symbol denoting the ith reactant
Concentration of the ith reactant
Total collision frequency for molecules of species A
Activity of the ith component
Hypothetical activity of the activated complex
Impact parameter
Molecular speed
Velocity of light
Concentration of the ith component
Velocity of the ith particle
Velocity of the center of mass
Constant volume heat capacity per molecule
Hard-sphere diameter (for an A-B collision)
Electronic charge, energy per mole
Error function
Three-dimensional velocity distribution for molecules of species A as a function of speed
Frictional force on a particle
Degeneracy factor for the ath degree(s) of
freedom
Planck's constant
Current density attributable to the ith ionic
species
Boltzmann's constant, rate constant for reaction
Rate constants for reaction
Rate constant for irreversible decomposition
Rate constant for association (dissociation) in a
diffusion-limited reaction
Limiting rate constant at high (low) pressure in
unimolecular decomposition
Energy-dependent rate constant for reaction
Electron mass
Mass of the ith particle

xiv

Notation
ni

Number density of the ith component

n+ Hypothetical number density of the activated

P
Pi
p(x)

pCb, E, i, j), pCb, E)

qA

q+
qrx
r
r rx
,i

'M

u
u(r)

v
w

x, y
x, y, z
Zi

ZA(C)

rex)
L1

'l'

complex
Pressure
Partial pressure of the ith component
Position-dependent scattering probability for
molecules in a beam
Reaction probability as functions of molecular
states i, j, collision energy E, and impact parameter b
Molecular partition function for species A
Reduced molecular partition function for the
activated complex
Partition function for the ath independent
degree(s) of freedom
Distance in spherical polar coordinates
Rate of the ath step in a reaction sequence
Position of the ith particle
Position of the center of mass
Time
x component of velocity, mobility in an electric
field
Intermolecular potential function
y component of velocity, volume per mole,
reaction velocity in an enzyme-catalyzed reaction, vibrational quantum number
Drift velocity in an external field
z component of velocity
Position components for the equivalent particle
in the dyIiamical model of Chapter 10
Position components in Cartesian coordinates
Charge of the ith ionic species
Speed-dependent collision frequency per molecule of species A
Gamma function
Reaction exo- or endoergicity for dynamical
model of Chapter 10
Equivalent conductance of the ith electrolyte
(at infinite dilution)
Time- and position-dependent wave function

xv

Notation

ai(A)

fJ
Y
Yi
y+

Y(A)

+
.j(v)
1]

()

()i

"
"i

AA
Ai (AiO)
fli

fli
flij

v
vA
Vi

Molar absorption coefficient of the ith component at wavelength A

IjkT, angle characterizing mass ratio in dynamical
model of Chapter 10
Thermal coupling constant for dissipation of
heat to surroundings
Activity coefficient of the ith component
Hypothetical activity coefficient for the activated
complex
Light attenuation coefficient per unit length at
wavelength A
Energy per molecule, dielectric constant, energy
parameter in Lennard-Jones potential
Classical barrier height per molecule in activated
complex theory
Molar extinction coefficient for the ith component at frequency v
Shear viscosity
Polar angular velocity III spherical polar coordinates, polar angle in spherical polar coordinates, angle of deflection, angle characterizing energy partitioning in dynamical model
of Chapter 10
Fractional surface coverage attributable to adsorption of the ith component
Thermal conductivity, transmission coefficient
Decay constant for the ith relaxation process
Wavelength, decay constant for an independent
kinetic process
Mean free path of species A
Equivalent conductance of the ith ionic species
(at infinite dilution)
Chemical potential of the ith component
Electrochemical potential of the ith component
Reduced mass for relative motion of particles i
and j
Frequency
Number of atoms of type A in a compound
Stoichiometric coefficient of the ith reactant,
vibrational frequency of the ith normal mode

Notation

XVI

v+ Frequency defined by curvature at the peak of

~(t)

~(E)

ei
ei(E)

eiu)
a

r(A)
cp

CP;.
CPI (cpQ)

X
1p

the reaction profile

Time-dependent progress variable
Depletion factor in the theory of unimolecular
reaction
Mass density of the ith component
Density of quantum levels in electronic state i at
energy E
One-dimensional velocity distribution
Specific conductivity, hard-sphere diameter for
a collision, length parameter in Lennard-Jones
potential, symmetry number of a molecule
Intrinsic lifetime of a photoexcited state
Azimuthal angular velocity in spherical polar
coordinates, azimuthal angle in spherical polar
coordinates, angle of deflection
Quantum yield at wavelength A
Fluorescence (phosphorescence) quantum efficiency
Polar angle in spherical polar coordinates
Electric potential, position-dependent wave function
Oscillation frequency of a kinetic process, azimuthal angle in spherical polar coordinates

Kinetic Theory
of Gases-Equilibrium
1.1. Introduction
The kinetic theory of gases represents the first truly successful effort
to construct a model which provides a mechanical basis for understanding
the properties of bulk matter. By combining the molecular hypothesis
with Newtonian mechanics and using a statistical approach, the theory
provides an explanation for the thermodynamic similarities between dilute
gases, makes numerous verifiable predictions, and suggests a number of;
unifying principles. In its simplest form the theory is based upon the following premises:
(1) a gas is composed of an enormous number of molecules;
(2) in the absence of external forces particles move in straight lines(l);
(3) particles collide infrequently;
(4) collisions, whether with the walls or with other molecules, are
elastic (momentum and kinetic energy are conserved).

In a dilute gas, collisions, although relatively infrequent, are very

important. They insure that molecules move in all directions with a continuum of speeds. Rather than attempting to determine the velocity of each
molecule (a hopeless undertaking considering that collisions continually
alter the velocities), one describes the gas by a velocity distribution. Such a
(1)

The interparticle forces actually cause trajectories to bend as molecules approach one
another. However, quite accurate predictions follow if billiard ball dynamics are assumed.

Chap. 1

Kinetic Theory of Gases-Equilibrium

description, which focuses upon the behavior of classes of molecules,

eliminates reference to the properties of individual molecules. While the
statistical picture provides only a limited amount of information about a
gas, it is in fact a description that corresponds precisely to the available
experimental data. (2)

1.2. The Maxwell-Boltzmann Distribution of Velocities

A macroscopic sample of a gas contains an enormous number of
molecules continually colliding with one another and with the walls of the
container. The effect of these collisions is to change the velocity of the colliding particles. The fundamental problem in kinetic theory is to determine
the distribution of velocities of the molecules in a gas. The answer was
first given by Maxwell for a gas in thermal equilibrium. Indicative of the
importance of this problem is the fact that, over a period of years, Maxwell
and Boltzmann presented four different derivations of the basic equation,
each somewhat more sophisticated than its predecessor. We shall consider
the two simplest ones.
The velocity of a molecule, c, is specified by its components u, v,
and w in the x, y, and z directions, respectively. If we were to attempt
to find the number of molecules with a specific velocity c, we would have
an immediate problem of measurement. The more precisely c can be
measured, the fewer the molecules that will be found with the prescribed
velocity. Rather than attempt unnecessary accuracy, (3) we focus on the
quantity F(c) = F(u, v, w), where F(u, v, w) du dv dw is the fraction of
molecules with velocity components between u and u + du, v and v + dv,
wand w + dw; the total number of molecules in this velocity domain is
then NF(u, v, w) du dv dw. If du, dv, and dw are chosen too small, we again
encounter the problem that few molecules will lie in the velocity range of
interest and the quantity F( c) might not be a smooth function of c. However,
as long as the system of interest is macroscopic, the velocity increments
may be prescribed as small as measurement permits; there is always an
enormous number of molecules in the specified velocity domain and F( c)
is a smoothly varying function.
(') Experiments using molecular beams provide a direct measure of the effect of interparticle collisions. Interpretation of such data requires the analysis of molecular trajectories
(Chapter 8).
(3) In fact there are limits, given by the uncertainty principle, as to how precisely one may
specify molecular velocities (problem 1.3).

Sec. 1.2

The Maxwell-Boltzmann Distribution

The quantity F(u, v, w) is the velocity distribution function. It is not

a probability but rather a probability density and represents the fraction of
molecules with the velocity components u, v, and w per unit volume in a
three-dimensional velocity space. (4) Since the total probability is I, the
distribution is normalized:

1=

f +OO f+oo f+oo F(u, v, w) du dv dw

-00

-00

-00

(1.1)

By integrating over y and z components of velocity we can compute ex(u) du,

the fraction of molecules with x component of velocity between u and
u+ du:
eiu)

f +OO f+oo F(u, v, w) dv dw

-00

-00

(1.2)

Naturally ex(u) is also normalized,

1=

f +OO eiu) du
-00

(1.3)

as is seen by combining (1.2) and (1.1).

As yet F(u, v, w) is unspecified; it may be determined using Maxwell's
original arguments. (5) Experience suggests that, as long as gravitational
effects are unimportant, a gas at equilibrium is isotropic. The orientations
of the coordinate axes are immaterial; the axes may be rotated arbitrarily.
Thus the velocity distribution cannot depend upon u, v, or w in any way
that distinguishes one direction from another. The only molecular property
with no orientational dependence is the magnitude of the velocity c =
(u 2 + v 2 + W 2 )1I2 so that an alternative description of the velocity distribution is as a function of conly, f(c). The equivalence of the descriptions
requires that
F(u, v, w)

f(c)

(l.4)

In addition, Maxwell introduced the simple hypothesis that information

about the distribution of x components of velocity provides no insight
into the distribution of either y or z components, i.e., the distributions in
(4)

(5)

The concept is analogous to the familiar idea of spatial density. The number of molecules in a volume element limited by x and x + dx, y and y + dy, and z and
z + dz is dN = e(x, y, z) dx dy dz; e(x, y, z) is the number density, i.e., the number
per unit volume.
J. C. Maxwell, Phil. Mag. 19, 19 (1860).

Chap. 1

Kinetic Theory of Gases-Equilibrium

u, v, and ware uncorrelated and F(u, v, w) can be factored:

F(u, v, w)

ex(u)evCv)ez(w)

(1.5)

The form of these functions is specified completely by (1.4) and (1.5).(6)

Take logarithms and differentiate both sides of these equations with respect
to u: the result, making use of the chain rule when differentiating lnf(e), is

( a In F(u, v,

au

din ex(u)
du

w) )
11,W

u dlnf(e)
e
de

(1.6)

Dividing both sides of (1.6) by u yields

din exCu)
du

d Inf(e)
de

(1.7)

If u is fixed the left-hand side of (1.7) is a constant, -2A. However, e may

still be altered independently by varying v or w; no matter what value e
takes, A does not change so that the quantity c 1 dlnf(e)/de is independent
of e. Thus f(e) is the solution of the differential equation

dlnf(e)
de

-2Ae

which is
Inf(e)

-Ae 2

+ In B

or in exponential form

fee)

B exp( -Ae2 )

(1.8)

The rationale for the negative sign in the exponential is now clear; in
this way infinite velocities are forbidden. Factoring (1.8), the separate
distributions in u, v, and w can be obtained; ex(u) is
(1.9)
with similar expressions for ey(v) and ezCw). Evaluation of A and B requires
further information. The normalization (1.1) or (1.3) provides one condition; in the next section the other is found.
Before determining A and B let us derive (1.8) in a way that does not
presume the factorization property (1.5). Consider collisions in a dilute
(6)

A more sophisticated derivation, not based upon (1.5), was formulated a few years later
[J. C. Maxwell, Trans. Roy. Soc. 157, 49 (1867)]. It is presented later in this section.

Sec. 1.2

The Maxwell-Boltzmann Distribution

gas of identical molecules. As interparticle distances are large only a relatively few molecules are interacting at any instant; of these interactions
the overwhelming majority involve pairs of particles. The trajectories of a
colliding pair are determined by the intermolecular forces between them.
As long as the forces are short range the details of the force law are unimportant. (7) The physical significance of such collisions in dilute gases
is that they provide a mechanism for momentum transfer between molecules.
In a typical collision molecules approach with initial velocities CI and C2 ,
interact, and then separate with final velocities c l ' and c2'. The details of
the interaction need not concern us; we only need to know the initial and
final velocities. Limiting consideration to structureless identical particles,
energy conservation requires that(S)
(1.10)
For the process considered the rate must be proportional to the fraction
of molecules of each velocity, i.e., to F(c l )F(c 2 ). The rate of the reverse
process, in which molecules initially with velocities c/ and c2 ' emerge with
velocities CI and C2 , is analogously proportional to F( cl')F( c2 '). In a system
at equilibrium the rate of any process and its inverse are equal. (9) Since,
as we shall now demonstrate, the proportionality constants determining
the two rates are equal, detailed balance requires that
(1.11)
which, with (1.10), has the unique solution (1.8).
To establish that the proportionality constants are the same, consider
the collision processes as viewed by an observer moving with the centerof-mass velocity (c I + c 2 )/2 = (c l ' + c2 ')/2. Combining this with (1.10)
we find that the relative speed is the same before and after collision,
I C I - C 2 I = I c/ - c 2 ' I = C 12 Collision accomplishes nothing more than
(7)

(8)

(9)

.r;;,

Short-range forces are those for which the integral

u(r)4nr 2 dr is finite [here u(r)
is the intermolecular potential]; ro is arbitrary. Ordinary van der Waals interactions,
which fall off as r- 6 , are short range. On the other hand, coulombic forces, for which
u(r) ex r-\ are long range. The consequence is that even in a dilute plasma one must consider the motion of the ions to be correlated.
Such processes, in which kinetic energy is conserved, are known as elastic processes.
In general, when the effect of molecular structure is considered, collisions are inelastic,
i.e., there is interconversion of translational and rotational or vibrational energy. Only
inelastic collision processes can lead to chemical reaction.
The correctness of this assertion, known as the principle of detailed balance, is not
immediately apparent. It will be discussed more fully in Chapter 4.

Chap. 1

(0)

(b)

(e)

(d)

Kinetic Theory of Gases-Equilibrium

Fig. 1.1. Dynamics of a binary collision

and its inverse as viewed in lab coordinates (a, b) and center-of-mass coordinates (c, d). In this example m. = 2ml
and cl ' = O. Note that in center-ofmass coordinates the collision and its
inverse are simply related by rotation
about the center of mass.

to reorient the relative velocity vector. Thus, in the center-of-mass system

two processes lead to the changes shown in Fig. 1.1. In each case the
particles are deflected through the same angle. There is no physical'difference between them. Thus, barring the effect of walls or interactions with
a third particle, there is no reason for preferring either. The absolute rates
must be equal and (1.11) is established.
To show that (1.10) and (1.11) are indeed equivalent to (1.8), specialize
to the case that cI ' = O. Then I c2 ' I = (C12 + C 22 )1I2 and with (1.4) we have
(1.12)
Using arguments similar to those which led from (1.5) to (1.8) again
establishes the form of f(c).

1.3. Determination of Pressure

The pressure of a gas is the force per unit area exerted due to molecular
collisions with the walls of the container. Since force is momentum change
per unit time, pressure is determined by computing the momentum transfer
per unit time per unit area. Consider a wall at x = L; if at each collision
there is perfect reflection the x component of velocity changes from u to -u
and the net momentum transfer is 2mu. While perfect reflection is not a
reasonable assumption we know that the gas has an isotropic velocity
distribution and from (1.8), F(u, v, w) = F( -u, v, w). Thus, on the average,
the fraction of molecules leaving the walls with an x component of velocity
-u is the same as that striking the wall with x component of velocity u.
The net effect is a mean momentum transfer of 2mu.
To determine the contribution such collisions make to the pressure
we must compute the number of molecules with x component of velocity

Sec. 1.3

Determination of Pressure

J-x

Fig. 1.2. Collision cylinders for molecules with different velocities VI and V.,
but the same x component of velocity,
striking a target area A in a time .dt.

between u and u + du that strike an area A in a time LIt. From (1.2) the
total number of such molecules is Nex(u) duo Only a small fraction of these
hit the wall during a limited time. From Fig. 1.2 we see that the molecules
within the x distance u LIt hit the wall during LIt. The values of v and w
are irrelevant. (10) The fraction that hits the designated area is the ratio of
volume, Au LIt, of the collision cylinder to the volume, V, of the container,
so that the total number of molecules providing a momentum transfer
2mu to the area A in the time LI t is
Au LIt

- V - Nex(u) du

(1.13)

The momentum transferred by these molecules is

Au LIt
(2mu) - V - Nex(u) du

Integrating over all u greater than zero (only molecules moving toward
the wall may collide) yields the total momentum transfer at x = L; dividing
by A and LIt we obtain the pressure
p

Since eiu)
(10)

= ex(-u)

VN

2m

foo u2ex(u) du
0

the range of the integration can be extended to

Molecules within this distance may collide with other molecules before striking the wall.
Although kinetic energy and momentum are conserved (collisions are elastic) each
molecule will generally change its x component of velocity. Simultaneously (i.e., during
.dt) molecules with different x components of velocity will collide and one of the collision products may then have an x component u. The rate of these processes must
be equal at equilibrium. Otherwise the distribution i!x(u) would be time dependent,
which, as we have seen, it is not. Thus, on balance, for each molecule with component u initially within u.dt of the wall, a molecule (not necessarily the same one) with
component u will strike the wall during .dt.

Chap. 1

f::

Kinetic Theory of Gases-Equilibrium

minus infinity, which indicates that

pV = Nm

u2ex(u) du

Nmu 2

(1.14)

where u2 is the average value of u2. As isotropy requires that u2 =

v2 = w2 = c2/3 an alternate form is
pV = tNmC2

= jNE = jE

(1.15)

where E is the average kinetic energy per molecule and E the total kinetic
energy. Dilute gases are ideal; thus pV = NkT, which, with (1.15), yields
two fundamental results relating mean kinetic energy and mean molecular
speed to temperature,

tmC2 =

ikT,

E= iNkT

(1.16)

The first of these defines the root mean square (rms) speed
Crms

= (C 2)1I2 = (3kTlm)1/2

(1.17)

To obtain the parameters A and B substitute (1.9) in (1.3) and (1.14).

The resulting integrals are
1

Bl/3

p VI N = kT =

f~ exp( -

BlI3

Au 2 ) du

f~ mu2 exp( -

(1.18)

Au 2) du

which are examples of those commonly encountered in applications of

kinetic theory. They are of the general form

2(AB+l)1I2

0OO

US exp( -Au2) du

Some basic properties of the

rex

+ 1) =

J'' O X(S-1)/2 r
0

dx

S + 1 )
== r (--2-

(1.19)

r function are
r(1) = 1,

xr(x),

(1.20)

from which one can establish the important special cases

+ 1) =
r(n + t) =
r(n

(11)

nt,
(2n)!

0,1,2, ...
n

0, 1,2, ... (11)

(1.21)

For an extensive discussion of the r function including a tabulation of values of r(x),

for 1 ::; x ::; 2, see M. Abramowitz and I. A. Stegun, eds., Handbook of Mathematical
Functions (New York: Dover Publications Inc., 1965), pp. 253-293.

Sec. 1.4

Properties 0/ the Maxwell Distribution

Using these results we find that

which indicate that

A = m/kT,

B1I3 = (m/2nkT)1I2

Now, using (1.8) and (1.9) the velocity distribution function is completely
specified:
(lx(U) = (m/2nkT)1I2 exp( -mu 2/2kT)
(1.22)
F(c) = F(u, v, w) = (m/2nkT)3!2 exp( -mc 2/2kT)

1.4. Properties of the Maxwell Distribution

The one-dimensional velocity distribution is plotted for N2 at 100 K
and 300 K in Fig. 1.3. As a consequence of the isotropy of the gas it is
an even function of u, i.e., (lx(u) = (lx( -u). At low temperature the distribution is more sharply peaked, a corollary of the fact that higher molecular speeds are less probable at low temperatures.
The three-dimensional velocity distribution can be expressed in a
number of ways. We have, so far, considered a Cartesian description of the
velocity vector. If, however, we switch to spherical polar coordinates in
which
w = cos
u = c sin ecos cp,
v = c sin sin cp,

2.5

200

400

600

800

u (ms')

Fig. 1.3. One-dimensional velocity distribution function for N2 at 100 K and 300 K.

Chap. 1

10

Kinetic Theory of Gases-Equilibrium

the fraction of molecules with speed between c and c + dc oriented in a

solid angle between e and e + de and between q; and q; + dq; can be found
using (1.22) and the relationship between differential volume elements in
Cartesian and spherical polar coordinates

F(c, e, q;) dc de dq;

(mJ2nkT)3/2 exp( -mc2/2kT)c 2 sin e dc de dq;

(1.23)

In many cases the molecular orientation is not of interest and we need

consider only the distribution of speeds

F(c)

f "f2" F(c, e, q;) de dq; =

0

4n(m/2nkT)3/2c2 exp( -mc2/2kT)

(1.24)

This distribution is plotted in Fig. 1.4 in terms of the dimensionless parameter c(M/2RT)1/2. Such a plot adjusts the speed scale to account for effects
due to mass and temperature variation.
The maximum in the speed distribution, which occurs at the most
probable speed cmp (M/2RT)1/2 = 1, corresponds to speeds of 422 m S-1
and 244 m S-1 for N2 at 300 K and 100 K, respectively; plotted on the same
scale the speed distribution at 100 K would be more sharply peaked than
that at 300 K, just as was the case for the one-dimensional velocity distribution of Fig. 1.3. The maximum is due to two opposing effects. As the speed
increases, the exponential factor f( c) decreases, reflecting the consideration
that a higher speed means a higher energy; on the other hand the factor
4nc 2 increases, reflecting the fact that higher speeds can be attained in many
2

cmed

3
cJM/2RT

Fig. 1.4. Speed distribution F(c) as a function of dimensionless speed. Various measures
of molecular speed are indicated.

Sec. 1.4

Properties of the Maxwell Distribution

11

more ways than lower ones. (12) The result is the most probable speed,
which is found by calculating dF(c)/dc and setting it equal to zero. This
measure of molecular speed is somewhat different from the rms speed
already calculated, another consequence of having a distribution. A third
common measure of speed is the average (or mean) speed, c, given by the
cF(c) dc; using the formulas (1.19)-(1.21) yields
integral

f:

c=

(SRT/nM)1I2

(1.25)

intermediate between cmp and Crms'

In addition to the various averages that can be obtained, the speed
distribution may be used directly to determine the fraction of molecules
within a given speed range. As F(c) dc is the probability that a molecule
has a speed between c and c + dc, the fraction of molecules with speeds
between Cl and C u is
CU F(c) dc

CI

In terms of the dimensionless parameter, Y

required fraction is

= c/cmp =

c(M/2RT)1/2, the
(1.26)

demonstrating that the ratio of the molecular energies of interest to the

thermal energy of the gas determines this property of the gas. The integral
is related to a common transcendental function, the error function(13)
erf(x)

2 fX exp( _y 2 ) dy
--m

F or many purposes the fraction with speeds above the cutoff c I is of interest.
Integrating by parts and setting Yu =
the result is

n:/2 [Yl exp( -yl)

=

(12)

(13)

f:

exp( _y 2 ) dy]

--m
Yl exp( -YI2) +
n

[1 - erf(YI)]

(1.27)

Molecules with speed c have their velocity vector c on the surface of a sphere of
surface area 4:n:c 2 The larger the value of c the larger this area is.
This function can be looked up in mathematical tables; it is defined so that erf(O) = 0
and erf(oo) = 1. Using these tables is little more trouble than using ordinary tables
of trigonometric functions. See M. Abramovitz and I. A. Stegun, eds., Handbook of
Mathematical Functions (New York: Dover Publications Inc., 1965), p. 310.

Chap. 1

12

Kinetic Theory of Gases-Equilibrium

which can be used to provide yet another measure of molecular speed, the
median speed; by definition half the molecules move faster and half slower
than the median. To compute the median refer to tables of the error function
and find the value of yz for which (1.27) equals 0.5; the result is YI = 1.0876
or Cmed = 1.538(RT/ M)1I2. The relative values of the four measures of
molecular speed are then
Cmp : Cmed :

C:

Crms =

1 : 1.088 : 1.128 : 1.225

which are also indicated in the plot of F(c). When the cutoff speed is large,
(1.27) is simplified. For 10% accuracy the integral term can be ignored
whenever Y I ;C 3; if 1% accuracy is required Y I ;C 7.
The dimensionless parameter which characterizes the speed distribution
is the ratio of the molecular kinetic energy to kT. Thus we may determine
a Maxwellian energy distribution by defining E = mc 2/2 from which dE =
mc dc and (1.24) becomes
F(e) de

4n(m/2nkT)3/2(2E/m)1I2 e -dkT dE/m

= F(E) dE

2n(l/nkT)3/2E1I2e-<lkT dE

(1.28)

which measures the fraction of molecules with energy between E and E + dE.
A plot of F(E) would be similar to that of F(c) in Fig. 1.4. Of more
interest is the fraction of molecules with energies greater than a specified
value Eo. This quantity, F*(Eo), is foc
F(E) dE. Integrating by parts the
<0
result, expressed in terms of a dimensionless parameter a o = Eo/kT, is
(1.29)

which could equally well be obtained from (1.27) by setting Yz = (ao)1I2

and Y = a 1l2 . It is plotted in Fig. 1.5; note how rapidly F*(a o) decreases
for a o > 1. The fraction of molecules with energy greater than 6.9 x 10- 21 J
(which is kT at 500 K) is indicated for a number of temperatures; it falls
precipitously at the lower temperatures. While (1.29) and Fig. 1.5 precisely
describe only a three-dimensional dilute gas, the qualitative behavior is
general, extending even to the molecular energy distribution in liquids
and solids, which has important ramifications, particularly in chemical
kinetics. As only molecules with energies greater than a threshold energy,
Eo, can react this suggests that the reaction probability should be proportional to exp( -Eo/kT). Thus a basis for understanding the enormous
temperature dependence of the rate of chemical reactions can be provided.

Sec. 1.5

Molecular Flux and Effusion of Gases

13

lOOK

Fig. 1.5. Fraction of molecules with energies greater than Eo. The fractions with energy
greater than 6.9 x 10- 21 J at various temperatures are indicated.

1.5. The Molecular Flux and Effusion of Gases

A direct measure of molecular speed can be obtained from the process
of effusion. In such an experiment a pinhole is made in the wall of a container
and gas leaks into an evacuated collecting vessel. The rate of gas flow is
determined from the pressure in the collection chamber. If the pinhole
is sufficiently small so that leakage does not perturb the gas in the container,
the effusion rate is the same as the rate at which molecules collide with a
wall. In our determination of the pressure we computed this quantity.
The molecular flux, G(u) du, is defined as the number of molecules
with x component of velocity between u and u
du which strike unit area
in unit time, a quantity already calculated in Section 1.3. Using the same
arguments which led to (1.13) yields

A LltG(u) du

(Au LltJV)Nez(u) du

G(u) du

(NJV)uez(u) du

or
(1.30)

For effusion into vacuum there is no return flow and the total flux is found
by integrating over all values of u greater than zero (only molecules with

Chap. 1

14

Kinetic Theory of Gases-Equilibrium

these velocities are moving toward the hole). Using (1.22) and (1.19)-(1.21)
the total flux is

G=

VN foo0 uex(u) du =

nc/4

(1.31)

with n = N/ V. If the area of a pinhole can be measured, an effusion experiment provides a direct measure of the mean speed; if not, effusion
can be used to compare the mean speeds of different gases, (14)

GI

nIcI

7J; = n2 c2

nl (TI /m l )1/2
n2(T2/m 2 )1/2

(1.32)

Problems
1.1.

Show that (1.12) implies (1.8).

1.2.

A gas is in a container which undergoes uniform translational motion in

the x direction with velocity Vx , e.g., the container is in an airplane. Assuming that the gas is at equilibrium, express the Maxwell distribution with
respect to a stationary origin, e.g., the airport.

1.3. A I-liter cube contains N2 at 300 K and 10- 3 atm. Using the uncertainty
principle compute the minimum uncertainty in the specification of x component of velocity u. Accepting' this quantity as the minimum possible
value of .d u, compute the fraction of molecules with x component of velocity
between u and u + .du for (a) u = ii = 0, (b) for u = Urms = (kT/m)1/2.
How many molecules are in the two velocity domains?
1.4.

Show that the speed distribution (1.24) and the energy distribution (1.28)
are both normalized.

1.5.

What fraction of molecules in He and CO have kinetic energies greater

than 1.0 eV at 500 K and at 2000 K?

1.6.

In the treatment of surface phenomena the speed and energy distributions

in two dimensions are of interest. Transform to polar coordinates u =
c cos rp, v = e sin rp and show that
F(u, v) dv du = (!x(u)(!y(v) du dv
"" F(c, rp) de drp = (m/2nkT) exp( -mc 2/2kT) e de drp

(14)

The effect was put to practical use in the Manhattan project. In order to construct an
atomic bomb 235U had to be separated from its far more abundant isotope 238U. This
was originally accomplished by building a huge gas-effusion system for the separation
of gaseous 2lI5UF. from .38UF.

15

General References
1.7.

Integrate F(c, q;) dc dq; over all q; and show that the two-dimensional speed
distribution is
F(c) dc

(m/kT) exp( -mc 2/2kT) c dc

Now transform to energy as a variable and obtain the two-dimensional

energy distribution
F(E) dE = (l/kT)e-<lkT dE

c,

1.8.

For a two-dimensional gas what are c mp ,

1.9.

What is the average translational kinetic energy of particles effusing through

a pinhole? (Assume that the idealized conditions for an effusion experiment
are met.)

Cmed,

and crma ?

1.10. Given the following isotopic abundance

H: 0.99972
235U: 0.00715

D: 0.00028
23BU: 0.99285

compute the isotopic composition when the following naturally occurring

isotopic mixtures effuse through a pinhole:
(a) molecular hydrogen
(b) uranium hexafluoride
1.11. To construct an atomic bomb based on uranium, the fuel must be nearly

pure 23U. How many passes through an effusion apparatus are required
to enrich naturally occurring uranium to 95% 235U? to 99% 235U? The
effusing gas is UF 6

General References

s.

Golden, Elements 0/ the Theory 0/ Gases (Reading, Mass.: Addison-Wesley, 1964),

Chapters 3 and 4.
W. Kauzmann, Kinetic Theory o/Gases (New York: Benjamin, 1966), Chapter 4.
J. E. Mayer and M. G. Mayer, Statistical Mechanics (New York: John Wiley, 1940),
pp. 1-18.
R. D. Present, Kinetic Theory 0/ Gases (New York: McGraw-Hill, 1958), Chapters 1,2,
and 5.

Kinetic Theory
of Gases-Transport
2.1. Introduction
According to the kinetic model, molecules move very rapidly. At 300 K
the rms speed of gaseous bromine is 125 m sec-I. Yet, if a small amount of
bromine is added to nitrogen at 300 K and I atm the rate at which the
bromine diffuses through the gas is much slower, about 10-3 m sec-I. The
presence of the nitrogen impedes the movement of the bromine, reducing
the effective rate of travel by a factor of 105 from that which occurs in the
absence of nitrogen. The most obvious possible cause for such retardation
is the effect of molecular collision.
As long as the dimensions of the container are large enough so that
molecules undergo collisions in traversing the vessel, a simple, although
imperfect, analogy is found in the new amusement park ride of miniature
hovercraft supported above a rink by air jets. Imagine that all drivers are
blindfolded, the steering wheels of their craft locked so that they move in
straight lines, and that they are required to cross the arena. If there are
few craft the chances are excellent that no collisions will occur and that
they will traverse the floor without incident. As more craft participate
collisions become more likely and the distance traveled between them decreases. Each collision redirects the craft, which goes charging off on its
new path almost surely to collide with another vehicle. The possibility of
making headway is enormously reduced and the time required to cross
the floor increases greatly. On the molecular level the difficulties are much
the same. Thus, the time required for a molecule to traverse a container
17

Chap. 2

18

Kinetic Theory of Gases-Transport

Direction
of be om of
A molecules

Fig. 2.l. Model for calculating the A-B

collision probability when a beam of A
molecules is directed at a gas of B molecules. One A-B pair is shown making a
glancing collision. The dashed circles indicate the effective A-B collision diameters.

AX

of gas is related to the distance traveled between collisions or, equally well,
to the collision frequency.

2.2. Mean Free Path

An approximate determination of the mean free path, i.e., the average
distance traveled between collisions, can most simply be done by considering
the following experiment in which a beam of A molecules, originally of
intensity h(O), is attenuated as it passes through a gas containing B molecules. The beam intensity at any point in the gas is hex). To compute hex)
requires determining the probability, p(x) dx, that an A molecule is scattered from the beam between x and x + dx. The decrease in intensity
(the number ejected from the beam) in this interval is the product of the
scattering probability and the intensity
dIA(x)

-h(x)p(x) dx

(2.1)

To compute p(x) consider a cross section of the gas of thickness Llx, as

shown in Fig. 2.1. The B molecules provide the targets for the beam of
A molecules. Assuming that the molecules are rigid spheres(1) of radii
rA and rB, the effective area of each target, known as the collision cross
section, is nCrA + rB)2; stated differently, if the path of the center of an
A molecule passes within a distance r A + rB of the center of a B molecule,
a collision occurs. The probability of a collision is the ratio of the area
covered by target molecules to the area of the beam, a. If B is so dilute
that individual target areas do not overlap, this probability is

(1)

The hard-sphere assumption is a considerable restriction, and most certainly an unrealistic one. See footnote (1) in Chapter l.

Sec. 2.2

Mean Free Path

19

where nB is the number density of B and dAB = rA + rB; the number of B

molecules in the specified volume is nBa Llx. Simplifying the expression
yields p(x), the probability per unit distance of a molecule in the beam
colliding with the target gas,
(2.2)
Combining (2.1) and (2.2) we obtain

dlA

dx = - Th(x)

which has the solution

(2.3)
In the beam experiment the mean free path, AA, is the average distance
traveled by an A molecule before it undergoes a collision. Thus AA is the
integrated product of the fraction of A molecules ejected between x and
x + dx multiplied by the distance the molecules have traveled:
AA -

A ]
f x~oo [--dl
-- X

x~O

h(O)

Using (2.1), (2.2), and (2.3) the result is

AA

fool_A_ _ X dx =
f oo p(x) [leX)]
o

IA(O)

e-xllx dx = I

(2.4)

If the target gas were a mixture of A and B the analysis would have to
be generalized to account for A-A collisions. Then p(x) and I would no
longer be given by (2.2). Assuming the target gas mixture is dilute enough
so that the collision probabilities are additive the appropriate generalization is
(2.5)
with

d AA = 2rA.

There is an obvious flaw in the arguments used to derive (2.4) even

granting the assumption of rigid spherical molecules and the limitation to
low-density gases. No account has been taken of the variation in the relative
velocity of the molecules making up the beam and the target, an effect
which alters the numerical factor in (2.4) but not the qualitative behavior;
AA still decreases as the gas density or the collision cross section increases.

20

Chap. 2

Kinetic Theory of Gases-Transport

Since molecular diameters, determined from the volume of condensed

matter, range between 0.2 and 0.5 nm, the mean free paths can be estimated.
If we use (2.2) and (2.4) the mean free path of Br2 molecules in N2 gas at
300 K and 1 atm can be estimated. Assuming 0.4 nm for the N2-Br2 distance
of closest approach we find AB r 2 ,....., 100 nm. Since at 300 K the mean speed
of Br 2 is,....., 200 m sec-I, the typical bromine molecule makes "-' 2 X 109
collisions per sec. Each collision redirects the molecule; thus it is no surprise
that diffusion through a molecular maze is a much slower process than
unimpeded effusion.

2.3. Collision Frequency

An accurate calculation of mean free path requires computing the
average distance a molecule travels between collisions, a quantity which
depends upon the speed of the molecule. The mean free path of stationary
molecules is zero since they do not move until struck by other molecules.
To very fast molecules the remainder of the gas appears as immobile target
molecules and the calculation of the previous section is exact. We thus
expect (2.4) to provide an upper bound for the speed-dependent mean free
path.
In a gaseous mixture of A and B at equilibrium a representative A
molecule with speed C travels a distance c Lit in a time Lit. During this
period such a typical molecule makes ZA(C) Lit collisions(2) [ZA(C) is the speeddependent collision rate]. The average distance between collisions, AA(C), is
(2.6)

and the mean free path is found by averaging over speeds

(2.7)
where FA(C) is the speed distribution function (1.24).
As a representative A molecule moves it sweeps out collision cylinders
with respect to the other molecules in the gas. In Fig. 2.2 such a collision
(2)

As previously noted, collisions alter the speed of the individual molecules. However,
as the gas is at equilibrium, the total number of A molecules with a given speed is
invariant. It is thus correct to consider the properties of molecules of a particular
speed even though the members of this class are continually changing. See Section
1.2 and footnote (10) in Chapter 1.

Sec. 2.3

Collision Frequency

21

Fig. 2.2. Collision cylinder for A-B

collisions.

cylinder is shown for interaction with B molecules with velocity between

C 2 and C 2 + dc 2 . (3) The cross-sectional area is ndl B , just as in the model
discussed in Section 2.2. The height of the cylinder depends upon the relative
speed of the colliding molecules; those approaching slowly sweep out a
smaller volume and have a smaller chance of undergoing a conision.
Denoting the velocity of the A molecule as C1 , the collision cylinder described in a time L1 t has the volume
(2.8)
where C12 = I c1 - C 2 I is the relative speed of the colliding pair. As the
collision probability is the ratio of (2.8) to the total volume V and the
number of B molecules in the specified velocity domain is NBFB( c2 ) dc 2 ,
the A-B collision rate is
(2.9)
By integrating over alI values of the velocity C2 , ZAB(C), the collision frequency of a representative A molecule of speed C with B molecules of
arbitrary velocity can be found. The calculation is left as a problem. The
result is
(2.10)
where a = c(mB/2kT)1I2 and the integral is related to the error function
introduced in Section 1.4. In a binary mixture the total collision frequency is
(2.11)
which, when combined with (2.10) and (2.6), determines AA(C), The speeddependent mean free path is an involved function of the relative masses
of the two components, their densities, and their collision cross sections.
(3)

This is a shorthand notation; it indicates a velocity vector with components in a restricted

domain. In Cartesian coordinates the x, y, and z components lie between u and u + du,
v and v + dv, and wand w + dw, respectively. In spherical polar coordinates the speed
is between c and c + dc, and the solid angle is bounded by e and e + de and rp and
rp

+ drp.

Chap. 2

22

Kinetic Theory of Gases-Transport

1.0

0.8
2

nnd >-(0.)

0.6

0.4

ex. = c.JM/2RT
Fig. 2.3. (nnd')}.(a), the ratio of the speed-dependent mean free path to its value at
infinite speed, in a one-component system as a function of the scaled speed u.

In a one-component system there is considerable simplification; A is plotted

in Fig. 2.3 as a function of the scaled velocity c(Mj2RT)1I2. Our expectations
are verified; the relatively slow molecules have short mean free paths while
relatively fast ones have values of A which approach the bound calculated
in the previous section. If we use (2.7), AA can be determined by numerical
or graphical means; the result is
(2.12)
about i of its limiting value. To determine AA in a mixture requires a
separate calculation for each composition.
An alternative approach, which handles mixtures as readily as it
does pure substances, is to approximate (2.7) by averaging numerator and
denominator independently,
(2.13)
an expression which, while often defined as the mean free path, does not
properly account for the speed variation illustrated in Fig. 2.3. To determine
ZA requires the sum of the average frequency of A-A and A-B collisions,
the analog to (2.11). The mean collision rate of an A molecule with any B
molecule is found by multiplying (2.9) by the probability that an A molecule
has its velocity between Cl and C l + dCl and then integrating over all
values of C l and C2 ; the result is
(2.14)

Sec. 2.3

Collision Frequency

23

The integral in (2.14), which is the mean relative speed CAB, is most
easily evaluated by transforming to center-of-mass and relative coordinates,
(2.15)
To reexpress the product FA( cl)FB( c 2) note that the exponential factor
is dependent upon the total kinetic energy of the two particles, (mAc 12 +
mnc 22)/2. If (2.15) is solved for Cl and C2 and these values substituted in the
expression for the total kinetic energy, it takes the form (MCM 2 + fJc 2)/2,
where M is rnA + mB, and fJ, the reduced mass for an A-B pair, is mAmB/M.
Using these expressions along with (1.22) we find that
FA(c l )FB(C2)

)3/2( 2nkT
fJ )3/2
exp[-(MCM2 + fJc 2)/2kT]

M
( 2nkT

FM(CM)FR(C)

(2.16)

Note that the transformed representation is a product of Maxwellian distributions of the center-of-mass and relative kinetic energy. To complete
the transformation of (2.14) requires expressing dC l dC 2 in terms of the new
coordinates. The result, obtained by the Jacobian method outlined in the
Appendix, is dC l dC 2 = dCM dc so that the integral in (2.14) is the product

FM(CM) dCM

FR(C)C dc

which is easily interpreted. The first factor is a normalization integral for

particles of mass M and the second factor is the average speed for particles
of mass fJ so that CAB is
(2.17)
which when substituted in (2.14) yields
ZAB = nBn dlB[8kT(mA

+ mB)/nmAmB]1I2

(2.18)

Note that CAB is always larger than the mean speed of either molecule.
The mean collision frequency for an A molecule in a binary mixture
is thus
(2.19)

since cAA is 2l / 2CA. When combined with (2.13) the mean free path is
approximated by
(2.20)

24

Chap. 2

Kinetic Theory of Gases-Transport

Compared to the exact one-component result, (2.12), the approximation

overestimates A by 4%. Computer integration of (2.7) allows the same comparison to be made for binary mixtures. The results indicate that the
maximum error occurs when rnA = rnB; in practice the approximation
(2.20) is excellent.
For the example of a dilute mixture of Br2 in N2 gas at 1 atm and
300 K considered in Section 2.2, we find that the mean relative speed is
516 m sec-I. Assuming, as before, that dAB '"'-' 0.4 nm, we obtain a collision
frequency'"'-' 6.4 X 10 9 sec- 1 and a mean free path '"'-' 31 nm. It is worth
noting that ABr2 is much larger than the mean interparticle distance, which
is only 3.4 nm at this density.
In addition to quantities already calculated, the total collision rate
for all A molecules per unit volume is useful for application to chemical
kinetics. It is
(2.21 )
which is essentially nA times the molecular collision frequency (2.19);
the factor of 1/21/2 in the first term of (2.21) corrects for the fact that this
procedure counts each A-A collision twice, i.e., ZAA = nAzAA/2.

2.4. Macroscopic Equations of Transport

If a system is displaced from equilibrium by creating a temperature
gradient, heat flow takes place in the opposite direction to reestablish
equilibrium. This is a typical transport process; energy flows from one
region of the system to another. In a one-component system at constant
pressure there is a simple relationship between the heat flux and the temperature gradient. As long as the gradient is small, Fourier's law of heat
conduction is applicable. For a temperature gradient in the z direction the
heat flux, Jq , is
dT
J = -x-(2.22)
q
dz
where x is the thermal conductivity.
For mixtures with concentration (number density) gradients, matter
flows are established to restore homogeneity. There can be two contributions
to this flux, one due to motion of the system as a whole (center-of-mass
motion), the other due to the relative motion of the various components
of the mixture. It is the second of these, the diffusive flux J, which serves

Sec. 2.4

Macroscopic Equations of Transport

25

to homogenize the system. In a binary mixture at constant temperature

with a concentration gradient in the z direction, Fick's first law of diffusion
states that

-D dnA

(2.23)

dz

where D is the diffusion coefficient. If the system is also at constant pressure

the center of mass is stationary so that the total flux and the diffusive
flux are the same, and, since nA + nB = n, which is constant, JA = -JB .
In addition to temperature and concentration gradients, transverse
velocity gradients may also be established in a fluid. Figure 2.4 illustrates
the velocity profile in a system where there is viscous flow. For molecules
near the stationary plate the mean x component of velocity, ii, is zero; at
the moving plate it is bU. The gradient in ii, dii/dz, is opposed by the drag
force required to keep the plate in motion. The larger the area of the plate,
the greater this force. Poiseuille's law describes the relation between the
force per unit area in the x direction, P x' and the velocity gradient
(2.24)
where rJ is the shear viscosity. Since the viscous force per unit area has the
dimensions of pressure it can equally well be interpreted as the momentum
flux established to counter a velocity gradient.
The three formulas (2.22)-(2.24) are examples of linear transport
equations; they relate the response of a system (the flux) to a small perturbing force (the gradient). The transport coefficients D, rJ, and u are the
parameters of proportionality, to be determined experimentally. A familiar
transport equation is Ohm's law. Here voltage is the force, current the
response, and conductivity (the reciprocal of resistance) the transport
coefficient. In general, equations of transport are not as simple as these.
In a two-component system with a temperature gradient, Fourier's law
states that there is only heat flow. However, if the masses of the components

Tz
Fig. 2.4. Velocity profile in a gas undergoing viscous flow. The lower plate
is stationary while the upper one has a
velocity Vx = t5U.

Moving plate
6U
--------~--------+.~=

------+.

v. = z6U/L

------------- v = 0
Stationary plate

26

Chap. 2

Kinetic Theory of Gases-Transport

differ significantly, so does the mean speed of each one. There is no reason
to expect that the flux of each is the same. In fact, the temperature gradient
leads to a flux of matter, the phenomenon of thermal diffusion. (4) Similarly
a concentration gradient causes heat flow as well as the matter flow demanded
by Fick's law.
Rigorous phenomenological justification of transport equations is
the domain of irreversible thermodynamics. (5) By extending the entropy
concept to systems slightly displaced from equilibrium and requiring that
entropy increases with time, it is possible to generalize flux-force relations
like (2.22)-(2.24) to account for coupled phenomena such as thermal
diffusion. An important consequence is that there is an inherent correlation
between reciprocal processes such as mass flux due to a temperature gradient
and heat flux due to a concentration gradient. The transport coefficient
for the process and its inverse are not independent. Such constraints on the
transport coefficient for a process and its inverse are quite general and are
known as Onsager reciprocal reiations.(6)

2.5. Solution of the Transport Equations

As presented the transport equations do not explicitly involve time.
To include temporal effects consider a system such as the diffusion cell of
Fig. 2.5 in which the A molecules are initially confined to the narrow
region between 0 and a; when the barriers are removed they diffuse into the
(essentially infinite) region containing B molecules. To determine the concentration of A as a function of both position and time consider a part of
the cell of cross-sectional area S between z and z + LIz. The rate of change
of the number of A molecules in this volume is the difference between
the entering and leaving flows:
( faNA)
it z

Sh(z) - Sh(z

+ LIz)

Since the volume is S LIz the number density of A is NA/S LIz and because
(4)

(5)

(6)

Thermal diffusion has been used as a basis for isotope separation techniques [K. Clusius and O. Dickel, Naturwiss. 26, 546 (1938)].
For an introduction to this subject see I. Prigogine, Thermodynamics of Irreversible
Processes, 3rd ed. (New York: Interscience, 1967).
The theorems demonstrating the interconnections were first proved by L. Onsager,
Phys. Rev. 37, 405 (1931); 38, 2265 (1931).

Sec. 2.5

27

Solution of the Transport Equations

Regions initially
containing
pure B

Region to which
A was originally
confined

Fig. 2.5. Schematic model of a diffusion cell.

J(z

+ LIz) =

J(z)

+ LIz aJ/az

we find, with (2.23),

or, if D is independent of position,

= D( a2n~ )
( anA)
at z
az
t

(2.25)

which is Fick's second law of diffusion. It determines the concentration profile

for given boundary conditions.
A similar equation for the temperature can be derived from Fourier's
law (2.22). Arguments similar to the ones just given show that

ae )
( a2T)
(at
z =" az
t

(2.26)

where e is the energy per unit volume. Since e

=
( ~)
at z

aCE/V) )
at

E/ V,

aCE/V) ) (~)
aT

at

~ (~)
V

at

and the equation for heat conduction is

( aT )

at

z =

" ( a 2T )
az 2 t

cv! V

(2.27)

Similarly the velocity profile is determined by the equation

(2.28)
where e is the mass density.

28

Chap. 2

Kinetic Theory of Gases-Transport

The three equations (2.25), (2.27), and (2.28) all have the same form.
While general solutions for arbitrary boundary conditions present complicated mathematical problems, (7) certain one-dimensional problems are
simple and instructive. Consider a cylinder of which one end is held at
temperature T* and the material inside is initially at temperature To.
This is a model for heat flow in a long metal rod or in a tube containing
liquid or gas. Making suitable correspondences it also describes diffusion
in a long tube where the solute concentration at one end is kept constant.
Ignoring effects due to the surface of the cylinder, the boundary conditions
are
z = 0, all t
T= {T*,
(2.29)
z> 0, t = 0
To,
The solution to (2.29), which may be verified by direct differentiation, is

T- To _
T* _ To -

2
n 1/ 2

f= exp( -y ) dy -

2_

1 - erf(s)

(2.30)

where s = z/[2(D*t)1I2], D* = "V/Cv , and erf(s) is the error function

discussed in Section 1.4. Since erf(O) = 0 and erf(oo) = 1, the boundary
conditions (2.29) are obviously satisfied.
The temperature profile is a function of the single composite variable s.
This observation implies that:
(a) the time required for a point to reach a given temperature is proportional to the square of the distance from the heat source, inversely proportional to the thermal conductivity, and directly proportional to the heat
capacity;
(b) the distance to which a given temperature has penetrated is proportional to t 112;
(c) the heat flux required to keep the z = 0 surface at constant temperature is proportional to t- 1/ 2 ; this follows by substituting (2.30) in (2.22)
and setting z = O. (8)
As an application of (2.30) consider a piece of copper tubing which
has one end maintained at 500C (easily done with a propane torch).
Let us estimate the time required for a point 0.3 m (about 10 in.) from the
end to reach lOOC (which is too hot to hold) assuming an initial temperature
(7)

(8)

For general solutions see J. Crank, The Mathematics of Diffusion (Oxford: Clarendon
Press, 1975).
J. Crank, The Mathematics of Diffusion (Oxford: Clarendon Press, 1975), p. 87.

Sec. 2.6

Kinetic Theory of Transport

29

of 20e. Numerical substitution indicates that s is the solution of the

equation erf(s) =!; referring to tabulations of the error function we
find that s '"" 0.98. For copper at room temperature :x; = 420 J m -1 sec- 1 K-1
and Cv/V = 3.46 X 10 6 J m- 3 K-1 so that D* = 1.21 X 10-4 m 2 sec- 1 and
thus
Z2

Z2

t = 4D*S2 = 2.07 X 103 S2 = 2.15 X 103z 2

A point 0.3 m from the end of the rod reaches 100C in approximately
193 sec or a little over 3 min. (9)

2.6. Kinetic Theory of Transport and Postulate of Local

Equilibrium
The linear relations between forces and fluxes which were discussed
in the previous section are empirical. To provide a justification based upon
kinetic theory and simultaneously to obtain expressions for the transport
coefficients l-e, 'Yj, and D requires developing a molecular model for transport
in dilute gases and extending the concept of a velocity distribution to nonequilibrium systems.
For the systems considered the gradients are very small. Thus, in any
region of the gas, concentration, temperature, and mean velocity are welldefined experimental quantities. There is no objection, either in principle
or practice, to inserting a probe which can measure the local values of
these macroscopic parameters (as, for example, one might measure temperature at different points in a room). As the gas is not at equilibrium the
velocity distribution is positionally dependent but the mean molecular
properties change little over distances comparable with the mean free path
(estimated in Section 2.2 to be between 30 and 300 nm for a gas at 300 K
and I atm). In any region, extending over many mean free paths, the
properties of the gas can be characterized by the local values of the macroscopic observables.
(9)

The solution to the diffusion equation (2.30) implies that the temperature decreases
monotonically with z at all times. It is interesting that this uniform behavior is not
a property of ultrapure materials. Instead, heat is propagated as a wave, in a fashion
similar to the propagation of sound. This phenomenon, known as second sound, was first
discovered in liquid helium at very low temperatures. It has since been observed in solid
helium and has been reported to occur as well in the isotopically and chemically pure
materials NaF and Bi at very low temperatures. For a discussion of the effect see B.
Bertram and D. J. Sandiford, Scientific American 222(5), 92 (1970).

30

Chap. 2

Kinetic Theory of Gases-Transport

Consider a gas at I atm in which the substantial temperature gradient

of I K mm- 1 is maintained. In traveling 1 mm a molecule undergoes
""' 104 collisions. If the mean temperature of the gas is ""' 300 K then the
mean speed differs by 0.2% for regions I mm apart. On the other hand, the
relative dispersion in the speed is about 20%. Thus the variation in c due
to the temperature gradient is far less, in the example considered, than the
normal dispersion of the distribution. We can therefore imagine that in
each region of the gas the velocity distribution is Maxwellian, but with the
temperature characteristic of the particular region. If there are also gradients
in concentration and velocity it is natural to extend this assumption and to
assert that the positionally dependent velocity distribution is Maxwellian
in every region of the gas; however the distribution is characterized by the
local values of temperature, velocity, and concentration. It must be emphasized that this postulate of local equilibrium is only a first approximation.
To improve upon it, which is necessary to develop an exact theory, requires
elaborate physical and mathematical analysis.

2.7. Simplified Transport Theory: Dilute Hard-Sphere Gas

Depending upon the density, transport in a gas may be either effusive
or diffusive in nature. If the gas is so dilute that the mean free path, A, is
greater than the length of the vessel, L, a molecule rarely undergoes a collision while traversing the vessel. It is much more likely to hit the wall.
Transport in this effusive domain, known as Knudsen flow, (10) is extremely
sensitive to the nature of the interactions between molecules and the walls
of the container. While sensitive to gas-surface interactions, Knudsen
flow does not depend on the forces between gas molecules.
At high pressures A < L and the gas is in the diffusive domain. To
construct a molecular model which accounts for collisional effects, consider
Fig. 2.6; the macroscopic gradients are in the z direction. If we assume
looal equilibrium is established in the volume element dr, particles in that
region have a Maxwellian speed distribution and are moving randomly
with mean speed c. (11) Molecules in dr have properties different from those
(10)

(11)

For consideration of transport when A > L see M. Knudsen, The Kinetic Theory of
Gases (New York: John Wiley and Sons, 1950).
This is not quite accurate; there are small perturbations to the distribution function
induced by the macroscopic gradients. To treat such effects properly requires a far
more elaborate treatment than is given here. See R. D. Present, Kinetic Theory of Gases
(New York: McGraw-HilI, 1958) for details.

Sec. 2.7

Simplified Transport Theory

31

z
Direction of
macroscopic
gradients

Direction of
macroscopic
fluxes

--~~~----~-------x

Fig. 2.6. Coordinate geometry for determining the differential molecular flux through a
reference plane at zo.

in the vicinity of a reference plane located at Zo. The further dr is from zo,
the greater the differences. If molecules from dr reach Zo without undergoing
a collision they will on average, contribute to the macroscopic flux through
Zo in a manner determined by the density, momentum, and temperature
in dr. Of course the further dr is from Zo the greater is the chance of a
collision which leads to reequilibration in a new region. We assume that
such reequilibration is established with each collision, i.e., that molecules
are only aware of the last stimulus to which they have been exposed. Thus
to compute the flux of any property at Zo requires determination of the
differential flux of molecules originally in dr that reach Zo without an intermediate collision. The differential flux multiplied by the mean local energy
or momentum yields the differential energy or momentum flux. Then
summing over all regions dr yields the corresponding macroscopic flux.
The total number of particles in dr is n(z) dr, where n(z) is the number
density, which may be position dependent. Of these only the fraction
moving in the direction of r can pass through the specified area A in the
reference plane Zo. Those molecules originally in dr that have not undergone
a collision spread out uniformly over a sphere of area 4nr2. Since the projection of A onto this sphere is A 1cos X I, the fraction which could pass
through A is A cos X 1/4nr2. Collisions reduce the flux through A. In
Section 2.2 we found that the fractional change in flux intensity per unit
distance traveled is A-Ie-riA. Then, since molecules, on the average, travel
a distance eLl t during an interval Ll t, the total number of molecules initially
in dr that pass through A in a time t without undergoing an intermediate
collision is
n(z) dr(A cos X 1/4nr2)(A- I e- rIA )c Llt
1

Chap. 2

32

Kinetic Theory of Gases-Transport

Since molecules crossing the reference plane from above are moving in the
-z direction the differential molecular flux at Zo is
dJ = -n(z)(cos xI4nr 2A)ce-

r, ). dr

(2.31)

When 0 < X < nl2 (molecules arriving from above zo) the differential
flux is negative; when nl2 < X < n it is positive. (12)
Application of (2.31) is straightforward. To obtain Poiseuille's law
compute the momentum flux for a system in which the mean value of the
x component of velocity is position dependent,
Px

(2.32)

f mii(z) dJ

Expanding ii(z) in a Taylor's series about Zo and setting Zo

ii(z)

ii(zo)

+ z -dii
- I + -Z2
dz
2
zo

d 2ii
dz

-2-

0 we find

I + ...
Zo

(2.33)

Since viscous flow occurs at constant temperature and density, the momentum flux, with the geometry of Fig. 2.6, is
mnc
cos
P = - - foo f" f2" ( -X
- e- rl )' )
x
4:n;}..
0
0
0
r2
X

( iio

+ r cos Xiio' + r2 cos

2

iio"

+ ... ) r2 sin X dr dX dw

where we have substituted ii(zo) = iio , etc., and z = r cos X. As long as

gradients are small, higher-order terms in the ii(z) expansion may be ignored.
The integrals are easily done. Because of the dependence on cos X only
the term proportional tOiio' is nonzero; the result is
p _

x-

mnAc dii I
- - - 3 - dz

(2.34)

Zo

precisely Poiseuille's law (2.24). The viscosity is

n
(12)

mnU
3

_2_ (nmkT)1I2
3n
nd 2

(2.35)

It is possible to modify (2.31) to account for the distribution of molecular speeds and
thereby to consider the speed dependence of the mean free path. However, such modification does not incorporate the effect of macroscopic gradients on the distribution
function. Furthermore, it is still restricted to billiard ball dynamics. Thus, even though
it is conceptually more in tune with the previous analysis we do not pursue it here.

Sec. 2.7

Simplified Transport Theory

33

where the second expression is determined by introducing C from (1.25)

and A from (2.20), where d = dAA . (13)
Fourier's law is almost as easily established. The energy flux is
Jq =

(2.36)

E(Z) dJ

where E(Z) is the z-dependent mean energy. Since heat flow occurs in a
system at constant pressure, the density is not uniform, n(z) = p/kT(z).
Furthermore the mean speed is also position dependent, C = [8kT(z)/nm]1I2.
Finally the mean energy of an ideal gas is cvT(z), where Cv is the constant
volume heat capacity per molecule. Substituting into the heat flow integral
(2.36), expanding T(z) in a Taylor's series about zo, and integrating, the
result is
(2.37)
which is Fourier's law (2.22). The thermal conductivity may be identified as
nCvAC
6

Cv (nmkT)1/2
-'-----,='-m
nd 2

" = - - - = - - -3n

(2.38)

The simple mean free path approach is not adequate for describing
mutual diffusion in a binary system at constant temperature. Only in
one case is the theory self-consistent-when the molecules are mechanically
indistinguishable (self-diffusion). By integrating (2.31) we obtain the total
flux of one component, h = f dh. Since the only position-dependent
quantity is nA(z), the analysis used in the derivation of Poiseuille's law
yields
(2.39)
precisely Fick's law (2.23) from which we identify
(2.40)
The same arguments may be applied to component B; the analogs to (2.39)
and (2.40) are obtained. From (2.23) we note that there is only a single
diffusion coefficient, i.e., DA = DB. In general AACA
ABCB so that the

*-

(13)

This expression for Ais used for internal consistency. Its derivation, like this treatment
of viscosity, did not consider the variation of mean free path with speed.

Chap. 2

34

Kinetic Theory of Gases-Transport

Table 2.1. Comparison of Various Theoretical Formulas for the Transport

Coefficients of Hard-Sphere Gases
Transport coefficient x[nd2(nmkT)1!2]
Property

mnD

Section 2.6

Exact

0.159

0.375

0.159

0.3125

0.0796cv

m'X

0.3125cv

+ 0.7031k

mean free path technique is internally inconsistent. The inconsistency may

be corrected within the framework of this approach; however, the revised
formulas are neither simple nor illuminating nor precise. (14) Instead we
limit consideration to the one case where the matter flux computed using
(2.39) is zero. When the molecules are mechanically indistinguishable,
i.e., m == mA = mB, d = dAA = dAB = dBB , (2.40) becomes
D

2
3n

(nmkT)1!2
nmnd 2

= - - -'-----;:::0--

(2.41)

where n = nA + nB.
Our treatment is limited to a very simple model, a dilute gas of hard
spheres. Nonetheless the results are only approximate. The expressions
for'Y) and D, (2.35) and (2.41), differ from the results of an exact theory of
hard-sphere transport by a numerical factor. The expression for :;c, (2.38),
is in error in an additional way. The derivation does not account for possible
differences in the rate of transport of translational energy (center-of-mass
kinetic energy) and of rotational and vibrational energy (internal energy).
Exact theory does; the results, assuming internal energy is also equilibrated
with each collision, are summarized in Table 2.1.
Both treatments lead to similar qualitative conclusions. The viscosity
of a gas is predicted to be independent of the density, an astonishing result
when first discovered by Maxwell. Experiment has confirmed the theory,
which is readily understood qualitatively. As the density increases the
molecular flow increases proportionately. On the other hand, the mean free
path decreases and collisions occur more often. The two effects exactly
(14)

For a discussion of this point see J. E. Mayer and M. G. Mayer, Statistical Mechanics
(New York: John Wiley, 1940), p. 30.

Sec. 2.8

Comparison with Experiment

35

balance. Another surprise was the prediction that gas viscosity increases
with temperature, a result opposite to the more familiar situation in liquids,
where viscosity decreases rapidly as temperature increases.
For the same qualitative reasons K, like 'f}, is density independent. The
thermal conductivity is more temperature sensitive than the viscosity
because of the dependence on heat capacity. Its behavior is also quite
different from that observed in liquids. In gases K increases with T, while
in liquids the opposite behavior is normal.
Unlike 'f} or K, D is density dependent since diffusion, as discussed in
the introduction to this chapter, is essentially a direct measure of the mean
free path. The temperature dependence is a consequence of the fact that
faster molecules diffuse more rapidly. From Table 2.1 we see that the ratio
mnDj'f} is predicted to be constant in dilute gases; this is completely different
from Walden's rule in liquids, D'f} f':::i const.

2.B. Comparison with Experiment

The predictions of an exact treatment of transport for the dilute gas
of hard spheres are summarized in Table 2.2. The correction for internal
degrees of freedom is included. The effects of mean speed and collision
cross section always occur in the combination TI12jnd 2 ; thus the ratios

(O.4cv

mK

mnD = 1.2
'f}

(2.42)

25

(2.43)

+ O.9k)'f)

are predicted to be constants, independent of temperature or of the particular gas. These ratios are tabulated for a number of dilute gases at
Table 2.2. Exact Theoretical Expressions for Transport Coefficients of Hard
Spheres
Transport coefficient, x[nd2(nmkT)1I2]

Flux

Gradient

Matter

Concentration

= 3(8mn)

Momentum

Velocity

'YJ

= 5(16

Energy

Temperature

= 5 [(cv

+ 2.25k)(16m]

Chap. 2

36

Kinetic Theory of Gases-Transport

Table 2.3. The Ratio mnD/'f/ for Gases at Various Temperatures a

Ne

Ar

N.

O.

77.7

1.29

1.26

1.33

1.33

1.60 @ 90.2 K

273.2

1.37

1.31

1.34

1.39

1.43

353.2

1.39

1.34

1.39

1.43

1.36

Temperature (K)

CR.

a Data from H. H. Landolt and R. Bornstein, Zahlenwerte und Funktionen, 6th ed. (Berlin:
Springer, 1969), Vol. II, Part 5a, pp. 3, 516.

different temperatures in Tables 2.3, 2.4, and 2.5. From Table 2.3 it is
clear that mnD/'Y) does not vary greatly. It is closer to 1.2 for monatomic
gases; deviation from the predicted ratio becomes larger as the molecules
become more complex. The data indicate that the hard-sphere model
accounts for most of the observed phenomena but that it is certainly not
precise. Table 2.4 emphasizes the reliability of the simple theory for monatomic gases (for these molecules O.4cv + 0.9k = cv ); there is very little
deviation from the predicted value of 2.5. Table 2.5 indicates that the correction for internal energy is extremely important; only if this is made do
all dilute gases behave in roughly the same fashion. Had we assumed that
kinetic and internal energies were transported in the same way, the invariant
ratio would have turned out the same as for monatomic gases, i.e., mx/ cv'Y),
a number which varies greatly. The variation of the ratio (2.43) provides
a test of the assumption that internal energy is equilibrated with each
Table 2.4. The Ratio xm/'f/cv for Monatomic Gases at Various Temperatures a
Temperature (K)

Re

Ne

Ar

90

2.44

2.46

2.49

195

2.45

2.52

2.51

2.5

273

2.45

2.50

2.48

2.50

373

2.44

2.51

2.53

2.50

2.43

2.47

2.47

2.47

491

Kr

a Data from H. H. Landolt and R. Bornstein, Zahlenwerte und Funktionen, 6th ed. (Berlin:
Springer, 1969), Vol. II, Part 4, pp. 398--451; Part 5a, p. 3; Part 5b, pp. 45-53.

Sec. 2.8

Comparison with Experiment

37

Table 2.5. The Ratios um/,'Icv (in Parentheses) and um/'f}(OAcv

for Polyatomic Gases at Various Temperatures a

Temperature

+ O.9k)

H_

0_

100

2.45
(2.31)

2.37
(1.74)

200

2.50
(2.04)

2.50
(1.92)

2.28
(1.58)

2.55
(1. 78)

2.39
(1. 75)

273

2.50
(1.93)

2.53
(1.92)

2.51
(1.67)

2.60
(1.76)

2.46
(1.86)

300

2.51
(1.94)

2.58
(1.96)

2.55
(1.69)

2.65
(1.79)

2.50
(1.89)

(K)

CO_

CH.

NO

2.42
(1.80)

a Data from H. H. Landolt and R. Bornstein, Zahlenwerte und Funktionen, 6th ed. (Berlin:
Springer, 1969), Vol. II, Part 4, pp. 398-451; Part 5a, p. 3; Part 5b, pp. 45-53.

collision; its approximate character is apparent since the ratio is much

more temperature and species dependent for polyatomic than for monatomic
gases.
A better test of the model is given by considering the temperature
dependence of the transport coefficients. If molecules were truly hard
spheres the values of d, computed using the expressions of Table 2.2 and
experimental values of ", 1], or D, would be constant. In Figs. 2.7 and 2.8
the apparent temperature dependence of d- 2 (which is proportional to the
transport coefficient) is illustrated. Clearly d- 2 is not a constant; it increases
with temperature. The temperature dependence varies considerably from
gas to gas. Furthermore there are also differences in the effect of temperature on d- 2 depending upon which transport coefficient is used to compute it. The differences are most severe for polyatomic molecules, where the
hard-sphere model is least reliable.
The naive model is not adequate. However, a bit more can be extracted
from it. Table 2.6 quotes measurements of 'Y} and ". Included are three
estimates of d, two calculated from transport data using the expression for
'Y} and" of Table 2.2. The third is determined from critical point data via the
van der Waals bo: sincevc = 3bo andb o = tNo(4nd 3J3) thend = (vcJ2nNo)l/3.
In addition the mean free path is evaluated using (2.12) with the value of
d determined from viscosity data. The estimates of d from transport and
critical point data differ considerably. Nonetheless they are of the same
order of magnitude, which suggests that the model is not unreasonable.

Chap. 2

38

Kinetic Theory of Gases-Transport

He

20
_ _--Ne

_-----H2

10

200

400

600

T(K}
Fig. 2.7. Reciprocal of apparent value of d 2 as determined from thermal conductivity using
hard-sphere transport theory (Table 2.2). Transport coefficient data from H. H. Landolt
and R. Bornstein, Zahlenwerte und Funktionen, 6th ed. (Berlin: Springer, 1969), Vol. II,
Part 5b, pp. 45-53.

To improve the description a more realistic intermolecular potential

must be used. The following qualitative argument was given by Sutherland. (15) Since molecules are not hard spheres, increasing their relative
kinetic energy allows them to approach one another more closely. Thus
an increase in temperature leads to a decrease in the mean collision diameter.
This may be formulated as
(2.44)

where d= and A are constants, with d= the hard-core diameter at infinite

temperature. While accounting for deviations from the p/2 prediction
the Sutherland modification is inadequate to explain the observed temperature dependence or the differences between the transport coefficients
evidenced in Figs. 2.7 and 2.S. To accomplish this a quantitative picture
is required.
(15)

W. Sutherland, Phil. Mag. 36, 507 (1893).

Sec. 2.8

Comparison with Experiment

20-

,...... . - '

39

-------- -.---------.-

/~

------

---

_ - - - - - - Ar

. . . .-----. . ~-==--.:::::----~
CH
....- -- ---

//

/.::;.-

Ne

.-----

-.:=

--

'-C02

200

400
T(K)

600

Fig. 2.8. Reciprocal of the apparent value of d' as determined from viscosity ( - ) and selfdiffusion coefficients (- -) using hard-sphere transport theory (Table 2.2). Transport coefficient data from H. H. Landolt and R. Bornstein, Zahlenwerte und Funktionen, 6th ed.
(Berlin: Springer, 1969), Vol. III, Part 5a, pp. 3, 516.

Table 2.6. Molecular Properties of Gases and Related Transport Data Measured
at 273.2 K and 1 atm"

Gas

Viscosity,
Thermal con- Mean free
ductivity, u x 10' path, A
1) x 105
(kg m- 1 sec 1 ) (J K -1 m- 1 sec- 1 )
(nm)

Molecular diameter, d (nm)

From 1)

Fromu

From critical
point data

168

0.218

0.218

0.250

He

1.85

Ne

2.97

4.60

120

0.258

0.258

0.224

Ar

2.11

1.63

61

0.364

0.365

0.273

H,

0.845

108

0.272

0.269

0.260

0,

1.92

2.42

62

0.360

0.358

0.272

CO,

1.36

1.48

37

0.464

0.458

0.296

CH.

1.03

3.04

47

0.414

0.405

0.299

14.3

16.7

a Data from H. H. Landolt and R. Bornstein, Zahlenwerte und Funktionen, 6th ed. (Berlin:
Springer, 1969), Vol. II, Part 5a, p. 3: Part 5b, pp. 45-53.

Chap. 2

40

Kinetic Theory of Gases-Transport

Agreement with experiment can be obtained by constructing a rigorous

theory for the positionally dependent velocity distribution function; such
theories, which provide the tools to determine the deviations from local
equilibrium, are based upon analysis of the dynamics of binary collisions.
As already pointed out in the derivation of the Maxwell distribution
presented in Section 1.2, the net change that occurs in an elastic binary
collision is reorientation of the relative velocity vector. The reorientation,
which rotates the vector C12 through the scattering angle X' is determined
by the intermolecular potential, the collision energy of the two particles,
!,ud2, and the impact parameter, b. This latter quantity is simply the distance
of closest approach for noninteracting point molecules with the given
initial velocities.
To calculate a trajectory requires knowledge of the intermolecular
potential, which is not readily measured. To circumvent this difficulty
a guess is made as to the form of the potential which is then used to compute
the transport coefficients, the second virial coefficient, and a few other
properties dependent upon two-particle interaction only. Good agreement
between theory and experiment provides a posteriori justification for the
assumed potential. A simple model potential was suggested by LennardJ onesY6) It is
(2.45)

where

is the depth of the potential well and (f the distance at which

O. The attractive term, r- 6 , has some theoretical justification; the
choice of r- 12 dependence to describe the repulsion which is dominant at
small separations was made on grounds of computational simplicity. As
an illustration of the differences between the hard-sphere model and a
more realistic model like the Lennard-Jones potential, consider Fig. 2.9.
The hard-sphere diameter is chosen to be the Lennard-Jones 11. In the
hard-sphere case the collision is impulsive; the scattering angle is uniquely
determined by the impact parameter. At high collision energy the LennardJones potential leads to only slightly different results. However, at low
collision energy, the colliding pair is moving relatively slowly as interaction
becomes important. Deflections are large and scattering is very different
from that of hard spheres. For hard spheres no deflection occurs if b > 11,
a statement that is incorrect for the Lennard-Jones potential. In particular,
if the collision energy is small relative to the well depth E, there is significant
u(r)

(l6)

J. E. Lennard-Jones, Proc. Roy. Soc. AI06, 463 (1924).

Sec. 2.B

Comparison with Experiment

+
(0 )

(e)

41

-+-__-J\

~+
(b)

(d)

Fig. 2.9. Trajectories of a particle which interacts with a center of force via a LennardJones potential for various reduced impact parameters b* = bla. As b* increases from 0,
the path changes from a head-on to a glancing collision. Plots are for different reduced
collision energies, T* = f!c~2/2E: (a) 0.8, (b) 4, (c) 20, and (d) a hard-sphere potential.
For the smaller b* the trajectories are not greatly affected by changes in T* although the
deflection angle is energy dependent. For the larger b* there are qualitative differences,
most evident in (a), where T* is the smallest. [Data from J. O. Hirschfelder, C. F. Curtiss,
and R. B. Byrd, Molecular Theory of Gases and Liquids (New York: John Wiley, 1954),
pp. 1132-1146.]

deflection even if b is substantially greater than (j. It is thus apparent that

mean free path is an imprecise concept when realistic intermolecular
potentials are considered. There can be considerable deflection, even at
large b, when the collision energy is small. Thus the collision diameter
depends, in effect, upon the collision energy.
Trajectory analysis is used to compute the transport coefficients. Since
momentum transfer in collisions accounts for transport, it is apparent that
calculations based upon the Lennard-Jones potential (or any other that
incorporates intermolecular attraction) lead to rather different predictions
than did the hard-sphere model. The differences are naturally most significant at low temperatures, where the hard-sphere model is particularly
inappropriate, a phenomenon which is marked by the temperature dependence of the apparent value of d- 2 (Figs. 2.7 and 2.8). The greater variation of d at low temperature reflects the fact that the mean collision energy

Chap. 2

42

Kinetic Theory of Gases-Transport

Table 2.7. Parameters for the Lennard-Jones Potentiala

From viscosity
Gas

Ne
Ar
Kr
N.
O.
CO.
CR,

Elk
(K)

35.7

a
(nm)

0.2789

From second virial coefficient

Elk
(K)

35.60

a
(nm)

0.2749

124

0.3418

119.8

0.3405

190

0.361

171

0.360

91.5

0.3681

95.05

0.3698

113

0.3433

117.5

0.358

190

0.3996

189

0.4486

137

0.3882

148.2

0.3817

a Data from J. O. Hirschfelder, C. F. Curtiss, and R. B. Byrd, Molecular Theory o/Gases and Liquids (New York: John Wiley, 1954), pp. 1110-1112.

is small so that particle trajectories deviate substantially from those of

rigid spheres (Fig. 2.9). By astute choice of and (J the potential (2.45)
can be adjusted to fit either transport or vi rial data. The corresponding
Lennard-Jones parameters are tabulated in Table 2.7 for a few gases. The
difference between the parameters best suited to describe the viscosity and
those which fit the second virial coefficient is small; the molecular diameters
of Table 2.6 show much greater sensitivity to the method of determination.
On the other hand, the fact that different parameters are needed indicates
that (2.45) does not properly describe the intermolecular interaction.
With the development of high-speed digital computers it has become
possible to determine a single potential which correlates all two-particle
properties. For argon, where the most extensive experimental data are
available, such a potential was determined by Dymond and AlderY7) It is
shown in Fig. 2.10; for comparison the Lennard-Jones potential based on
viscosity data is also plotted. The curves are significantly different. In the
attractive region [u(r) < 0] the Dymond-Alder function has a deeper,
narrower potential well, while, in the repulsive region, it increases much
more slowly. A Lennard-Jones potential using virial parameters would
differ very little from the one using viscosity parameters. Neither accurately
approximates the Dymond-Alder potential for any value of r. Thus the
(17)

J. R. Dymond and. B. J. Alder, J. Chem. Phys. 51, 309 (1969).

Sec. 2.8

Comparison with Experiment

43
5

200

I
I

I
I

u(r)/k
0

0.2

10 .4

~\

V.

"

100

(K)

log u/k 3

0.2

0.4

r(nm)

0.6

r (nm)
-100

Fig. 2.10. Comparison of Dymond-Alder (-) and Lennard-Jones (- -) potential-energy

functions for argon. The insert compares them in the domain where repuisive forces
dominate.

agreement of viscosity and virial based parameters in Table 2.7 can be

misleading. It would be incorrect to assume that using (2.45) with the
two sets of parameters would provide bounds for the true intermolecular
potential.
Determining the numerical potential of Fig. 2.10 is a formidable task,
even with modern computers. However, once the potential is known, it is
relatively simple to use it to calculate thermodynamic and transport properties. It is therefore natural to hope that the argon potential could be
expressed in a parametric form and used to compute properties of other
gaseous systems. This would be particularly important for determining gas
properties in regions of high temperature not easily reached in the laboratory. While the available data are insufficient to provide a completely
reliable test, it appears that in a two-parameter form the potential of
Fig. 2.10 can be used to correlate the properties of dilute rare gases. (18)
An alternate approach to determination of the pair potential u(r)
is to calculate it using quantum mechanics. Perturbation calculations,
valid at large intermolecular separations, have been carried out for many
atomic and molecular systems. At small interparticle separations the
problem is greatly complicated since an atomic pair can no longer be
(18)

For an improved, simpler u(r) see R. A. Aziz and H. H. Chen, J. Chern. Phys. 67, 5718
(1977).

Chap. 2

44

Kinetic Theory of Gases-Transport

described as weakly interacting. Electron indistinguishability, resulting

in exchange interactions, must be specifically incorporated; hence, for
N-electron atoms, a 2N-electron wave function, which is not a simple
product function, is required. As a result the Ar-Ar potential is still too
complicated for accurate calculation to be possible. The He-He and Ne-Ne
potentials represent the present limit of accurate theory. (19)

Appendix. Coordinate Transformation by Jacobian Method

To express the differential volume element in terms of new variables
is most readily done using Jacobians. (20) Given the coordinate transformation
x = x(u, v),
y=y(u,v)
the relation between dx dy and du dv is
dx dy

I lex, y I u, v) I du dv

where the Jacobian of the transformation is the determinant

lex, y I u, v)

(~~t (~~
(~~t (~t

The formalism may readily be extended to transformations involving more

variables by constructing the analogous determinants. As an example
consider the transformation from Cartesian to polar coordinates
x

r cos 0,

y = r sin 0

The corresponding Jacobian is

lex, y I r, 0)

which indicates that dx dy

cos 0 -r sin 01
sin 0
rcosO

r dr dO, the familiar result.

H. Margenau and N. R. Kestner, Theory of Intermolecular Forces, 2nd ed. (Oxford:

Pergamon Press, 1971), Chapters 2 and 3; J. N. Murrell, in Rare Gas Solids, M. L. Klein
.and J. A. Venables, eds. (London: Academic Press, 1976), Vol. I, Chapter 3.
(20) For proof of the relations to be stated see H. Margenau and G. M. Murphy, The Mathematics ofPhysics and Chemistry, 2nd ed. (Princeton: Van Nostrand, 1956), pp. 192-194.
(19)

45

Problems

Problems
2.1.

What is the probability that an A atom in air at 300 K and 1 atm travels
1 [Lm without making a collision? 10 [Lm? (Use the data of Table 2.6 and
assume N2 and O. have the same effective molecular diameters.)

2.2.

Show that integration of (2.9) yields (2.10). [Choose the polar axis in the
direction of C1 so that C12 = (Ci' + co' - 2c 1c. cos 0.)1 /'.]

2.3.

Use the Jacobian method to show that the differential volume element in
spherical polar coordinates is c' sin 0 dc dO drp.

2.4.

Use the Jacobian method to prove the statement made in Section 2.4,
= dCM dc; the coordinate transformation is given by (2.15).

dC 1 dc.

2.5.

Modify the analysis of Section 2.2 and show that in a two-dimensional

gas the limiting mean free path [the analog to (2.5)] is 0.5 [nA d AA +nB dAB]-'.

2.6.

By constructing the two-dimensional analog to (2.18) show that the collision frequency in a two-dimensional gas is
2(nRT/MA)l/'[nA d AA

+ nB d AB (1 + MA/MB)l/']

This is the analog to (2.19).

2.7.

Imagine that it is possible to construct and maintain a system with the

sinusoidal temperature profile T(z) = To + bTsin(mnz/L), where L is the
width of the container and m is an integer. What is the heat flux at the edges
of the container? Show that if the system is allowed to decay to equilibrium
the time-dependent temperature profile is
T(l, z) = To

mn ). "Vt ]
mnz
bTexp [ - ( ---y;- -c;;- sin---y;-

2.8.

Demonstrate that (2.30) satisfies (2.27) and fits the boundary conditions
(2.29).

2.9.

Another important, and soluble, one-dimensional diffusion problem is the

example for which there is constant matter flow into one end of a cylindrical
tube. The boundary conditions for this problem are

ac
Daz

c = co,

all z,

= -J = const

'

t = 0

z = 0, all

Demonstrate by direct differentiation that the solution to (2.25) consistent

with these constraints is

c
where s

Co

+ -Jz
D

z/[2(Dt)l/'].

[exp( -s')
nl/'s

2
- -nl/'

~ exp(-y')dy

Chap. 2

46

Kinetic Theory of Gases-Transport

2.10. At 15C and 1 atm the viscosity of NH. is 9.6 x 10- 6 kg m- ' sec- ' and the
heat capacity is 0.52 cal g-l. Estimate both D and x. The observed value
of x is 0.040 J K -1 m- 1 sec ' ; comment on any difference between your
estimate and the experimental result.
2.11. The Lennard-Jones potential can be used to compute a temperaturedependent effective hard-sphere d in the following fashion. Assume that
d is the value of r at which u(r) is the mean thermal energy, 1.5kT; this
accounts, in a rough way, for the Sutherland effect. Make this assumption
and compute d(T) for A at 100, 200, 300, 400, and 500 K. Then compare
your results with the data in Fig. 2.8. Why do you think this approach works
better at high temperatures?
2.12. At 273 K the viscosity of gaseous Br, is 1.46 x 10-' kg m- 1 sec- ' ; estimate
the molecular diameter of Br, molecules. Using the data in Table 2.6 determine the m~an free path of Br, molecules in 0, gas at 273 K and 1 atm.
2.13. The diffusion coefficient in a binary mixture is D = !(I/n""nd~B)(n""kT/2)l/'
where"" is the reduced mass of an AB pair. Estimate the diffusion coefficient
of Br, in air at 273 K and 1 atm assuming that N, and 0, have the same
diameter and a mean mass.
2.14. Assume that a tube filled with air at 273 K and 1 atm is placed in contact
with a reservoir containing Br, at its vapor pressure, 0.0844 atm. How long
will it take for Br2 to become visible at a distance 0.5 m down the tube?
This concentration is approximately equivalent to a pressure of 0.006 atm.
Use the results of Problem 2.13 to solve this problem.
2.15. The speed-dependent differential flux can be determined by extension of
the arguments that led to (2.31). It is

dJ

-n(z)F(e; z)[cos x/4nr2Jc(e)]ee-r/),ICI dr de

where F(e; z) is given by (1.24) with a position-dependent temperature.

Since the kinetic energy flux is
JqlKI

!me 2 dJ

show, by expanding T(z) in a Taylor's series around z - zo, that

Jql K I

no_
_
3To

f Fo(e)Jc(e) -1 me' ( -me- 2

2kTo

5)]
de -dT
dz

where no == n(zo), etc.

2.16. The internal energy of a molecule is E - !me 2; thus the mean internal
energy of a molecule in the volume element dr is l(z) - !kT(z). Assuming

47

Generai References

that internal energy is equilibrated at each collision and that

where C1 is ihe molecular internal heat capacity show that

JqiIl =

nOc1
[-

f Fo(c)Jc(c)c (------;:;:;mc' - -3)]

dT
dc -

2k1o

dz o

Comparison of JolKl and Joill indicates that the transport of kinetic and
internal energy involves different averaging processes. Thus x is not simply
proportional to the total molecular heat capacity, C1 + l.5k.
2.17. What are the conditions for Knudsen flow in a vessel of length 0.1 m?
Assume molecular diameters of ~0.4 nm. To what pressure does this
correspond at 298 K?
2.18. Demonstrate that (2.37) follows from (2.36).

General References
Theory of Gas-Phase Transport
L. Boltzmann, Lectures on Gas Theory, translation by S. G. Brush (Berkeley: University
of California Press, 1964).
S. Chapman and T. G. Cowling, Mathematical Theory of Non-Uniform Gases, 3rd ed. (Cambridge: The University Press, 1970), Chapters 5-8.
J. O. Hirschfelder, C. F. Curtiss, and R. B. Byrd, Molecular Theory of Gases and Liquids
(New York: John Wiley, 1954), Chapters 1, 7 and 8.
M. H. C. Knudsen, The Kinetic Theory of Gases, 3rd ed. (New York: John Wiley, 1950).
W. Kauzmann, Kinetic Theory of Gases (New York: Benjamin, 1966), Chapter 5.
J. E. Mayer and M. G. Mayer, Statistical Mechanics (New York: John Wiley, 1940),
pp. 18-30.
R. D. Present, The Kinetic Theory of Gases (New York: McGraw-Hili, 1958), Chapters 3,
4, 7, and 8.

N onequilibrium Thermodynamics
D. D. Fitts, Nonequilibrium Thermodynamics (New York: McGraw-Hili, 1962), Chapters
1-3.
S. R. de Groot and P. Mazur, Nonequilibrium Thermodynamics (Amsterdam: NorthHolland, 1962), Chapters 2--4.
I. Prigogine, Thermodynamics of Irreversible Processes, 3rd ed. (New York: Interscience,
1967), Chapters 1-5.

Electrolytic Conduction
and Diffusion
3.1. Introduction
There are a number of features which distinguish kinetic problems in
solutions from those in the gas phase. The most obvious is that the chemical
entities being studied are different in the two phases. A sodium ion in the
gas phase is a bare ion. In solution it carries with it a sheath of solvent,
i.e., it is solvated. This difference is apparent in the equilibrium properties
of solutions. In most instances there are both enthalpy changes and volume
changes upon solvation. These effects clearly indicate that the solvated
species and the pure substance differ significantly, differences which also
affect the kinetic properties.
The solvent has another important effect. Unlike the gas phase, in
which the medium is a vacuum through which molecules travel in linear
paths except for the occasional instance of interaction, collision, and
deflection, in solution the molecules of the species of interest are always
interacting with solvent. This interaction radically alters the molecular
motion. Instead of traveling in straight lines the molecules move randomly
through the maze provided by the solvent. One can describe the average
motion of molecules in solution by means of the laws of diffusion; the
motion of individual molecules is subject to such a variety of influences
that precise description is hopeless.
Solvents of high dielectric constant like water stabilize ionic species.
Thus much interest in kinetic properties in solution necessarily focuses
on the behavior of ions. This significantly complicates the discussion
49

50

Chap. 3

it

Electrolytic Conduction and Diffusion

since the coulombic interaction extends over a long range. The requirement
of electro neutrality means that ions cannot diffuse through a solution
independently of one another even in very dilute solution.
Before we treat the phenomenology of chemical kinetics, assuming
we wish to discuss ionic reaction in solution, we must consider the nature
of electrical conduction in solutions. To this end we first review some properties of electrolytes and the mechanism for transport of electric current
in solution. We shall find that electrical conduction and solute diffusion are
processes that are inextricably coupled, a coupling that can be used to
relate the transport coefficient describing electrical conduction, the conductivity A, with that describing diffusion, the diffusion coefficient D.
By incorporating some ideas from the general theory of fluid flow we shall
be able to relate A (or D) to molecular properties of the solvated species.
Conductance measurements provide a way to determine D, thus establishing
limits for the reaction rate in solution. In addition, when carried out in
high electric fields, such measurements can be used to imply the existence
of ion pairs, species that are different from both solvated molecules and
solvated ions. For simplicity, and because it is the case of greatest chemical
interest, our discussion will be limited to constant temperature and constant
pressure processes.

3.2. Ionic Conduction in Solution-a Review

In metals, current flow is due to electron motion. This is not the case
in electrolytic solution; here a potential difference leads to net ionic migration. It is the motion of the ions which accounts for charge transfer in
solution.
For consideration of ionic conduction in solution, Ohm's law is most
conveniently formulated in a form analogous to the transport equations
presented in Section 2.4,
j = -a' 17'lf' = a'E

(3.1)

Here j is the current density, or equivalently, the charge flux. The electric
field E, which is the gradient of the potential 'If', provides the driving force
for conduction. The transport coefficient a' is the specific conductivity,
expressed in ohm-I m- I . While (3.1) may appear unfamiliar, it reduces
to the common expression I = VIR in the special case of conduction in a
wire of uniform cross section (see Problem 3.1).

Sec. 3.2

Ionic Conduction in Solution

51

To relate the transport coefficient a' to the properties of the solvated

species requires recognizing that even pure solvent contains some ions.
Thus, the results of conductivity measurements are corrected for the
specific conductivity of the cell in the absence of electrolyte, ao; the conductivity attributed to electrolyte is then

a' - ao

(3.2)

Electrolytic conduction is due to the presence of ions; thus a is a

function of the concentration of equivalents of ions, C. (1) Strong electrolytes
like NaCI, HN0 3 , etc., are fully dissociated while weak electrolytes like
HCN are not. In the former case C is the same as the normality of the
electrolyte C; in the latter it is not. As a consequence the conductivity of
a strong electrolyte is roughly proportional to C while that of a weak one
is not. It is convenient to define the equivalent conductance which accounts
for the concentration dependence

A
Since

C is not always easy to

alC

(3.3)

measure the apparent equivalent conductance

Aapp = alC

(3.4)

is also useful. The concentration dependence of Aapp is illustrated in Fig.

3.1. For strong electrolytes Aapp decreases slowly with increasing C while
for weak electrolytes the decrease is precipitous.
Strong electrolytes at the same concentration obey an approximate
additivity rule in A app , e.g.,
(3.5)
a relation which becomes precise in the limit of infinite dilution. If each
ionic species contributes independently to the conductivity and if each is
completely dissociated it is easy to rationalize (3.5). Thus we decompose
A as
(3.6)
where the A are the ion equivalent ~onductances. Now, since C = C for
strong electrolytes (complete dissociation) the data of Fig. 3.1 require that
11)

A 0.1 M solution of NaCI is also 0.1 N. However, a 0.1 M solution of Na 2 SO. is 0.2 N;
it contains 0.2 equivalents of positive (and negative) charge per liter.

Chap. 3

52

Electrolytic Conduction and Diffusion

1.5

I\OPP

x 10

(m 2 ohm-I eq-I)

HAc

0.4

0.5

Fig. 3.1. Apparent equivalent conductance, Aapp, of various electrolytes as a function of

concentration.

A itself be concentration dependent, an observation that is generally

consistent with the equation

AO - aC1I2

(3.7)

where AO is the equivalent conduction at infinite dilution. Here a is a speciesdependent constant. A model accounting for the concentration dependence
is discussed in Section 3.7.
From (3.6) and (3.7) it is apparent that a quantity of basic interest in
electrolytic conductance is the ion equivalent conductance at infinite
dilution. Ordinary conductance measurements cannot separate the contribution of individual ions. By designing experiments that distinguish
between ion migration toward positive and negative electrodes in a currentcarrying cell it is possible to measure A+ and L separately. A tabulation
of the results of such measurements is given in Table 3.1. (2)
(2)

A variety of techniques is available for determining electrolyte and ion conductivity.

For a description of the methods and their drawbacks see R. A. Robinson and R. H.
Stokes, Electrolyte Solutions (London: Butterworths, 1955), Chapter 5.

Sec. 3.3

Nernst-Einstein Relation

53

Table 3.1. Ion Equivalent Conductivities at Infinite Dilution,

at 25 0 ca
Cation

Anion

in Aqueous Solution

A" x 10'
(m 2 ohm- 1 eq-l)

H+

3.498

OH-

1.978

Li+

0.3866

F-

0.544b

Na+

0.5011

CI-

0.7635

K+

0.7352

Br-

0.7820

Rb+

0.778

1-

0.769

Cs+

0.773

NO a-

0.7144

Ag+

0.6192

CIO a-

0.646

NH.+

0.734

BrOa-

0.558

!Mg2+

0.5306

10a-

0.405

!Ca2+

0.5950

Acetate-

0.409

!Sr2+

0.5946

HSO.-

0.50b

!Ba 2+

0.6364

HCO a-

0.445 b

!Cu2+

0.54

CN-

0.78 b

iZn +

0.53

!C,O!-

0.240

iNi2+

0.49 b

iSO!-

0.800

!Fe 2+

0.54b

tFe(CN)."-

1.00

!Pb2+

0.70b

!Fe(CN).4-

1.11

tFe3+

0.684b

,1.0 x 10'
(m 2 ohm- 1 eq-l)

,1.0,

tCr"+

0.67 b

tLa3+

0.695

Except where noted data from H. S. Harned and B. B. Owen, The Physical Chemistry of Electrolyte Solutions, 3rd ed. (New York: Reinhold, 1958), p. 231.
H. H. Landolt and R. Bornstein, Zahlenwerte und Funktionen, 6th ed. (Berlin: Springer, 1969),
Vol. II, Part 7, pp. 258-267.

3.3. Nernst-Einstein Relation

According to Ohm's law, applying an electric field to an ionic solution
leads to an ion current which creates a concentration gradient. This gradient
immediately sets up a diffusive flow to attempt to restore homogeneity.
Both processes involve ion flow; the diffusive flux and the charge flux are

Chap. 3

54

Electrolytic Conduction and Diffusion

inextricably coupled. The consequence, as we shall see, is that the IOn

equivalent conductance is related to the ion diffusion coefficient.
Consider a single cationic species of equivalent conductance ,1+, ion
concentration n+ and ionic charge +z+. From (3.3) and (3.6) such a species,
at low concentration, contributes z+n+A+ to the specific conductance of
the system, and, from (3.1), the cationic contribution to the current density is
(3.8)
The ion flux required to produce this current density is (j+/z+F), where F,
Faraday's constant, is the number of coulombs per equivalent of charge.
Thus the total cation flux in a potential field, found by combining (3.8)
with Fick's law (2.23) is(3)

. + z+F
i+ =

J+ = J+(dlff)

-D+

[7

n+A+ [7
n+ - - F - 1p

(3.9)

The corresponding anion flux is

J-

-D_ [7 n_

[7
+ -n_A_
F - 1p

(3.10)

where the sign reversal in the second term reflects the fact that a potential
gradient drives positive and negative ions in opposite directions.
To establish the relationship between D and A assume, for simplicity,
that the anionic species is essentially immobile (as would be the case were
it very large) so that J_ = O. Now assume further that the potential gradient
can be adjusted, if only in principle, to balance the concentration gradient
so that J + = O. The necessary condition, from (3.9), is
(3.11 )
With these constraints the system remains electrically neutral throughout
as there is no net ionic flow.
(3)

This formulation is based upon the simple form of Fick's law (2.23). In fact the flux
of a species depends not only on its own concentration gradient but also on the
concentration gradients of all other species in solution. A more complete theory would
have to account for such phenomena. While the precise expressions may be affected,
the general conclusions are not significantly different from those based upon (3.9)
where cross-diffusion terms are ignored. This and other coupled phenomena are discussed
in S. R. de Groot, Thermodynamics of Irreversible Processes (Amsterdam: North-HoIland, 1963).

Sec. 3.3

Nernst-Einstein Relation

55

Consider the potential energy of the cations due to the electric potential;
it is simply

where e is the electronic charge. As 1p is position dependent, so is U +.

For the example being treated-no ionic flow, constant temperature,
constant pressure-the cation properties may be treated from an equilibrium
viewpoint. Thus the concentration profile of the positive species is related
to its potential energy by a Boltzmann distribution
(3.12)
where n+o is a reference state concentration which is not position dependent.
Taking the gradient and substituting in (3.11) immediately yields

from which we obtain the Nernst-Einstein relation(4)

(3.13)
Had the cations been assumed immobile a similar formula would be found
relating A_ to D _L.
The derivation of (3.13) was based on the assumption that only cations
are current carriers. However, the result is general. Ion equivalent conductance is a property of the individual ionic species, independent of the nature
of the counterion (at least in dilute solution). Choosing a counterion for
which D = A = 0 simplifies the argument; it does not affect the result.
Limiting consideration to dilute solution, (3.6) and (3.13) can be combined
and we find(5)
(3.14)

14)

15)

We can now recognize why only the cation is viewed from an equilibrium standpoint.
Its electrochemical potential is fi+ = 11+" + RTlnn+ + z.FIJ! which, with (3.12), is
fi+ = Il+" + RTln n+". Since both Il+" and n+" are position independent so is fi+, as
it must be in equilibrium. The electrochemical potential for the anion is fi- = IC" +
RTln n_ - z_FIJ!. Since electroneutrality demands that z+n+ = Ln_ we find, using
(3.12), that fi- = 11_" + RT In n_" - (z+ + z_}F1J! which, since IJ! is position dependent, is not constant. Thus the anions are not at equilibrium and we see the importance
of the assumption of anionic immobility for establishing (3.13).
W. Nernst, Z. Physik. Chem. 2, 613 (1888).

Chap. 3

56

Electrolytic Conduction and Diffusion

3.4. Electrolyte Diffusion

According to the Nernst-Einstein relation (3.13), the conductance
and the diffusion coefficient are proportional and the data of Table 3.1
can be used to calculate single-ion diffusion coefficients. Thus D C1- ' "
2D Li +, and, were a concentration gradient established in a solution of
LiCI, CI- would initially diffuse twice as fast as Li+. Such a situation cannot
last; charge separation creates an electric field accelerating Li+ and braking CI-.
To determine the electrolyte diffusion coefficient combine (3.9) and
(3.10) with 0.13),
(3.15)
The plus signs refer to cations and the minus signs to anions. With a binary
electrolyte (A H+ )v/B-z-)v_ at concentration n we have
(3.16)
and
(3.17)
To maintain e1ectroneutra1ity the ion fluxes must be coupled,
(3.18)
Combining (3.15)-(3.18) we obtain the local electric field due to the coup1edion flow

which when substituted in (3.15) leads, with (3.16), (3.17), and the observation that J = J +/v +, to alternative forms

a result also obtained by Nernst. (6) The electrolyte diffusion coefficient is

an average, always intermediate between D + and D _, which is reasonable
since the local field induced by the difference in ion mobility accelerates
the less mobile ion and slows up the faster one.
16)

W. Nernst, Z. Physik. Chern. 2, 613 (1888).

Sec. 3.5

Stokes' Law

57

3.5. Stokes' Law and the Microscopic Interpretation of 1

One of the early results of continuum fluid mechanics was the determination of the frictional force exerted on a spherical particle as it moved
through a viscous fluid. Stokes' law relates the force on the particle f,
to the particle radius r, its drift velocity v, and the viscosity of the medium 'Yj,
f = 6Jt'YjTv

(3.20)

a result derived assuming that the fluid close to the particle is moving with
the same velocity as the particle, i.e., the fluid sticks to the particle. A great
body of experimental evidence demonstrates that (3.20) accurately accounts
for the behavior of macroscopic particles. How and why (3.20) can be used
to describe the motion of ions in a solvent is unclear. From the standpoint
of the ion the solvent is certainly not a continuous medium; it has a
molecular structure. (7) Nonetheless we shall assume that (3.20) describes
the average force acting on ions migrating in a solution. A partial justification is found in the molecular insights that will be obtained. However,
the application of (3.20) to ion migration must be considered to be a
postulate, not a consequence of fluid mechanics.
In an electric field E = -17"'1/', the force on a representative ion of
charge z is zeE; if the ion is not being accelerated this force is balanced
by friction in the same direction so that
zeE = 6Jt'Yjrv

(3.21)

The ion mobility u, defined as velocity per unit electric field strength, (8)
is therefore
(3.22)
u = vjE = zej6JtYjf
To relate u to A,o consider an ionic species with mean velocity v; the ion
(7)

(8)

It is a question of some interest whether the solvent can be presumed to stick to the
solvated ion as it diffuses through the solution. If instead it is assumed that the solvent
slips by the solute the factor 6.n in (3.20) is reduced to 4.n. This has no great qualitative
effect on the picture to be developed. For a derivation of Stokes' law see R. M. Fuoss
and F. Accascina, Electrolyte Conductance (New York: Interscience, 1959), pp. 53-59.
Theoretical studies on hard-sphere liquids indicate that Stokes' law applies if it is
assumed that solvent slips by the molecules [B. J. Alder, D. M. Gass, and T. E. Wainwright, J. Chern. Phys. 53, 3813 (1970)].
Since E, v, and j are oriented in the same direction, they are proportional to the same
unit vector n = vii v I = Ell E I = j/l j I. It is thus permissible to "divide" the vectors since their n dependence cancels.

Chap. 3

58

Electrolytic Conduction and Diffusion

II

Table 3.2. Comparison of Crystal Radii, rc , with Ionic Radii in Aqueous Solution,
r A , as Estimated from (3.23)
Species

H+
Li+
Na+
K+
Rb+
Cs+
Mg'+

rA

rca

(nm)

(nm)

0.026
0.23
0.18
0.12
0.12
0.12
0.35

0.06
0.10
0.13
0.15
0.17
0.07

Species

CaH
SrH
Ba'+
OH-

ClBr1-

rA

rca

(nm)

(nm)

0.31
0.31
0.29
0.046
0.12
0.12
0.12

0.10
0.11
0.14
0.18
0.20
0.21

a L. Pauling, The Nature of the Chemical Bond, 3rd ed. (Ithaca: Cornell, 1960), p. S18.

flux is J = nY, where n is the ion concentration. Recognizing that the

current density is
j = zFJ = zFnv
we find, using (3.22) and Ohm's law (3.1),
a = j/E

= nzFv/E = nzFu = nz2eF/6nw

from which, using (3.3), the ion equivalent conductance at infinite dilution
can be calculated,
;'0

= a/e = a/nz = Fu = zeFj6n'Y)r

(3.23)

Thus ;'0 can be used to determine ionic radii in solution. Some values of r
are given in Table 3.2; for comparison, crystal radii of the ions are also
inc1uded.(9l Except for H+ and OH-, for which r appears to be absurdly
small, the Stokes' law values are reasonable and comparable to the crystal
radii. However, the trends in both the alkali and alkaline earth ion series
are opposite. In part this may be rationalized by recognizing that a small
ion, like Li+, binds its water of solvation more strongly than a larger ion,
like Cs+; this suggests that the effective ionic radius of solvated Li+ could
be larger than that of Cs+. However, the major source of the discrepancy
is the use of Stokes' law itself. Applying a formula designed to describe
the forces on macroscopic objects in continuum fluids to a molecular
(9)

Using the slip condition (see footnote (7) in this Chapter] the values of r estimated
from (3.23) increase by 50%.

Sec. 3.5

Stokes' Law

59

problem cannot possibly be correct; the fact that it works as well as it

does is somewhat surprising.
An important consequence of (3.23) is Walden's rule
'rjA O = (const)/r

(3.24)

Therefore, if the radius of the solvated ion is an invariant, the product

'rjAo is constant. Assuming that r is constant for both anion and cation,

a more general form is

(3.25)
Figure 3.2 illustrates that Walden's rule accounts for most of the temperature
dependence of ion conductivity in water solution; over the range considered
the viscosity of water decreases tenfold. Again H+ and OH- are anomalous.
The more highly charged ions deviate the least from Walden's rule; a possible interpretation is that such ions bind their water of solvation more
strongly and are less affected by temperature variation.
Assuming no change in the nature of solvation, Walden's rule can be
used for crudely estimating conductivity in different solvents. One example,
for NaCI and KCI in water-methanol solution, is given in Table 3.3. Here,
(3.25) appears accurate as long as the solution is rather dilute (25 wt. %
methanol is 16 mole %), but, as the methanol concentration increases,
deviations are substantial suggesting major changes in solvation. A second
example, for KI in a variety of solvents, is given in Table 3.4. The Walden's
rule product is roughly constant for the organic solvents but water and
S02 differ greatly.

La 3 +
1.0

-----------Ca2 +

-------____

'1 (t)

).0(t)

,,\(0) ).(0)

Fe(CN):- - - - - K+

ow

0.5

O~~----~~----~~------i-----o
50
100
150

t(Oc)
Fig. 3.2. Test of Walden's rule. The ratio [1](t) ).Ct)/1](O) ).(0)] for a series of ions as a
function of temperature.

60

Chap. 3

Electrolytic Conduction and Diffusion

Table 3.3. Tests of Walden's Rule. The Product AOrJior KCI and NaCI in MethanolWater Mixtures at 25 0 ca
A1) x 10"

Methanol (wt. %)

1)

x 103

(kg m- i sec-i)

0
25

(ohm- i m kg sec- i eq-i)

KCl

NaCI

0.8949

1.341

1.131

1.475

1.334

1.150

50

1.54

1.15

1.01

75

1.15

0.896

0.808

100

0.541

0.565

0.527

a H. S. Harned and B. B. Owen, The Physical Chemistry of Electrolyte Solutions, 3rd ed. (New
York: Reinhold, 1958), p. 706.

Table 3.4. Tests of Walden's Rule. The Product A1) for KI in Various Solvents
at 25 0 ca
Solvent

Ax102
(ohm- i m 2 eq-i)

1) x lOS
(kg m- i sec-i)

AI'j X 105

Acetonitrile

1.982

0.345

0.684

Acetone

1.855

0.316

0.586

Nitromethane

1.240

0.611

0.758

Methanol

1.148

0.546

0.627

Ethanol

0.509

1.096

0.560

Furfural

0.431

1.490

0.642

Acetophenone

0.398

1.620

0.644

Waterl'

1.5038

0.8937

1.344

S02c

2.65

0.394

1.044

a J. O'M. Bockris and A. K. N. Reddy, Modern Electrochemistry (New York: Plenum Press,
1970), Vol. 1, p. 386.
b H. S. Harned and B. B. Owen, The Physical Chemistry of Electrolyte Solutions, 3rd ed. (New
York: Reinhold, 1958), p. 697.
c At 00C.

Sec. 3.6

Mobility

0/ H+ and OH- in Water

61

3.6. The Mobility of H+ and OH- in Water

In the previous section we saw that Stokes' law could be used to estimate
reasonable values of ionic solvation radii and that 'fjAo is only slightly
dependent upon temperature for a variety of ions in aqueous solution.
There are two glaring exceptions, H + and OH -; their solvation radii,
0.026 and 0.046 nm, respectively, are ridiculously small. The temperature
dependence of 'fjA o is very great.
A possible explanation for these large conductivities (small radii)
is given if we consider the structure of water as shown schematically in
Fig. 3.3. Proton diffusion in water can proceed by a cooperative method
that does not require ion migration over large distances. First a hydrogen
bond is broken and reformed as in Fig. 3.3a; the distance traveled is
;S 0.05 nm. This then is repeated as in Fig. 3.3b. The original water molecule
may then rotate as in Fig. 3.3c and be prepared to accept another proton.
The net effect is the displacement of a proton by ",0.3 nm without any
particular H + actually moving that distance. This mechanism is unique
to H+ (or OH-) in water. Ionic migration of any other species requires
significant solvent motion to permit the physical displacement of the ion;
the net effect of such motion must be that a solvent molecule and a solute
molecule exchange positions in the fluid.
The proton-jump model for diffusion is based upon the premise that
the structure of liquid water derives from an icelike hydrogen-bonded
network; in addition to the two normal OH bonds in water each oxygen
atom is linked to neighboring ones via hydrogen bonds. The linkage is
imperfect and there are defects in the network. These imperfections, known
H
+

H-O:H --..

O-H

H
I
O-H
(a)

H-O

O-H

H
1+
H-O-H

H
1

H-O

H
1
O-H
(b)

H
H-b:H
(e)

O-H
1
H

O-H
I
H

O-H
I
H

H
I
O-H

H
I
O-H

H
1
O-H

Fig. 3.3. Proposed mechanism for proton diffusion in water. (a) Proton jump. (b) Proton
jump. (c) Rotation of water molecule.

62

Chap. 3

Electrolytic Conduction and Diffusion

as Bjerrum faults, (10) are of two types: a pair of oxygen atoms with two
hydrogen atoms between them or a pair of oxygen atoms with no hydrogen
atom between them. In some steps of the diffusional process such faults
must be formed as shown in Fig. 3.3c, where molecular rotation has broken
the hydrogen-bond network by creating an (O-H H-O) domain. While
no theory offault dynamics in water has been shown to correlate its electrical
properties, the Bjerrum model is known to account for the anomalous
electrical properties of ice. (11)

3.7. The Concentration Dependence of A

From the data presented in Fig. 3.1 it is clear that, at low concentration,
the equivalent conductance det:reases with increasing concentration

AO -

aCl/2

(3.26)

which can be accounted for if the effect of the ion atmosphere is taken
into consideration. At equilibrium an ion in solution is surrounded by
a spherical distribution of charge, just balancing the ionic charge itself,
as shown in Fig. 3.4a. When an electric field E is applied the ion migrates
in one direction while the ionic atmosphere, being oppositely charged,
moves in the opposite direction. Since the ion cloud is large it may be
treated as if it were a charged colloidal particle moving under the influence
of an electric field in a viscous fluid. In colloid chemistry this is known as
electrophoresis. As the ion atmosphere migrates it naturally tries to carry
its central ion with it; thus due to the electrophoretic effect there is an
effective field EE in the opposite direction to the applied field which decelerates the central ion.
In addition to the electrophoretic drag force, there is an additional
effect because a moving ion deforms its own ion atmosphere. As shown in
Fig. 3.4b an ion tends to pile up charge in front of it as it moves and tends
to leave a diminishing wake behind as the atmosphere tries to readjust
and reform the spherical distribution about the ion. The effect is to separate
charge which produces an electric field called the relaxation field E R , in the
direction of, but opposing the applied field. Considering both electrophoresis
and relaxation the local field at the ion is E - EE - ER . Since the mean
(10)

(11)

N. Bjerrum, Science 115, 385 (1952).

L. Onsager and M. Dupuis; in Electrolytes, B. Pesce, ed. (New York: Pergamon, 1962),
pp.27-46.

Sec. 3.7

Concentration Dependence 0/ A

Fig. 3.4. Deformation of ion atmosphere due to ionic motion. (a)

Spherical ion cloud surrounding representative ion at equilibrium. (b)
Oval ion cloud surrounding representative migrating ion.

63

Spherical
ion atmosphere

Oval ion
atmosphere

(b)

(a)

ionic velocity, from (3.21), is proportional to the field we have

Thus, using (3.22), the ion mobility is

where UO is the mobility at infinite dilution. (12) Using the same arguments
which led to (3.23) yields an expression for Awhich accounts for the effect
of the ion atmosphere,
(3.27)
At very low electrolyte concentration the ion atmosphere is very diffuse
and both EE and ER are weak. As the concentration increases EE and ER
will also increase. Thus the ratio (EE + ER)/E increases as concentration
increases, accounting qualitatively for (3.26).
Estimates of EE and ER in dilute solution can be obtained using DebyeHuckel theory in its simplest form. Assume the electrolyte to be point
charges interacting via coulomb forces only. Ignore the molecular structure
of both solvent and ions. Since the electrophoretic effect is due to the migration of the ion cloud through the fluid, the reduction in ion mobility can be
calculated using Stokes' law (3.21)
v
E

ze

UE=-=--

6nfjr

(3.28)

where UE is the mobility of the ion atmosphere; r is the radius of the ion
atmosphere. From the Debye-Huckel theory of ion activity coefficients, (13)
(12)

(13)

Since the vectors E, EE, and ER are parallel they can be treated as scalars in division.
See footnote (8) in this chapter for discussion of a similar problem.
See, e.g., T. L. Hill, Introduction to Statistical Thermodynamics (Reading, Mass.: Addison-Wesley, 1960), p. 321ff.

Chap. 3

64

Electrolytic Conduction and Diffusion

'0'
(\J

4>

'E

0)(

<

.c
o

(\J

1.31

1.27

1.23

.02

.04

.06

fC (MI/2)

Fig. 3.5. Comparison of equivalent

conductance, A, of various electrolytes as predicted by Debye-HiickelOnsager equation (3.30) with those
observed experimentally. [Adapted
from J. O'M. Bockris and A. K.
N. Reddy, Modern Electrochemistry
(New York: Plenum Press, 1970),
Vol. 1, p. 436.]

r can be identified as

(3.29)
where E is the dielectric constant of the solvent and ni are the ion concentrations. Combining (3.28) and (3.29) shows immediately that UE ex: CI12.
The effect of the relaxation field is far harder to compute and we shall not
attempt it here. Qualitatively, this field is produced by distortion of the
ion atmosphere. The corresponding ionic displacements depend upon
individual ion mobilities. However, as in the problem of electrolyte diffusion
(see Section 3.4), ionic motions are coupled. As a result the precise dependence of ER upon the ion equivalent conductance is rather involved.
Onsager(14) developed a quantitative description of the relaxation field;
he demonstrated that, like EE, ER is also proportional to x and thus to C1I2.
Onsager's final expression for the equivalent conductance has the form
(3.30)
where A depends on 'Yj, E, and T while B is a function of A+, A_, E, and T.
A comparison of (3.30) with experiment for various electrolytes is shown
in Fig. 3.5. A more sensitive test is to compare the Onsager coefficient
(14)

L. Onsager, Z. Physik 28, 277 (1927).

Sec. 3.B

Wien Effects

65

Table 3.5. Observed and Calculated Values of A + EAO from (3.30) in Aqueous
Solution at 25 0 ca
Electrolyte

LiCI
NaNO.
KBr
KCNS
CsCI
MgCI 2
Ba(NO')2
K2SO.

Observed slope x 10"

(ohm- t m7/2 eq-3/2)

Calculated slope x 10'

(ohm- t m7/2 eq-3/2)

2.56
2.60
2.78
2.42
2.40
4.56
5.08
4.43

2.30
2.35
2.54
2.46
2.54
4.60
4.76
5.04

a J. O'M. Bockris and A. K. N. Reddy, Modern Electrochemistry (New York: Plenum Press,
1970), Vol. 1, p. 437.

A + BAo with that determined experimentally from the slope of the equivalent conductance; some values are tabulated in Table 3.5. It is clear that
the Debye-Hiickel-Onsager theory accounts satisfactorily for the behavior
of A at low concentrations. As the ion concentration increases into regions
where Debye-Hiickel theory no longer accurately describes electrolyte
activity, there are also severe deviations from (3.30).
The theory of conductivity has been improved by eliminating some of
the assumptions implicit in its derivation. (10) The coupling of the relaxation
field and the electrophoretic field has been accounted for. The effect of
finite ion size and of ion-pair formation has also been taken into consideration.

3.8. The Wien Effects

Ohm's law is a nearly ideal example of the linear relation between
force (the electric field) and flux (the current). However, in aqueous solution
at field strengths ;<: 105 V m- I , Ohm's law no longer holds; the conductance
becomes noticeably field dependent. (16) In solutions of strong electrolytes
(15)

(16)

R. M. Fuoss and L. Onsager, J. Phys. Chern. 66, 1722 (1962); 67, 621 (1963); 68, 1
(1964).
In nonpolar solvents Ohm's law breaks down at much lower field strengths; measurable
deviation occurs at E ~ 10' V m- 1 as (the dielectric constant) -+ 1.

66

Chap. 3

Electrolytic Conduction and Diffusion

the conductance increases 5-10% in high fields and approaches a limiting

value A= for fields ~ 2x 10 7 V m- I (first Wien effect).(I7) In solutions of
weak electrolytes the conductance at high fields is proportional to the field
strength. No saturation occurs; in some instances there is even a quadratic
dependence of conductance on field strength (second Wien effect)YS)
The first effect arises because the ion atmosphere is greatly modified
at high fields. The ion velocity is very large and the ion traverses distances
comparable to that of the radius of the ion atmosphere in times so short
that the atmosphere cannot reform. As a consequence the relaxation field
is destroyed and the electrophoretic field significantly altered. There is
less retardation; It increases, approaching a high field limit.
The second effect is altogether different. Since the conductance IS
proportional to field strength, no high field saturation is found. It appears
that the number of charge carriers increases at high fields. This may be
understood if we postulate a second nonconducting species, the ion pair.
For a weak electrolyte AB there are then two equilibria(19):
Ko

AB = (A+, B-)

K1

A+

+ B-

(3.31)

In the ion pair (A+, B-) the ions are not yet individually solvated but are
partially separated by interpolated solvent. The equilibrium constant Ko
is not field dependent. However, high fields naturally favor separation of
the ion pair; more ions are produced, i.e., KI is an increasing function of
the field. For this reason the field dependence of conductance in weak
electrolytes is also known as the dissociation field effect. Arguing from (3.31),
the apparent equilibrium constant is
[A +][B-]

[ABlto t

(3.32)

where [ABltot = [AB] + [(A +, B-)]. A quantitative theory describing the

effect of field variation has been developed by Onsager. (20)
The Wien effects are more than just curiosities. They significantly
affect the kinetics of electrode processes, particularly in solvents of low
(17)

(18)
(19)

(20)

M. Wien, Phys. Z. 29, 751 (1928).

M. Wien and L. Schieb, Phys. Z. 32, 545 (1931).
For a discussion of the statistical mechanics of ion-pair formation see N. Davidson,
Statistical Mechanics (New York: McGraw-Hill, 1962), pp. 507-512. A kinetic derivation of K1 is given in Section 9.13.
L. Onsager, J. Chern. Phys. 2, 599 (1934).

Problems

67

dielectric constant. Since there is always a potential gradient near an

electrode due to the formation of an electrical double layer, there may be
significant non-Ohmic behavior in the vicinity of an electrode. (21) Such
phenomena must be considered when interpreting the kinetics of electrode
reactions.
Another nontrivial application of the Wien effect is in discriminating
between CO 2 (aq) and H 2C0 3 in order to accurately measure the true
equilibrium constant for the aqueous ionization(22)

The equilibrium constant for the hydration reaction

is not field dependent whereas K[ is. The apparent ionization constant is

given by (3.32) with Kl = K[ and Ko = K H ; high-field conductance measurements permit distinction between the two forms of dissolved CO 2,

Problems
3.1.

Consider a wire of cross section A and length I in which there is a potential

drop V and a current f. Show that (3.1) reduces to V = fR, where R =
IfAa'.

3.2.

A weak electrolyte is only partially dissociated. The equilibrium AB

+ B- has equilibrium constant K. Show that

A+

In an approximate form, valid when C is very small, A is set equal to AO;

the resulting expression for K is the Ostwald dilution law.

3.3.

(21)

(22)

The electrolyte diffusion coefficient, (3.19), may be related to ion equivalent

conductance using (3.13). Show that an alternative form for D is

There is an enormous literature on electrode kinetics; see P. Delahay, Double Layer and
Electrode Kinetics (New York: Interscience, 1965).
D. Berg and A. Patterson, Jr., J. Am. Chern. Soc. 75, 5197 (1953); K. F. Wissbrun, D.
M. French, and A. Patterson, Jr., J. Phys. Chern. 58, 693 (1954).

Chap. 3

68

Electrolytic Conduction and Diffusion

3.4.

Tracer diffusion studies on NaCI in water at 2SoC show that D(Na+) =

1.32 x 10- 9 m' sec- 1 and D(CI-) = 1.98 x 10- 9 m' sec-- 1. Using (3.19) compute D(NaCI); the measured value is I.S3 x 10- 9 m' sec- 1. Compare the three
values of D with those calculated using (3.13) and the result of Problem 3.3.

3.S.

Determine the local electric field in a cell of length 0.1 m containing an

aqueous solution of NaCl at 2SC. Assume a linear concentration profile
with C maintained at 0.01 M at one end of the cell and O.o1S M at the
other end. Estimate the drift velocity of the ions due to this field. Make
sure you handle units properly.

3.6.

Consider a membrane of thickness d across which there is a potential difference LltP. Show that the current density across the membrane is
j

(j

LltP

G LltP

where G is the conductivity per unit area of membrane surface. Now show
that
where C* is the concentration of charge carriers in the membrane.
3.7.

Use the data of Table 3.4 to estimate the ionic radius of K + in the various
solvents listed. You may assume, at this level of approximation, that K +
and I-have the same ionic radii. Can you suggest any explanation for the
changes observed?

3.8.

Liquid ammonia is an ionizing solvent with a high dielectric constant. Its

structure is probably like that of water, governed by a hydrogen-bonded
network, in this case with one hydrogen bond and three normal bonds per
ammonia molecule. Unlike the proton mobility in water, the proton mobility
in NH3 is not anomalously high. The value of ,10 is typical of univalent
cations in this solvent. Can you suggest any reasons which might account
for this difference?

3.9.

For 1-1 electrolytes in water solution at 2SoC the values of A and B in the
Onsager equation are A = 1.90 X 10-4 ohm- 1 m' eq-1 (m3 eq-1)1/2 and
B = 7.24 X 10- 3 (m3 eq-1 )1/2. Using these values and data from Table 3.1
calculate:
(a) the specific conductivity of a 0.1 M solution of KCI; the experimental
value is 1.2896 ohm- 1 m- 1;
(b) the specific conductivity ofa 0.1 M solution of HAc (KA = 1.8 X 10- 5 M).

3.10. For a binary electrolyte the expressions for A and B in the Onsager equation
are
A = 9.218 x 1O- 6 (z+ + L)3!2/,YJ(T)l!2

3.13

1O- 4 (z+

+ L)1!'w/(T)3/2

General References

69

where all quantities are in SI units and w is a dimensionless function of

the ion equivalent conductances:
w = 2qz+z_/(l

Z+L(A+

+ ql/2)

+ L)/[(z+ + L)(Z+A_ + LA+)]

By using mixed solvents the dielectric constant of the medium can be varied
continuously. The values of E in H20-CHaOH mixtures at 25C are
CHaOH (wt. %):
E:

25

50

75

100

78.54

67.8

56.3

44.9

32.66

Using these data and those of Table 3.3 calculate A for lO- a M and 0.1 M
solutions of KCl for the various solvents. Do you think the estimates of A
at the higher concentrations are reliable?
3.11. Show that (3.32) follows from (3.31).

General References
J. OM. Bockris and A. K. N. Reddy, Modern Electrochemistry (New York: Plenum Press,

1970), Vol. 1, Chapters 4 and 5.

H. S. Harned and B. B. Owen, The Physical Chemistry of Electrolyte Solutions, 3rd ed. (New
York: Reinhold, 1958), Chapters 4, 6, and 7.
R. A. Robinson and R. H. Stokes, Electrolyte Solutions (London: Butterworths, 1955),
Chapters 5, 6, 7, 10, and 11.

4
Determination of Rate Laws
4.1. Introduction
Altering the constraints on a chemical system poses two distinct
questions. What is the new equilibrium configuration? How rapidly does
the system approach this new state? The first is a problem of applied
thermodynamics; the second is the central problem of chemical kinetics.
The transport phenomena discussed in Chapters 2 and 3 are the simplest
examples of kinetic phenomena, if not the most familiar. For the cases
considered, a single macroscopic property (number, charge, momentum,
or energy density) was displaced from its equilibrium value. This perturbation caused a flux in the opposite direction, proportional to the displacement.
The proportionality constant is the transport coefficient. For simple gaseous
systems the relations between displacement and flux and the transport
coefficient were determined using a hard-sphere model to describe molecular
interaction.
Reacting systems are much more complex. Corresponding to the linear
phenomenological transport equations (Fick's, Fourier's, Ohm's, or Poiseuille's laws) are the nonlinear rate laws for chemical reactions. The
rate constant is the analog of the transport coefficient. The goal of theory
is to develop a molecular model which accounts for the rate law and which
provides an expression for the rate constant. These problems are, in many
ways, still poorly understood; a unified molecular theory such as was
developed for treating transport in gases does not yet exist. Instead, reaction
mechanisms, which are proposed as models for the underlying molecular
events, are constructed to be consistent with the observed rate laws. Then
very approximate theories for the rate constants are formulated in terms of
71

Chap. 4

72

Determination of Rate Laws

the elementary steps which comprise the mechanism. General theories,

based upon molecular distributions, are either approximations of uncertain
validity (if computationally tractable) or fairly rigorous (but applicable
to only the simplest systems).
In practical terms the study of chemical kinetics involves three separate
problems. The first is to establish, experimentally, the rate law which
describes the kinetic observations. The second is to determine a mechanism
(molecular model) which is consistent with the rate law. (1) The third is to
develop a theory for the rate constant in terms of the specific mechanism.
In this chapter we discuss three general problems. First, we consider
the rate law's relationship with and distinction from both reaction stoichiometry and reaction mechanism. Second, we describe some of the experimental techniques used in establishing rate laws. Finally, we analyze
some specific examples.

4.2. Stoichiometry, Rate Law, and Mechanism

The reactions between hydrogen and the halogens have the stoichiometry
(4.1)
However, when Bodenstein(2) studied the kinetics of the iodine and bromine
reactions he found that the experimental data led to completely different
formulas to describe the rate of production of HX
(4.2)
d[HBr]
dt

1 + kif [HBr]/ [Br 2]

(4.3)

These equations, discovered after repeated experimentation, and the way

in which the parameters k, k', k" depend upon the external conditions
(temperature, pressure, etc.) provide the data which are used to construct
a mechanism. A simple bimolecular mechanism, closely related to the
As we shall see in the next chapter there are always many mechanisms which are consistent with the data and a choice between them is made on various bases.
(') M. Bodenstein, Z. Phys. Chern. 29, 295 (1899); M. Bodenstein and S. C. Lind, Z. Phys.
Chern. 57, 168 (1907).
(1)

Sec. 4.3

73

Elementary Rate Laws

stoichiometry, was proposed to account for (4.2)<3)

(4.4)
For many years this was believed to be an accurate molecular description.
No simple mechanism accounted for (4.3). The following complex series
of steps, which could not have been guessed from the stoichiometry (4.1),
Br2

2Br

+ H2 --' HBr + H
H + Br2 --' HBr + Br
H + HBr --' H2 + Br
Br

(equil. dissoc.)
(slow)
(fast)

(4.5)

(fast)

was finally shown to be consistent with (4.3). Later research suggested that
(4.2) and (4.4) were deceptively simple; the first step in the formation of
HI is very likely the dissociation of the halogen. The subsequent steps are
different from (4.5) and there are still numerous unanswered questions.
The fact that seemingly simple kinetics like those in the formation
of HI could sti\l be an active subject for controversy for three-quarters of
a century is indicative of a major problem in mechanistic assignment.
It is impossible to prove a mechanism correct. Rather one shows that various
plausible mechanisms are inconsistent with the data. By amassing sufficient
chemical evidence one hopes to show that all but one chemically reasonable
mechanism is impossible. However, this can never really be done since
there are always more complex mechanisms which can account for the
data. Thus, even given compelling evidence, there remains a chance that
further experimentation will require revising the model which we call a
mechanism.

4.3. Elementary Rate Laws and the Principle of Mass Action

A general expression for the stoichiometry of a chemical reaction is
(4.6)
where Vi are the stoichiometric coefficients and Y i represent the species
which react. The Vi are positive for products and negative for reactants;
(3)

We use the notation A + B = C to describe the stoichiometry of a reaction. The notation A + B ~ C is used to mean a specific molecular encounter.

74

Chap. 4

Determination of Rate Laws

in these terms the hydro halogenation reaction (4.1) becomes

where VI = V 2 = -1 and V3 = +2. If the reaction represented by (4.6)

proceeds in a single step, the concentration changes of the species involved
are simply related at any time,
(4.7)
where [Yd is the concentration of species i at time t and ~(t) is the progress
variable for the reaction. (4) A kinetic formulation in terms of a single
progress variable is only correct if the reaction involves just one step.
This is clear if we refer to the mechanism (4.5); the important reaction
intermediates, Br and H, would not even appear in (4.7) since they do not
occur in the overall stoichiometry.
Because the rates of change of all the components are simply related
at all times in single-step reactions, the rate of reaction may be defined as
(4.8)
At equilibrium R is zero, which we interpret to mean that the rate of reaction
and its inverse precisely balance. To determine the rate of the forward
process in a single-step reaction note that a reaction occurs in one step
only if all reactants are close together at the same time. Stated differently
the reacting molecules must collide essentially simultaneously. Since collision probability is proportional to the product of the number density of
the molecules which participate in a kinetic event, the equilibrium forward
and reverse reaction rates are

Rieq)

kiT,p)

RrCeq)

kr(T,p)

n
n

react.
prod.

[Yd;;i

(4.9)
[Yi]~~

where the k's are the rate constants.(5) Since the two rates are equal at
(0)

(5)

For a discussion of progress variables and chemical equilibria see, e.g., F. T. Wall,
Chemical Thermodynamics, 3rd ed. (San Francisco: Freeman, 1974), Chapter 10.
The pressure dependence of k in single-step reactions is usually small; this problem is
treated by C. A. Eckart, Ann. Rev. Phys. Chem. 23, 239 (1972). Gas-phase unimolecular
reaction, for which k is a sensitive function of p, is not a single-step process (Section
5.4),

Sec. 4.3

Elementary Rate Laws

75

equilibrium, (4.9) can be rearranged to yield

(4.10)
In ideal gases or ideal solutions the right-hand side of (4.10) is the
equilibrium constant for reaction Keq(T, p). More generally it is activity,
not concentration, which specifies a concentration-independent equilibrium
constant. Defining the activity as

where Yi is the activity coefficient, (6) we find from (4.10)

K (T)
eq
,p =

TI ai

v
l

= kf(T,p)
k (T)
r ,p

TI Yi'

(4.11 )

The ratio of forward and reverse rate constants can be identified as the
Yi Vi = 1, a condition satisfied in ideal
equilibrium constant whenever
systems such as dilute gases. In addition, by redefining the standard state
of the reacting species, the constraint may be satisfied if the Yi do not
vary significantly with reactant concentration (see Problem 4.2). It is
clearly to the kineticist's advantage to study systems under conditions
where the activity coefficient product is unity. The activity correction
requires that considerable care be exercised when comparing kinetic parameters determined under grossly different conditions.
Assuming that away from equilibrium the reaction rates of the forward
and reverse steps are still given by (4.9),(7) we expect that R j *- Rr and that
the overall rate of reaction is

(4.12)
which is precisely of the form (4.2) as long as kr or [HI] are small. From
(4.11) we note that if Keq varies with temperature or pressure one (or both)
of the rate constants must do so too. Furthermore, the activity dependence
indicates that the rate constants may be concentration dependent. The
(61

(71

For a discussion of activity, activity coefficient, and their relationship to the equiiibrium
constant see F. T. Wall, Chemical Thermodynamics, 3rd ed. (San Francisco: Freeman,
1974), pp. 365-366, 419--420.
This is equivalent to assuming that the collision frequency (2.18) is unaltered in a
system displaced from equilibrium.

Chap. 4

76

Determination of Rate Laws

rate law (4.12) is consistent with the single-step molecular model

react.

-viYi ~

prod.

ViYi

Complicated sequential mechanisms such as (4.5) require a collection of

progress variables, one for each elementary reaction. With such models,
and assumptions about the magnitude of the rate constants, it becomes
possible to provide molecular interpretations of rate expressions such as
(4.3). In multistep sequences often only the forward rate law is established
experimentally. However, as the concentration of reactants and products
are related via the equilibrium constant and as Rieq) = Rr(eq), knowledge
of the forward rate law can be used to limit the possible rate laws for reverse
reaction (Problem 4.20).
The rate law (4.12) is unlike any of the linear transport equations
discussed in Chapters 2 and 3. The response is the reaction rate { However,
the right-hand side of (4.12) bears little apparent resemblance to the thermodynamic driving force for reaction, the affinity, A, defined as

where /1i is the chemical potential. (8) The condition for chemical equilibrium
is that A be zero; the chemical transport equation is thus
~ = LA

(4.13)

with L the corresponding transport coefficient. The two expressions for ~,

(4.12) and (4.13), are not obviously equivalent and L cannot be immediately
related to the rate constants. In fact the equations are not the same except
in the vicinity of equilibrium. (9) Further from equilibrium the linear relation
(4.13) is inadequate and it must be augmented to include all powers of A.
As a result the standard form (4.12) is by far the more practical in application. The relationship between L and the rate constants is usually complicated by activity effects. However, if these may be ignored, the result is
simple; L is proportional to the equilibrium reaction rates, (4.9),
L
(8)

(9)

Rieq)jRT = Rr(eq)jRT

The affinity is discussed by F. T. Wall, Chemical Thermodynamics, 3rd ed. (San Francisco: Freeman, 1974), Chapter 10.
For a simplified demonstration of this limited equivalence see A. Katchalsky and P. F.
Curran, Nonequilibrium Thermodynamics in Biophysics (Cambridge, Mass.: Harvard
University Press, 1965), pp. 91-97.

Sec. 4.4

Reaction Order and Molecularity

77

4.4. Reaction Order and Molecularity

To distinguish between a rate law determined experimentally and one
proposed on the basis of an assumed mechanism we use decimal numbers
as exponents in experimental rate laws and integers or fractions as exponents
in mechanistic rate laws. In the presence of argon the rate law for the
reaction
determined experimentally is
(4.14)
On the other hand, an assumed mechanism for the aqueous reaction
1-

+ CIO- =

10-

+ CI-

predicts a rate law(1O)

d[IO-] ex [I-)1[ClO-]I[OH-]-l
dt

Equally important is the molecularity of an elementary reaction. One

of the fast steps in the hydrogen bromine kinetic sequence (4.5) is
H

+ Br

->.

HBr

+ Br

(4.15)

which is a statement about the molecular processes that lead to reaction.

When an atom of hydrogen interacts with a bromine molecule, there is a
probability of the transformation to HBr and Br. According to (4.15) the
rate of HBr production is predicted to be

The process is unimolecular in both Hand Br 2. Since elementary reactions

such as (4.15) represent molecular collision processes the exponents in the
predicted rate expressions must always be positive integers. Thus experimental rate laws like those for the H2 + Br2 reaction (4.5) or the H2 + D2
reaction (4.14) indicate mechanisms which involve more than one elementary
step.
(10)

This reaction is discussed in Problem 5.10.

Chap. 4

78

Determination of Rate Laws

4.5. The Principle of Detailed Balance

We have seen that dynamic equilibrium requires that rate constants
be related to the equilibrium constant in a single-step reaction. For sequential reactions the overall equilibrium constant is the product of the equilibrium constants for the individual reactions, and, in an n-step process,
the extension of (4.11) is
(4.16)
A more interesting situation arises in systems of which the following
is a well-known example:
(A::;;:::'::: B)
(B : ; : :': : C)
(C::;;:::,::: A)

+H0
CO (aq) + H 0::;;:::'::: H+ + HCO aH+ + HCO a-::;;:::,::: H CO a
H 2CO a ::;;:::,::: CO 2 (aq)
2

(4.17)

It would appear possible that equilibrium could be established by means

of the cycle A ---' B ---' C ---' A even if the forward and reverse steps in
each elementary step did not proceed at the same rate. An expression
analogous to (4.16) could be derived for an equilibrium constant even
though (4.11) might not be true for the individual steps. It is precisely
such a possibility that is forbidden by microscopic reversibility.(ll)
Newtonian mechanics and quantum mechanics have in common the
property of time-reversal invariance. The Newtonian equations of motion
for a system of N interacting particles are
for all x, y,

Z,

i = 1, ... , N

(4.18)

Changing t ---' -t does not affect the accelerations Xi; on the other hand the
velocities Xi do change sign. As a consequence, for every solution of (4.18)
there is an identical one in which all particles have their velocities
reversed.o 2 ) For classical point particles the dynamics of a colliding pair
The following treatment is adapted from N. Davidson, Statistical Mechanics (New
York: McGraw-Hili, 1962), pp. 230-235.
(12) The dynamics are more complicated when magnetic fields are present. Since the field
couples with the velocity by means of terms like v x B it is also necessary to reverse
the field direction when the velocities are reversed. In this way time-reversal invariance
can be generalized.

(11)

Sec. 4.5

Principle of Detailed Balance

79

are specified by their velocities, CI and C2 , and impact parameter b. Due to

time-reversal invariance the transition probability for the change of state
(c l , C 2 , b) --' (c/, cz', b') must equal that for the change (- c/, - c2 ', - b')
--' (-CI' -C 2 , -b).(13)
Quantum mechanics describes the change of state which occurs in
an A-B collision in terms of the wave functions of the iateracting particles; the wave function for the pair changes from Pl(A, B) --' P2(A, B).
Ignoring magnetic fields, the equations of motion can be used to demonstrate that the time-reversed collision is one involving particles with reversed momenta and angular momenta. Denoting that change of state
as P 2 -(A, B) --' PI-(A, B) and defining the transition probability for the
change from state 1 to state 2 as w(l --' 2), time-reversal invariance requires that
w(l --' 2)

w(2- --' 1-)

Since neither collision nor time reversal affects energy, the states 1,2,2-,
and 1- are isoenergetic and therefore equally probable in an equilibrium
system. As a consequence the number of pairs making the transition 1 --' 2
exactly equals the number which go from 2- --' 1-. This is the quantum
version of detailed balance. (14)
We can now show that each reaction in the system (4.17) equilibrates
individually. On the molecular level in an equilibrium system the number
of A --' B transitions, each molecule being in a specified state, is exactly
equal to the number of back transitions for the time-reversed molecules.
Since a molecular state and its time-reversed partner are chemically indistinguishable (the two states are degenerate) the overall effect, when averaged over a Boltzmann distribution of energies, is that the rate of forward
and backward reaction in the individual steps of (4.17) must be equal at
equilibrium. In addition detailed balance implies a similar dynamic equilibrium among molecules undergoing transitions between any subset of
states.

(13)

(14)

For particles which interact via spherical potentials a more restricted principle can be
demonstrated. Velocity reversal is not required for equivalence of transition probability
so that the transition (c 1 ', co', b') ~ (c1 , c2 , b) and its inverse (c1 , c2 , b) ~ (c1 ', co', b')
are equally probable, a result demonstrated in the derivation of the Maxwell distribution, Section 1.2.
Microscopic reversibility is the transition probability constraint. Detailed balance is
its consequence in an equilibrium system. See B. W. Morrissey, J. Chern. Ed. 52, 296
(1975).

Chap. 4

80

Determination of Rate Laws

4.6. Experimental Methods for Determining Rate Laws

A rate law is established by monitoring the time dependence of the
concentrations of the species that take part in the reaction and by determining how this dependence is ~ffected by altering the constraints on the
reacting system. Since temperature usually has a considerable effect upon
the reaction rate, the reaction vessel is always thermostatted. While temperature control is rarely difficult unless the reaction is highly exothermic
and simultaneously very rapid, the accurate measurement of time may pose
a significant problem, particularly for fast reactions. In most chemically
interesting systems there are two major, closely related, experimental
problems. The first is to find a property of the system that is tightly coupled
to the concentration of one (or more) reactant and which is easy to monitor
continuously. The second, more directly related to the problem of mechanism, is to be able to detect, identify, and monitor the trace species that
can be intermediates in the reaction sequence.
Many different techniques have been used to detect trace intermediates.
These include:
(a) observation of characteristic emission spectra, either ultraviolet,
visible, or infrared;
(b) titration of the intermediate with a specific reagent followed by
observation of the product;
(c) observation of a free radical by means of electron paramagnetic
resonance;
(d) observation of gaseous atomic intermediates by atomic resonance
spectroscopy;
(e) observation of an unusual oxidation state by polarographic
methods.
Each of these techniques has drawbacks. For example, if a reactant has an
emission band in the same region of the spectrum as the intermediate,
spectroscopic detection will fail. Other difficulties arise because the speed
of the reaction may limit the time available for observation.
An almost unlimited number of ways have been proposed to monitor
reaction. Among the more common are:
(1)
(2)
(3)
(4)
(5)

pressure (in some gas-phase reactions);

dielectric constant or index of refraction;
light absorption at a specific frequency;
EMF with respect to an ion-specific half-cell;
volume changes (dilation);

Sec. 4.6

Rate Laws: Experimental Methods

81

(6) conductance;
(7) optical rotation;
(8) concentration (by removmg an aliquot, quenching the reaction,
and titrating);
(9) vapor chromatography;
(10) mass spectrometry;
(11) electron paramagnetic resonance spectroscopy.
It is not always simple to determine the concentrations from the measured

property. Consider light absorption in a multicomponent system. The

frequency-dependent absorbance A(t, '1') is related to the concentrations
by the Beer-Lambert law
A(t, '1') =

II Ei(Y)Ci(t)

(4.19)

where I is the path length, Ei(Y) the extinction coefficient of species i, and
Ci(t) its concentration. If only one species absorbs, A is directly proportional to the concentration of that species. If more than one species absorbs,
measurements are required over a range of frequencies to determine the
Ci(t). Similar difficulties exist no matter what monitoring procedure is
used. The general time-dependent response of an evolving system, R(t), is
R(t)

aiCi(t)

(4.20)

where the ai are determined by the properties of species i and by the measurement procedure used(15) (Problems 4.21 and 4.22).
In all cases the time required for a measurement (the response time
of a particular technique) must be short compared with the time during
which significant composition changes occur. For example, method 8 is
well suited to reactions with half-times on the order of hours but requires
great ingenuity when applied to reactions with short half-times. (16) The
practical limits of various techniques are summarized in Fig. 4.1.
For reactions which are essentially complete between periods ranging
from 1 msec to 1 sec rapid mixing techniques must be used. Two such
I thank Professor L. Parkhurst for pointing out to me the practical significance of this
problem.
(16) Quenching, by spraying a reacting mixture into a cold liquid or onto a cold plate,
can be completed within milliseconds. Combining this approach with flow methods allows method 8 to be applied to reactions with half-times ;s 100 msec. See B. Chance,
Q. H. Gibson, R. H. Eisenhardt, and K. K. Lonberg-Holm, eds., Rapid Mixing and
Sampling Techniques in Biochemistry (New York: Academic Press, 1966), Chapter 6.
(15)

82

Chap. 4
-flow-

10 6

Determination of Rate Laws

__ electric field - +
p_ jump _ _ _
pulse

T - j u m p - - - - - -...

--+.

sound

10- 6

10 0

tube
- - flash --...;.......

- --

proton
_ _nmr
_

'~

10-12 1

shock.

classical --+

I limit
of
I chemistry
II
t(sec)

pulsed laser
flash
~

1
1
I

esr
_ _fluorescence
_ _ _- J '

~y~

competition with

Fig. 4.1. Practical range of applicability of various methods for measuring the rates of
chemical reaction.

procedures are the stopped flow and the continuous flow methods. In
stopped flow(17) the reactants are driven into a small reaction chamber in
such a way that mixing is essentially complete within 0.1 msec. The reaction
is then monitored automatically using sensitive recording techniques
synchronized with the arrest offluidflow. In continuous flow(18) experiments
reactants continuously enter a reaction chamber, mix, and are forced into
a flow tube where the reaction occurs. Depending upon the rate of flow,
each point in the tube corresponds to a different time after the onset of
reaction. By monitoring at different points the degree of reaction at different times can be determined.
When the reaction time is less than 1 msec, flow methods are too slow
to insure homogeneity. Shock tube(19) methods are useful for studying
high-temperature gas-phase reactions which take place in time spans of
1 msec to 1 [lsec. Here a driver gas (usually He or H2 at a few atmospheres
pressure) is constrained by a diaphragm at one end of a long tube; the
remainder of the tube is filled with reactant mixture at low pressure (typically
10-3 atm). When the diaphragm is ruptured the driver gas expands abruptly,
(17)

(18)

(19)

The basic idea for stopped flow was introduced by B. Chance, J. Franklin Inst. 229,
455, 613, 737 (1940). For a modern apparatus see Q. H. Gibson and L. Milnes, Biochem. J. 91, 161 (1964).
This method was pioneered by Rutherford for studying gas-phase reactions [E. Rutherford, Phil. Mag. 44, 422 (1897)]. The mixing methods that are now in common use
were designed by H. Hartridge and F. J. W. Roughton, Proc. Roy. Soc. (London) AI04,
376 (1923), and by G. A. Millikin, Proc. Roy. Soc. (London) A155, 277 (1936).
Shock tubes were invented by M. P. Vieille, Compt. Rend. 129, 1228 (1889). For a discussion of modern design see H. T. Nagamatsu, Fundamental Data Obtained from Shock
Tube Experiments, R. Ferri, ed. (New York: Pergamon, 1961), pp. 86-136.

Sec. 4.6

Rate Laws: Experimental Methods

83

producing a shock wave which travels down the tube at supersonic speed.
The shock front is very sharp. As it passes any point in the reactant mixture
the reactant gases are compressed, heating them to temperatures "-' 103-10 4
K; this compression, which takes "-' 1 [isec, provides one limit for the
technique's applicability. Reaction begins as the gases are heated with
passage of the shock front. To study the system one relies on the fact that
the driver gas lags behind the shock front, taking a few milliseconds to flow
down the tube. An observation point is established near the end of the tube.
The first changes to be seen take place when the shock front passes. The
gas behind the front has been reacting from the instant the shock wave
passed by. It is pushed past the monitoring station by the driver gas. The
later the reactants pass the point, the longer they will have had to react.
Thus a time profile can be established.
Shock tubes are of limited utility. A more general approach to the
study of reactions which are complete in the range I msec-I nsec is to use
fast reaction methods. (20) An equilibrium system is perturbed by an external
stimulus applied for a very short time (always less than the half-time for
reestablishing equilibrium). A common approach is to effect a temperature
jump in the system by a brief burst of heating. If the equilibrium is temperature sensitive the concentration of reactants must readjust; by synchronizing an automatic recording technique with the onset or termination
of the heating pulse the relaxation to the new equilibrium state can be
followed. There are many other stimuli that can be used to perturb the
system. These include dilation (pressure jump), electric field (Wien effect),
etc. Any method that can perturb the system very rapidly is potentially
useful for such an experiment.
Perturbation methods displace a chemical equilibrium but no new
species are formed. In flash photolysis novel, unstable chemical species are
produced by irradiation. Their reactions with the species initially present
are the domain of photochemistry (treated in Chapter 6). Such reactions
(20)

The original fast reaction techniques were based upon flash photolysis [G. Porter,
Proc. Roy. Soc. (London) A200, 284 (1950)] and the Wien effect [M. Eigen and J.
Schoen, Z. Elektrochem. 59, 483 (1955)]. An enormous literature exists on the subject.
For recent discussions of some techniques see literature on: (1) temperature jump,
G. W. Hoffman, Rev. Sci. Instr. 42, 1643 (1971), and D. H. Turner, G. W. Flynn, N.
Sutin, and J. V. Beitz, J. Am. Chern. Soc. 94, 1554 (1972); (2) pressure jump, A. Jost,
Ber. Bunsenges. Phys. Chern. 70, 1057 (1966); (3) electric field methods, L. C. M. de
Maeyer, in Methods in Enzymology, Vol. 16, K. Kustin, ed. (New York: Academic
Press, 1969), p. 80; (4) ultrasonic methods, K. G. Plass, Acoustica 19, 236 (1967/68);
(5) flash photolysis, G. Porter and M. R. Topp, Proc. Roy. Soc. (London) A315, 163
(1970).

Chap. 4

84

~========'~~~--~EJE
I

~'

,00

Determination of Rate Laws

Fig. 4.2. Schematic diagram of apparatus

for nanosecond flash photolysis. The
components are: (A) pulsed laser source;
(B) beam splitter; (C) movable mirror;
(D) scintillation solution; (E) reaction
vessel; (F) spectrograph. (-) Path of
exciting pulse. (- -) Path of pulse to
scintillation solution (varied by moving
C). ( ... ) Path of analyzing pulse.

are completely different from those studied using perturbation methods.

Flash photolysis is designed to investigate fast photochemical reaction.
An alternate approach to the study of fast reactions is by means of
competition methods. Here the rate of a process is measured separately
in systems where chemical reaction can and cannot occur. If the process
is perturbed by the reaction, the difference in the rates is a measure of
the rate of reaction. For example, an unstable species may be formed
photochemically by irradiation and its lifetime determined. The experiment
is then altered by adding a substance which can react with the unstable
species; it will thus have a shorter lifetime from which the rate of the
chemical reaction can be deduced. To be a useful technique the chemical
and physical processes must proceed at roughly similar rates which limits
the applicability of this approach, as is emphasized in Fig. 4.1. Other competition experiments (and the corresponding observables) have been made
using polarography (current), nuclear magnetic resonance (line width),
electron spin resonance (line width), and diffusion (diffusion coefficient).(21)
Ultrafast flash photolysis ("-' 1 nsec) can be studied using pulsed
lasers. (22) Here a pulse of radiation of duration "-' I nsec is formed and
passed through a beam splitter, as illustrated in Fig. 4.2. Half of the beam
is sent through the reaction vessel as an initiating flash. The other half of
the beam is reflected from a mirror into a solution containing a scintillation
material which fluoresces almost instantly producing a nanosecond pulse
of continuous radiation which passes through the reactant vessel. Concentrations are determined by ordinary spectroscopic means; the scintillation
radiation provides the exciting light. The time at which the concentration
(21)

(22)

For an introduction to these methods see D. N. Hague, Fast Reactions (London: WileyInterscience, 1971), pp. 47-75. Photochemical competition (quenching) is discussed in
Chapter 6.
G. Porter and M. R. Topp, Proc. Roy. Soc. (London) A315, 163 (1970).

Sec. 4.7

Integrated Rate Laws-First Order

85

is monitored is the lag between the times when the initiating flash and the
scintillation radiation pass through the reactant vessel. The time lag is
specified by the paths of the light beams and is, therefore, fixed by the
geometry of the optical system. By varying the position of the mirror that
sends light to the scintillation solution, different time lags are produced
(0.3 m == 1 nsec). The technique has only one serious limitation: pulsed
lasers cannot produce exciting radiation at every frequency of interest.
Thus not all systems are amenable to study. Still faster reactions have been
studied using a variation of the pulsed-laser technique; the picosecond
range is now accessible. (23)

4.7. Integrated Rate Laws-First Order

Most kinetic measurements monitor the time dependence of the concentration of the reacting species, not the reaction rate itself. Thus the rate
law is derived from the observed time dependence of the concentrations.
Simple rate laws have characteristic integrated forms. For a first-order
rate law,
d[A]

dt

-k[A]l.O

(4.21 )

The integrated expression is

where Ao is the concentration of A at t = o. Exponential time dependence

always implies first-order kinetics. (24) If the experimentally determined
decay constant is independent of the concentration of any other species
present, it is a true rate constant and the data provide strong evidence for
a first-order reaction, even though (4.21) may not describe the kinetics.
To see why caution is needed consider the possible experimental rate law
d[A]

dt

= _ d[B] = -k [A]1.0
dt

+ L[Bp.o

(4.22)

P. M. Rentzepis and C. J. Michele, Anal. Chem. 42(14), 20A (1970); P. M. Rentzepis,

Science 169, 239 (1970); T. L. Netzel and P. M. Rentzepis, Chem. Phys. Lett. 29, 327
(1974).
(241 As we shall see exponential decay is characteristic kinetic behavior in many cases. By
itself it is not adequate evidence for the rate law (4.21). The observed decay constant
is seldom a true rate constant; it is generally a complicated combination of the rate
constants for individual steps.
(231

86

Chap. 4

Determination of Rate Laws

which is characteristic of a reversible reaction. The product B may itself

react to reproduce A. In terms of the progress variable $,

[AJ

Ao - $,

[BJ

Bo

+$

which when substituted in (4.22) and integrated, yields

Here the time dependence is also exponential but the decay constant is
the sum of the forward and reverse rate constants. Thus, for exponential
relaxation, either of the rate laws, (4.21) or (4.22), is consistent with a
concentration-independent decay constant. The individual rate constants
may be evaluated if the equilibrium constant is also known. Ignoring
activity effects, L = k +K;;J and we find
(4.23)
where x is the decay constant. If Keq ~ I, k+ is essentially equal to the
decay constant; the reverse step in (4.22) is unimportant and the two rate
laws are kinetically indistinguishable.
First-order processes are very common. The most familiar is radioactive decay, e.g., 14e ---->. 14N + e, the nuclear reaction used in carbon
dating for which k = 1.24 X 10-4 yr- 1 An interesting chemical example
is the gas-phase decomposition of N 20 5 which obeys the stoichiometry

but follows the rate law

with k = 3.38 X 10-5 sec- 1 at 298 K. This suggests a first step of fragmentation of an N 20 5 molecule; it provides no information about the
fragments produced or the subsequent steps in the reaction. To detail
the further course of reaction more experiments are needed. Examples
of the reversible first-order rate law (4.21) are found in many reactions
involving rearrangement, isomerization, or conformational change. An
interesting case is the chair 1 ---->. chair 2 rearrangement in cyclohexane shown
in Fig. 4.3. The rate law is
d[chair ]1
dt

-k[chair]l

+ k[chair12

Sec. 4.8

Isolation Methods

87

~2~

Fig. 4.3. Chair! ~ chair. interconversion in cyc10hexane showing exchange of axial and equatorial hydrogen atoms.

(2)

since the specific rate constants for both forward and backward steps must
be equal; at 206 K the rate constant is 52.5 sec- 1 .(25)

4.8. Isolation Methods

While integrated rate expressions can be found for second- and higherorder rate laws, in practice this is rarely useful. To see why, consider the
five second-order rate laws in Table 4.1. If one monitors the concentration
of A, all five rate laws lead to the same general time dependence. In terms
of the progress variable (4.7), these rate laws all have the form
-~ = a2;2

+ a + ao
1;

where the coefficients ai depend upon the initial concentrations and stoichiometric coefficients of the reacting species and upon the rate constants
in a variety of complicated ways. Integration of the rate expression always
yields a result that involves three parameters,
[A]

-A
=1+
o

ae(l - e- At )
At
1
ee-

(4.24)

where a, e, and A are determined by the aj. (26) To derive and use such expressions for eliciting a rate law is a time-consuming, demanding, and
error-prone task. (27)
F. R. Jensen, D. S. Noyce, C. H. Sederholm, and A. J. Berlin, J. Amer. Chem. Soc. 84,
386 (1962). The experiment is particularly interesting. Reactants and products are chemically indistinguishable but it is still possible to investigate their rate of interconversion.
The reaction rate is deduced from nuclear magnetic resonance line-width measurements.
(26) The same form is deduced if the rate laws of Table 4.1 describe nonelementary systems
for which autocatalysis (A or B as products) or autoinhibition (C or D as reactants)
occur. However, in some such instances A may be negative or imaginary; as a consequence unchecked growth or oscillatory behavior may be predicted. In Chapter 7 we
shall see that such behavior cannot describe a reacting system at all times although
it may be approximately correct for some periods in systems far from equilibrium.
(27) For a compendium of integrated rate laws see A. A. Frost and R. G. Pearson, Kinetics and Mechanism, 2nd ed. (New York: John Wiley, 1961), Chapters 2 and 8.
('5)

88

Chap. 4

Determination of Rate Laws

Table 4.1. Possible Elementary Reversible Second-Order Rate Laws

k+[A]2.0 - L[Cl LO
k+ [AJ1o [B]1.0 - L [Cl1.0

-d[A]

k+[A]2.0 - L[Cl 2.0

dt

k+ [A]1.0 [B]1.0 - L [Cl2.0

k+[A]1.0[B]1.0 - L[C]1.0[D]1.0

Rather than determining a rate law under conditions where all concentrations vary simultaneously, it is easier to isolate one species and hold
the concentration of the others constant. Consider a reaction which goes
to completion and follows the rate law
(4.25)
where a, b, and c are the unknown exponents. If the reaction mixture is
flooded with Band C so that during reaction their concentration is invariant,
the apparent rate law is
(4.26)
which may be easily integrated. If a = 1.0, the kinetics are pseudo-firstorder; the dependence of " on the initial concentrations Bo and Co then
establishes the exponents band c. When a -::/= 1.0, decay is not exponential;
the integrated form of (4.26) is
[Ap-a

(a - l)"t

+ Aol-a

(4.27)

Plots of [Ap-a as a function of t then determine a by trial and error; a is

varied until a straight-line plot is obtained. As seen in Fig. 4.4 if a is chosen
too large, curvature is positive and vice versa if a is too small. Once a is
known, " is evaluated from the slope of the line, and, by varying Bo and Co,
the other exponents can be determined.
Flooding is useful even if the rate law is more complicated. Consider
two examples

d~~]

-k1[A]a[B]b _ k 2 [A]a'[C]c

(4.28)

d~]

-k+[A]a[B]b

+ L[C]C[D]d

(4.29)

Sec. 4.9

a
I

89

Relaxation Methods

= I. 5

=2

= I. 25

J[A)4

f[A]4

2
2

2
t

234

Fig. 4.4. Plots of [A]l-a vs. time for various choices of a in (4.27). A linear relation is
assumed when a = 1.5.

In (4.28) there are two parallel reactions which consume A. If conditions

can be adjusted so that little B is present while C is in great excess only
one path is significant; the rate law is approximately
d[A]
dt

-k C c[A]a l
2

and normal isolation experiments can be carried out. The rate law (4.29)
includes a step in which A is produced. If initially there is a large excess
of B the reaction, in its early stages, appears to be of order a; as reaction
progresses it is affected by the reverse step and isolation conditions no
longer apply. To maintain isolation one of the reaction products must be
removed, as is possible if it precipitates or is volatile.
A practical indication that isolation conditions have not been established is when the trial and error method for determining the exponent a
outlined in Fig. 4.4 fails. Then concentrations other than that of A vary
significantly during the course of reaction. If no changes in the experimental
conditions are adequate to circumvent this problem a different approach
must be used.

4.9. Relaxation Methods

The observation central to the use of relaxation methods is simply
that, sufficiently close to equilibrium, all reactions follow pseudo-first-order
kinetics. Consider the rate law (4.29); near equilibrium
(-[A]

+ Aeq)/VA =;,

(-[B]

+ Beq)/VB =;,

etc.

(4.30)

90

Chap. 4

Determination of Rate Laws

where ~ is a progress variable which equals zero at equilibrium. As long as

the reaction is sufficiently close to equilibrium the only terms which significantly contribute to the reaction rate are those linear in ~. Introducing
(4.30) into (4.29), expanding, and keeping only linear terms we find

~;

-k+A~qB~q + LC~qD~q - ~[k+( -aVABeq - bVBAeq)A~;;lB~;;l

+k_(cvcDeq
=

+ dVDCeq)C~~lD~;;l]
(4.31)

-x~

The first term in (4.31) vanishes since, at equilibrium, the rate of reaction
is zero. When integrated (4.31) is
(4.32)
a result only valid when ~o is small. The decay constant x is the sum of two
terms, characteristic of the forward and of the reverse step in the reaction.
In general we can write
(4.33)
where Xi is the contribution to the decay constant from the forward step
in the reaction, etc. In Table 4.2 expressions for xi are given for a number
of rate laws; with these results it is possible to construct the decay constant
that describes any relaxation experiment close to equilibrium. The relation
between the order of a particular step in the reaction and the concentration
dependence of the corresponding term in x is particularly simple. For a
first-order step x is concentration independent, for a second-order step x
Table 4.2. Contribution of Forward Step in Reaction to Overall Decay Constant
for Different Rate Laws
Forward rate expression

Xf

(defined in text)

k+[A]l.O

k+

k+[AP' o

4k+Aeq

k+ [AP' o[B]l.O

k+(Aeq

k+[A]3.o

9k+A~q

k+ [A]"'o [B]l.O

k+(4AeqBeq

k+ [A]l.O [B]l.O [Cl l . O

+ Beq)

+ A~q)
k+(AeqBeq + AeqCeq + BeqCeq)

Sec. 4.9

Relaxation Methods

91

is linear in concentration, etc. By measuring" as a function of concentration

the order of the reaction and the specific rate constants can be established.
Systems for which there is only a single kinetic process are the exception, not the rule. In general there is a spectrum of relaxation times. As
long as these are well separated the individual decay constants are easily
measured. However, deducing the coupled rate laws and the rate constants
is more difficult. (28) If the kinetic processes proceed at similar rates considerable sophistication in data analysis is needed to obtain reliable values
of the decay constants. (29)
Relaxation methods were pioneered by Eigen and by Norrish and
Porter(30) as a method of studying fast reactions in systems where the
difficulty of obtaining homogeneity forbids the use of ordinary rapid mixing
techniques. However, by using sophisticated electronic monitoring devices
they can be applied more generally. Any system can be perturbed by
adding a reactant, then stirred continuously, thermostatted, and monitored
in the region near equilibrium. The concentration changes are small but
accurately measurable using modern analytical methods.
Perturbation approaches such as temperature jump, pressure jump,
etc., are based on a simple principle. For a system initially at equilibrium
the forward and reverse rates of reaction are equal. In a temperature-jump
experiment the system is heated so rapidly that the temperature increases
faster than the concentrations can respond. The initial reaction rate is
determined by the rate constants at the new temperature T and concentrations are set by the equilibrium at the original temperature To; the initial
reaction rate is then

for the rate law (4.29). Unless [k+(T)jL(T)] = [k+(To)jL(To)] the rates
of forward and reverse steps are no longer equal and the system must
readjust to a new equilibrium state. Analysis in terms of the progress
variable leads to the previous result, (4.32). Naturally there must be a
(28)

(29)

(30)

Fora general treatment of the concentration dependence of decay constants for a system
of sequential reactions see G. w. Castellan, Ber. Buns. Gesell. 67, 898 (1963).
A useful method for separating many overlapping exponentials is given by R. D. Dyson
and I. Isenberg, Biochemistry 10, 3233 (1971). Where only two processes overlap a
simple subtraction procedure can be used; see D. S. Honig and K. Kustin, Inorg. Chern.
11, 65 (1972) for an example.
M. Eigen, Disc. Faraday Soc. 17, 194 (1954); R. G. W. Norrish and G. Porter, Disc.
Faraday Soc. 17, 40 (1954).

92

Chap. 4

Determination of Rate Laws

detectable change. For essentially irreversible reactions there are measurable

concentrations of either reactants or products but not both. Perturbation
does not alter this situation so the technique is not applicable. (31) An
instance where perturbation is not practical is in the study of cyanidemyoglobin binding; cyanide binds too strongly with myoglobin. (32) However
the azide-myoglobin system, in which the binding is much weaker, can be
studied using temperature jump. (33)
Inhomogeneity problems are avoided using this method; the system
is initially uniform and a sharp increase in temperature will not introduce
concentration gradients. However, care must be taken to avoid temperature
gradients. In addition, the perturbation must be rapid enough so that
heating is completed before reaction commences. The heating pulse must
be synchronized with the monitoring device so that a record of the relaxation
may be obtained. If reaction is so rapid that significant changes occur before
heating is complete it is no longer permissible to consider the heating pulse
instantaneous. The observed response is a function of both the heating
curve and the chemical relaxation. (34)
Relaxation methods have wide applicability. The temperature-jump
study of the carbonyl addition reaction(35)
H

hO

'c'"

0
~N 1

Pyridine-4-aldehyde
(P)

+
Piperazine
(A)

Carbinolamine
(C)

monitored spectrophotometrically, was shown to proceed

pathways
P + A ~ C and P + AH + ~ CH +
(31)

(32)

(33)
(34)

(35)

VIa

parallel

The general constraints which limit the application of perturbation methods are derived
by M. Eigen and L. de Maeyer, in Techniques of Organic Chemistry, 2nd ed., A.
Weissberger, ed. (New York: Interscience, 1963), Vol. 8, Part 2, pp. 929-941.
P. George and G. I. H. Hanina, Disc. Faraday Soc. 20, 216 (1955).
D. E. Goldsack, W. S. Eberlein, and R. A. Alberty, J. Bioi. Chern. 240, 2312 (1965).
For analysis of the consequences of different heating pulses see M. Eigen and L. de
Maeyer, in Techniques of Organic Chemistry, 2nd ed., A. Weissberger, ed. (New York:
Interscience, 1963), Vol. 8, Part 2, pp. 917-928.
H. Diebler and R. N. F. Thornley, J. Am. Chern. Soc. 95, 896 (1971).

Sec. 4.10

Determination of Rate Law from Data

93

where AH + and CH + are the protonated forms of piperazine and carbinolamine. Furthermore, the possible mechanism
PH++A~CH+

involving the protonated species PH + could be ruled out.

Another example is the hemoglobin ~ oxyhemoglobin equilibrium

A dilution-jump technique, in which oxyhemoglobin is injected into an

anaerobic buffer, (36) requires the establishment of a new equilibrium via
deoxygenation of Hb0 2 ; the rate of reaction is monitored spectrophotometrically.
It is even possible to consider using perturbation methods to study
biological processes in vivo by means of concentration jump (injection of
a reactant into the cellular material) and then monitoring the response of
the system. Such an experiment could be done nond~structively and would
be more representative of the biological process than any in vitro experiment.
By changing the perturbation one would determine which reactants affect
the process of interest as well as the rate of reaction.

4.10. Determination of Rate Law from Data-Two Examples

Many sources of error complicate the deduction of a rate law from
kinetic data. In each experiment the time dependence of the concentration
has a specific functional form which is usually not self-evident from the
data unless isolation or relaxation methods are used. In addition no data
ever precisely fit a trial function. However, if all errors in the experiment
are random, probabilistic methods can be used to determine whether the
trial function is reasonable and to estimate the parameters of the function.
As long as only a single chemical process is significant, isolation and relaxation data are most readily treated using linear least-squares analysis,
described in the Appendix. This procedure provides the most reliable
estimate of the decay constant. Then, by varying experimental conditions
the concentration dependence of the decay constant can be obtained. With
such information probabilistic methods are again useful. A presumed rate
136)

G. G. Hammes, ed., Investigations of Rates and Mechanisms of Reactions, Part. II

(New York: Wiley-Interscience, 1974), p. 57.

Chap. 4

94

Determination of Rate Laws

law can be tested for consistency with the kinetic measurements and the
rate constants can be determined.
In general the kineticist is faced with the problem of deducing rate
laws that describe consecutive reactions involving many steps. The corresponding rate equations are coupled, and, in most cases, the functions
that are to be fit cannot be determined using linear least-squares analysis.
A vast literature of nonlinear function-fitting methods exists to treat these
problems. (37)

4.10.1. A Concentration-Dependent Rate Constant

The electron-transfer reaction
Eu(lI)

+ Fe (III) =

Eu(III)

+ Fe(lI)

is representative of systems for which the kinetics depend upon species

other than those indicated by stoichiometry. Studies carried out between
pH I and 3 were consistent with the binary rate law
_ d[Eu(II)]

dt

= k'[Eu(U)][Fe(III)]

(4.34)

However the apparent rate constant k' is pH dependent.(3S) Values of k'

determined at three temperatures are given in Table 4.3. Since k' decreases
as [H +] increases, a possibility worth testing is that k' might be a linear
function of [H+]-l. Figure 4.5 shows that the data are well represented by
such a plot suggesting that
(4.35)

While the fit is gratifying, it is hardly conclusive since many other functions
are qualitatively similar, e.g.,
k' = a/(b

+ [H+])

(4.36)

However, when this hypothesis is tested (see Problem 4.10), it is not in

accord with the data. The parameters ko and Ll of (4.35) can be estimated
(37)

(38)

For a brief discussion, with examples from enzyme kinetics, see J. T. Wong, Kinetics
of Enzyme Mechanisms (London: Academic Press, 1975), Chapter 11. A more extensive
development is given by Y. Bard, Nonlinear Parameter Estimation (New York: Academic Press, 1974).
D. W. Carlyle and J. H. Espenson, J. Am. Chem. Soc. 90, 2272 (1968).

Sec. 4.10

Determination of Rate Law from Data

95

Table 4.3. Apparent Binary Rate Constant, k',/or the Reaction of Eu(JI) and Fe(JII)
as a Function of [H+] and Temperature a
T= 1.4C
k' x 10-4

[H+]
(M)

(M-I sec-I)

0.031
0.04
0.05
0.1
0.2
0.5
0.879
0.939

8.14
6.92
4.92
2.70
1.33
0.729
0.616
0.613

T= 15.8C
k' x 10-'

[H+]
(M)

(M-I sec-I)

0.031
0.036
0.05
0,07
0.1
0.13
0.2
0.876
1.00

22.9
17.9
13.4
8.54
6.86
5.91
3.30
1.51
1.25

T= 25.0oC
k'

10-4

[H+]
(M)

(M-I sec-I)

0.03
0.04
0.05
0.0667
0.08
0.111
0.167
0.2
0.4
0.953

51.7
38.9
28.8
24.6
18.1
13.3
9.14
7.65
4.08
2.11

a D. W. Carlyle and J. H. Espenson, J. Am. Chern. Soc. 90, 2272 (1968).

Fig. 4.5. Temperature and hydrogen-ion dependence of apparent binary rate constant in
Fe(III) + Eu(II) system. 0, 25.0C. 0, 14.8C. /::., 1.4C. [Adapted from D. W. Carlyle
and J. H. Espenson, J. Am. Chern. Soc. 90, 2272 (1968).]

96

Chap. 4

Determination of Rate Laws

by visually fitting the straight line or by carrying out a linear least-squares

analysis (see Appendix). Using the latter procedure, the values of the
parameters and their corresponding standard deviations at 1.4C are
ko = (2.49 0.07) x 10 3 M-I sec- I and LI = (2.45 l.1)x 103 sec-I. The
correlation coefficient is 0.997, a result, given eight data, that would occur
by chance less than 1 in 106 One may wonder, given such apparent significance of the correlation, why the standard deviation in k-I is so large.
First, it must be remembered that comparison is being made with chance
occurrence, not with another proposed model. (39) Second, Fig. 4.5 indicates
that values of k determined for large values of [H+]-I are not sufficiently
consistent to precisely establish the intercept of the line. Even with the
numerical uncertainties it is clear that (4.35) represents the data quite well
and that, with (4.34), the rate law is similar in form to (4.28). There are
two kinetically significant pathways; the complete rate law is
(4.37)

4.10.2. A Relaxation Experiment

The complexation reaction between Rh 2 (OAC)4 2H 20 and 5' -AMP
(adenosine monophosphate) has been investigated using temperature
jump. (40) It provides an interesting example of relaxation kinetics because
two equilibria are maintained. Denoting the dirhodium complex as M and
the 5'-AMP as L, the important equilibria are
M . 2H 20
M L(H 20)

kl
+ L~
M
k_l

. L(H 20)

+ H 0,
2

k.
+ L~M
2L + H 0,
k_.

KI = 1893 M-I

(4.38)

where the equilibrium constants were determined at 25C. (41) The approach
to equilibrium was followed by a spectrophotometric technique which was
sensitive to the concentration of the dirhodium complex M 2H 20. The
decay constant was determined at 25C for a variety of total concentrations
of both M and L, CM, and CL; the data are given in Table 4.4.
(39)

(40)
(41)

When (4.36) is assumed to represent k, the correlation coefficient is 0.964, a result that
would occur by chance only once in 104 ; of course, the hypothesis (4.35) correlates
substantially better. The inappropriateness of (4.36) is only demonstrated when k- 1
is plotted vs. [H+] (see Problem 4.10).
K. Das, E. L. Simmons, and J. L. Bear, Inorg. Chern. 16, 1268 (1977).
L. Rainen, R. A. Howard, A. P. Kimball, and J. L. Bear, Inorg. Chern. 14,2752 (1975).

Sec. 4.10

Determination of Rate Law from Data

97

Table 4.4. Concentration Dependence of the Decay Constant in the

Rh 2 (OAc). 2H2 0-(5'-AMP) System a
CM

CL

xx 10-3

(mM)

(mM)

(sec-I)

[M]
(mM)

[L]
(mM)

1.94

1.96

5.88

0.79

0.67

1.92

2.91

6.25

0.50

1.21

1.92

3.38

7.14

0.41

1.51

1.91

3.85

7.69

0.33

1.84

1.90

4.31

8.70

0.27

2.19

1.89

4.76

9.09

0.23

2.54

a K. Das, E. L. Simmons, and J. L. Bear, Inorg. Chern. 16, 1268 (1977).

To establish the rate law we must determine whether the observed

relaxation is a consequence of reestablishing the first or second equilibrium
in (4.38). If we assume that the experiment is monitoring the reversible
elementary reaction M + L :;;:': ML, the decay constant can be found
using Table 4.2. The result is
(4.39)
A similar expression can be found for x 2 The kinetically significant variables for Xl are the concentrations of unreacted M and L which are determined by the equilibrium relations
[ML]
[M][L]

= Kl

(4.40)

and the stoichiometric constraints

CM =
CL =

+ [ML] + [ML
[L] + [ML] + 2[ML
[M]

2]

(4.41 )

2]

The values of [M] and [L] are also included in Table 4.4. To test whether
(4.39) adequately accounts for the data, X is plotted against [M] + [L] +
IIKI in Fig. 4.6. According to (4.39) we expect a straight line passing through
the origin in good agreement with the data. If it is assumed that the line
passes through the origin, linear least-squares analysis (see Appendix)

98

Chap. 4

Determination of Rate Laws

Or-----,-----~------.-----,

OL-----~I----~2------3~--~4

[M] + [L] + 11K, (mM)

Fig. 4.6. Dependence of the decay constant on concentration for the Rh 2(OAc).-5'-AMP
complexation reaction [K. Das, E. L. Simmons, and J. L. Bear, Inorg. Chern. 16, 1268
(1977)].

yields the rate constants kl = (2.85 0.03) X 10 6 M-I sec-I, Ll = (1.51

0.02) X 103 sec-I. Had we assumed that the ML
L:;;:::'::: ML2 reaction was
responsible for the observed relaxation no correlation would be found
(Problem 4.11).

4.11. Temperature Dependence of Rate Constants

The experimental characterization of a chemical reaction is not completed by deducing the rate law. In addition the way in which the rate
constants depend upon temperature, pressure, etc., must be determined.
The variation of k with temperature is striking; repeated experimentation
in the late nineteenth century showed that it could be represented in the
form
(4.42)
k(T) = A exp( - Ea/ RT)
Arrhenius(421 was the first to interpret Ea as an activation energy and present,
as qualitative justification for this point of view, the rationalization based
(42)

S. Arrhenius, Z. Phys. Chern. 1, 110 (1887).

Sec. 4.11

Temperature Dependence of Rate Constants

99

upon kinetic theory discussed in Section 1.4. To determine A and Ea linear

least-squares analysis is used; T-l is the independent variable and In
is the dependent one.
In some cases, either on experimental grounds or to effect comparison
with a theoretical model for the rate constant, one wishes to express k(T)
in the form
k(T) = A'Tme-OIT
(4.43)

ken

If m is known on theoretical grounds, A' and () may be estimated by the

least-squares method if In k - m(ln T) is treated as the dependent variable.
A direct extension of the least-squares approach is not possible if m is
unknown because In T and liT are not independent quantities. However,
by determining Ea for both high- and low-temperature data, m may be
estimated. Since (4.42) defines Ea as

Ea

RP d(ln k)
dT

-R d(ln k)
d(lIT)

it is possible to relate Ea and () via (4.43)

Ea

R()

+ mRT

(4.44)

from which we note that

Ea(high T) - Ea(low T)

= ---''-;;R:-::(~hi;-g7""h-';:;T;:;----:1'=-0'':''''w--;T:;:;):---=--

(4.45)

With this and nearby estimates of m it is possible, using the least-squares

procedure for known m, to determine both A' and () as functions of m.
The optimal set of parameters may be defined as that which leads to the
largest correlation coefficient. An example for which there is definite
curvature in a plot of In k vs. liT is the reaction Cl(2P) + CH 4 --' HCI +
CH 3 , illustrated in Fig. 4.7. (43)
An alternate way to determine the parameters A', m, and () is to fit
the data using a nonlinear least-squares procedure. However, all such
methods require reasonable initial choices for the parameters so the method
outlined in the previous paragraph is a necessary preliminary. (44)
Temperature-dependent activation energies are discussed by W. C. Gardiner, Jr., Acc.
Chem. Res. 10, 326 (1977).
(44) For details see N. R. Draper and H. Smith, Applied Regression Analysis (New York:
John Wiley, 1966), Chapter 10.
(43)

Chap. 4

100

Determination of Rate Laws

9r-r--------.------~r-------._,

..

(/)

::2:
.::.t.

cii
E

.002

.003

.004

.005

rl(K)

Fig. 4.7. Arrhenius plot of the rate constant for the reaction Clep) + CH 4 ~ HCI + CH.
Note the definite curvature [D. A. Whytock, J. H. Lee, J. V. Michael, W. A. Payne, and
L. J. Stief, J. Chern. Phys. 66, 2690 (1977)].

Comparing (4.11) and (4.42) it is clear, when activity effects are

properly taken into account, that the change in internal energy during
reaction, LJ E rxn , is
(4.46)
and therefore that Ea is a measure of the energy required to effect a chemical
transformation. It may never be less than LJErxn ' For an exothermic reaction
(LJErxn < 0), the reactants have sufficient energy to form products. One
might then expect that Ea ~ 0 for such systems. However, in many cases
Ea is substantial, which indicates that the reactants must go through a
highly energetic intermediate in order to form products. Similarly, for
endothermic reactions (LJErxn> 0), Ea need only be as large as LJErxn
but in many cases it is substantially greater, again indicating a highly energetic intermediate. The relative energies of the reactants, products, and
intermediate are shown schematically in Fig. 4.8.
Care is required when comparing thermodynamic reaction energies
with kinetic activation energies. Rate constants, as they refer to concentrations, presume measurements at constant volume. Equilibrium constants

Fig. 4.8. Schematic diagram of the activation

barrier for the reaction R(eactants)~ P(roducts) via an I(ntermediate) which is more
energetic than either R or P.

Appendix

101

are generally determined at constant pressure; as a result thermodynamic

data provide the enthalpy of reaction, not the energy of reaction. Transformation is required to obtain LlErxn (45)
In rare instances Ea < O. Almost invariably this indicates that the rate
law determined experimentally does not describe an elementary step.
However, if 0 ;S Ea ;S -10 kJ mol- l it is possible that the rate law could
describe an elementary step. Small negative Ea may also be due to experimental error.

Appendix. Linear Least-Squares Analysis

A major problem in interpreting data is to determine whether experimental quantities are correlated on the basis of an assumed theoretical
model. Since there are always errors of measurement, correlation is never
exact and a method for judging whether correlation is significant is required.
From probability theory random errors have a Gaussian distribution. (46)
If all errors are random, the best procedure for fitting data to a model is
the method of least squares.
As an example, consider the data presented in Fig. 4.6; experimental
values of the decay constant" are plotted against experimental values of the
quantity [M] + [L] + llKl . If there are N data with ordinates Yi and
abscissas Xi the problem is to test whether the functional form
(4A.1)

y=kx

is consistent with these data, to determine k, and to estimate the error in k.

For any value of k the difference
(4A.2)

need not be zero. If the functional form (4A.l) is correct and all experimental
errors are random the best estimate of k is found by minimizing the quantity

Li di 2 = I
(45)

(46)

Yi 2 -

2k I

XiYi

+ k2I

Xi 2

(4A.3)

See F. T. Wall, Chemical Thermodynamics, 3rd ed. (San Francisco: Freeman, 1974),
Chapter 3.
For an introduction to data analysis see H. D. Young, Statistical Treatment of Experimental Data (New York: McGraw-HilI, 1962), in particular, pp. 101-132. A comprehensive treatment is given by N. R. Draper and H. Smith, Applied Regression Analysis
(New York: John Wiley, 1966), Chapter 10.

Chap. 4

102

Determination of Rate Laws

with respect to k. The result is

and we obtain
(4A.4)

More generally we have no reason to expect that the assumed function

passes through a fixed point and we search for a fit to the function
Y

kx

+q

(4A.5)

Proceeding as in the previous paragraph but minimizing L di 2 with respect

to k and q yields
k

(NL XiYi - L Xi LYi)jiJX

(L Xi 2 LYi - L XiYi L Xi)jiJx

(4A.6)

where
(4A.7)

The questions of the significance of correlation and of the statistical

error in k and q may be answered using probabilistic arguments. (47) Significance is tested via the correlation coefficient, with iJy analogous to
(4A.7),
(4A.8)

When r = 1 there is perfect correlation, i.e., (4A.5) is exact for each datum;
when r = 0 there is no correlation. Since 1 > r > 0 in any series of experiments we may refer to tables of correlation coefficients(48) for the
answer to this question: What is the probability that the same value of r
would be found by chance if the variables were unrelated? Finally the
standard deviation in k and q can be calculated (49):
(h = a(NjiJx)1I2,

a
(47)

(48)

(49)

aq

--,--

a(L Xi 2jiJX)1I2

[L(Yi - kXi - q)2jN]1/2

(4A.9)

See H. D. Young, Statistical Treatment of Experimental Data (New York: McGrawHill, 1962) for details.
See H. D. Young, Statistical Treatment of Experimental Data (New York: McGrawHill, 1962), p. 164 for such a table.
If q = 0 then correlation is via (4A.l). In this case Gq == 0, r = ~ XiY.J(~ Xi2~ y.j2)1I2,
and Gk = ~(yj - kXj)2/~ Xj2]1/2.

103

Problems

Problems
4.1.

Recent research [J. H. Sullivan, J. Chem. Phys. 46, 73 (1967); 39, 3001 (1963)]
has indicated that the H. + I. reaction may proceed by one of the two following sets of elementary steps:
I.
(i)

21

(equilibrium dissociation)

H. + 21~2HI

(ii) H. + 1 = H.I
H.I

(equilibrium addition)

k'

+I~2HI

(a) Show that either pathway accounts for the overall stoichiometry.
(b) Write the rate law for each pathway and show that each is consistent
with Bodenstein's experimental results (4.2).
(c) Relate the overall equilibrium constant to the rate constants of the
individual steps. You may need to include some other reactions.
4.2.

Kinetics of ionic reactions are studied in solutions of constant ionic strength

(usually 0.1 M or greater).
(a) Refer to Debye-Hiickel theory and show that, if the reactants are in
mM concentration, their activity coefficients are independent of reactant
concentration.
(b) If fli* is the reference chemical potential of Yj at infinite dilution and
Yi(l) is the activity coefficient at ionic strength I show that an equilibrium
constant at ionic strength I is
(kflk r ) = Keq(T, p, l) =

where ili =
4.3.

fli*

IT exp( -vdidRT)

+ RT In yj(l).

At 25C the forward rate constant for the aqueous reaction

Co(NH a).Br2+ + Hg2+ + H.O

Co(NH.),(H.O)3+ + HgBr+

has the approximate ionic strength dependence, log k f = log kl+4.1(I 1 /').
[J. N. Bronsted and R. Livingston, J. Am. Chem. Soc. 49, 435 (1927)].
Use Debye-HiickeI theory to estimate the activities and thus obtain the
ionic strength dependence of kr .
4.4.

The kinetics of the hydrogen peroxide oxidation of sodium dithionite were

studied by monitoring the disappearance of dithionite ion spectrophotometrically under different initial concentrations of H 20 2 [CO Creutz and N.
Sutin,Inorg. Chem. 13,2041 (1974)]. The quantities tabulated are the times
after which [S.O.'-] has dropped to specific values. At a pH of 6.5 the data
given in Table 4.5 are representative:
(a) Determine the dependence of the rate law on [S.O.'-J for each value
of [H 20.] and find the decay constants.

104

Chap. 4

Determination of Rate Laws

Table 4.5
t (sec)
0.0450a

0.0899

0.135

0.180

0.5

0.45

0.55

0.25

0.15

0.15

0.4

1.15

0.55

0.35

0.25

0.35

1.80

0.85

0.55

0.45

0.3

2.50

1.20

0.75

0.60

0.25

3.20

1.55

0.95

0.80

0.2

4.05

1.95

1.20

1.00

Initial concentration of [H,O,l o (M).

(b) From the decay constants determine the dependence of the rate law
on [H 20,].
4.5.

Show that T (temperature) jump is a useful kinetic probe only if ;JHrxn =F- O.
Show that P (pressure) jump reqUires ;J V,rcn
O. (Consider the effect of
the perturbation on the equilibrium constant.)

4.6.

An "old chemists' tale" is that, near 300 K, the reaction rate doubles with
each 10 K increase in temperature. What is the "universal" Ea?

4.7.

Show that in a tracer exchange reaction of the type

"*

AX*

+ BX~ AX + BX*

the rate of exchange of X* is first order regardless of the details of the

reaction mechanism. Assume t; ,at [AX*] ~ [AX] and that there is no
isotope effect, i.e., that labeled and unlabeled molecules react at the same
rate.
4.8.

Many reactions can lead to a variety of products. The simplest case is

(a) What is the rate law for the disappearance of A?

(b) If B(O) = C(O) = 0, what are B(t) and C(t)?
(c) How can both kl and k, be determined experimentally if there is no way
to monitor B or C; if Band/or C can be monitored?
4.9.

The reaction of pyridoxal phosphate (PLP) with glutamate (Glu) in the

presence of Cu H has been studied by measuring the initial rate of reaction

Problems

105
Table 4.6

[PLPl o

[Glulo

[Cu 2 +1 o

(mM)

(mM)

(mM)

Initial rate
(11M sec-')

0.1
0.1
0.2
0.3
0.5
0.5
0.2
0.2
0.2
0.2
0.2

8.0
8.0
8.0
8.0
8.0
8.0
0.7
1.9
3.2
8.0
11.0

0.2
1.0
0.2
1.0
1.0
0.2
0.4
0.4
0.4
0.4
0.4

0.12
0.11
0.24
0.33
0.56
0.59
0.023
0.054
0.094
0.23
0.31

[M. E. Farago and T. Mathews, J. Chern. Soc. A., 609 (1969)]. Some results
are tabulated in Table 4.6 for 25C and pH 4.0.
(a) Deduce the rate law.
(b) What is the rate constant?
(c) What is the standard deviation in the rate constant?
4.10. Fe(III) exists in two forms in acid solution; these are readily interconverted
according to the equilibrium

(a) If the only species that reacts with EU(lI) (see Section 4.10.1) is FeOH~:q)
show that the apparent rate constant for the electron-transfer reaction
would be
k'

k*Kal([H+]

+ Ka)

(b) Test this possibility by plotting 11k' vs. [H+] using the data in Table 4.3
for T = 1.4C.
(c) Use a least-squares procedure to determine k* and Ka.

+ L ~ ML2 were being monitored for

the system discussed in Section 4.10.2 the decay constant would be

4.11. (a) Show that if the reaction ML

k 2([ML]

+ [L] + I/K

2)

(b) Use the data in Table 4.4 and the values of K, and K2 to compute
[ML].
(c) Test the hypothesis of part (a) both graphically and via a least-squares
procedure.

Chap. 4

106

Determination of Rate Laws

4.12. (a) Use the data of Table 4.3 to compute ko and k-l for the Eu(I1)
reaction at 15.8OC and 25C.
(b) What is Ea for the two pathways?

+ Fe(III)

4.13. (a) Show that if for all data, Yi = kXj + q, the correlation coefficient
r (4A.8) is unity.
(b) When variables are uncorrelated the value of x has no influence on
that of y. Construct an argument to show that, when N is large, for such
variables,

and thus that r

O.

4.14. The electron-transfer reaction

where R is 4,7-dimethyl-I,10-phenanthioline, has been studied using temperature jump [J. Halpern, R. J. Legare, and R. Lumry, J. Am. Chem. Soc.
85, 680 (1963)]. The experiments were carried out under conditions where
the initial concentrations of the reactants were equal, i.e., [FeR3'+]0 =
[lrCl 0 2-]0 = Co. The results suggested that the decay constant is proportional to Co.
(a) Assume a simple bimolecular reversible mechanism and show that

(b) The equilibrium constant for the reaction is K. Show that

x = 2Cok+/Kl/' = 2Cok_Kl/' = 2Co(k+k_)1/'

Table 4.7
t

x 10 (sec)

(0C)

Cox 10 (M)

10

0.55

10
7.5
5

33
55
66

18

0.40

10
7.5
5

32
42
59

30

0.28

10
7.5
5

23
31

44

Problems

107

(c) The data were obtained (r "" x- 1 ) as shown in Table 4.7. Determine
k+ and k_ at the three temperatures.
(d) Determine A and Ea for both forward and reverse steps.
4.15. (a) For the system of Problem 4.14 write the second-order rate law.
(b) In an ordinary kinetic experiment with initially equal reactant concentrations Co, show that the rate of product formation is

X=

k+Co' - 2k+CoX

+ (k+

- k_)X'

(X is the concentration of either product).

(c) Refer to a table of integrals and show that

(1 - K)y + K
(K - l)y - K

K1/'
K 1/'

K 1 /'
K 1 /'

+K

= e"t

with y "" X/Co and x given by one of the formulas in problem 4.14b. This
expression can be rewritten to yield an equation for y,
yet) = ( 1

K 1 /' coth

xt

)_1

(d) If the experiment is to be carried out using a stopped-flow apparatus

which can measure decay times <; 1 msec, what initial concentrations are
required at 4C? at 50C?
(e) Would it be possible to determine the rate constants using techniques
which only measure decay times <; 1 sec?
(f) Plot yet) for the initial conditions Co = 10- 7 M, t = 4C.
4.16. The reaction of Cr(VI) with 1- has been studied by measuring the initial
rate of disappearance of [1-] [K. E. Howlett and S. Sarsfield, J. Chem. Soc.,
683 (1968)]. The reactive form of Cr(VI) is the HCrO.- ion. The data
in Table 4.8 describe the initial conditions at 25C and an ionic strength
of 0.1 M.150)
The analysis of the data is somewhat complicated since Cr(VI) may exist
in a number of forms; the following equilibria are important:
H.O
H+

+ Cr.0 =
+ HCrO.- =
7 '-

0.018 M

2HCrO.-,

Kl

H.CrO.,

K. = 5 M- 1

(a) Show that, after equilibration of Cr(VI) and H+, but before reaction
with 1-, the hydrogen-ion concentration, h, is

150)

After R. G. Wilkins, The Study of Kinetics and Mechanism of Transition Metal Compounds (Boston: Allyn and Bacon, 1974), p. 56.

Chap. 4

lOS

Determination of Rate Laws

Table 4.8
[H+]o

[Cr.O,~]o

[1-]

-(d[1l/dt)o

(mM)

(mM)

(mM)

(ftM sec-I)

20.0
20.0
20.0
20.0
20.0
20.0
20.0
20.0
19.8
40.0
60.0
80.0
139.7

10.0
20.0
30.0
40.0
30.0
30.0
30.0
30.0
30.0
30.0
30.0
30.0
30.0

0.14
0.55
1.27
2.13
0.052
0.76
4.97
7.10
0.45
0.80
1.02
1.20
1.65

20.0
20.0
20.0
20.0
4
16
40
49.5
12.0
12.0
12.0
12.0
12.0

(b) Set [HCrO. -]0 = x and show that [HCrO. -]0 is found by solving the
equation
2x'

+ K,(1 + K.[H+].)x

_ 2K,[Cr(V1)].

K,[H+]0(K2 x)'
1 + K 2x

(c) Since the right-hand side of the equation in (b) can be neglected show
that
[HCrO.]o- = x = 4K, [{ (1

+ K.[H+]o)' +

16[Cr(VI)]0
}'/' - 1 - K.[H+]o ]
K,

(d) Use the results from (c) and (a) and show that the rate law is d[I-]/dt
= -k[HCrO.-][I-]'[H+p.
(e) What is the rate constant, k?
4.17. The stoichiometry for the reaction in Problem 4.16 is
2HCrO.-

+ 61- + 14H+ =

2Cr 3 + + 31.

+ SH.O

and a more detailed study of the reaction kinetics leads to the rate law

(a) What experimental approach can you devise for confirming this result?
(b) Could it be established by monitoring the initial rate of the back reaction?

Problems

109

4.18. The reaction

was studied using excess Fe3+ and H+ [J. H. Espenson and S. R. Helzer,
Inorg. Chem. 8,1051 (1969)). Small changes in [FeCrO.+] were monitored
with the result that
dln[FeCrO.+]
---d'--t---

where k was found to be proportional to [Fe 3 +]/[H+]. What is the rate

law?ISI)
4.19. For the reaction CI("P) + CH. ~ CH. + HCI some of the values of k,
are [D. A. Whytock, J. H. Lee, J. V. Michael, W. A. Payne, and L. J. Stief,
J. Chem. Phys. 66, 2690 (1977)]:

k,

1014 ( cm 3 molecule- I sec-I):

90.6

57.1

37.7

2.34

1.91

1.46

T (K):

500

447

404

220

210

200

Determine Ea in the two temperature regions. Then estimate m. Use your

value of m to estimate () and the correlation coefficient. If you have access
to a computer vary m and refine your results. Statistical analysis of the 14
data reported in the paper led to m = 2.5, () = 608.
4.20. Sometimes the law of mass action can be used to deduce possible forms for
the reverse rate law if the forward rate law is known. Consider the reaction
2HgH

+ 2Fe'+ =

Hg,H

+ 2Fe +
3

for which the forward rate law is R j = k j [Hg2+][Fe'+]. At equilibrium forward and backward rates are equal. Formulate the equilibrium constant,
determine some different expressions for the quantity [Hg'+HFe'+], and
show that Rr may have the form (among others)
Rr ex [Fe 3 +HHg, '+ ]I/2

ex [Fe3+]'[Hg2H]/[Fe H HHg H ]
ex [Fe3+]" [Hg, H ]3/2/ [Fe H ]' [Hg'+]'

(i)
(ii)
(iii)

4.21. Show that absorbance measurements made at a single frequency determine

the decay constants when relaxation methods are used, even if more than
one constituent absorbs at the monitoring frequency.
4.22. A system contains two components, A and B, which absorb in the same
frequency range. The differential absorbance LlA(v) = A(v, t) - A(v, 00)
was measured at two frequencies. The molar extinction coefficients are
(51)

After R. G. Wilkins, The Study of Kinetics and Mechanism of Transition Metal Compounds (Boston: Allyn and Bacon, 1974), p. 57.

Chap. 4

110

Determination of Rate Laws

Table 4.9
Frequency

Vi

1000

2000

Vs

1500

500

Table 4.10
t (sec):

0.02

0.04

0.08

0.12

0.16

0.20

0.24

0.28

.1A(v, ) :

7.5

8.9

9.5

9.5

8.7

7.8

6.3

5.3

4.3

.1A(v.) :

13.9

12.3

10.8

8.5

6.8

5.8

4.3

3.5

2.8

given in Table 4.9, and the measured differential absorbances are given
in Table 4.10.
(a) Determine [A(t)] - [A( 00 )], [B(t)] - [B( 00 )], the path length of the cell is
0.1 m.
(b) Are the data consistent with a single relaxation process?
4.23. The rate of halogen hydrolysis X. + H 20 ~ X- + H+ + XOH, where
X = CI, Br, or J, could not be determined until the advent of fast reaction
techniques, especially temperature-jump [M. Eigen and K. Kustin, J.
Arner. Chern. Soc. 84, 1355 (1962)]
(a) Derive the expression relating the reciprocal relaxation time (l(T) to
the forward and reverse rate constants and concentrations in the halogen
hydrolysis reaction.
(b) Calculate (within 10% accuracy) the relaxation time, T, for bromine
hydrolysis when total bromine concentration is 3 x 10- 3 M; at 293 K and
ionic strength 1.0 M, Keq = [BrOH][Br-][H+]f[Br.] = 6.9 x 10-9 M2 and
k+ = 110 sec-i.

General References
Microscopic Reversibility and the Principle of Detailed Balance
R. K. Boyd, Chem. Rev. 77, 93 (1977).
S. R. deGroot and P. Mazur, Non-Equilibrium Thermodynamics (Amsterdam: NorthHolland, 1962), pp. 92-100.
D. ter Haar, Elements of Statistical Mechanics (New York: Rinehart, 1954), pp. 381-382.
A. Messiah, Quantum Mechanics, translation by J. Potter (Amsterdam: North-Holland,
lQIlIl) nn.664-675.

General References

111

Experimental Methods in Chemical Kinetics

C. F. Bernasconi, Relaxation Kinetics (New York: Academic Press, 1976), Chapters 11-16.
J. N. Bradley, Shock Waves in Chemistry and Physics (London: Methuen, 1962), Chapters
4 and 10.
A. Ferri, ed., Fundamental Data Obtained/rom Shock Tube Experiments (New York: Pergamon, 1961), Chapters 3 and 4.
G. G. Hammes, ed., Investigations 0/ Rates and Mechanisms 0/ Reactions, Part II (New
York: John Wiley, 1974).
K. Kustin, ed., Methods in Enzymology, Vol. 16 (New York: Academic Press, 1969).
P. M. Rentzepis, Adv. Chem. Phys. 23, 189 (1973).

Relaxation Theory
C. F. Bernasconi, Relaxation Kinetics (New York: Academic Press, 1976), Chapters 1-5.

Integrated Rate Laws

s.

W. Benson, The Foundations 0/ Chemical Kinetics (New York: McGraw-Hill, 1960),

Chapters 2 and 3.
A. A. Frost and R. G. Pearson, Kinetics and Mechanism, 2nd ed. (New York: John Wiley,
1961), Chapters 2 and 8.

Pressure Dependence of Rate Constants

C. A. Eckert, Ann. Rev. Phys. Chem. 23, 239 (1972).

Stationary State Mechanisms

5.1. Introduction
It is not enough to establish rate laws. The information must be interpreted to yield insights into the elementary steps by which molecules
interact, react, and recombine. This is done by constructing mechanistic
models consistent with the phenomenological rate laws. Such mechanisms
are rarely unique and guidelines are needed to construct them. The fundamental ground rule is simplicity. With this principle it becomes possible
to choose between competing models, each of which agrees with the data.
It is obvious that our notion of simplicity is redefined if new intermediate
species are discovered.
In many cases the rate law is a direct indication of the only important
elementary step which contributes significantly to the rate of reaction.
Examples occur in both gas-phase and solution kinetics. Typical of the
gas-phase reactions for which this is believed to be the case are the bimolecular reactions(1)
NO + 0 3
N0 2 + O 2
----"0

C2H 4

+ H2

----"0

C2H 6

the termolecular recombination reactions(2)

1 + 1 + He
Br + Br + A
(1)

(2)

----"0

----"0

12 + He
Br 2 + A

A tabulation of such reactions can be found in A. A. Frost and R. G. Pearson, Kinetics

and Mechanism, 2nd ed. (New York: John Wiley, 1961), p. 104.
A. A. Frost and R. G. Pearson, Kinetics and Mechanism, 2nd ed. (New York: John Wiley, 1961), p. 108.

113

Chap. 5

114

Stationary State Mechanisms

and the unimolecular reactions(S)

N 20 4 - ' 2N0 2
(CHSCO)20 -' CHsCOOH

+ CH 2CO

cis-stilbene -' trans-stilbene

In solution kinetics an important distinction between solvent and
reactants must be maintained. The solvent is continually interacting with
the reactants; if such interactions were incorporated in the mechanistic
model all solution mechanisms would perforce be multimolecular. By
considering the solvent as a medium and not as a participant in the reaction
(unless, of course, solvent actually takes part in a reaction step), the problem
of mechanism is greatly simplified. In this sense isomerizations, rearrangements, and conformational changes, like the chair! :;.::':: chair 2 interconversion
in cyclohexane, are first-order reactions for which the empirical rate law
is a direct indication of the only important elementary step. Most solution
reactions proceed via bimolecular steps. There are countless examples for
which only one such step is needed and for which the rate law reflects that
process.
In this chapter we first consider a mathematically tractable model
mechanism and demonstrate that, depending upon the relative magnitudes
of the rate constants, there are two chemical approximations that may be
appropriate for simplifying analysis: the preequilibrium and the steadystate assumptions. We then demonstrate how hypotheses based upon
these simplifications are used to interpret rate law data and to develop
chemically reasonable mechanistic descriptions for gas- and solution-phase
reactions. Finally we consider the problem of catalysis, i.e., how addition of
trace amounts of an intermediate permits a sluggish or kinetically forbidden
reaction to become rapid if a new mechanistiC pathway can be created.

5.2. Consecutive Reactions-First Order

Most reactions take place via consecutive steps. If each is first order
the general mechanism is
(5.1)
(S)

A. A. Frost and R. G. Pearson, Kinetics and Mechanism, 2nd ed. (New York: John Wiley, 1961), pp. 110-111.

Sec. 5.2

Consecutive Reactions-First Order

115

The corresponding rate equations are

+ k_[B]

d~]

-k+[A]

d~]

+k+[A] - (L

+ k+')[B] + L'[C]
(5.2)

One is redundant; conservation of matter requires that the sum [A] +

[B] + ... + [Z] be constant. The general solution to (5.2), obtained by
the matrix method outlined in Appendix A, is a superposition of n exponential relaxations
[A]

L ai(O) exp(-xit),

etc.

(5.3)

i~l

where the Xi are decay constants determined by the rate constants and the
coefficients ai(O) are fixed by both the rate constants and the initial conditions.
Deduction of the mechanism (5.1) and of the rate equations (5.2)
is greatly simplified if the various decay processes occur at vastly different
rates. Analysis of the case n = 2 exhibits all chemically significant features
of the general mechanism. To introduce the ideas of time-scale separation
and of a trace intermediate assume that the reactions go to completion,
i.e., k_ - k_' == O. Then, if initially only A is present, the concentrations
of A, B, and C are found by direct integration of the rate equations
[A]

Ao exp( -k+t)

[B]

k+Ao[exp( -k+t) - exp( -k+'[)]/(k+' - k+)

[C]

Ao{k/[l - exp( -k+t)] - k+[l - exp( -k+'t)]}/(k/ - k+)

(5.4)

Further simplification occurs if there is a separation of time scales, i.e., if

the rate. constants are vastly different.
When k+ ~ k+' the conversion of A to B is completed before B starts
to decompose. Only 1% of the original amount of A is left after a time
t = 4.6/k+, which is much less than l/k+', the relaxation time for the further
reaction of B. Then the approximate time dependences of [B] and [C] are
[B]

Ao exp( -k/t),

[C]

Ao - [B],

Chap. 5

116

Stationary State Mechanisms

+-

o....
C
Q)
u

c
80.5
Q)

>

+-

Q)

0::

2.0

1.0

3.0

time
Fig. 5.1. Relative concentrations of A, B, and C for the mechanism (5.1) assuming
k/ = 2k+ and k_' = L = O. The time scale is arbitrary.

which is readily detectable since the rapid buildup of the intermediate B

should be obvious in an experiment. Furthermore the disappearance of
A is much more rapid than the appearance of C.
If k+' ?> k+ then whatever B is formed reacts immediately; no significant concentration can develop. We find that

so that [B] is always a fixed (small) fraction of [A]. Here it may be difficult
to establish that the reaction proceeds in two steps since B is present in
very low concentration at all times. Monitoring of neither A nor C would
suggest that the reaction proceeds via an intermediate; other evidence
would be needed.
If k+ ~ k/ all three species are present simultaneously as shown
graphically in Fig. 5.1. Experimentally one infers the two-step mechanism
by observing C. Were only a single first-order process important, such as
in the two limits considered, then In [Ao - C] would be a linear function
of t with a slope determined by the significant reaction. For the intermediate
case, as shown in Fig. 5.2, this is not true until t is very large.
In general, reactions are reversible. If the rate equations (5.2) are
analyzed using the matrix method of Appendix A for the special case
n = 2, the two decay constants of (5.3) are /{:
(5.6)

where
(5.7)

Sec. 5.2

Consecutive Reactions-First Order

117

3.0
2.0

-In(i-

[el )
Ao

1.0

OL-~--~~--

1.0

__~=-____~~__
2.0

time

3.0

Fig. 5.2. Plot of In (1 - [C]/Ao) for the mechanism (5.1) assuming k+' = 2k+ and
L' = k_ = O. The plot is nonlinear at short times since both reactions are significant.

a and a' would be the decay constants for uncoupled A ~ Band B ~ C

equilibria, respectively. The two decay constants x+ and x_ correspond to
fast and slow processes, respectively. As long as the quantity under the
square root in (5.6) is large, the time scales are separable, i.e., x+ ~ x_.
One relaxation is complete before the other process starts, as illustrated
in Fig. 5.3. There is further simplification if certain elementary steps of
(5.1) are much faster than others.
Simplification occurs if one of the reactions equilibrates rapidly
(preequilibrium) or if a trace intermediate forms (steady state). There
is preequilibration if a ~ a', i.e., if A ~ B equilibration is much faster than
0.8.-------.----r-----,r----,----.-------.

[8]
Ao

OL--~-~~~-~~--~--~~--~

. 0.1

0.2

2.0

4.0

6.0

8.0

time - Fig. 5.3. Plot of [BJ/Ao for the mechanism (5.1) assuming k+ = 20, L = 10, k+' = 0.1,
and L' = 0.2. Note the clean separation of time scales. There is rapid growth of the
intermediate B governed by the fast decay (u+ = 30.034) which is followed by a slow decay
(u_ = 0.266) to equilibrium. The individual decay constants can be determined by analyzing the early growth and the late decay separately. Note that for 0.15 :'S t :'S 0.5, [BJ essentially is constant. At equilibrium [BJ/Ao = 0.5.

118

Chap. 5

Stationary State Mechanisms

the B ~ C equilibration. With this constraint and approximating the square

root in (5.6), as done in Appendix B, the decay constants can be computed:
(5.8)
The fast process always has a decay constant k + + k _, as expected since
the A ~ B interconversion is a first-order reversible reaction. Since k+/k_ =
K, the equilibrium constant for the A ~ B reaction, 'L IS
(5.9)
When K ~ 1 the equilibrium favors B and the slow process only measures
the rate of B ~ C interconversion; the decay constant is k+' + k_'. When
K ~ 1 there is never much B present. The B ~ C reaction is then operationally irreversible; whatever B is formed by the slow reaction of C
continues to react and form A. Thus, the rate ofthe slow process is governed
not only by the rate at which Band C interconvert but also by the amount
of B present. In the event a' ~ a, the B ~ C equilibration is rapid. A slow
A ~ B equilibration follows. This is the converse of the limit just considered.
When either k/ or L (or both) are large compared with k+ or L',
B reacts as rapidly as it is formed; no significant population of this transient
ever exists. The corresponding approximations to (5.6) are carried out in
Appendix B with the results
(5.10)

The fast process measures the rate of disappearance of B. The smaller

decay constant is a complicated function of the rate constants for the
individual elementary steps.
A different derivation of (5.10) may be obtained directly from the rate
law (5.1) with n = 2. When k_ or k/ are large, B decomposes almost
immediately upon formation. The species A and C form a reservoir which
slowly convert to B. The net effect of the two opposing processes is that a
steady-state population of B is formed after a short induction period
governed by "'+, the fast decay. After a time period that is long compared
with x+ -1 we can make a steady-state assumption, that d[B]/dt ~ 0 in
comparison with both k+[A] + k_'[C] and (k_ + k+')[B] in (5.1). Hence
(5.11 )

Sec. 5.2

Consecutive Reactions-First Order

119

1.01------~

0.8

[A]

0.6

AO

0.4
0.2

/--~
/

0.006

y_--- -----

0.004

Ao

\[8J

[8]

AO

0.002

O~-~--~I---~--~--~-~

0.1

0.2

12

16

timeFig. 5.4. Plot of [Al/A. and [BI/A. for the mechanism (5.1) assuming k+ = 0.1, L = to,
= 20, and k_' = 0.2. These are conditions under which the steady-state assumption is
valid. There is a rapid induction period governed by the fast decay (x+ = 30.0) during
which a steady state is produced. This is followed by a slow approach to equilibrium
(x_ = 0.133). During the induction period, [A] is constant. In the steady-state domain
(t;<; 0.2), [B] is not constant; however, there is never much B present.

k+'

It is important to understand that the steady-state assumption does not

demand that d[B]fdt = 0, as illustrated in Fig. 5.4. Rather, it asserts that
the net rate of change of [B], d[B]jdt, is much smaller than either its instantaneous rate of formation, k+[A] + k_' [e], or its instantaneous rate of

decomposition, (L + k+')[B]. Neither of the latter two quantities need

be small, only their difference. Making use of the steady-state hypothesis
and substituting (5.11) in (5.1) we find
deC]
dt

(5.12)

a rate law indistinguishable from one for a simple reversible first-order

mechanism. The decay constant is, therefore, exactly the x_ of (5.10).
The most complicated situation arises when all the rate constants
are comparable in magnitude. Then x+ and x_ do not differ greatly. All
three species are present simultaneously and experimental concentrationtime plots are similar to Fig. 5.2; a single exponential decay does not
account for the data. Analysis requires using the complete expressions (5.6),
a complication which arises when k+ r::::: k_, i.e., if A and B have similar

120

Chap. 5

Stationary State Mechanisms

thermodynamic stability, etc. While many chemical reactions involve

species with markedly different stability there are many problems for which
there is no separation of time scales.
We limited our analysis of (S.l), considering only two-step mechanisms.
In a more complicated sequence there would be more relaxation times,
the scale of which might be overlapping. The net result would be greater
problems in interpreting the data. While time-scale separation, as we have
seen, simplifies analysis because each relaxation can be observed independently, it can lead to misinterpretation if each is not observed. Reconsider the two-step process with the decay constants of (S.10). If only x_
is measured one might reasonably infer a single-step mechanism. However,
in some special cases the temperature dependence is sufficient to demonstrate greater complexity. Depending on the magnitudes of the four elementary rate constants, an Arrhenius plot of the observed decay constant can
yield an apparent activation energy that is either negative or quite temperature dependent (see Problems S.3 and S.22). Such behavior is an immediate indicator that the observed decay does not reflect an elementary step.
In general, lacking direct evidence of an intermediate, it is hard to infer
its existence on kinetic grounds alone.
In general, independent of any time-scale separation, decay constants
are complicated combinations of rate constants for the elementary steps.
Only in the most rapid step of a reaction sequence is the decay constant
a sum of rate constants, as is evident from (S.8) or (5.10). The complications
increase if the mechanism involves more steps.

5.3. Consecutive Reactions-Arbitrary Order

In contrast to the first-order mechanisms just considered, the integrated
rate laws for most multistep second(or higher)-order elementary processes
can only be determined numerically, using computer methods. However,
in many cases of interest the ideas of preequilibrium and steady state are
applicable and the kinetic analysis is much simplified.
Consider the gas-phase decomposition of ozone

for which the empirical rate law is

d[03] _
k
[0 3]2
d { - - applOJ

(S.13)

Sec. 5.3

Consecutive Reactions-Arbitrary Order

121

A possible two-step mechanism(4) is

(5.14)

If the first step is rapid preequilibration of atomic and molecular oxygen

with ozone then [0] is determined by the values of lOa] and [0 2 ],
(5.15)
where K

k1/k_ 1. The rate expression for the second step is

which with (5.15) becomes

d[Oa]
dt

-k K [Oa]2
2
[0 2]

+k

-2

[0]2
2

(5.16)

precisely the form observed experimentally if the back reaction in the

second step does not occur. The experimental rate constant, k app , is to be
interpreted as the product of the equilibrium constant for the first step
and the forward rate constant of the second step.
Alternatively it is reasonable to assume that the concentration of atomic
oxygen is small and to apply the steady-state hypothesis to the mechanism
(5.14). The result is

which when substituted in the rate expression for either step yields the
approximate rate law

-klk2[Oa]2 + L 1k_ 2 [02]a

L 1 [02] + k 2 [Oa]

(5.17)

This, but for the factor \12, is equivalent to (5.16) if kz[Oz] ~ k_1[Oz].
The discrepancy arises because (5.15) is an approximation; the last two
terms in d[031 dtJare not exactly equal. Stated in chemical terms, the 0 +
0 3 reaction must be slow, precisely the condition for rapid preequilibraC,)

We shall see in the next section that, strictly speaking, the initial decomposition of
ozone must also involve a bimolecular step.

122

Chap. 5

Stationary State Mechanisms

tion of 03, O 2 , and O. In this example both the steady-state and preequilibrium hypotheses are consistent with the empirical rate law (5.13),
and therefore adequate to demonstrate that the mechanism (5.14) is
plausible. Unambiguous demonstration of (5.14) is more difficult since, in
the preparation of 0 3 , atomic 0 is always present.(5)
If the proposed mechanism involves many steps or if some of the
elementary reactions are third (or higher) order, exact analysis can become
quite involved. The attempt at chemical simplification proceeds similarly.
On chemical grounds one postulates that certain steps involve rapid equilibration and/or that some intermediates are in.a steady state. The predicted
rate law is then compared with the one determined experimentally. If they
agree the mechanism has passed its first test.
It is not always possible to solve kinetic problems using simple approaches. Even if the mechanism only involves a few steps the data may
not readily suggest the coupled rate laws. Chemical intuition is required
to propose a plausible mechanism which then must be tested. The natural
procedure is to integrate the mechanistic rate equations and to compare
the predicted concentrations with the observed ones. If a set of kinetic
parameters can be found that satisfactorily accounts for the data, the
mechanism is a possible one. Such calculations always involve extensive
numerical work. Thus reasonable starting points are needed, even if modern
high-speed computers are used.
While closed-form solutions for irreversible two-step mechanisms
involving combination of uni- and bimolecular steps have been formulated, (6) in general the rate laws must be integrated numerically. Straightforward approaches such as the Runge-Kutta methods are usually adequate
to insure rapid convergence. (7) The process must be repeated for each set
of rate constants to be tested. The most efficient procedure is usually to
program a computer to carry out the calculations. In some instances the
simple numerical integration algorithms converge slowly, or not at all.
This is a particular problem for the complex mechanisms that describe
systems for which there are multiple steady states, oscillations, or ex(6)

(6)

(7)

For detailed discussion of the mechanism see S. W. Benson, Foundations of Chemical

Kinetics (New York: McGraw-'HilI, 1960), pp. 400-406.
J. Y. Chien, J. Am. Chem. Soc. 70, 2256 (1948); R. G. Pearson, L. C. King, and S. H.
Langer, J. Am. Chem. Soc. 73, 4149 (1951); A. S. Baladin and L. S. Liebenson, Dokl.
Akad. Nauk. SSSR 39,22 (1943); T. Kelen, Z. Phys. Chem. N. F. 58, 268 (1968); 60,
191 (1968); Acta Chim. Acad. Sci. Hung. 59, 323 (1969); 60, 87 (1969).
See D. D. McCracken and W. S. Dorn, Numerical Methods and Fortran Programming
(New York: John Wiley, 1964), Chapter 10.

Sec. 5.3

Consecutive Reactions-Arbitrary Order

123

plosion. (8) Special numerical techniques, particularly suited for computer

applications, have been developed to circumvent these difficulties. (9)
The study of sequential reactions is often much simplified if relaxation
methods are applicable since such experiments are carried out near equilibrium where all rate expressions are first order in the progress variable.
The general treatment outlined in Section 4.9 is equally applicable if there
is a series of reactions. Corresponding to each independent reaction there
is a progress variable. By analogy to (4.30) expressions for deviations from
equilibrium can be written in terms of the progress variables. Then coupled
first-order differential equations which describe the evolution of the various
reactions can be found. These equations are the analogues to (4.31) and
have the same mathematical form as (5.2) with progress variables replacing
the concentrations. The matrix method of Appendix A can be used to obtain
decay constants. (10) If there are two independent chemical reactions there
are two exponential relaxations. The decay constants depend upon both
the rate constants and the equilibrium concentrations. By carrying out the
kinetic experiments under a variety of conditions the mechanism can often
be characterized.
Even if a proposed mechanism accounts for the kinetic data, it is not
necessarily correct. An important plausibility test is given by the magnitudes
of the rate constants required to account for the phenomenology. In both
gas- and solution-phase reactions there are natural upper bounds to the
rate constants. A mechanism requiring larger values of k cannot be correct.
These conditions are easily established for gas-phase reactions. The upper
bound for the rate of a second-order reaction is fixed by the total rate of
binary collisions, (2.21),
k2[A][B] = Rate

<

ZAB =

Non dlBcAB[A][B]

With reasonable values for molecular parameters this yields k 2 ;:S 1010_
1011 M-1 sec- 1 when T ,....." 500 K. A similar bound can be found for k3 if,
for the reaction to occur, it is assumed that two molecules must be within
a collision diameter when struck by a third. The result is k3 ;:S 108_1010
M-2 sec- 1 (Problem 5.23). In solution the rate constant is limited by the
rate of diffusion (Sections 5.7 and 9.13).
(8)

A few of these problems are discussed in Chapter 7.

(9)

C. W. Gear, Numerical Initial Value Problems in Ordinary Differential Equations (En-

(10)

glewood Cliffs, N. J.: Prentice-Hall, 1971), Chapter 11. For a general introduction
see D. Edelson, J. Chern. Ed. 52, 642 (1975).
C. F. Bernasconi, Relaxation Kinetics (New York: Academic Press, 1976), Chapters 3,
4, and 7; G. W. Castellan, Ber. Bunsen Gesell. 67, 898 (1963).

Chap. 5

124

Stationary State Mechanisms

5.4. Unimolecular Decomposition-Gases

Numerous gas-phase decompositions follow first-order kinetics over
a wide range of pressures. Typical examples are the reactions
N 20 s = 2N0 2

cyclopropane

+O

propene

Both reactions have activation energy barriers suggesting the question:

How can one reconcile first-order kinetics with an activation process which
must involve molecular collision? Lindemann(ll) proposed the following
mechanistic answer:
(5.18)
ka

A * ~ products
An A molecule is activated by collision with another A molecule. In the
collision process the relative kinetic energy of the two colliding molecules
is transformed, in part, to internal energy of one of the molecules, thus
"activating" it. Such inelastic collisions (in which kinetic energy is not
conserved) provide the mechanical basis for understanding the activation
process. (12) The excited molecule A * is stabilized in two ways. Either it is
deexcited by another inelastic collision with A or it decomposes to form
products. The rate expressions deduced from (5.18) are

d[~*]

k+[A]2 - L[A*][A] - kd[A*]

d[product] = _ d[A] = kd[A*]

dt
dt

(5.19)

It is reasonable to assume that the steady-state hypothesis is applicable to

A * since molecular collisions are constantly occurring. The rates of production and depletion of A * greatly exceed its net rate of change. Hence

(11)

(12)

F. A. Lindemann, Trans. Faraday Soc. 17, 598 (1922).

Inelastic processes are precisely those ignored by the postulates of elementary kinetic
theory; see Section l.l.

Sec. 5.4

Unimolecular Decomposition-Gases

125

and the rate of production formation is

d[product]
dt

k+kd[A]2
kd + L[A]

(5.20)

Depending upon the concentration of A, the predicted rate law varies

between first and second order. At very low pressure kd ~ k_[A] and the
rate law is second order with an apparent rate constant k+. However,
when the pressure is high and L[A] ~ kd the reaction is first order with
an apparent rate constant koo = kdK, where K is the equilibrium constant
for the activation process. To test the Lindemann mechanism express the
data in first-order form

d[product]
dt

k [A]

ex

and with (5.20) define a concentration dependent "first-order" rate constant

(5.21)
In Fig. 5.5, log(kex/k oo ) is plotted for the decomposition of cyclopropane.
While it appears to be a linear function of log [A] at low pressure, suggesting
that the low-pressure limit has been reached, the slope is closer to t than 1,
the value predicted by (5.21). Such differences are typical; the actual
falloff with pressure is more gradual than is consistent with the simple
Lindemann mechanism. Improvement requires considering the specific

o
00

0>0

c5b

0 000

~o

cf

-1.0

of

-1.5'-'-_ _-'-_ _--'-_ _--'-_ _--l...J

-3.0
-2.0
-1.0
0
1.0

log p (atm)
Fig. 5.5. Dependence of the isomerization rate of cyclopropane on the pressure of cyclopropane. [Adapted from H. O. Pritchard, R. G. Sowden and A. F. Trotman-Dickenson,
Proc. Roy. Soc. A217, 563 (1953).]

Chap. 5

126

Stationary State Mechanisms

energy dependence of the population of A* and of its decomposition rate.

A theory accounting for the energy effects is discussed in Chapter 9.
Since the activation process is collisional, similar effects are found if
an inert buffer gas is introduced into the system. In addition to the processes
of (S.18) there is a parallel activation-deactivation reaction with the buffer
gas, M,
k '

A+M~A*+M
k_'

Assuming steady-state conditions apply, the rate of product formation is

d[product] _ k
dt

k+[A]2 + k+'[AJ[M]
L[A] + L'[M] + kd

(S.22)

which reduces to first-order kinetics when either the buffer or the reactant
is at high pressure. At intermediate and low pressures this rate law is only
qualitatively correct. Nevertheless a measure of the relative rates at which
different gases activate A is given by the "first-order" rate constant derived
from (S.22) at low gas pressure [the analogue to (S.21)],

By changing both [A] and [M] the relative efficiency of activation by

Table 5.1. Relative Efficiency of Various Buffer Gases in Maintaining the Rate
of Isomerization in the Cyclopropane-Propene Reactiona
Buffer gas
Cyclo-C aH 6
He
Ar
H.

N.

CO.
CH 4
H 2O
Toluene.
Mesitylene .
Trifluorobenzene
a

Relative efficiency
1.0
0.060
0.053
0.24
0.060

0.011
0.007
0.03
0.003
o.on 0.009
0.27 0.03
0.79 0.11
1.59 0.13
1.43 0.26
1.09 0.13

H. O. Pritchard, R. G. Sowden, and A. F. Trotman-Dickenson, Proc. Roy. Soc. A217, 563

(\953).

Sec. 5.5

Radical Recombination-Gases

127

different gases can be measured. An example is the cyclopropane-propene

isomerization in the presence of different buffer gases. The results are
given in Table 5.1. The major obvious trend is that the more complex
the molecule, the more effective an activator it is. When we consider
molecular collision dynamics in Chapter 10, some of the mechanical
features which account for energy transfer in inelastic collisions will be
clarified.

5.5. Radical Recombination-Gases

"Unimolecular" decomposition is a classic instance where theory
demonstrates that a mechanism is more complex than the stoichiometry
suggests. Another instance is in the atomic recombination reaction

The experimental rate law is often second order in X but the mechanism
cannot be

Consider a collision of two X atoms; they have a relative kinetic energy

greater than zero. As the atoms interact total energy is conserved; their
total energy remains positive. In Fig. 5.6 the relative potential and kinetic
energy of a colliding atomic pair is plotted. At the turning point (distance

of closest approach) the potential energy is necessarily positive and so the

X2 molecule is unstable as formed. The atoms fly apart and no bond is
created.
Stabilization occurs if a third particle takes part in this process. (13)
Some of the excess vibrational energy may be transmitted to the third
particle leaving the X 2 molecule in a bound state. For the recombination
of iodine atoms in the presence of a nonreactive buffer gas, M, the observed
rate law is
(5.23)
(13)

Another mode of stabilization is conceivable. Before the unstable X. molecule dissociates it could spontaneously undergo a vibrational transition and be stabilized. This is
a most unlikely process, even less common than three-body collisions in a gas. It may,
however, be important in molecule formation in interstellar space; the formation of
CH. + from C+ + H. appears to follow this pathway. For an introductory review see E.
Herbst and W. Klemperer, Phys. Today 29(6), 32 (1976).

Chap. 5

128

f\

Stationary State Mechanisms

/T(r)

/'('

, __ _

U(rl

Figure 5.6. Kinetic and potential energies for atoms

approaching one another along the line of centers.
The total energy T(r) + U(r) is conserved; d is the
distance of closest approach.

suggesting the parallel mechanisms

21 + 12

kl

12 + 12

21+M~12+M
A third mechanism should also occur

but because the concentration of 1 atoms is always small, it could not

contribute significantly to (5.23).
The same considerations obviously apply to atomic reactions of the
type
x+Y=XY
but are less important in radical recombination such as
2C6HS ----' biphenyl
The distinction is that a polyatomic molecule has many vibrational degrees
of freedom. Temporary stabilization occurs if energy can be transferred
from the C-C vibration to some of the other vibrational modes of the
biphenyl molecule. Such a process occurs readily so that three-body collisions are not immediately required for the recombination of phenyl
radicals. Reactions such as
H

+ OH ----' H 0
2

2CH 3 ----' C 2H 6
are intermediate cases. There are fewer vibrational modes in water (3) and
ethane (18) than in biphenyl (60); as a result decomposition of the energetically unstable molecule is rapid. However, in all cases the molecule is

Sec. 5.5

Radical Recombination-Gases

129

formed with excess vibrational energy; collision with another molecule is

eventually required for complete stabilization.
A general mechanism for the recombination of X and Y in the presence of a buffer gas M can be proposed
k

X+Y ~XY*
k_

XY*+M~XY+M

(5.24)

X+Y+M~XY+M
There is a two-step pathway in which the metastable XY* is deactivated
by collision with M and the parallel one-step pathway involving direct
stabilization via three-body collisions. (14) If a steady-state assumption if>
made on XY*, the predicted rate law is
(5.25)
thus the kinetics should vary from second to third order depending upon
the pressure of buffer gas and the magnitude of the rate constants. (15) For
small molecules the decomposition rate of XY* must be very fast; thus
k_ ~ k2[M] at reasonable pressures of M and third-order kinetics are found.
For larger molecules k2[M] > k_ at moderate partial pressures of M in
which case (5.25) simplifies to
Q

Since three-body collisions (in the gas phase) are very unlikely, k+ > k3[M]
unless the partial pressure of M is quite large (Problem 5.23) and (5.25)
reduces to a second-order rate law. As an indication of the effect of molecular
complexity, the observed recombination rate of I atoms is always third
(141

(151

M is sometimes termed a "chaperone," the chaperone's function being to lead partners

into a stable union rather than leaving them in a highly excited state.
Equation (5.24) does not account for all possible pathways. Often the following
mechanism is important:
k+'

X+M~XM,

k2'

XM+Y~XY+M

Ie '

The additional term in (5.25) would be k+'k;[M][X][YJj(k_'"+ ko'[Y)) so that this

pathway is kinetically distinguishable form those contributing to (5.24) if the shift
from third- to second-order kinetics is observable.

Chap. 5

130

Stationary State Mechanisms

Table 5.2. Apparent Activation Energy jar the Reaction 1 + 1 + M ~ 12 + M"

Third body (M)
Helium
Argon
Oxygen
Carbon dioxide
Benzene

Ea (kJ mol-i)

Third body (M)

Ea (kJ mol-i)

-1.7
-5.4
-6.3
-7.3
-7.1

Toluene
Ethyl iodide
Mesitylene
Iodine

-11
-10
-17
-19

a Data from G. Porter and J. A. Smith, Proc. Roy. Soc. (London) A261, 28 (1961).

order, that of phenyl radicals is always second order, and that of methyl
radicals varies from third to second order as the buffer gas pressure changes
from 0.0013 to 0.06 atm.(l6)
At low or moderate pressure the direct three-body mechanism is not
important and the apparent rate constant is, from (5.25), k2k+/L = k2K,
where K is the equilibrium constant for formation of vibrationally excited
XY. The apparent activation energy is therefore
(5.26)

where LlE(U is the energy of reaction for formation of XY and Ea(2) the
activation energy for the stabilization step. Since the first step is slightly
exothermic (a weak bond is formed), LlE(1) < 0; the second step is highly
exothermic and requires no (or negligible) activation so Ea(2) ~ O. Thus
we expect Ea ~ LlE(l) < 0, which is the rule in recombination reactions.
The same arguments apply if reaction proceeds via the weakly bonded
intermediate XM (or YM). To establish which pathway is followed experiments are carried out using a range of buffer gases. Since Ea is set by
the bond energy of the intermediate it is independent of buffer if XY*
is the transient and variable if XM (or YM) is the transient. The effect
of buffer gas in the iodine recombination is illustrated in Table 5.2; clearly
the intermediates are 1M. Comparable data for the recombination of H
atoms or of Cl atoms or for the addition of H to NO show no effect of
buffer on Ea suggesting that XY* is the likely intermediate. (17)
The mechanism (5.24) is, in its deactivation step, the converse of the
Lindemann mechanism (5.18). Whereas unimolecular decomposition poses
(18)
(17)

G. B. Kistiakowsky and E. K. Roberts, J. Chem. Phys. 21, 1637 (1953).

V. N. Kondratiev, Rate Constants of Gas Phase Reactions, translation by L. J. Holtschlag
(Washington, D.C.: National Bureau of Standards, 1972).

Sec. 5.6

Complex Mechanisms-Gaseous Chain Reactions

131

the problem of adding energy to a molecule, bimolecular recombination is

a problem in energy removal. The sequence (5.24) does not account for
quantitative aspects of recombination kinetics. Unlike unimolecular reactions, which are well understood in terms of more elaborate theories, no
truly adequate treatment of the termolecular processes has yet been formulated. (18)

5.6. Complex Mechanisms-Gaseous Chain Reactions

The classic example of a chain mechanism is the H2 + Br 2 reaction
first studied by Bodenstein. (19) The experimental rate law is
d[HBr]
dt

(5.27)

where kif ""' 0.1 and k' <X e- 20 ,1001T for temperatures from 500 to 600 K.
The result defied interpretation for a number of years. Independently
Christiansen, Herzfeld, and Polanyi(20) considered the following reaction
sequence:
kl
+ Br2 ~
2Br + M
k2
Br + H2 ~ HBr + H
ka
H + Br2 ~ HBr + Br
k.
H + HBr ~ H2 + Br
k.
M + 2Br ~ Br2 + M

(1) M
(2)
(3)
(4)
(5)

(5.28)

which is a typical chain mechanism. (21) Reactive intermediates, in this

For an abbreviated discussion of these theories see W. C. Gardiner, Rates and Mechanisms
of Chemical Reactions (New York: Benjamin, 1969), pp. 141-148. A more rigorous treatment is given by D. L. Bunker, Theory of Elementary Gas Reaction Rates (Oxford: Pergamon Press, 1966), Chapter 4. A recent review article is H. O. Pritchard, Ace. Chem.
Res. 9, 99 (1976).
(19) M. Bodenstein and S. C. Lind, Z. Phys. Chem. 57,168 (1907).
(20) J. A. Christiansen, Kgl. Danske Vid. Se/s. Mat.-Fys. Medd. 1, 14 (1919); K. F. Herzfeld, Ann. Physik 59, 635 (1919); M. Polanyi, Z. Electrochem. 26, 50 (1920).
(21) Since the mechanism (5.28) was postulated before Lindemann's rationalization of unimolecular decomposition, buffer gas was not included in steps (1) or (5) of the original
proposal. Since there is preequilibration of Br2 and Br, the rate law does not reflect the
participation of buffer gas.
(18)

Chap. 5

132

Stationary State Mechanisms

case Hand Br, at low concentrations are postulated in order to continue

the' cycle. Steps (2), (3), and (4) maintain the concentration of the intermediates although only (2) and (3) lead to HBr formation; step (4) involves
inhibition by HBr which is necessary in view of the experimental rate law
(5.27). Step (1) is required to initiate the chain. Step (5) is a chain termination reaction. The buffer gas M in steps (1) and (5), required for activation
or deactivation, may be either H2 or Br 2 or an inert additive.
The rate of production of HBr is

To determine the concentration of Br and H atoms we apply the steadystate hypothesis. Since

d[Br]

~ =

-k2[Br][H2]

+ k 3[H][Br 2] + k4[H][HBr]

+2k 1 [Br 2][M] - 2kdBr]2[M]

I'=::::>

(5.30a)
(5.30b)

we find from (S.30b)

[H]

k2[H2][Br]/{k3[Br~]

+ k4[HBr]}

(5.31)

and, by adding (5.30b) to (5.30a),

(5.32)
Combining (5.31) and (5.32) yields

Substituting (5.32) and (5.33) into (5.29) then yields, after slight rearrangement,
d[HBr]
2k2(kl/k5)1/2[H2] [Br 2]1/2
(5.34 )
dt
1 + (k 4/k 3)[HBr]/[Br 2]
in exact accord with the empirical expression (5.27) if k' = 2k2(kl/k5)1/2
and k" = k4/k3.
It is possible to evaluate k2 since kl/k5 = Keq for the dissociation of
Br2. Then, considering the temperature dependence of k' and Keq , the

Sec. 5.6

Complex Mechanisms-Gaseous Chain Reactions

133

activation energy for step (2) in the mechanism (5.28) is found to be 73.6
kJ mol-I, which is certainly reasonable. Since this step is endothermic by
68.2 kJ mol-I, a comparable activation energy is required.
The mechanism (5.28) is consistent with the empirical rate law (5.27).
The activation energy for step (2) is quite reasonable. However, it is also
important to show that other possible steps such as
(6)

+ H2
M + HBr
Br + HBr
M + 2H
M + H + Br
M

(7)
(8)
(9)
(lO)

--->.

--->.

--->.

--->.

--->.

+M
H + Br + M
H + Br2
H2 + M
HBr + M
2H

are not plausible. If these were significant reaction pathways the rate law
would not be (5.34). There are also other reasons for eliminating them.
The bond energy of both H2 and HBr is much larger than that of Br 2.
Thus the activation energy of either (6) or (8) is much greater than that
of step (1) of (5.28); the dissociation rate of either H2 or HBr is very slow
in comparison with the dissociation rate of Br 2. Bond energy arguments
HBr reaction (8) has an activation barrier of at
indicate that the Br
least 173.2 kJ mol-I; thus it is not competitive with steps (2) or (4) of
(5.28), each of which has a much lower activation energy. The chain termination steps involving atomic H, (9) or (lO), can be eliminated because
the H-atom concentration is too small to be effective (Problem 5.7). (22)
The H2
Br 2 reaction is an example of a linear chain reaction. In
each propagation step one atom of reactive intermediate is produced for
each atom that is used up. From (5.34) we see that, if the temperature is
constant, the rate of HBr production is greatest near the beginning of the
reaction and drops off as H2 and Br2 are consumed. If the system is not
thermostatted, the temperature will increase because the reaction is exothermic, and the reaction is thus accelerated. If heating occurs with sufficient rapidity, depletion of reactants will not be sufficiently fast to keep
the reaction in check and the conditions for a thermal explosion are present.
Br 2 reaction, because heating is relatively
While not occurring in the H2
slow, an explosion can occur in the photoinitiated H2 + Cl 2 reaction if the
light intensity is high enough.

(22)

A. A. Frost and R. G. Pearson, Kinetics and Mechanism, 2nd ed. (New York: John
Wiley, 1961), p. 239.

Chap. 5

134

Stationary State Mechanisms

A completely different cause of explosion is found in reactions which

proceed by a branched-chain mechanism. Here more than one atom of
intermediate is produced in the chain propagation scheme for each atom
that reacts. The result, if chain carrier is produced faster in the initiation
and propagation steps than it is consumed in termination steps, is a rapid
increase in the concentration of chain carriers. As a consequence the reaction
velocity accelerates rapidly and explosions can occur. The best-known
mechanism of this type is the fission reaction in an atomic bomb. Chemical
examples abound and include the reactions of oxygen with hydrogen,
phosphorous, carbon monoxide, etc. Heating is an after effect of explosion,
not the cause. (23)
The simplest branched-chain mechanism requires initiation, branching,
and termination steps. If we assume a first-order chain termination step,
such as reaction of the chain carrier R at the walls of the container, and
let ... indicate any reactant or product other than chain carrier, the mechanism is
...

Initiation:
Branching:
Termination:

+ '"
R + ...

---->.

---->.
---->.

+ ...
aR +
nR

(5.35)

The various rates r may depend upon the concentrations of any reactant
species other than the chain carrier. As long as the rate of formation of R
is small, steady-state kinetics are applicable and

so that
[R]

nrJ[r t - (a - l)rb]

(5.36)

Since a> 1 the denominator in (5.36) is zero if rb = rt/{a - 1). Then

[R] ---->. 00 and the steady-state hypothesis cannot apply. However, the
important qualitative feature is that the concentration of chain carriers
increases very rapidly under such conditions. As a consequence there is an
enormous increase in reaction velocity with the possibility of explosion.
Explosion is a phenomenon which cannot be analyzed using the stationary state hypothesis. Due to chemical heating or unchecked population
growth an instability is created. Similar instabilities can lead to concentration
(23)

The two mechanisms for explosion are discussed at greater length in Chapter 7.

Sec. 5.7

Simple Mechanisms-Solution

135

oscillations (rather than relaxations) during the course of chemical reaction.

It is even possible for spontaneous, periodic spatial inhomogeneities to

form during chemical reaction. Further discussion of these topics is reserved

for Chapter 7.

5.7. Simple Mechanisms-Solution

The qualitative features of reaction mechanisms in solutions are substantially different from those in gases. Unimolecular processes still occur
via collisional activation. However, solvent molecules which can affect
activation are always in high concentration. In terms of a rate law such as
(5.22) the experiments are being carried out under conditions where
k_'[M] ~ kd or L[A] and k+'[M] ~ k+[A]; here [M] represents the
concentration of solvent, always in large excess. There is significant shortrange order in liquids as solvent molecules are loosely bound to one another
and form transient structures which reduce the mobility of the products of
a decomposition. Thus the rate at which products separate by diffusion must
limit the rate of unimolecular reaction in solution. (24) Unimolecular decomposition may still be considered a two-step process:
A

<

k+ ,
k_

BC* ~ B + C

The first step involves rapid equilibration of A with a BC* complex; in

the second step Band C diffuse apart. The rate law, assuming a steady
state of BC*, is always first order in A with an apparent rate constant
k+kd/(L

+ kd)'

Isomerization, rearrangement, and conformational changes do not

require diffusional separation of the products. Instead, a reasonable mechanism for the conversion of A to A' is

(24)

The solvent structure is often described as a "cage," a somewhat misleading term since
the "solvent cage" has no permanence. Different solvent molecules diffuse in and out
of the vicinity of the products so that the structure is being continuously broken and
reformed. However, the important feature which limits mobility is that the decomposition products cannot separate unless solvent molecules are displaced. The idea of
a solvent cage was introduced by E. Rabinowitch and W. C. Wood, Trans. Faraday Soc.
32, 1381 (1936).

Chap. 5

136

Stationary State Mechanisms

The corresponding rate law, assuming a steady state of the excited intermediate A *, is
d[A]
dt

(5.37)

typical of a reversible first-order reaction.

Most solution reactions take place via bimolecular steps. Collisional
deactivation of products occurs easily because of interaction with solvent.
Chemical reaction may be viewed as a multistep process
A

+ B ~ (A ... B)

(A ... B)
(C ...

(diffusion)

(C ... D)

(reaction)

+D

(diffusion)

D)~C

(5.38)

in which reaction is initiated by the formation of the encounter complex

(A ... B). The rate of encounter (or separation) is governed by the rate
of diffusion of reactants (or products). After encounter an (A B)
complex is constrained by the surrounding solvent and the reactants undergo many collisions before reacting or separating. The contrast with
gas-phase reaction is striking; there encounter and collision are identical.
In solution, reaction may occur rapidly after the formation of the encounter
complex; in that case the kinetics of the overall process are limited not by
the chemical step but by the diffusive ones.
It is possible to make quite accurate estimates of the rate constant for
the diffusion process

A+B~(A ... B)
by solving the diffusion equation (2.23) under appropriate boundary conditions. This was done by Debye(25); kdiff depends upon the diffusion coefficents of the reactants and thus, by Walden's rule, upon the solvent
viscosity. If A and B are ions, kdiff is a very sensitive function of the solvent
dielectric constant and the product of the ionic charges. (26) In aqueous
solution most solvated species have similar size and comparable diffusion
coefficients. It is found that kdiff is most sensitive to changes in the charge
of the reacting species; it drops precipitously for reactions involving similarly
charged ions.
(25)
(26)

P. Debye, Trans. Electrochem. Soc. 82, 265 (1942).

We shall consider the Debye theory in Chapter 9.

Sec. 5.7

Simple Mechanisms-Solution

137

Since kdiff provides an upper bound to the overall rate of reaction, any
kinetic scheme from which one deduces a rate constant greater than kdiff
must be excluded from further consideration. Thus such estimates are
extremely valuable. A listing of rate constants for diffusion-limited reactions
involving very different species (free radicals, solvated electrons, ions)
is given in Table 5.3. To be exactly comparable the data should all be taken
at the same ionic strength and temperature. At higher temperatures and
higher ionic strengths the importance of ionic charge is somewhat reduced.
Nonetheless the charge product effect remains striking. The H + + S042reaction (charge product, -2) is 200 times as fast as the OH- + HP2073reaction (charge product, +3). The effect of diffusion coefficient, for reactions involving species with the same charge product, is illustrated by
comparing the rate constant for reactions involving H + (D ""' 10- 8 m 2 sec-I),
OH - (D ""' 5 X 10-9 m 2 sec-I), and all other solvated species (D ""'2 X 10-9
m 2 sec-I). The comparison is most striking for the rates of the H+ + OHreaction and the H + OH reaction. The former is 20 times faster, which is
not an ionic effect (from the data presented, the difference in charge product
might account for a factor of 2).
An example where the criterion of the diffusion limit was used to
eliminate a possible mechanism was a study of carbonyl addition reactions
similar to that discussed as an application of temperature jump in Section
4.9. Reactions of the type
H

RNH2

I
+ R'CHO :;;::::: R-N-C-R'
I I
H

OH

were shown not to involve the protonated form of the aldehyde as an

intermediate. (27) From available experimental data the rate constant for
the hypothetical step

H H

+ R'CHOH+

RNH
2

---->.

I I
I I

R-NLC-R'
H

OH

R'=N02~
(27)

E. H. Cordes and W. P. Jencks, J. Am. Chern. Soc. 84, 4319 (1962).

UO.2+ + H 20

Cl- + . CH.C0 2-

CI- + . CH 2C0 2H

OH- + HP 20,8-

p.O,'- + H.O

285

298

283

293

298

298

298

0.1

0.1

0.1

0.1

10"

4.7 X 108

3xlO"

5.0 X 10"

~6x

~1.6x1010

1.2 X 10"

6.5 X 10"

1.4 X 10"
1.4 X 1010

4.3 X 1010

7x 10"

1.4 X 1011

4.7 X 1010
3.4 X 1010

a M. Eigen, W. Kruse, G. Maass, and L. de Maeyer, Prog. React. Kinetics 2, 285 (1964), where original references are given.
b J. K. Thomas, Trans. Faraday Soc. 61, 702 (1965).
C
M. S. Matheson and L. M. Dorfman, Trans. Faraday Soc. 61, 156 (1965).
a E. Hayon and H. A. Allen, J. Phys. Chern. 65, 2181 (1961).
M. Anbar and E. J. Hart, J. Phys. Chern. 69, 271 (1965).
f D. L. Cole, E. M. Eyring, D. T. Rampton, A. Silzars, and R. P. Jensen, J. Phys. Chern. 71, 2771 (1967).
g M. S. Matheson and L. M. Dorfman, Pulse Radiolysis (Cambridge: MIT Press, 1969), p. 104.

eaq + Fe(CN),'-

[Co(NH.),OH.]8+
Fe(CN)a8-

CO.2- + H.O

H+ + [Co(NH.),OH]2+

OH- + HCO.-

H+ + U0 2(OH)+

eaq + CICH.C0 2-

eaq + ClCH 2CO.H

CaH.O- + H 20
298

OH- + CaH,OH

298
298

NH.+

298

298

293

298

~1.0x1011

(M-1 sec-I)

Rate constant

0.1

(M)

(K)

293

Ionic strength

Temperature

2 CHi:HOH ~ 2,3-Butanediol

+OH~H.O

H+ + NH.

H 20

~.

H+ + OH-

NH.OH

H.CO.

HSO.-

OH- + NH.+

H+ + HCO.-

H+ + 80. 2-

Reaction

Table 5.3. Rate Constants for Some Diffusion-Limited Reactions in Aqueous Solution

Reference

~.

~.

fl

"'"

g
~

00

......
w

Sec. 5.8

Complex Mechanisms-Solution

139

would have to have been 2.7 X 1012 M-1 sec- 1 to account for the observed
kinetics. This is 50 times as large as the rate constant for any reaction in
Table 5.3 with the same charge product. It is 400 times as large as the rate
constant for reactions with zero charge product which involve neither H +
nor OH-. Since solution reactions cannot have k ~ kdiff' this pathway
was excluded. The temperature-jump study of similar reactions carried out
9 years later provided direct confirmation of this deduction.
The solvent affects reactivity in other ways than just reducing mobility.
A case in point is the photodissociation of iodine which occurs when
iodine is irradiated with sufficiently high frequency light,

In the gas phase the quantum yield (the number of iodine molecules dissociated per quantum of light absorbed) is I; each photon dissociates a
molecule. (28) In solution the quantum yields drop dramatically, a result
that can be interpreted in terms of the effect of solvent structure. (29) Upon
absorbing light the molecule dissociates. However, separation is not immediate. The iodine atoms are trapped within a transient cage of solvent
molecules to which they may transfer any excess kinetic energy. They
undergo many collisions before separating. The observed quantum yield
is determined by competition between the rate of diffusion and recombination. It depends upon both the mass of the solvent molecule (which affects
the efficiency of energy transfer from iodine atoms to solvent) and the
viscosity of the solvent (which governs the rate of separation of the
atoms).

5.B. Complex Mechanisms-Solution

Many rate laws for solution reactions are consistent with diffusion
control. In general, knowledge of the chemistry of the reactants is the key
to unraveling the mechanism. We consider two examples.

(28)
(29)

J. Franck and E. Rabinowitch, Trans. Faraday Soc. 30, 120 (1934).

E. Rabinowitch and W. C. Wood, Trans. Faraday Soc. 32,1381 (1936); R. M. Noyes,
J. Chern. Phys. 18, 658 (1950); F. W. Lampe and R. M. Noyes, J. Am. Chern. Soc. 76,
2140 (1954); H. Rosman and R. M. Noyes, J. Am. Chern. Soc. 80, 2410 (1958);
D. Booth and R. M. Noyes, J. Am. Chern. Soc. 82,1868 (1960); L. F. Meadows and R.
M. Noyes, J. Am. Chern. Soc. 82, 1872 (1960).

Chap. 5

140

5.8.1. The Fe(Ill)

Stationary State Mechanisms

+ EU(Il) System-The Effect of pH

In Section 4.10 the rate law for the electron-transfer reaction

+ Fe (III) =

Eu(II)

Eu(III)

+ Fe(II)

was determined for values of the pH between 1 and 3. More elaborate

data analysis indicates that the rate law is
_ d[Eu(II)]

dt

k[Fe(III)][Eu(II)]

(5.39a)

with the pH-dependent rate constant well represented by(30)

k

ko

k_ I

[H +]

)(

[H+]
)
[H +] + Ka

(5.39b)

To interpret this requires considering the equilibrium chemistry of the

ferric ion. Even in acid solution (see Problem 4.10) Fe (III) exists in the two
rapidly equilibrated forms Fe3+ and FeOH2+:
(equilibrium, K)
Using the equilibrium condition and the stoichiometry we find
[F 3+]
eaQ

[H+][Fe(III)]
[H+] + K '

[F OH2+]
e
aq

K[Fe(III)]
[H+] + K

(5.40)

Assuming parallel binary reactions between Eu(II) and each of the ferric
ion species, the postulated rate law is
-

d[E~;II)]

k'[Eu(II)][Fe!ci]

+ k"[Eu(II)][FeOH~;:-]

(5.41)

Substituting from (5.40) yields the experimental form (5.39) if we identify

k'

ko,

k"K=k_I'

K=Ka

At 1.4C and unit ionic strength the parameters are ko = 3.38 X 103
M-I sec-I, k-I = 2.24 X 103 sec-I, and Ka ""' 7 X 10-4 M so that kIf ""'
3.2x 10 6 M-I sec-I. Under the experimental conditions (pH 1-3), [Fem >
(30)

D. W. Carlyle and J. H. Espenson, J. Am. Chern. Soc. 90, 2272 (1968).

Sec. 5.8

Complex Mechanisms-Solution

141

[FeOHi~] and only at the highest pH is the FeOH~~ concentration significant. Nonetheless, since the hydroxy complex is so much more reactive,
the pathway involving that species predominates at all but the lowest pH.
Note that neither k' nor k" are comparable to their diffusion limits (about
2 X 10 7 M-l sec-1 and 2 X lOB M-l sec-I, respectively). The chemical process
of electron-transfer controls the rate of reaction of either Fe~~ or FeOHi~.

5.8.2. The Fe(III)

+ V(III)

Reaction-Inhibition

In studies of the electron-transfer process

V(III)

+ Fe(nI) =

V(IV)

+ Fe(n)

initial kinetic experiments were consistent with the rate law

_ d[Fe(III)]
dt

k o[Fe(I1I)][V(II1)]

(5.42)

More detailed study indicated greater complexity. The reaction rate depends
upon the product concentration; it is accelerated by V(lV) and inhibited
by Fe (II). The complete rate law is(31)
_ d[Fe(lII)]
dt

{k'

+ k"

[V(IV)] }[Fe(III)][V(III)]
[Fe(II)]

(5.43)

which cannot be observed if, initially, no products are present. Under

such conditions their concentrations are always equal and (5.43) reduces
to (5.42) with ko = k' + k".
As is the case in the Fe(III) + Eu(II) system the rate constants k'
and k" are pH dependent, which in all probability reflects the different
reaction rates of the various hydroxy forms of Fe (III). Even ignoring this
complication the rate law (5.43) requires a mechanism that postulates
parallel reactions. The first term in the rate law is consistent with the onestep electron-transfer reaction
Fe(III)

+ V(III) ~ Fe(II) + V(lV)

(5.44)

Since the second term in the rate law is minus first order in Fe(n), it must
(31)

w. C. E. Higginson, D. R. Rosseinsky, J. B. Stead, and A. G. Sykes, Disc. Faraday Soc.

29, 49 (1960).

Chap. 5

142

Stationary State Mechanisms

involve more than a single step. In some step V (IV) must appear as a reactant. Furthermore Fe(II) must appear as a product in a reversible step in
order for it to act as an inhibitor. Recognizing that vanadium can exist
in all oxidation states from +2 to +5, a simple pathway that incorporates
these constraints is
ko

V(IV) + Fe(III) ~ V(V) + Fe(II)

k-2

V(V) + V(III) ~ 2V(IV)

(5.45)

Assuming the steady-state approximation can be applied to [V(V)J the

rate law derived from (5.45) is

d[Fe(III)]
dt

kak2[Fe(III)) [V(IV)] [V(III)]

L2[Fe(II)] + k3[V(IIl))

which, if L2[Fe(II))?> k 3 [V(III)), reduces to the second term of (5.45)

with k" = k3k2/k_2. In (5.45) the V(IV) formed by direct reaction of
Fe(III) and V(III) is oxidized still further. The +5 and +3 forms of vanadium then react to reform the V(IV). Such pathways, where the oxidation
(or reduction) of one reactant temporarily proceeds beyond the thermodynamically stable state during the course of reaction, are common for

species which exist in many oxidation states. There is always a further step
which reduces (or oxidizes) the intermediate to the stable state.

5.9. Homogeneous Catalysis

The rate of a chemical reaction may be enhanced in many ways. These
methods include:
(1)
(2)
(3)
(4)

raising the temperature;

irradiating with light;
changing the solvent;
adding trace amounts of a reagent.

The first two methods are not catalytic since they are based on adding
energy to the system. The third involves gross alteration of the reaction
medium. Only the fourth, in which trace quantities of reagent markedly
affect the reaction rate, is catalysis.
Conventionally a catalyst is defined as a substance that alters the
speed of a chemical reaction without undergoing any chemical changes

Sec. 5.9

Homogeneous Catalysis

143

itself. Thus it should not appear in the overall stoichiometry of the reaction
nor should it combine other than transiently with either products or reactants. In practice this is a somewhat restricted use of the term; catalysts
often combine with reaction products or are slowly used up in side reactions
which occur along with the process that gives them their catalytic activity. (32)
A case where a catalyst accelerates reaction is the commercial process
for sulfuric acid. The first step is the oxidation of S02 to S03; uncatalyzed
this proceeds via the slow termolecular reaction
(5.46)
In the presence of NO a two-step pathway may be followed (the termolecular
NO-0 2 reaction has zero activation energy):

+ O 2N0 2
N0 2 + S02 ---' NO + S03
2NO

2 ---'

(5.47)

Each of these reactions is reasonably fast; the net rate of production of

S03 is greatly enhanced by addition of NO. The overall chemical result
of the sequence (5.47) is exactly the same as the sequence (5.46). The
catalyst provides a parallel pathway to reaction, one that is much faster
than the uncatalyzed one.
An example which illustrates a different aspect of catalysis is the Wacker
process for oxidizing ethylene to acetaldehyde. (33) The earlier technique
involved two steps and the isolation of an intermediate. However, consideration of the aqueous reactions(34)
C 2H 4 + PdCl 2

+H0 =
2CuCI + Pd(s) =
2

CH 3CHO
2CuCI

+ 2HCI + Pd(s)

+ PdCl

suggests that Cu(II) could be used to oxidize ethylene as long as both

An example of a catalyst that is slowly destroyed in the course of its catalytic action is
the catalytic converter in new automobiles. The Pt catalyst accelerates the conversion of
CO to CO 2 , At the same time it also aids in the formation of S03 and NO. , products
which, at elevated temperatures, will combine with the Pt. In addition it is poisoned
by accumulation of carbon on the surface.
133) J. Smidt, W. Hafner, R. Jira, J. Sedlmeier, R. Sieber, R. Riittinger, and H. Kojer,
Angew. Chem. 71, 176 (1959).
184) The specific species which participate are various complexes of Pd(II), Cu(II), and
Cu(I). They are not indicated.
132)

Chap. 5

144

Stationary State Mechanisms

Pd(II) and Pd(O) were present

Then, since CuCI is easily oxidized in HCI solution

one could imagine that by properly adjusting the conditions trace quantities
of Cu and Pd could catalyze a one-step oxidation in aqueous HCI
(5.48)

Thus judicious use of catalysts may provide a mechanism that is both

faster and requires isolating fewer intermediates. The elementary steps in
the Wacker process are quite involved; a succession of organopalladium
intermediates is formed. (3S) The chemistry of the reaction (5.48) is complex
but the individual steps are rapid.
Catalysis is, in many cases, irreversible; the mechanism may be represented as
A
A

kl

B ~ products

slow (uncatalyzed)

+ catalyst -~ X
k2

fast (catalyzed)

+ B -~ products + catalyst
k2'

(5.49)

fast

Assuming a steady state in the intermediate X, the reaction rate is

d(product]
dt

kl[A][B]

+k

[A][catalyst]

so that, if k2 ~ kl' a small amount of catalyst greatly accelerates the reaction. In addition (5.49) predicts that the rate law is first order in catalyst,
a common (though not universal) experimental situation. A counterexample is the N 20 S catalyzed decomposition of ozone which is i order
in both N 20 S and 03. (36)

(35)
(36)

For their characterization see A. Aquilo, Adv. Organornet. Chern. 5, 321 (1967).
H. J. Schumacher and G. Sprenger, Z. Physik. Chern. A140, 281 (1929); B2, 267
(1929).

Sec. 5.10

Enzyme Catalysis

145

5.10. Enzyme Catalysis

In biochemical systems a small amount of enzyme (E) catalyzes conversion of a substrate (S) to product (P). (37) The rate law observed, in
many cases, is
d[P]
kEo[S]
(5.50)
--crt" = Km + [S]
where Eo is the total enzyme concentration. The simplest model consistent
with (5.50) is the Michaelis-Menton mechanism(38)
E

+ S , kl ,
k_l

ES ~ E

+P

(5.51)

Assuming that ES is in a steady state we find

and the rate of product formation is

(5.52)

where K". = (k2

L1)/k1 is known as the Michaelis constant. Since the
uncomplexed enzyme concentration is generally not experimentally accessible, (5.52) is more useful when expressed in terms of the total enzyme
concentration Eo. Since Eo = [E) + [ES] we find

and hence

d[P]

--crt"

(5.53)

precisely the form observed experimentally (5.50). By varying the substrate

concentration both k2' known as the turnover number, and the Michaelis
constant can be determined. At high substrate concentration there is no
free enzyme (the enzyme is saturated) and (5.53) is exceptionally simple;
(37)

(38)

For an introduction to the biochemistry of enzyme mechanisms see A. L. Lehninger,

Biochemistry, 2nd ed. (New York: Worth, 1975), Chapters 8 and 9; J. T. Wong, Kinetics of Enzyme Mechanisms (London: Academic Press, 1975).
L. Michaelis and M. L. Menton, Biochem. Z. 49,333 (1913) used a preequilibrium formulation valid when k2 ~ k_ 1 The first steady-state treatment was by G. E. Briggs
and J. B. S. Haldane, Biochem. J. 19, 338 (1925).

Chap. 5

146

Stationary State Mechanisms

the rate of product formation is a constant, V = k 2E o. The Michaelis

constant can then be determined from the integrated form of (5.53)

by plotting In [S] vs. [S] + Vt; the slope is - Km. Alternatively the reaction
velocity v = d[P]/dt can be plotted as a function of substrate concentration
in a variety of ways. The most common are

V-I

l/V + (Km/V)[S]-I

(Lineweaver-Burk)

V - Km(v/ [S])

(Eadie)

[S]/v

Km/V + [S]/V

. (Hanes)

Of these the Hanes transformation is most reliable for extracting values

of kinetic parameters. (39)
The Michaelis-Menton mechanism (5.51) is the simplest member of
a set of series-parallel reactions,

EH+

+ S~X2~EH+ + p

1l

1l

(5.54)

1l

This particular mechanism takes into account the fact that enzymes can
be protonated which alters the specific rate constants for the various
elementary steps. Similar, but more complicated, schemes arise if either
substrate or product (or both) can be protonated. In addition substances
other than hydrogen ion may complex with enzyme, substrate, or product.
The reaction network would be analogous to (5.54). While a direct steadystate treatment of (5.54) is complicated, a systematic method for determining the rate law in terms of the specific rate constants has been worked
out.(40)
(39)

(40)

J. T. Wong, Kinetics of Enzyme Mechanisms (London: Academic Press, 1975), pp.

239-240.
E. L. King and C. Altman, J. Phys. Chem. 60, 1375 (1956). A clear exposition of the
method is given by K. M. Plowman, Enzyme Kinetics (New York: McGraw-HilI,
1977), pp. 30-38, 156-164. For a formulation that lends itself to computer treatment
see A. R. Schulz and D. D. Fisher, Can. J. Biochem. 48, 922 (i970).

Sec. 5.11

Heterogeneous Catalysis

147

In addition to varying pH, enzyme kinetics can be studied by adding

other substances which react with the enzyme in a variety of ways. The use
of inhibitors is often an aid to understanding the nature of the enzymatic
activity. In competitive inhibition the inhibitor complexes the enzyme and
renders it inactive. Assuming the dissociation equilibrium
EI~E

+ I,

KI

[E ][1]

[EI]

competes with the first step in (5.51), the analog to (5.53) is

d[P]
dt

[S]

+ Km{l + [I]/Kr}

(5.55)

Comparison with (5.53) indicates that the dependence of reaction rate on

substrate concentration is unaltered; however, the apparent Michaelis
constant is now a function of inhibitor concentration.
In uncompetitive inhibition the inhibitor is presumed to complex the
reaction intermediate ES,
ESI~ES

+ I,

KI' = [ES][I]/[ESI]

and thus compete with the product formation step in (5.51). The reaction
rate is now
d[P]
(5.56)
dt
[S]{l + [I]jKr'} + Km
Again the dependen~e on [S] is unaffected; however, the limiting velocity
at high substrate concentration is now a function of inhibitor concentration.
The rate laws (5.55) and (5.56) are kinetically distinguishable and permit
differentiation between these possible modes of inhibition. Another important, distinguishable process is noncompetitive inhibition (Problems
5.13 and 5.14).

5.11. Heterogeneous Catalysis

When metal catalysts are used to accelerate gaseous reactions the
mechanism involves five steps:
(1) diffusion or convection of reactant to the surface;
(2) adsorption of reactant;

Chap. 5

148

Stationary State Mechanisms

(3) elementary reactions of adsorbate, either with gas molecules or

with a nearby adsorbed molecule;

(4) desorption of product;
(S) diffusion or convection of product from the surface.

Any of these may be rate limiting. Measurements of diffusion rates allow

(1) and (S) to be characterized separately. Adsorption and desorption may
often be studied separately as well, allowing the surface reaction (3) to be
studied separately.
Except at very low pressure diffusion does not limit the rate of reaction;
however, adsorption is very important. If the surface has a total of So
adsorption sites a general mechanism is

+ S:;:=:: A . S
A . S + ... ~ products
A

(equilibrium, K)

k2

(S.S7)

Whether A . S itself reacts or whether it combines with a gas-phase molecule

to yield product, (S.S7) is mechanistically equivalent to the MichaelisMenton mechanism (5.S1); it differs from the mechanism (S.49) in that
the adsorption process is reversible. The surface sites replace enzyme
molecules as catalysts; if [A] is too large the surface is saturated. Adapting
the arguments that led to (5.53) yields the Langmuir adsorption isotherm(41)
which is the fraction of occupied sites (J == [A . S]/ So,
(J

= K[A]/{l

+ K[AJ) =

bp/(l

+ bp)

(S.58a)

where, assuming the ideal gas law applies,

b

KjkT

Pursuing the enzymatic analogy, inhibition arises if another constituent

of the gas is adsorbed also,
B+S~BS

(equilibrium, K')

The same analysis which led to (S.S8) now indicates that

(JA = K[A]/{l

+ K[A] + K'[BJ)

where ()A is the fraction of sites occupied by A.

(41)

I. Langmuir, J. Am. Chern. Soc. 38, 2221 (1916); 40, 1361 (1918).

(S.59)

Sec. 5.11

Heterogeneous Catalysis

149

A variety of mechanisms for the reaction of adsorbate can be imagined:

k2

A . S ~ product

(5.60a)

+ B ~ product

(5.60b)

A . S

ko

A S + B

ko

S~product

(5.60c)

In (5.60a) adsorbate undergoes a unimolecular transformation; an example

is the formation of atomic H at the walls of a high-temperature reaction
vessel containing H 2 In (5.60b) adsorbate reacts with a gas molecule as
is found in some surface-catalyzed combinations of atoms and free
radicals. (42) In (5.60c) adsorbed molecules on neighboring sites interact,
as seems to be the case in the exchange reaction between D2 and NH 3 (43)
Corresponding to the mechanisms are the rate equations
R

== -

d[A]
(Jt =

k 2 KSo[A]
1 + K[A] + K'[B]

k2KSO[A][B]
K'[B]

1 + K[A]

(5.61b)

k 2KK' SorA] [B]

(1

(5.61a)

+ K[A] + K'[B])2

(5.61c)

In every case a steady-state concentration of adsorbate is assumed. The

pressure dependence of the reaction rates is extreme. For the case involving
reaction of adsorbed species the rate varies from second order to zeroth
order as surface coverage increases. If species A is preferentially adsorbed
the reaction rate, at high [A], is

R= k 2K' S ~
K

[A]

(5.62)

If a nonreactive inhibitor is added the situation is more c{)mplex (see

Problem 5.19). For the cases considered desorption of product is presumed
to be very fast. If it is not, there is product-induced inhibition which must
be overcome for practical catalytic application (see Problem 5.18).
Reactions at surfaces are dependent on the phenomenon of adsorption.
The treatment in the preceding paragraphs has assumed all surface sites
(40)

One example is the surface-catalyzed recombination of H and OH, W. V. Smith, J.

(43)

Chern. Phys. 11, 110 (1943).

J. Weber and K. J. Laidler, J. Chern. Phys. 19, 1089 (1951).

150

Chap. 5

Stationary State Mechanisms

are similar which is not true. Elegant experiments, in which various gases
are used to etch polished metal spheres, have shown that different crystallographic surfaces are etched at very different rates. (44) In an extreme example,
where copper was exposed to oxygen, the {100} face reacted 17 times as
fast as the {311} face. (45) With the development of ultrahigh vacuum techniques (p ;S 10-12 atm) it has become possible to study the details of the
adsorption process. At such ultralow pressures, adsorption is slow enough
that reproducible surface crystallographic studies of an adsorbed monolayer
are feasible using low-energy electron diffraction (LEED).(46)
Depending upon the binding energy, adsorption is described as physical
adsorption or chemisorption. In physical adsorption binding is weak
(10--30 kJ mol-I) and the adsorbate-substrate separation is large, typical
of van der Waals interaction. In chemisorption, binding is stronger (>60
kJ mol-I) and the separations are short, typical of chemical bonds. The
surface structures formed by physically adsorbed and chemisorbed species
differ markedly.
Physical adsorption of inert gases on metals is studied at temperatures
between 10 and 78 K. At too high a temperature the adsorbed layer boils
off the surface. At too low a temperature an adsorbed gas molecule does
not migrate on the surface after striking it; the surface structure is random
and does not anneal to reflect the energetics of adsorbate-substrate interaction. In the temperature range for which surface equilibration occurs,
the structure of the surface layer is independent of the inert gas adsorbed
and of the metal surface exposed. Whether Xe is adsorbed on graphite, (47)
Pd, (48) Ir, (49) or CU(50) the surface layer develops a structure of hexagonal
symmetry, an arrangement which corresponds to the closest-packed plane
of crystalline Xe. Similar results are found in the binding of Ne or Ar to
the {100} face of Nb. (51) In each instance the structure of the surface layer
is determined by adsorbate-adsorbate interaction; the adsorbate-substrate
binding is too weak to have a significant effect. (52)
A. T. Gwathney and R. E. Cunningham, Adv. Catal. 10, 57 (1958).
R. W. Young, Jr., J. V. Cathcart, and A. T. Gwathney, Acta Met. 4, 145 (1946).
(46) For an introductory review see J. C. Buchholz and G. A. Somorjai, Ace. Chern. Res.
9, 333 (1976).
(47) J. J. Lander and J. Morrison, Surf Sci. 6, 1 (1967).
(4S) P. W. Palmberg, Surf Sci. 25, 598 (1971).
(49) A. Ignatiev, A. V. Jones, and T. N. Rhodin, Surf Sci. 30, 573 (1972).
(50) M. A. Chesters and A. Pritchard, Surf Sci. 28, 460 (1971).
(51) J. M. Dickey, H. H. Farrell, and M. Strongin, Surf Sci. 23, 448 (1970).
(52) The surface layer is oriented in a limited number of ways with respect to the underlying
metal, indicating some specificity in the adsorbate-adsorbent interaction.
(44)
(45)

Sec. 5.11

Heterogeneous Catalysis

151

Physically
Adsorbed State

2M-X

Chemisorbed
State

Figure 5.7. Schematic diagram for the variation of potential energy as a function of distance
from a metal surface for dissociative adsorption of a molecule X 2 Qp is the heat of physical
adsorption for the molecule, Qc is the heat of chemisorption, and QD is the heat of dissociation of the free molecule. The binding energy per chemisorbed atom is (QD + Qc)/2.
Note that for I Qc I sufficiently small, chemisorption no longer occurs. [T. N. Rhodin
and D. L. Adams, Treatise on Solid State Chemistry, N. B. Hannay, ed. (New York:
Plenum Press, 1976), Vol. 6A, p. 346, reprinted by permission.]

Catalysis, which typically requires bond rupture of the adsorbate

(e.g., H2 -----' 2H), is a problem in chemisorption. Molecular dissociation is
described as a two-step process
M + X 2 --' M - - X 2
M - - X 2 -----' 2M - X

(physical adsorption)
(chemisorption)

(5.63)

In the first step X 2 is weakly bound at large separation; if M-X interaction

is sufficiently large, dissociation occurs and a chemisorbed species is formed.
The energetics of the model, with its two distinct steps, is illustrated in
Fig. 5.7. (53) The binding energies of the chemisorbed species are large; as a
result the equilibrium surface structure reflects the periodicity of the metal
surface. Binding is very site specific and chemisorbed atoms, to maximize
their interaction with substrate atoms, occupy positions of high coordination
number. (54) Bonding appears to be basically covalent. A comparison of
(53)
(54)

The model was proposed by J. E. Lennard-Jones, Trans. Faraday Soc., 28, 28 (1932).
S. Anderson and J. B. Pendry, J. Phys. C. 5, L41 (1972); J. E. Demath, D. W. Jepsen,
and P. M. Marcus, Phys. Rev. Lett. 31, 540 (1973); 32, 1182 (1974).

152

Chap. 5

Stationary State Mechanisms

adsorbate-metal atom distances with those estimated by summing covalent

radii indicate few significant differences. (55)
Nondissociative molecular chemisorption has been less exhaustively
studied. In the one case where a full LEED analysis is available, it appears
that geometric constraints are very significant. Acetylene binds to the
Pt{lll} face at a triangular site(56); here the n-orbitals interact effectively
with exposed Pt atoms without molecular deformation being required. (57)
Chemisorption is too effective if adsorbate-substrate binding is stronger
than metal atom binding at the surface. In such cases the adsorbate atom
is actually incorporated in a reconstructed metal surface so that adsorbate
and metal atoms are intermixed in the surface layer. Such processes destroy
catalytic activity. They may be the first step in a solid-state reaction leading
to oxide, carbide, or nitride formation. Surface reconstruction has been
observed when C or 0 are chemisorbed on W, (58) Ni, (59) or Fe(60) at high
temperatures.
Modern surface chemistry remains terra incognita. Recent work,
designed to identify surface features that promote catalytic activity, suggests
that there are significant differences between smooth and stepped surfaces
of transition metals.(6I) On smooth Pt surfaces, like the {IOO} and {llI}
faces, neither H2 nor O 2 chemisorb readily; however, on staircaselike
surfaces both appear to chemisorb (and dissociate) at relatively low temperatures. (62) Similar effects may control the rate at which C-C and C-H
bonds are broken using Pt catalyst.(63) The interior and exposed edges of
stepped surfaces may bind molecules preferentially.(64) Other discontinuities
which affect the electron densities of the surface atoms could also effect their
binding properties. As a result, dislocations might alter catalytic afficiency. (65)
(55)
(56)

J. C. Buchholz and G. A. Somorjai, Ace. Chern. Res. 9,333 (1976).

L. L. Kesmodel, P. C. Stair, R. C. Baetzold, and G. A. Somorjai, Phys. Rev. Lett.

36, 1316 (1976).

Such structures need not be simple. Recent work indicates that benzene, when bound on
the Pt {Ill} face, does not bond parallel to the surface but is tilted [Po C. Stair and G.
A. Somorjai, J. Chern. Phys. 67, 4361 (1977)].
(58) M. Boudart and D. F. Ollis, The Structure and Chemistry of Solid Surfaces, G. A. Somorjai, ed. (New York: John Wiley, 1969).
(59) A. U. MacRae, Science 139, 379 (1963).
(60) J. E. Boggio and H. E. Farnsworth, Surf Sci. 3, 62 (1964).
(61) B. Lang, R. W. Joyner, and G. A. Somorjai, J. Catal. 27, 405 (1972).
(62) J. C. Buchholz and G. A. Somorjai, Acc. Chern. Res. 9, 333 (1976).
(63) J. C. Buchholz and G. A. Somorjai, Acc. Chern; Res. 9, 333 (1976).
(64) B. Lang, R. W. Joyner, and G. A. Somorjai, Surf Sci. 30, 454 (1972); M. Salmeron,
R. J. Gale, and G. A. Somorjai, J. Chern. Phys. 67, 5324 (1977).
(65) K. Besocke and H. Wagner, Surf Sci. 53, 351 (1975).
(57)

Appendix A

153

Appendix A. Matrix Solution to the Coupled Rate Equation (5.2)

The set of coupled equations (5.2) can be readily solved using matrix
methods. (66) The advantage of this approach is that it can easily be generalized to treat any number of coupled first-order (or pseudo-first-order)
rate equations. For simplicity consider a two-step process and define a
column vector constructed from the concentrations
(5A.I)
Then (5.2) can be written as
dX =K.X
dt

(5A.2)

where the corresponding rate constant matrix is

(5A.3)

From the results of matrix algebra it is known that a transformation

matrix T can always be constructed to generate a new vector
Y= T X

(SA.4)

which has the property that

dY
dt

-=

-AY

(SA.S)

where A is diagonal, i.e.,

(5A.6)

The transformation T has decoupled the rate equations and the components
1&6)

See, e.g., H. Margenau and G. M. Murphy, The Mathematics of Physics and Chemistry
(Princeton: Van Nostrand, 1956), Chapter 10.

Chap. 5

154

Stationary State Mechanisms

of (SA.S) are the simple first-order equations

which have the solution

(SA.7)

so that the Ai are the decay constants.

The rules for determining A and T are well known; A is found by
solving the algebraic equation which is formed by expanding the determinant
(SA.8)
I K + All = 0
where I is the unit matrix,

(1 0 0)

1= 0 1 0
001

The roots of (SA.8) are just the Ai' To determine T multiply (SA.2) by T
which yields, with (SA.4),
dY
-=TKX
dt

Introducing (SA.4) into (SA.S) we find another relation between dY/dl

and X
dY
- = -A T X
dt

Comparing these two equations yields a matrix equation for T

T .K

-A T

(SA.9)

which is really a set of simultaneous linear equations in the unknown elements of the T matrix, lij' Introducing the expression for A, (SA.6), this
becomes
(SA. 10)
I limkmj = -A4ij
m

Thus once the

(SA. 10).

Ai

are determined from (SA.8) the

lij

can be found from

Appendix A

155

In most instances only the decay constants are of interest. Combining

(5A.3) and (5A.8) and expanding the determinant, the equation which
determines the decay constants for the rate equations (5.2) is

This has three roots, one obviously A = 0, and the others are found by
applying the quadratic formula. Defining the quantities

the roots are

Al

A2

Aa

+ a' - [(a {a + a' + [(a {a

+ 4Lk/P/2}j2
a')2 + 4Lk+'p/2}j2
a')2

(5A.l!)

The root A = 0 is a restatement of the condition of conservation of mass.

From (5A.7) a certain linear combination of concentrations, Yi(t), is constant. The other roots determine the temporal behavior of the system;
in terms of the notation of Section 5.2,

and x_, being the smaller decay constant, governs the long-term behavior
of the system.
If, for some reason, the concentrations Xi(t) are also needed, they
may readily be computed. From (5A.4) we obtain

T-I . Y

==

5 Y

(5A.12)

so that with (5A.7)

(5A.13)
To find yiO) apply (5A.4) at

t =

and using (5A.13) we have

(5A.14)

Chap. 5

156

Stationary State Mechanisms

or, in matrix notation,

X(t)

T-l F(t) T X(O)

(5A.I5)

where
(5A.16)

Appendix B. Approximations to u+ and u_

If a ~ a', (5.6) or (5A.ll) can be greatly simplified. We can write

a' )
"' -_ Ta {( 1 + -a

a' )2
4k_k/ ]1/2}
+ (-a+ -a---:

2a'

1 - -a-

2,-'-

(5B.1)

Since the terms under the square root are much less than 1 we can use the
approximation
(1

+ X)1I2 = 1 + -2 - -x8 + ...

2

(5B.2)

and find that quantity to be

1 [
2a'+ ( -a'
1+-2
a
a

)2 + 4k_k+' ] - -1 ( -2a'
- )2
a2

(5B.3)

The second term in this expansion is needed because there are quadratic
terms in the quantity being approximated. Consolidating (5B.3) yields

which may be com:bined with (5B.1)

"'+

a (1 a
+ -'a

r=:::::; -

"'_ r=:::::;.!!...-

+1-

- a'

2k -k +')
a2

(1 + ~a _ 1+ ~a _ 2Lk+')
a
2

a
(5B.4)

a' _ k_k+'
a

Since a' ~ a by hypothesis, the term Lk+'/a satisfies the following inequalities

and thus has been neglected in comparison to a in "'+.

Problems

157

The approximation to (5.6) when k+' or L (or both) are large is found
in a similar way. Define

so that with (5.7) the expressions for the decay constants are
2r + r2+
x
=2q-{I +qr- [ I -
q
q2

4(k +k

+ k +'k-' q2

k +k ') ]1/2}

(5B.5)
Approximating the square root using (5B.2) yields
I _ ~
q

2(k+L

+ k/L' q2

k+L')

Combining this with (5B.5) leads to

if:+ = q

rq - k+k_ - k+'k_'
q

+ k+k_'

(5B.6)

The term k+L'/(k+' + L) has been dropped in the expression for if:
because of the inequality

which must be satisfied since, by hypothesis,

Problems
5.1.

Assume a two-step mechanism of the form (5.1) for which k+ = 106 sec-l,
k_ = 103 sec-\ k+' = 103 sec-\ and k_' = 1 sec-l. Assume initially only
A is present.
(a) Calculate "+ and ,,_ using (5.6).
(b) Determine A(t), B(t), and e(t) by substituting expressions of the form
A(t) = a+ exp( -,,+t) + a_ exp( -,,-t) + a eq into the rate equations.

Chap. 5

158
5.2.

Stationary State Mechanisms

The reversible decomposition of phosgene COCl 2(g) into CO (g) and CI.(g)
has been interpreted according to two distinct mechanisms l ' )
Cl2 = 2CI

(equilibrium, K , )

+ CO = COCl
ka
COCl + CI ~ COCl2
k_a

(equilibrium, K 2 )

(i)

CI

(ii)

CI

+ Cl 2 =

CIa

+ CO ~ COCl 2 + CI

(equilibrium, K 4 )

CIa
k5

k_5

Derive the expression for the rate of formation of COCl, according to both
mechanisms. Are they kinetically distinguishable?
5.3.

The activation energy for the reaction

2NO

+ O2 =

2N02

is -41 kJ. Suggest a mechanism to account for this.

5.4.

Bimolecular association reactions involving simple species generally have

Arrhenius A-factors much less than the collision frequency. Interpret this
in terms of the discussion in Section 5.5. [For typical A-factors see S. W.
Benson, The Foundations of Chemical Kinetics (New York: McGraw-Hill,
1960), p. 302.]

5.5.

The pyrolysis of ethane may be accounted for by the Rice-Herzfeld mechanism [F. O. Rice and K. Herzfeld, J. Amer. Chem. Soc. 56, 284 (1934)]:

(0 C 2H. ~2CH3
(ii) CH, + C,H, ~ CH,

(iii)
(iv)
(v)

C2H. ~ C2H 4 + H
H + C2H. ~ C2H.
H + C2H. ~ C 2H.

(initiation)

+ C'H')

+ H2

(propagation)
(termination)

(a) Show that d[C 2H 4]/dt ex d[H 2 ]/dt ex -d[C 2H.]/dt ex [C 2H.].
(b) How can this mechanism be experimentally distinguished from unimolecular decomposition?
(c) Use the bond energies for C-C, C=C, C-H, and H-H (342, 614,
415, 435 kJ mol-I, respectively) to estimate lower bounds for the activation
energy in each step.
(d) Use these estimates to show that
-d[C 2 H.]/dt

d[H 2 ]/dt

d[C 2H 4]/dt

(k,kak 4/k.)li2[C 2H 6]

(e) Estimate the apparent activation energy in the overall rate law.
I.')

After I. Amdur and G. G. Hammes, Chemical Kinetics (New York: McGraw-Hili,

1966), p. 84.

159

Problems
5.6.

Another pyrolysis whose kinetics is described by the Rice-Herzfeld mechanism is the decomposition of acetaldehyde:
(initiation)
(i) CH 3 CHO ~ CHs + CHO
CHs + CH 3 CHO ~ CH 4 + CH,CHO }
(propagation)
(iii) CH,CHO ~ CO + CHs
(termination)
(iv) 2CH s ~ C,H o
(ii)

(a) Determine the rate laws for the formation of the stable products and the
decomposition of acetaldehyde.
(b) Use the bond energy data of Problem 5.5 and estimate the activation
energy for each of the apparent rate constants determined in (a).
5.7.

In the formation of HBr, termination steps involving atomic hydrogen were

ignored because [H] is small. Make whatever approximations you feel are
reasonable and demonstrate the truth of this assertion.

5.8.

In the Christiansen-Herzfeld-Polanyi mechanism only chemical processes

,were included. In addition to recombination of Br atoms, the chain may
be terminated by adsorption at the walls of the reaction vessel. Consider
the additional steps in (5.28)
k.

Br ~ adsorbed atom
H

k7

adsorbed atom

and derive the analogue to (5.34). What reaction conditions would emphasize
adsorption?
5.9.

The observed second-order rate constant for the reaction

semicarbazide

+ RCHO --+ semicarbazone

kObo = ko + 10 [H+] M- sec-

1
1 while at
4 33
is pH dependent. At low pH,
1
1
13
O
high pH, kObs = ko' + lO [H+] M- sec- for reaction with p-chlorobenzaldehyde [E. H. Cordes and W. P. Jencks, J. Amer. Chern. Soc. 84,
4319 (1962)]. The equilibrium constant for the protonation reaction

RCHO

H+ = RCHOH+

is K = 10- 7 26
(a) Assume the slow step in the reaction is
semicarbazide

k.

+ RCHOH+ ~ protonated

semicarbazone

and show that the pH-dependent term in kObO would be k,K[H+].

(b) Evaluate the k2 required if the pathway in (a) is important at low pH,
high pH. Compare your values with those for diffusion limited reactions.
Is the pathway in (a) tenable?

Chap. 5

160

Stationary State Mechanisms

5.10. A proposed mechanism for the 1- + ClO- exchange reaction is

ClO- + H.O = HClO + OHHClO +

k.

I-~HIO

k_.

(equilibrium, K,)

+ Cl-

HlO + OH- = 10- + H.O

(equilibrium, Ka)

(a) Show that

d[Cl-]/dt

{k.K,[ClO-][I-] - L.Ka-'[IO-][Cl-]}f[OH-]

(b) At 298 K, Kl = 3.1 X 10- 7 MI and K J - 1

served rate law in basic media is

= 4.3

d[Cl-]/dt = x[ClO-][I-]/[OH-],

10-4 M. The ob-

x = 60 sec-'

What can be deduced about k. and L.?

5.11. Suggest a plausible mechanism for each of the following reactions based
upon the rate law. All are in aqueous solution. In some instances there may
be more than one reasonable mechanism. ( )
(a) [J. H. Baxendale and P. George, Trans. Faraday Soc. 46, 736 (1950).]
Fe(H.O).3+ + 3bipy(ridine) = Fe(bipy)aa+ + 6H.0
d[Fe(bipy).3+]/dt

k[Fe(H.O).3+][bipy]3

(b) [A. B. Hoffman and H. Taube, Inorg. Chern. 7, 1971 (1968).]

(NH.).CoO.Co(NH a)54+ + 4V(II) + 14H+
-d[Co complex]/dt

2Co(II) + IONH.+
+4V(III) + 2H.0

k[Co complex]

(c) [0. J. Parker and J. H. Espenson, Inorg. Chern. 8, 1523 (1969).]

Fe(III) + Cu(I) = Fe(II) + Cu(lI)
-d[Fe(I1I)]/dt = k[Fe(III)][Cu(l)]f[H+]
(d) [W. C. E. Higginson, D. R. Rosseinsky, J. B. Stead, and A. G. Sykes,
Disc. Faraday Soc. 29, 49 (1960).]
2Hg(II) + 2V(IIl)
-d[V(III)]/dt

(.8)

Hg.(I) + 2V(IV)

[Hg(II)][V(I1I)]'{(a[V(IV)] + b[V(III)J)-'
+(c[V(lV)] + d[Hg(II)])-'}

After R. G. Wilkins, The Study of Kinetics and Mechanism of Reactions of Transition

Metal Compounds (Boston: Allyn and Bacon, 1974), pp. 117-118.

Problems

161

5.12. The N 20s-catalyzed decomposition of 0 3 follows the unusual rate law

-d[03]/dt = ka[03]"/3[N 20s]"/3

[H. J. Schumacher and G. Sprenger, Z. Physik. Chem. A140, 281 (1929);

H2, 267 (1929)]. Recognizing that N 20 s is in equilibrium with N0 2 and

NO. suggest a mechanism to account for the observed rate law. (The
simplest one involves two binary kinetic steps in addition to the N,Os
equilibrium. )
5.13. When an inhibitor acts simultaneously competitively and uncompetitively,
this is known as noncompetitive inhibition. Derive the analog to (5.55) and
(5.56). Is this kinetically distinguishable?
5.14. The invertase-catalyzed hydrolysis of sucrose was studied by measuring the
initial rate of reaction as a function of [sucrose] [A. M. Chase, H. C. v.
Meier, and V. J. Menna, J. Cell Compo Physiol. 59, 1 (1962)]. The following
data were obtained(6.):
[Sucrose] (M):
Initial rate (M sec 1):
Initial rate (M sec- 1 )
in 2M urea:

0.0292
0.182

0.0584
0.265

0.0876
0.311

0.117
0.330

0.146
0.349

0.175
0.372

0.234
0.371

0.083

0.111

0.154

0.182

0.186

0.192

0.188

(a) Determine the apparent Michaelis constants and the saturation rate
for the two sets of experiments {plot (initial rate)-l vs. [sucrose]-l}.
(b) What is the nature of the urea inhibition? Estimate KJ and/or K/.
5.15. In general the reverse reaction

cannot be ignored in the Michaelis-Menton scheme. Show that if this is

included in the mechanism (5.51) the reaction rate is
-d[S]
d[P]
(k,[S]/Km - L1[P]/Km')Eo
= -d-t- = ----:1:-+---=[=-:S]:-:-/K=-=-m----:+-:[=P-::-:]/-=-=K:-m,'----dt

5.16. The fumarase-catalyzed reaction

fumarate

+ water = I-malate

follows the general Michaelis-Menton mechanism of Problem 5.15. At

25C, pH 7, and ionic strength of 0.01 the following parameters were found:

V+/Eo = 1.2 x 10 3 sec-"

V_lEo = 0.93 x 103 sec-"
(69)

= 4.7 X 10-6 M
KM' = 15.9 X 10- 6 M
/('d

After W. J. Moore, Physical Chemistry, 4th ed. (Englewood Cliffs, N. J.: PrenticeHall, 1972), pp. 418-419.

Chap. 5

162

Stationary State Mechanisms

[R. A. Alberty and W. H. Pierce, J. Am. Chem. Soc. 79, 1526 (1957)],
where V+ and V_ are the saturation velocities for the forward and reverse
reactions, respectively. Compute LlGobS and as many of the four rate constants as the data permit. (70)
5.17. Demonstrate (5.58a) and (5.59).
5.18. Consider the possibility that product desorption is significant in catalysis.
The simplest possible mechanism is

Derive the rate law for the disappearance of A and compare with the result
of Problem 5.15. Why do the results differ?
5.19. Consider the mechanism (5.60c) with the additional complication that
there is an inhibitor present which adsorbs on the surface according to the
reaction
In. + S ~ 2In . S
Kr
Derive the analogue to (5.61c). What is the apparent rate law when [In.]
is large?
5.20. In Problem 4.20, three reverse rate laws were shown to be consistent with
the stoichiometry and the forward rate law for the reaction
2Hg(II)

+ 2Fe(II) =

Hg. +2

+ 2Fe(III)

Deduce possible mechanisms to account for each of the reverse rate laws.
5.21. Laser T-jump has been used to study the reaction

[D. H. Turner, G. W. Flynn, N. Sutin, and J. V. Beitz, J. Am. Chem. Soc.

94, 1554 (1972)]; k+ = 6 X 109 M-l sec-l. Would you expect the rate
constant for the reaction

to be similar to, greater than, or less than k+? Explain.

(70)

After F. Daniels and R. A. Alberty, Physical Chemistry, 4th ed. (New York: John
Wiley, 1975), p. 355.

General References

163

5.22. Show that for the mechanism (5.1) under conditions where the steady state
applies, the apparent activation energy for the slow relaxation is significantly temperature dependent if
(a) at some T, k/ 0::::; L but E/
(b) A+' 0::::; A_ and E+' 0::::; E_.

*' E_.

5.23. (a) Show that the ratio of ternary to binary collisions in a gas is Za/ Z2
0::::; d/ A, where A is the mean free path.
(b) Assuming all three-body collisions are reactive obtain a numerical
upper bound for ka in a gas.
(c) At what pressure does termolecular reaction compete with bimolecular
reaction, assuming in both cases that reaction occurs with each collision?
To what density does this correspond?

General References
Gas-Phase Reactions
I. Amdur and G. G. Hammes, Chemical Kinetics (New York: McGraw-Hili, 1966), Chapter 3.
S. W. Benson, The Foundations of Chemical Kinetics (New York: McGraw-HilI, 1960),
pp. 252-264, 290--305, 308-312, 319-424.
J. Nicholas, Chemical Kinetics (New York: Halstead, 1976), Chapters 6 and 7.
A. A. Westen berg, Ann. Rev. Phys. Chem. 24, 77 (1973).

Solution Reactions

s.

W. Benson, The Foundations of Chemical Kinetics (New York: McGraw-Hili, 1960),

Chapter 14 and 15.
K. J. Laidler, Chemical Kinetics, 2nd ed. (New York: McGraw-Hili, 1965), Chapter 10.
R. G. Wilkins, Kinetics and Mechanism of Reactions of Transition Metal Complexes
(Boston: Allyn and Bacon, 1974), Chapter 2.

Catalysis
I. Amdur and G. G. Hammes, Chemical Kinelics (New York: McGraw-Hili, 1966),
Chapter 7.
K. J. Laidler, Chemical Kinetics, 2nd ed. (New York: McGraw-Hili, 1965), pp. 256-286,
Chapter 9.
P. B. Weisz, Ann. Rev. Phys. Chem. 21, 175 (1970).
J. T. Wong, Kinetics of Enzyme Mechanisms (London: Academic Press, 1975).

Other Topics
Crystallization Kinetics:
M. Kahlweit, Ann. Rev. Phys. Chem. 27, 59 (1976).

Kinetics of Heterogeneous Reactions:

M. E. Wadsworth, Ann. Rev. Phys. Chem. 23, 355 (1972).

6
Photochemistry
6.1. Introduction
In ordinary kinetic studies the required activation energy is introduced
via intermolecular collisions. When the activation barrier is large, reaction
must be run at high temperature or progress is imperceptible. Moreover,
thermal activation is quite undiscriminating. Energy is partitioned among
the various internal degrees of freedom of the reactants according to a
Boltzmann distribution. There is no procedure to concentrate the energy
of the reagents in the vibrational degree of freedom which most favors
reaction. Furthermore, thermal processes always proceed by low-energy
pathways. It is thus not possible, by heating, to select another mode of
reaction requiring vastly greater activation energy.
Photochemical activation is a powerful kinetic tool. A large amount
of energy is pumped into a molecule when it undergoes an electronic transition. Photoexcited molecules are far from equilibrium and their excess
energy is lost as the system tries to equilibrate. Such molecules are exceptionally reactive. Irradiation may accelerate reaction; it may permit reaction
by one (or many) new pathways.
A problem with ordinary photochemical activation is that too much
energy is absorbed. As a consequence the possibilities for reaction are
usually greatly augmented but there are often so many photochemical
pathways that this approach can be quite unselective. A new method, based
on vibrational excitation of specific reagent normal modes, gives promise
of allowing the excitation energy to be introduced in such a way that
enhances selectivity. By tuning a CO 2 laser to a specific absorption frequency
of the molecule being studied the vibrational energy might be introduced
in a way that favors a particular reactive pathway.
165

Chap. 6

166

Photochemistry

6.2. Measurement of Intensity

The fundamental photochemical variables are the wavelength A and
the intensity Io(A) of the incident light. According to the Beer-Lambert
law, light is attenuated as it passes through a sample. The intensity a
distance x into the sample cell is
(6.1)

where
(6.2)
[Yd is the concentration of species i and ai(A) is the molar absorption
coefficient of that species. The decrease in intensity for a cell of path length
Lis IaCA) = Io(A) - I(A; L) and the attenuation attributable to a particular
species j is
(6.3)

An ordinary spectrophotometer measures the relative attenuation of

the beam I(A; L)/Io(A), not the absolute intensities of incident and transmitted light. However, in photochemical applications absolute intensity is
important since the fundamental quantity of interest is the quantum yield CP;.,

CP;.

rate of primary product formation

(rate of absorption of light quanta of wavelength A)
by the species of interest

(6.4)

Since the quantum flux J;. is proportional to the intensity J;. = I(A)/(hco/A),
where Co is the speed of light, the absorption rate per unit volume, Ra(A),
for a beam of area A passing through a cell of volume V is

R (A)
a

~ _A_
V

hco

[1 (A) - I(A; L)] =

0

AAIa(A)
Vhc o

(6.5)

Accurate measurement of the output of a photochemical source may be

particularly difficult since I(A; L) is often much less than 10 , Reliable
measurements are made using chemical actinometers, systems whose photochemical conversion rates are accurately known as a function of wavelength. In the range 200 nm < A < 550 nm the ferrioxalate actinometer
is an accurate quantum counter. (1) Irradiation of a solution of potassium
(1)

For a discussion of actinometry see J. N. Demas, in Creation and Detection of the Excited
State, W. R. Ware, ed. (New York: Dekker, 1976), Vol. 4, pp. 6-31.

Sec. 6.3

Spectroscopic Review

167

ferrioxalate leads to the reduction

Fe(III) -

hv

Fe(II)

and the evolution of CO 2 The frequency-dependent quantum yield of

Fe(lI) is well known so that analysis for Fe(II), which can be done with
great accuracy even at low [Fe(II)], determines the rate of absorption of
light quanta from (6.4). Then, from (6.5), laC).) may be calculated and,
since I(A; L)/lo(A) is known, one can determine the light intensity incident
on the actinometer.

6.3. Spectroscopic Review

In conventional photokinetic studies, radiation effects an electronic
transition from the ground state of the molecule to some electronically
excited state. Coupled with the electronic transition there are changes in
the vibrational and rotational state of the molecule. For a transition to
occur the Bohr frequency condition, LfE" = hv = hCo/A, must be obeyed.
Absorption at a specific wavelength is determined by the molar absorption
coefficient which, for photochemically interesting problems, is proportional
to the square of the electric dipole transition moment
(6.6)
"Pi and "Pf are the wave functions for the initial and final states, respectively,

and M is the electric dipole moment operator. (2) The magnitude of the
transition moment is thus a function of the initial and final states.
There are selection rules which determine whether the transition
moment is nonzero. (3) In most photochemical applications three features
of the wave functions are of major importance. The first is spin multiplicity.
As long as the molecule to be photo excited is composed of lighter atoms
(atomic number ;S 36) spin conservation in a transition is a good approxsinglet or triplet
triplet transitions are intense
imation. Only singlet
enough to be accessible photochemically; excitation is between states in
L--'>.

(2)

(3)

L--'>.

For a precise determination of the relationship between a(Je) and the transition moment see W. Kauzmann, Quantum Chemistry (New York: Academic Press, 1957),
pp. 577-583, 645-646.
For an introductory treatment see I. N. Levine, Molecular Spectroscopy (New York:
John Wiley, 1975), Chapter 7.

Chap. 6

168

Photochemistry

the singlet (or triplet) manifold. (4) In molecules containing heavier atoms
spin conservation is no longer rigorous; it is less reliable as atomic number
increases, breaking down totally for Z <: 54.
In addition to spin, the transition moment is strongly influenced by
the geometric symmetry of the initial and final states. In molecules of high
symmetry, e.g., benzene, quinone, oxygen, relatively few electronic transitions are allowed, even within a spin manifold. A particularly important
restriction affects molecules with a center of symmetry. In these molecules
the electronic wave function either changes sign (is odd) or remains unaltered (is even) when the positions of the nuclei are inverted through the
center of symmetry. Only odd "-" even transitions are permitted.
While the occurrence of a photochemical excitation is fundamentally
governed by electronic considerations, the shape of the absorption band
and the nature of the primary photoproduct is determined by the vibrational
part of the wave functions. (5) According to the Franck-Condon principle
electronic transitions take place without significant changes in nuclear structure. A semiclassical justification follows from the fact that electrons are
very light compared with nuclei. The electron distribution may thus be
altered in an interval that is short compared to the time required for significant vibrational motion. (6)
In photochemical studies photoproducts are often characterized by
their emission spectra. The same qualitative considerations govern emission
and absorption; however, since emission does not require an energy input,
it may occur spontaneously. The larger the population of excited-state
molecules the greater the emission intensity. There are usually many ways
to deexcite a species A* ; each radiative process may be described in chemical
terms as

A*~A+hv
where k is a wavelength-dependent rate constant. The analog to (6.5) for
emission is the rate law
RiA) = k(A)[A*] = T-l(A)[A*}
(4)

(5)

(6)

(6.7)

If, as in the case of NO, the ground state is a doublet, excitation is within the doublet
manifold.
Sometimes electronically forbidden transitions are vibronically allowed, i.e., the coupling
of electronic and vibrational motion relaxes a symmetry constraint. Such transitions
are never intense. In no case does vibronic coupling affect spin considerations.
For an in-depth discussion of the Franck-Condon principle see G. Herzberg, Spectra
of Diatomic Molecules, 2nd ed. (Princeton, N. J.: Van Nostrand, 1950), pp. 194-203,
420-432.

Sec. 6.4

Primary Processes

169

T(A) is known as the intrinsic lifetime for emission. When the transition
dipole moment is large emission is rapid and T is short; for strongly allowed
transitions T is in the range I fLsec to I nsec. As the transition moment
decreases T becomes correspondingly larger until, for spin-forbidden transitions, intrinsic lifetimes of the order of seconds are not uncommon. Formulas have been developed that permit T to be calculated if the transition
is also observed in absorption. (7) If a transition is sharp, its intrinsic lifetime
is determined by the extinction coefficient and the line width. If transition
is from a broad band, as is common in photochemistry, it is no longer
k(A) that is significant but a rate constant found by integrating over the
whole absorption band. The extinction coefficient and band shape determine
the corresponding T. The Strickler-Berg result, valid for broad bands, is

where

Here n is the index of refraction of the medium, v the frequency, gl and gu

the degeneracies of the lower and upper electronic states, E(V) the extinction
coefficient, and l(v) the normalized fluorescence intensity.

6.4. Primary Processes

Photoexcitation adds a substantial amount of energy to a molecule;
in its decay to stable photoproducts all the excess energy must be accounted
for. The distinction between primary and secondary relaxation processes
is a bit arbitrary. Primary processes involve deexcitation or dissociation of
the excited molecule. Secondary processes are the chemical reactions of the
excited molecule or of its dissociation fragments. A simplified energy-level
scheme (Jablonski diagram) illustrating excitation and some of the primary
pathways for energy release is given in Fig. 6.1.

6.4.1. Vibrational Relaxation

Because of Franck-Condon requirements photoexcited molecules are
usually produced in a distribution of vibrational states. Collisions induce
(7)

S. J. Strickler and R. A. Berg, J. Chern. Phys. 37, 814 (1962).

Chap. 6

170

Photochemistry

T;
T2
~.
J

TJ

4
2

S0
So
Fig. 6.1. A simplified Jablonski diagram illustrating some pathways for energy release.
The ground state is a singlet, So. Excited states are either singlet, Si' or triplet, Ti
(i ~ 1). Vibrational excitation is indicated by a superscript v, i.e., So". Radiative processes
are denoted by straight lines, radiationless ones by wavy lines. Those illustrated are:
(1) absorption; (2) resonance radiation; (3) fluorescence; (4) phosphorescence; (5) vibrational relaxation; (6) intersystem crossing; (7) internal conversion followed by vibrational relaxation.

vibrational-translational (V- T) and vibrational-vibrational (V-V) energy

transfer, leading, if other processes do not occur, to a Maxwellian distribution of energies among the vibrational levels of the excited electronic state.
The V- T process is most inefficient, particularly if the excited molecule is
small, because of the large changes in translational state that are required.
The probability per collision for the process
CO(v= 1)

+ Ar ----' CO(v=O) + Ar

is < 1.8 X 10-8 at room temperature. (8) However, V-V energy transfer
occurs much more easily, as few as 10-1000 collisions being required in
near-resonant processes ;(9) the more complex the molecules the more
readily thermalization occurs by either pathway.
As a result, in solution where the collision frequency is "-' 1013 sec- 1 ,
(8)
(9)

W. H. Green and J. K. Hancock, J. Chem. Phys. 59, 4326 (1973).

C. B. Moore, Adv. Chern. Phys. 23, 41 (1973). The closer the vibrational changes are to
being isoenergetic, the less the translational states must change. At resonance V- T
transfer is not required.

Sec. 6.4

Primary Processes

171

excited molecules are vibrationally thermalized within their various electronic states before any other process occurs. In the gas phase large
molecules provide their own heat bath and a redistribution of vibrational
energy occurs after vibronic excitation. However, for small molecules at
low pressures collisions are infrequent; the initial vibrational energy distribution does not change unless the intrinsic lifetime is long.
6.4.2. Fluorescence and Phosphorescence

The simplest way to release the excess energy is for the molecule to
re-emit light. In resonance radiation the emitted and absorbed quanta
have the same frequency; the molecule returns directly to its initial state.
Except in dilute gases this is uncommon. More often there has been vibrational relaxation. (10) If both ground and excited state are within the same
spin manifold the radiative lifetime is short, 1 nsec ;:S T ;:S 1 f1-sec; the
emission process is termed fluorescence. When the transition required for
deexcitation does not conserve spin T is much larger, 1 msec ;:S T ;:S 10 sec;
the process is phosphorescence. All three radiative processes are illustrated
in Fig. 6.1.
In many photochemical problems the reactions of a photoexcited
species are studied. As we shall see, in order to characterize the kinetics
the measured lifetime in the absence of reaction, To, must be known. To
establish To one measures the quantum yield for fluorescence, ((Ii A),
defined as
(6.8)

where Rf is the total rate of fluorescence emission throughout the spectrum

and Ra(A) is the absorption rate at the irradiation wavelength, (6.5). Even
under the most favorable circumstances ({If(A) is rarely unity. From Fig. 6.1
it is clear that excitation may be to any vibronic state. However, due to
radiationless relaxation processes to be discussed later fluorescence (phosphorescence) from a state other than Sl(T1 ) is rarely seen. Assuming the
kinetic pathways for deexcitation of an excited molecule A * can be represented as
(6.9a)
(all other paths)
(10)

This process is often called internal conversion. However, we shall limit use of the latter
term to describe radiationless singlet-singlet electronic deexcitation.

Chap. 6

172

Photochemistry

Table 6.1. Fluorescence Quantum Efficiencies of Selected Molecules

Molecule

Excitation
wavelength (nm)

'Pf

Process reducing 'Pf

Benzene (p < 1.3 X 10- 3 atm)

-350

0.39 a

Intersystem crossing

NO. (gas phase)

<398

O~

Predissociation

Anthracene (benzene solution)

-350

0.26c

Intersystem crossing

a E. M. Anderson and G. B. Kistiakowsky, J. Chern. Phys. 48, 4787 (1968); A. E. Douglas and
C. W. Mathews, J. Chern. Phys. 48, 4788 (1968).
~ R. G. W. Norrish, J. Chern. Soc. 1929, 1611.
C Quoted in W. C. Gardiner, Rates and Mechanisms a/Chemical Reaction (New York: Benjamin,
1969), p. 262.

while the excitation kinetics are

A(So)

rapid vib.
+ hVa~ A*(S1V)--~,
A*(S)
relaxation
1
ka

(6.9b)

the steady-state population of A *(S1) is

(6.10)

If there is little attenuation of the exciting radiation, the absorption rate

Ra = ka[A] so that with (6.7) and (6.8) the fluorescence quantum efficiency is
(6.11)

Since the transition probabilities for spontaneous emission and stimulated

absorption are proportional(ll) and since kil) is related to the absorption
coefficient y(l),!12) the absorption spectrum can be used to establish kf .
The fluorescence quantum yield then determines k'. The measured lifetime
in the particular experiment is, using the definition (6.7),
(6.12)

Some quantum efficiencies are given in Table 6.1. Note that even at low
pressures molecules may be deexcited without fluorescing.
(11)
(12)

W. Kauzmann, Quantum Chemistry (New York: Academic Press, 1957), pp. 647-650.
J. P. Simons, Photochemistry and Spectroscopy (London: Wiley-Interscience, 1971),
p.174.

Sec. 6.4

Primary Processes

173

6.4.3. Radiationless Conversion

From Fig. 6.1 it is apparent that molecules in different vibronic states
may have the same energy. For example, the ground vibrational state of
S1 is nearly isoenergetic with a high vibrational state (possibly in the
continuum) of T 1. Similarly S2(V=O) is isoenergetic with S1 v (v large);
again, the vibronic state of S1 may be in the continuum. Since energy is
conserved, it is possible for excited molecules to undergo nonradiative
transitions such as
S1(V=O) -" T1V(V large)
(6.13)
S2(V=O) -" S1V(V large)
The first, involving a spin change, is called intersystem crossing; the second,
conserving spin, is internal conversion. Since vibrational relaxation from
highly excited vibrational states is relatively efficient (the separation of
adjacent levels being small), collisional deactivation rapidly makes the
reverse processes nonresonant. Both processes of (6.13) compete with
fluorescence and reduce CPt.
The considerations governing the rate of internal conversion and
that of intersystem crossing are similar. (13) These rates are proportional
to the density of vibronic levels in the final state etCE) and the square of
the interaction matrix element that couples the two vibronic states, V,
(6.14)
Here V is computed as

where Ii is the complete Hamiltonian of the system and 'lfJi and 'lfJt are the
wave functions of the initial and final vibronic states. (14) In polyatomic
molecules etCE) may be extremely large since there are many possible
(13)
(14)

M. Bixon and J. Jortner, J. Chern. Phys. 48, 715 (1968).

Radiationless transitions occur even in dilute gases indicating that such processes are
possible in isolated molecules. If a molecule were excited to a stationary state of the Hamiltonian the matrix element V would, of necessity, be zero. However, excitation at a
precise energy is not possible unless exceptionally monochromatic radiation is used (as
could be the case with laser excitation). Hence, the energetic distinction between nearly
degenerate vibronic states is tost if the line width of the exciting radiation is sufficiently large. As a result transition may be to a nonstationary state. See A. H. Zewail,
T. E. Orlowski, and K. E. Jones, Proc. Nat. Acad. Sci. U.S.A. 74, 1310 (1977).

174

Chap. 6

Photochemistry

final state

Fig. 6.2. Radiationless conversion in a molecule with a high density of vibronic states.
As (!j ~ (!i, k+ ~ k_. See text for estimates of (!j as a function of the number of atoms
in a molecule.

vibrational modes. A prototype calculation assuming conversion to a

vibronic state with 96 kJ mol- 1 vibrational energy where the mean vibrational separation is 12 kJ mol- 1 was carried out for molecules with different
numbers of atoms. Varying N from 3 to 4 to 5 to 6 to 10 increases rAE)
by the factors of 1 to 70 to 800 to 7000 to 7 X 10 6 (15)
An energy level diagram for radiationless conversion is illustrated
in Fig. 6.2. Since eiE) increases and V decreases with increasing E, the
dependence of the rate constant k on the separation energy L1E is not immediately apparent. It is usually found that transition is more likely when
L1E is small. A case in point is intersystem crossing in anthracene for which
the energy level diagram is illustrated in Fig. 6.3. The transition 8 1 ----' T 2v
is favored over the transition 8 1 ----' T 1v even though e(E) for the latter case
is much larger. (16)

6.4.4. Dissociation
Figure 6.4 illustrates for a diatomic molecule, the possible excitedstate potential energy curves that permit dissociation. As mentioned
previously (the Franck-Condon principle) electronic transitions occur so
rapidly that the nuclei are, in effect, clamped during transitionY7) Quantum
mechanics requires that the transition occur between states which have
large probability amplitudes at the same internuclear separation. The
vibrational wave function of the ground state is maximum at the center of
the well; those of excited states have major maxima near the classical
(15)
(16)

(17)

M. Bixon and J. Jortner, J. Chern. Phys. 48, 715 (1968).

As quoted in J. P. Simons, Photochemistry and Spectroscopy (London: Wiley-Interscience, 1971), p. 185.
If there are nearly degenerate vibronic states, the concept of a well-defined nuclear
configuration is not precise. This, as already mentioned, is important in understanding
radiationless pathways in isolated molecules. For an analysis of the applicability of
the idea of "molecular structure," see R. G. Woolley, J. Am. Chern. Soc. 100, 1073
(1978).

Sec. 6.4

Primary Processes

175

---r---

T2

--r--+--- 1j

320 kJ mol-!

178 kJ mor!

312 kJ mol-!

so--L-----------~~~---

Fig. 6.3. Electronic energy le~'els in anthracene. The radiationless conversion rate SI ~ T2
is much larger than that for SI ~ Tl reflecting the effect of energy differences.

(b)

(c)
Fig. 6.4. Ground- and excited-state potential energy curves illustrating possible dissociation pathways. For a diatomic molecule r is the internuclear distance; in a polyatomic
molecule r is a normal mode displacement. Vibrational separations are greatly exaggerated.
The vertical arrows correspond to Franck-Condon allowed transitions. Where possible,
molecular examples are indicated. (a) Repulsive excited state-immediate dissociation.
(b) Crossing of bound and repulsive excited states-intersystem crossing leads to predissociation. (c) Metastable excited state-tunneling leads to dissociation. (d) Bound stateexcitation energy greater than dissociation limit.

Chap. 6

176

Photochemistry

Table 6.2. Some Photodissociation Reactions

+0
CHBra ~ CHBr. + Br
H,O ~ HO + H
Cyclobutanone ~ CO + Cyclopropane
H 0. ~ OH + OH
7 CH. + H. (preferred)
CH.
.
'" CHa + H
Octatetraene ~ Benzene + C.H.
NO,

NO

turning points. (18) To accommodate the various constraints the excitation

arrows in Fig. 6.4 are drawn at the equilibrium internuclear separation of
the ground state; they extend only to the edge of an upper potential curve,
the point which corresponds to the classical turning point.
If excitation is to a repulsive state (Fig. 6.4a), dissociation is immediate,
occurring within one vibrational period, t;S 10-13 sec. Another rapid
process involves predissociation (Fig. 6.4b). Here the molecule is excited
to a bound state which, because there is a resonant continuum level of the
repulsive electronic state, undergoes radiationless conversion followed by
immediate dissociation. Excitation could also be to a metastable vibronic
state (Fig. 6.4c) which dissociates via tunneling. The dissociation probability would be very sensitive to the height and the width of the barrier
to be overcome, and the decomposition rate greatly enhanced by vibrational
excitation in the excited state. Finally, as in Fig. 6.4d, excitation may be
to a bonding state but into a continuum level above the dissociation limit;
decomposition is again immediate. (19)
Figure 6.4 may be transcribed for polyatomic molecules. A similar
diagram exists for each vibrational mode where a normal mode displacement
corresponds to the interparticle distance. The situation is complicated
further because the vibrational energy is rapidly partitioned among many
(18)

110)

For a discussion of the properties of the nuclear wave functions see G. Herzberg, Spectra of Diatomic Molecules, 2nd ed. (Princeton, N. J.: Van Nostrand, 1950), pp. 76-79,
93-94.
Dissociation and predissociation in diatomic molecules are discussed in detail in G.
Herzberg, Spectra of Diatomic Molecules, 2nd ed. (Princeton, N.J.: Van Nostrand,
1950), pp. 387-434.

Sec. 6.4

177

Primary Processes

Fig. 6.5. Schematic diagram indicating a

method whereby photoexcitation may lead
to isomerization. Transitions are assumed to
obey the Franck-Condon principle. Q is a
normal mode which distorts molecule from
one isomer to the others, e.g., rotation about
C=C in an olefin. Vibrational separations are
much exaggerated.

vibrational modes. Polyatomic illustrations of Figs. 6.4c and d have not

yet been convincingly established. (20)
Photodissociation is a common primary process. Some examples are
given in Table 6.2. The photoproducts may be characterized in many ways.
Stable species are observed by chemical analysis. Excited ones are identified
spectroscopically and reactive ones by their participation in secondary
reactions.

6.4.5. Isomerization
The relaxation mechanisms that lead to photoinduced isomerizations
such as
cis-olefin + hv -' -' trans-olefin (numerous steps)

r:r + h,~~ @

(num"ou, "'p')

involve vibrational relaxation of the excited state sometimes preceded by

radiationless conversion. In the ground electronic state the isomers are
separate vibrational minima where the vibrational barrier to interconversion
is large. Excitation often leads, either directly or indirectly, to a state with
a different equilibrium geometry, as schematically shown in Fig. 6.5. If
the molecule fluoresces isomerization is assured because of Franck-Condon
requirements. If it is stabilized by radiationless conversion, the consequent
vibrational relaxation greatly favors the formation of the more stable
isotope.
(20)

For a discussion of dissociation and predissociation in polyatomic molecules, see G.

Herzberg, Electronic Spectra of Polyatomic Molecules (Princeton, N.J.: Van Nostrand,
1966), pp. 445-484.

Chap. 6

178

Photochemistry

6.5. Secondary Processes

6.5.1. Quenching Measurements
When a photoexcited molecule undergoes chemical reaction the
fluorescence quantum efficiency is reduced, i.e., it is quenched. Fluorescence
quenching provides an excellent example of how competition between
chemical and physical processes may be used to establish both rate law
and mechanism. The reaction of photoexcited acridine (A *) with amines
in aqueous solution has been thoroughly studied. (21) Addition of amines
reduces the fluorescence efficiency. Data for an analogous system are
presented as a Stern-Volmer plot in Fig. 6.6. The relative fluorescence
efficiency, cpAc=O)/cptCc), is equal to the ratio of fluorescence intensities
without and with quenchers IAc=O)/IAc). If this latter quantity is graphed
against amine concentration, c, it appears that
(6.15)
a result that is readily interpreted if we assume that A * reacts with amine
A * + amine ~ products
k.

(6.16)

Modifying the arguments which led to (6.11) to account for the chemical

process yields
(6.17)
or with (6.12)
CPj(O)/cpAc)

(kj

+ k' + k2C)/(kj + k') =

+ k2ToC

(6.18)

precisely the Stern-Volmer form (6.15). It is now obvious why determination of To is important. Only if it is known can rate constants for reaction
be established; otherwise only relative rate constants are accessible. For
aqueous acridine To is 13.5 nsec and values for k2 may be calculated. (22)
There are two significant limitations in using competition experiments,
of which fluorescence quenching is representative, to establish rate laws
and mechanisms. The methods monitor the depletion of a single species,
in this case photoexcited acridine. As such they provide information about
the first step in a reaction and, by themselves, do not characterize the
1'1)
122)

A. Weller, Prog. React. Kinetics 1, 187 (1961).

A. Weller, Prog. React. Kinetics 1, 187 (1961). Absolute lifetimes were determined using
the method due to A. Schmillen, Z. Phys. 135, 294 (1953).

Sec. 6.5

Secondary Processes

179

5,------.-----,,-----,------,
4

0.02

0.04
c(M)

0.06

0.08

Fig. 6.6. Stern-Volmer plot of quantum efficiency in quenching of anthracene fluorescence

by N-N diethylaniline in toluene solvent [H. Knibbe, Ph. D. Thesis (Free University of
Amsterdam: 1969), p. 44].

products or whether further chemical transformation takes place. (23) In

addition the rate of the competition reaction must be comparable to the
rate of the physical process. For example, if k2 < 7 X 105 M-l sec-I, the
quenching ratio (6.18) would not drop to 0.95 unless c ~ 5 M. Thus,
using excited acridine as a probe, it is impossible to study processes with
k z ;S 10 6 M-l sec- 1 since enormous quencher concentrations are required.
The converse may also prove to be a problem, i.e., if a reaction is too rapid
quenching may be essentially total at negligible quencher concentrations.
However, this is not a limitation in the acridine quenching (see Problem
6.3) experiments.

6.5.2. Sensitization and Excitation Transfer

Many molecules that quench the fluorescence of a primary photoproduct undergo a transition to an excited electronic state. The general
mechanism is, using D for donor and A for acceptor,
D*

+ A -" D + A*

(6.19)

where the D* is the molecule which absorbed the radiation. This process
is extremely useful; photosensitization permits the production of A *
molecules in cases where direct photoexcitation of A is either impossible
(a forbidden transition) or inconvenient. The requirements for radiationless
energy transfer are:
(1) the process is nearly resonaLf (small increments of internal energy
can be accommodated by changes in translational energy);
(23)

The reaction of amines with photoexcited acridine actually follows a multistep pathway, in part diffusion controlled. See A. Weller, Prog. React. Kinetics 1, 187 (1961).

Chap. 6

180

Photochemistry

(2) spin angular momentum is conserved (not necessary if the photosensitizer is a heavy atom).

In sensitization involving atoms and/or diatomic molecules the resonance

condition severely limits the processes which may occur. However, if both
o and A are polyatomic molecules the energy restriction reduces to the
constraint that the electronic energy of 0* be greater than that needed
to form A *; any difference is made up by vibrational and rotational excitation of A *.
The rate of excitation transfer is very dependent on whether donor or
acceptor change spin states. If both molecules remain in their original spin
manifolds, coupling of the transition dipoles leads to very efficient energy
transfer. The long-range dipole-dipole interaction is significant even at
distances as great as 6-7 nm. Molecules need not be in immediate contact
for the reaction

to occur. (24) Diffusion no longer limits the rate constants; k is not viscosity
dependent and, in the most efficient aqueous processes, is ,-...,3 X 1011 M-1
sec-I, many times the ordinary diffusion limit of ,-..., 1010 M -1 sec- 1 (Section
5.7). A quantitative theory for k yields
k ex - I6

J l(v)(v)

1 dv
v

-4

where (v) is the extinction coefficient of the acceptor, l(v) is the normalized
fluorescence of the donor, and R is the distance between donor and acceptor.(25) The R-6 dependence has been directly verified by constructing
systems in which donor and acceptor are substituent groups within the
same molecule, but widely separated by a rigid, nonconjugated system. (26)
Values of R determined from k are in good agreement with those deduced
from measurements on molecular models.
Reactions such as
30*

+ 1A

30*

+ 3A

----'0

10

+ 3A*

----'0

10

+ 1A*

are spin allowed. However, the molecules' electron distributions must

(24)
(25)
(26)

T. Forster, Disc. Faraday Soc. 27, 7 (1959); Naturwiss. 33, 166 (1946).
T. Forster, Disc. Faraday Soc. 27, 7 (1959).
s. A. Latt, H. T. Cheung, and E. R. Blout, J. Am. Chern. Soc. 87, 995 (1965).

Sec. 6.5

Secondary Processes

181

interact strongly before energy transfer is probable. (27) The sensitization

rate is now diffusion limited and thus viscosity dependent; an example
where the limit is reached occurs in glycerol solvent with naphthalene as
donor and l-iodonaphthalene as acceptor. (28)
The photosensitized acceptor may relax by any primary process as
well as by further reaction. Some instances where A * is formed in a weakly
bound, dissociative, or predissociative state are(29)

Hg(T2)
Hg(T2)
Hg(T2)

+ H 2(So) ---' ---' Hg(So) + 2H

+ H 20(So) ---' ---' Hg(So) + H + OH
+ CH CHO ---' ---' Hg(So) + CH + CHO
3

In addition to excitation transfer, the donor and acceptor may associate

and form a stable excited complex, an excimer, which is characterized by a
fluorescence spectrum different from any observable in 0* or A *. The
phenomenon was discovered while studying the fluorescence of P(yrene )PO)
At low concentrations the fluorescence is due to the Sl ---' So transition.
At high concentrations a new, lower-frequency fluorescence develops which
is associated with the reactions

6.5.3. Some Reactions

The photochemical literature is replete with photolysis studies. The
rates are measured via quenching of either fluorescence or phosphorescence.
The photoproducts are characterized either chemically or spectroscopically.
A few examples suffice to suggest the scope of such investigations.

6.5.3.1. Photolysis of Ketene

CH 2CO

+ hv(}. < 473.5 nm) ---' CH CO(Sl

/ CH (Sl or To) + CO (predissociation)
2

V)

CH 2CO(Sl V )
CH 2
(27)
(28)
(29)

(30)

~ CH 2CO

+ CH 2CO ---' C H. + CO

(vibrational relaxation)

D. L. Dexter, J. Chern. Phys. 21, 836 (1953).

A. D. Osborne and G. A. Porter, Proc. Roy. Soc. A284, 9 (1965).
See J. P. Simons, Photochemistry and Spectroscopy (London: Wiley-Interscience, 1971),
pp. 231-247 for details.
T. Forster and K. Kaspar, Z. Elektrochem. 59, 976 (1955).

Chap. 6

182

Photochemistry

In this example no resolvable structure is found in the absorption spectrum

of CH 2CO as long as A < Ao = 473.5 nm, suggesting rapid predissociation.
As the frequency of excitation increases the quantum yield for predissociation, qJdiss' increases; it is unity regardless of pressure for A < A1 = 270 nm.
When A1 < A < Ao , qJdiss is pressure dependent, dropping as p increases.
As p
0 one would expect qJdiss = 1 since the unstructured spectrum indicates predissociation. However, this is not observed: at A = 366 nm,
qJdiss
0.04 as p
O. No totally convincing explanation of the pressure
dependence of qJdiss exists. The net decomposition efficiency is twice qJdiss
since CH 2 reacts further with ketene. (31)
---->.

---->.

---->.

6.5.3.2. Photolysis of N0 2
N0 2 + hv(320 < A < 397.9 nm)
O(To)

+ N0 2

---->.

N0 3 *

...--?

NO

---->. ---->.

NO (Do)

+ 02(V ;S 12)

+ O(To)

~N03

Irradiation produces excited N0 2 which predissociates into doublet NO

and atomic oxygen. The 0 atoms react further with N0 2 either producing
more NO or transient NO a . When N0 2 produced in automobile exhaust
emissions is photolyzed the further course of atmospheric reaction is very
complex. Some of the reactions are(32):

o + O2

0 3 + N0 2

---->.

---->.

0 3,

N0 3
N 20 5

03

+O
+H 0

+ NO

N0 2 + O 2

N0 3 + N0 2 :;;=: N 20 5

2,
2

---->.

---->.

2HN0 3

The unimolecular reactions and the recombination reactions of course

Recent studies indicate that the mechanism of ~etene photodissociation is wavelength
dependent; as the energy of the exciting radiation increases, new primary processes
become energetically allowed [W. L. Hase and P. M. Kelley,f. Chern. Phys. 66, 5093
(1977)].
(32) The kinetics of smog production is complicated, involving products not only of nitrogen
oxide photolysis but also of hydrocarbon oxidation. With a catalog of the elementary
reactions that may be involved in the overall process it is possible, using the numerical
methods mentioned in Section 5.3, to solve the coupled rate equations. Comparison of
calculation and experiment then provides a test of whether the proposed reaction scheme
is plausible. See T. A. Hecht and H. J. Seinfield, Envir. Sci. Tech. 6, 47 (1972). It is worth
reiterating that finding a multistep mechanism which accounts for the data does not
demonstrate uniqueness; it only indicates consistency.

(31)

Sec. 6.5

Secondary Processes

183

require participation of an inert molecule to satisfy energy-transfer constraints.

6.5.3.3. Isomerization of Mesitylene

Irradiating mesitylene with near-ultraviolet light leads to a shift of
methyl groups. Isotopic labeling experiments have shown that the apparent
methyl migration is really the ring rearrangement(33)

hv

thus suggesting as intermediate a form of Dewar benzene

Similar experiments on benzene-I,3,5 d3 showed the same migration pattern.

6.5.3.4. Photolysis of Trapped Species

Gas- or solution-phase photolysis is always complicated by the fact
that the primary photoproducts diffuse and undergo secondary reactions.
To avoid such difficulties the technique of matrix isolation was developed.
Here a dilute mixture of, e.g., CH 4 in N2 is condensed from the vapor phase
onto a transparent window maintained at temperatures of 4--20 K. Then
the CH 4 can be photolyzed and the resultant primary photoproduct,
. CH 3, studied at leisure. (34) A variant of this technique is to use a reactive
matrix. When HBr or HI, condensed in a CO matrix, is photolyzed the
primary process is dissociation of the excited HX. The H atom then reacts
and HCO is observed. (35)

(33)

(34)
(35)

L. Kaplan, K. E. Wilzbach, W. G. Brown, and S. S. Yang, J. Am. Chem. Soc. 87, 675
(1965); K. E. Wilzbach, A. L. Harkness, and L. Kaplan, J. Am. Chem. Soc. 90,
1116 (1968); and L. Kaplan and K. E. Wilzbach, J. Am. Chem. Soc. 90, 3291 (1968).
D. E. Milligan and M. E. Jacox, J. Chem. Phys. 47,5146 (1967).
G. E. Ewing, W. E. Thompson, and G. C. Pimentel, J. Chem. Phys. 32, 927 (1960).

Chap. 6

184

Photochemistry

6.6. Chemiluminescence
Chemiluminescence is, in a general sense, the opposite of photodissociation. Here a molecule is formed in an excited state by means of chemical
reaction. The new species then fluoresces to deexcite. It is the process which
accounts for the characteristic colors of flames. An example is found in the
combustion of CO where atomic oxygen is produced, most likely as the
first step in the oxidation of CO. Then the following sequence of reactions occurs(36):

The final step in the relaxation is emission of chemiluminescent radiation

of wavelength ,,-,310-380 nm.
The chemiluminescence spectrum contains considerable information.
When highly exothermic reactions of the type
A

+ BC --" AB* + C

are studied the product AB* is always formed in a vibrationally excited

state which, in order to be stabilized, must either emit radiation or be
collisionally deactivated. At normal pressures the latter process dominates
since the radiative lifetime for vibrational fluorescence is "-' 10-3 sec. At
low pressure (10-3-lO-5 atm) the inherent inefficiency of V- T transfer
allows observation of the infrared fluorescence of AB* with an internal
energy distribution produced during the bond-forming process. The chemiluminescence spectrum may be analyzed since the vibrational spectrum of
AB is known; then the vibrational and rotational energy distribution in
AB* formed in the collisional process may be deduced. If the exothermicity
of the reaction is known, the fraction which appears as translational,
rotational, or vibrational energy of the products can be estimated. Some
data are given in Table 6.3. The differences are considerable. As we shall
see in Chapter lO product energy distributions may be used to determine
gross features of the ABC interaction potential.

(36)

A. G. Gaydon, The Spectroscopy of Flames (London: Chapman and Hall, 1957), Chapter 6; R. N. Dixon, Proc. Roy. Soc. (London) A27S, 431 (1963).

Sec. 6.7

Orbital Symmetry Correlations

185

Table 6.3. The Fraction 0/ Reaction Exothermicity Appearing in Various Degrees

0/ Freedom a

Exothermicity
(kJ mol-I)

Reaction

Mean fraction of exothermicity

appearing in t(ranslation),
r(otation), and v(ibration)

It

fr

Iv

H + CI.

HCI + CI

203

0.54

0.07

0.39

D + CI.

DCI + CI

208

0.50

0.10

0040

+ Br.

HBr

+ Br

183

0.39

i).05

0.56

F+H.

HF+H

145

0.26

0.07

0.67

F+D2

DF+D

144

0.25

0.06

0.69

CI + HI

HCI + I

142

0.16

0.13

0.71

CI + DI

DCI + I

142

0.15

0.14

0.71

a T. Carrington and J. C. Polanyi, in MTP International Review of Science, Physical Chemistry,

Series 1, Vol. 9, J. C. Polanyi, ed. (London: Butterworths, 1972), pp. 149-150.

6.7. Orbital Symmetry Correlations: Woodward-Hoffman

Rules
Both thermal and photochemical isomerization of disubstituted transbutadienes to disubstituted cyclobutenes are stereospecific; however, the
products are different:

Y\<

(Conrotatory)
X

(Disrotatory)

Similar stereochemical control is found in countless reactions, both unimolecular and bimolecular. The explanation is based upon three assump-

Chap. 6

186

'

.'

TT4

'

TT4

IT*

TT3
~400

TT3

0\

Photochemistry

cyclobutene

trans - butod i ene

r\
t::\
r.\

kJ mol-I

TT2

TT
(J

TTl

ill

m
H

Fig. 6.7. Energy levels and nodal structure of one-electron molecular orbitals (MO's) which
control the trans-butadiene ~ cyc10butene isomerization. The electron density in the various n-MO's is perpendicular to the carbon skeleton. The electron density in the a-MO's
is coplanar with the carbon skeleton. The spacing of energy levels is approximate. The
dashed lines indicate the position of the nodes in the various orbitals.

tions. (1) Reaction proceeds in a concerted, single-step pathway: the a-bond

and the n-bond in cydobutene are formed simultaneously. (2) An adiabatic
principle is obeyed: as the nuclei move the electron distribution adjusts
instantaneously. (3) The activation energies associated with the different
concerted nuclear motions are enormously different: the higher-energy
pathways are absolutely forbidden. (37)
To illustrate the molecular basis for this selectivity consider the molecular orbitals of butadiene and CYclobutene which are involved in the electronic rearrangement. (38) These are illustrated in Fig. 6.7; their relative
energies are nl < n 2 < n3 < n 4 and 0' < n < n* < 0'*. Now consider
disrotatory motion around the C-C bonds (rotation in the opposite
direction) as shown in Fig. 6.S. The two orbitals n 1 and n3 must correlate
with two orbitals of cydobutene. One of these is obviously n since both
nl and n3 are in phase and therefore bonding between C 2 and C3 ; the other
is 0' since in both n 1 and n3 disrotatory motion creates a bonding a-orbital
between C 1 and C 4 Similar arguments can be used to establish the pair(37)

(38)

R. B. Woodward and R. Hoffman, J. Am. Chem. Soc. 87, 395,2046,2511,4388,4389

(1965); H. C. Longuet-Higgins and E. W. Abrahamson, J. Am. Chem. Soc. 87, 2045
(1965).
For details on molecular orbital calculations see F. L. Pilar, Elementary Quantum
Chemistry (New York: McGraw-Hill, 1968), Chapter 18.

Sec. 6.7

Orbital Symmetry Correlations

187

Conrotatory

Disrotatory

2~3

,rf-

-~4

,~

~4

Fig. 6.8. Effect of different concerted rotations about the C1-C, and C 3-C. bonds
on the 71:1-MO of trans-butadiene. Conrotation distorts 71:1 ~ 71:; disrotation distorts
71:1 ~ a. Orbital correlation follows a minimum energy pathway.

wise orbital correlations

(a, :n*)

Conrotatory:

(:n2' :n4)

Disrotatory:

(:n 2, :n4) --->. (:n*, a*)

--->.

The orbital structure changes continuously as rotation occurs and the

energy levels of a correlated pair cannot cross since the orbitals have the
same symmetry properties. (39) Thus the individual orbital correlations are
Conrotatory:
Disrotatory:

Now consider the two isomerizations. In the thermal process transbutadiene is in its ground electronic configuration :n12:n22. Conrotatory
motion yields cyclobutene in the ground electronic configuration a 2:n 2
while disrotatory motion leads to the highly excited configuration a 2 (n*)2;
the activation barrier for disrotatory motion forbids this path. The lowest
photoexcited state of trans-butadiene is n 12n 2:n 3 which, assuming conrotatory
displacements, correlates with the highly excited a:n 2a* state of cyclobutene.
In a disrotatory displacement cyclobutene is formed in its first excited
state a 2:n:n*. The latter requires little or no activation; the former requires
a great deal and is forbidden. The stereochemical discrimination found in
(6.20) is thereby accounted for. (40)
This is a direct consequence of quantum mechanics. Any analysis which suggests that
the ordering of the energy levels of one-electron orbitals of the same symmetry may be
inverted as a molecule distorts is approximate. Important interactions must have been
ignored if such a result is predicted. If these interactions are accounted for the levels
no longer cross. See W. Kauzmann, Quantum Chemistry (New York: Academic Press,
1957), pp. 536-539 for a discussion of a similar problem: the change from ionic to
covalent binding in NaCl as the Na-Cl separation is varied.
(,0) For an extended discussion see R. Hoffman and R. B. Woodward, Science 167, 825 .
(1970).

(39)

188

Chap. 6

Photochemistry

Similar arguments have been developed to explain stereochemical

control in other systems. In each case the fundamentals are the same.
Reaction proceeds via a single-step pathway. The orbital structure adjusts
instantaneously as the nuclei move. The energy levels of correlated pairs
of orbitals do not cross.

6.8. Laser Activation

Ordinary photochemical excitation, initiated by absorption in the
visible or ultraviolet, introduces an enormous amount of internal energy
into a molecule. Radiation at 400 nm corresponds to an energy of 300 kJ
mol-I, enough to activate most reactions. As we have seen, such activation
is rarely discriminating. A range of products is formed depending upon the
relative efficiency of the various primary and secondary processes. The
ideal approach to selective bond breaking would be to pump energy into
the particular vibrational mode associated with the chemical process of
interest. With the development of high-power infrared lasers, it is possible
to imagine accomplishing this task.
The unique features of a laser, its high intensity and monochromaticity,
have led to its application in countless problems of scientific interest. We
shall only consider one. Infrared laser photochemistry is a very new field
(it began to develop in the early 1970s); as such there are numerous unresolved problems as well as ambiguities and contradictions. The following
discussion is designed to suggest the scope of the field, not to be comprehensive. (41)
The most versatile infrared source is the CO 2 laser; it can be tuned to
emit light at about 100 discrete wavelengths between 9.2 and 11.0 [Lm
C. . ., 12 kJ mol-I). (42) If used to provide a continuous source of photons the
power output is ,.....,,100 W cm- 2 However, if properly focused, a CO 2
laser can emit pulses of nanosecond duration with a power flux of up to
100 MW cm -2; at these wavelengths the pulse contains ,....." 1019 photons
cm- 2, which corresponds to a flux of ,......-10 28 cm- 2 sec-I! While the pulse
is on, the photon density in the sample is ,....." 1017 cm -3, which is comparable
to molecular densities when p ,....." 0.01 atm and T,....." 300 K. In addition to
(H)

(42)

It is based on two state-of-the-art reviews: A. L. Robinson, Science 193, 1230 (1976);

194, 45 (1976).
For a discussion of laser sources see S. R. Leone and C. B. Moore, in Chemical and Biochemical Applications of Lasers, C. B. Moore, ed. (New York: Academic Press, 1974),
Vol. 1, Chapter 1.

Sec. 6.8

Laser Activation

quasi - continuum

y= 4

y=3

--

y= 2
y=1
y= 0

189

--

------

In SF6 mean spacing becomes

so small that absorption at
nearly all frequencies
becomes possible

(2,J-I) -

(3,J)

t. = h"'23+ 2Jr

(I,J-I) -

(2,J-I)

t. = hW ,2

(O,J) -

(I,J-I)

t. = hw01 - 2J r

Fig. 6.9. Vibrational-rotational energy levels for a polyatomic molecule. Resonance is

established by change in rotational level to compensate for vibrational anharmonicity. Rotational separation is exaggerated.

providing enormous photon fluxes the laser, because of its very narrow
line widths, provides highly selective excitation. A properly tuned CO 2
laser has been used to selectively excite either IOBCl 3 or 11 BCI 3 (43) At
present only two other regions of the infrared spectrum are accessible with
high-power lasers-the CO and HF lasers which emit in the vicinity of
5 [Lm ("-'24 kJ mol-I) and 2.7 [Lm ("-'45 kJ mol-I), respectively.
In laser-induced gas-phase reactions the first step is absorption of a
vibrational quantum which, at moderate pressures, would only heat the
gas since V - T energy transfer occurs. If this process were dominant the
laser would be no more than a fancy (and expensive) Bunsen burner.
At low pressures conditions are more favorable for multiple-photon absorption. The photon flux is large enough so that more than one quantum
can be absorbed before significant V - T transfer occurs. The only trick
is to maintain resonance between the laser frequency and the various transitions to be excited. If only pure vibrational transitions were involved
anharmonicity would ensure that a laser tuned to the fundamental frequency
(VO->I) would be nonresonant for other transitions. However, since there
are also changes in rotational quantum number resonance may be reestablished. The resonance conditions are illustrated in Fig. 6.9.
An example of multiple-photon absorption is found in the laserinduced decomposition of D 3 BPF 3 irradiated with a continuous source
(43)

R. V. Ambartzumian, N. V. Chekalin, V. S. Doljikov, V. S. Letokhov, and E. A. Ryaboy, Chem. Phys. Lett. 25, 515 (1974).

Chap. 6

190

Photochemistry

producing "-' 150 W cm -2 at a photon energy of 11.7 kJ mol-I. At pressures

between 2 x 10- 6 and 7 x 10- 6 atm the room-temperature decomposition

Ea

123 kJ mol- I

appears to proceed via sequential absorption of three photons. (44) Instead

of requiring the collisional activation energy 123 kJ mol-I, the apparent
collisional activation energy in the irradiated sample was reduced to 14.5
kJ mol-I. While the precise meaning of activation energy for nonequilibrium
energy distributions is far from obvious, it does appear that laser activation
leads to far more efficient energizing of a molecule.
One important conclusion emerges. A small amount of vibrational
excitation may be exceptionally effective in enhancing reactivity. Under
thermal conditions '"" 120 kJ mol- I are needed if collision is to induce
reaction. This process, which might be termed T-v transfer, is neither
efficient nor selective. Even if molecules have a high collision energy there
is no reason, as we shall see in Chapter 10, for it to be readily converted
into internal energy. Furthermore, the internal energy absorbed need not
go into vibrational modes favoring B-P bond weakening. In the laserinduced reaction the energy apparently excites P-F vibration and then is
redistributed among modes with similar frequency weakening the P-B
bond; a small amount of collisional energy transfer then completes the
job. Thus relatively little energy, introduced selectively, is much more
effective than a thermal sledgehammer. (45)
At higher pressures and in liquids collisions are frequent. Energy deposited in a single vibrational mode does not remain localized. Collisional
processes first redistribute energy within the molecule and then convert it
into relative translational energy. By using ultrahigh-power lasers the
absorbing Ipolecules can be multiply excited before collision is probable.
In addition to lowering the overall activation barrier for reaction the
multiple-photon process may be used to excite molecules to selected vibrational states which then react to produce specific products. An example
is the laser-driven reaction

(44)

(4')

K-R. Chien and S. H. Bauer, J. Phys. Chern. 80,1405 (1976); E. R. Lory, S. H. Bauer,
and T. Manuccia, J. Phys. Chern. 79, 545 (1975).
For more recent work, emphasizing the effectiveness of laser activation in the reaction
NO + 0 3 ~ products, see K.-W. Hui and T. A. Cool, J. Chern. Phys. 68, 1022 (1977)
and references cited therein.

Sec. 6.8

Laser Activation

191

The thermal reaction yields B2H6 as well. (46) A related feature of laser
photoreactivity is the effect of varying wavelength. In the series of reactions

+ HBr =
B(CHa)2Br + HBr =
BCH 3Br 2 + HBr =
B(CHa)a

+ CH 4
BCHaBr2 + CH 4
BBr3 + CH
B(CHa)2Br

the first and third are accelerated by tuning the laser to A = 10.30 [Lm
while the rate of the last two is increased when A = 9.62 [Lm. (47)
Inasmuch as laser photochemistry is a relatively unexplored area of
research, it abounds with ambiguities and contradictions. Two instances
suffice. The decomposition of CF 2ClCF 2Cl was studied using a low-power
("'-'25 W cm- 2) laser at high pressure (",-,0.4 atm).(48) The decomposition
rate was very dependent on the irradiation wavelength; it was 160 times
faster when A = 10.86 [Lm than when A = 9.51 [Lm. Since both wavelengths
are strongly absorbed and since, under the experimental conditions, there
are",-, 104 collisions between each absorption event, V-T transfer should
lead to rapid laser heating and loss of selectivity which was not observed.
Speculation surrounds possible v-v transfer which may further excite
one molecule up the vibrational ladder. Frequency matching may be better
at the longer wavelength. However, there is presently no clear resolution of
the apparent contradictions.
Large-scale utilization of laser techniques would seem to require high
pressures (",-,0.1 atm) and high intensity (",-,10 6 W cm- 2). However, under
such conditions one important feature of laser photochemistry would appear
to be easily lost-selective vibrational excitation. While thermalization
via V-T transfer is slow, collisions still affect atomic motions and work to
redistribute energy localized in one mode among the molecules' other
vibrational degrees of freedom. In CH 3 F, "'-' 100 collisions are required to
complete the energy redistribution within the molecule; however, energy
transfer between nearly resonant modes requires ::S 10 collisions. (49) The
consequence for high-pressure laser activation is apparent; at p "'-' 0.1 atm
efficient energy redistribution should require < 50 nsec. However, studies
146)
147)

148)
149)

S. D. Rockwood and J. W. Hudson, Chern. Phys. Lett. 34, 542 (1975).

H. R. Bachmann, H. N6th, R. Rinck, and K. L. Kompa, Chern. Phys. Lett. 33, 261
(1975).
R. N. Zitter and D. F. Koster, J. Arn. Chern. Soc. 98, 1613 (1976).
E. Weitz and G. W. Flynn, J. Chern. Phys. 58, 2679 (1973).

192

Chap. 6

Photochemistry

of the decomposition of CCl 3F and CCIF 3 suggest otherwise. (50) In experiments using a SOO-nsec pulse the vibrational energy seems neither to be
thermalized by V- T transfer nor to be collisionally redistributed within
the molecules. Instead it seems to remain in the vibrational mode that was
excited initially. Again no clear resolution of the contradictory evidence
has been presented.

6.9. Laser-Controlled Isotope Separation

As mentioned in the previous section a CO 2 laser can be tuned to selectively excite the isotopically different species lOBC1 3 or llBCI 3. This phenomenon has been exploited to separate the two isotopes of boron. (51)
An isotope of BCl 3 is excited to a high vibrational state by an ultrahighpower CO 2 laser via multi photon absorption. It is then exceptionally reactive. In one series of experiments the excited BCl 3 dissociated and the boron
atoms reacted with oxygen in the cell. (52) The remaining BCl 3 was thereby
enriched in one or another isotope depending upon which species had
been photodissociated. The effect is not small. Figure 6.10 shows the chemiluminescence of BO under conditioIlB where the CO 2 laser was tuned to the
absorption frequency in 10BC1 3 or llBCl a . The isotopic composition of the
BO formed corresponds to the composition of the BCl3 excited. Another
set of scavenging experiments was done using H 2S. Here reaction appears
to occur with vibrationally highly excited BCl 3. Again the remaining BC1 3
was isotopically enriched. (53)
As a matter of practical interest the molecule that one wants to separate
isotopically is UF 6' However, its infrared active vibrational frequencies
do not fall in the 9-11 [Lm range and are thus not accessible to excitation by
a CO 2 laser. Thus much work has been done on the structurally similar
molecule SF 6 which has the infrared active fundamental V3 at 10.6 [Lm;
it has been possible to separate 32SF 6 from 34SF6 Here radiation is used to
dissociate 32SF 6 , probably to 32SF 4 and F 2; the 32SF 4 can be hydrolyzed
with water and the 32S0F2 separated. By using hundreds of laser pulses it
is possible to increase the naturally occurring 34Sj32S ratio by more than
(50)
(51)

(52)

(53)

D. F. Dever and E. Grunwald, J. Arn. Chern. Soc. 98, 5055 (1976).

For a recent review see V. S. Letokhov and R. V. Ambartzumian, Acc. Chern. Res.
10,61 (1977).
R. V. Ambartzumian, N. V. Chekalin, V. S. Doljikov, V. S. Letokhov, and E. A. Ryaboy, Chern. Phys. Lett. 25, 515 (1974).
S. M. Freund and J. J. Ritter, Chern. Phys. Lett. 32, 255 (1975).

Sec. 6.9

Laser-Controlled Isotope Separation

193

AI
/ I'
/

..,/

I
I

Excitation
of IOBCI 3

Excitation
of IIBCI3

A(A)

5070

5040

5010

Fig. 6.10. Chemiluminescence spectrum of BO formed by reaction of O 2 with laser-activated

BCI 3 when laser is tuned to absorption frequency of lOBCl 3 or "BCI,. In each case the corresponding isotope ofBO is formed preferentially [R. V. Ambartzumian and V. S. Letokhov,
Ace. Chern. Res. 10, 61 (1977), reprinted by permission, American Chemical Society].

Chap. 6

194

Photochemistry

103_10 4 (54) The enrichment ratio is sensitive to a range of variables: laser

frequency, pulse energy, pulse duration, pulse number, and gas pressure.
Naturally, increase of the pressure is rapidly counterproductive; above
,. . ",., 10-3 atm V- T transfer is sufficiently rapid so that SF6 cannot readily
be driven into a highly excited state and the enrichment ratio falls.
Efforts to maximize the enrichment ratio led to a variety of experiments,
some superficially unlikely to succeed. One of the latter had most unexpected results. The SF6 was irradiated using two lasers. One, at lower power
(50 kW cm- 2 to 2 MW cm- 2 ), was tuned to the excitation frequency;
the ultrahigh-power laser (,....",.,20 MW cm- 2 ) was not tuned. This mode of
operation led to a substantial increase in the enrichment ratio. (55)
Theoretical understanding of the laser-driven dissociation is far from
complete. Certainly multi photon absorption is required. However, rotational
compensation for vibrational anharmonicity, as illustrated by Fig. 6.9,
cannot possibly be operative all the way up the vibrational ladder. In SF6
more than 30 quanta at 10.6 [Lm are required to provide the dissociation
energy. The upper vibrational levels become so narrowly spaced that most
of the photon energy would have to go into rotational excitation in a
1. If, however, the mechanism of Fig. 6.9 accounts for
transition v ----" v
the initial excitation, the SF6 molecule may rapidly be driven to a vibrational
level where the vibrational-rotational states of one vibrational mode are
in the same energy range as those of another mode. The net effect is an
essentially continuous distribution of states all very anharmonic, the socalled "quasi-continuum." Ordinary selection rules no longer matter
greatly and photons of any energy may be absorbed to further energize the
molecule. (56) By means of this arbitrary division of the vibrational-rotational energy-level manifold, the two laser experiment can be rationalized.
The low-power tuned laser selectively excites SF6 into the quasi-continuum.
Then the high-power, untuned one drives only these excited molecules to
dissociation. Further study is required to verify this interpretation.

R. V. Ambartzumian, Yu. A. Gorokhov, V. S. Letokhov, and G. N. Makarov, Pis'rna

Zh. Eksp. Teor. Fiz. 21, 375 (1975), JETP Lett. 21, 171 (1975) (English translation);
J. L. Lyman, R. J. Jensen, J. Rink, C. P. Robinson, and S. D. Rockwood, Appl. Phys.
Lett. 27, 86 (1975).
(55) R. V. Ambartzumian, Yu. A. Gorokhov, V. S. Letokhov, G. N. Makarov, A. A. Puretskli, and N. P. Furzikov, Pis'rna Zh. Eksp. Teor. Fiz. 23, 217 (1976), JETP Lett. 23,
194 (1976) (English translation).
(56) R. V. Ambartzumian, Yu. A. Gorokhov, V. S. Letokhov, G. N. Makarov, and A. A.
Puretskli, Pis'rna Zh. Eksp. Teor. Fiz. 23, 26 (1976); JETP Lett. 23, 22 (1976) (English
translation).
(54)

195

Problems

The impetus for laser isotope separation studies is given by UF 6.

As mentioned in the previous section the presently known infrared lasers
emit in the vicinity of 2.7 [J.m (HF), 5 [J.m (CO), and 10 [J.m (C0 2 ). Thus
to employ the laser approach in uranium isotope separation requires the
development of a new high-power, low-frequency, laser source since the
V3 excitation mode used in the SF 6 experiments appears at 16.0 [J.m in

UF 6 (57)

Problems
6.1.

Quenching reactions with photoexcited species are subject to diffusion

limitation as discussed in Section 5.7. A more general mechanism than
(6.16) is
A*

+ B ~ A * . B ~ products
kdlff

k2

kdlss

(a) Modify the arguments leading to (6.18) and show that f{!o/f{! =
1 + ykdlflToC, where y = k2/(k2 + k dISS ).
(b) The phosphorescence lifetime of benzophenone is -160 sec. If a compound quenches benzophenone triplets at a diffusion-controlled rate
(kdlfl - 10 '0 M-1 sec- ' ) what concentration is required to quench 95% of
the phosphorescence?
6.2.

In addition to diffusion of A * and B the species A * . B may also be formed

by direct excitation of an A . B complex. Assume molar extinction coefficients E and E' for A and A . B, respectively, and the equilibrium
A

+B=

A B,

(a) Show that the concentration of A . B is [A . B]

where
a = Ao

+ Bo,

(l

= Bo - Ao, and R = l2K'

(aK

+1-

R)/2K,

+ 2aK + 1)'/'

(b) Show that the total rate of light absorption is proportional to

Ao(E

+ E'K[BD/(1 + K[BD

where [B] = lK - I + R)/2K.

(c) Now show that the fluorescence efficiency of A* is

f{!o

(57)

+ E'K[B]

+ (1 - y)'K[B]
1 + kdIffTo[B]

1 - ya

Infrared sources being developed are discussed by J. J. Ewing, in Chemical and Biochemical Applications of Laws, C. B. Moore, ed. (New York: Dekker, 1977), Vol. 2,
Chapter 6.

Chap. 6

196

Photochemistry

where y == (1 + k 2 TAB)/[1 + (k 2 + kd1ss)TAB], IX = (1 + j'K[B])-" and

TAB is the radiative lifetime of A * . B.
(d) Under what conditions will the Stern-Volmer plot be a linear function
of Eo?
6.3.

In acridine quenching by various amines in 0.03 M NaOH the experimental

values of yk diff vary from 2.8 x 107 M-l sec ' (NHa quencher) to 3.0 x 10"
M- l sec- l (ethyl aN quencher). Assuming that meaningful measurements
require that 0.95 ~ rpj rp" ~ 0.05, what quencher concentrations are needed?

6.4.

Which of the following pairs of molecules will be more likely to undergo

intersystem crossing?
(i) H 20 or H2Te.
(ii) Fluorene (Esl - ETI
25 kJ mol-').

6.5.

113 kJ mol-') or biacetyl (Esl - ETI =

The gas-phase fluorescence efficiency and spectral structure of benzene are

invariant from 5 x 10- 6 atm to 10-4 atm.
(a) What conclusion does this suggest about the collisional efficiency of
intersystem crossing?
(b) Assuming T ~ 300 K and a collision diameter of 0.4 nm estimate the
number of collisions required to induce an S, ~ T, transition.

6.6.

Given the information on triplet and singlet energies shown in Table 6.4,
which compounds in the table will sensitize or quench chrysene (a) fluorescence, (b) phosphorescence?

Table 6.4
Compound

Anthracene
Benzene
Benzophenone
Biacetyl
Chrysene
Fluorene
Naphthalene
Phenanthrene
Quinoline
Triphenylene

E(S,)

E(T,)

(kJ mol-I)

(kJ mol-I)

310
481
310
255
331
397
377
339
381
343

176
356
289
230
238
285
255
259
259
280

197

Problems

6.7.

Which of the following compounds will quench naphthalene fluorescence at

a diffusion-controlled rate: benzene, benzophenone, biacetyl, fluorene,
quinoline, or triphenylene?

6.8.

Describe the general shape of the vibronic spectrum observed for excitation
from the v = 0 level of the ground state for the potential functions of
Figs. 6.4a, b, and d.

6.9.

The phosphorescence decay of benzophenone in benzene solution is consistent with the rate law
d[T,J/dt

-k,[T,J - k 2 [T,l"

where, in 1 M benzophenone, k2 is close to the diffusion-controlled limit,

k2 = 1.1 X lO'o M-' sec-'.
(a) Estimate the distance between benzophenone molecules in aiM
solution.
(b) At intermolecular distances ;s 1.5 nm, "hopping" excitation transfer
T,

+ So' ~ So + T,'

may be important. At a separation R the probability of "hopping" is

proportional to r RiL , where L is a constant. Show that the "hopping" rate
ex [T,HSo]L".
(c) If "hopping" excitation transfer significantly increases the effective
mobility of triplet benzophenone, show that it contributes a term proportional to kH[So] to the second-order rate constant k 2
(d) What conclusions can you draw from the following measurements of
k, and k2 [T.-S. Fang, R. Fukuda, R. E. Brown, and L. A. Singer, J. Phys.
Chern. 82, 246 (1978)]?
[Benzophenone] (M):

0.001

0.005

0.01

0.015 0.5

1.0

to-sk, (sec-'):

1.3

1.4

1.3

1.6

2.9

5.0

lO-,ok 2 (M-' sec-'):

0.9

0.7

0.9

0.7

1.1

1.1

(e) Suggest experimental conditions that will quench diffusion but permit
hopping.
6.10. What is the likely product when the substituted cydohexadiene

undergoes rearrangement to form a 1,3,5-hexatriene via (a) a thermal

pathway; (b) a photochemical pathway?

Chap. 6

198

Photochemistry

Table 6.5

CClFa at 0.079 atm

EBbS

CClaF at 0.079 atm

CPFx 100

(kJ mol-')

EBbS

(kJ mol-')

CPFx 100

77.8

1.58

55.6

2.8

56.9

0.374

46.9

0.8

50.2

0.120

41.4

0.4

35.1

0.0060

28.9

0.009

6.11. In laser-activated reactions progress is defined in terms of conversion per

flash (CPF) and temperature corresponds to the energy absorbed per mole
of reactants, E. bs '
(a) Show that in a first-order process -d In [A]ldt ex (CPF).
(b) Show that activation energy corresponds to the slope of a plot of
In(CPF) vs. (-IIE. bs )' What assumptions are required to make this correspondence?
(c) Experiments on laser-induced decomposition of CClF3 and CCI3F
[D. F. Dever and E. Grunwald, J. Am. Chern. Soc. 98, 5055 (1976)] yielded
the results shown in Table 6.5. Estimate E. c ! and its variance for the two
decompositions.
(d) To what temperatures do the values of Eabs correspond?
6.12. The fluorescence efficiency of benzene solutions containing 5 : 1 mixtures
of l-C(hloroanthracene) and P(erylene) was studied for values of [P] ;S
0.006 M [E. T. Bowen and B. Brockelhurst, Trans. Faraday Soc. 49, 1131
(1953)]. The fluorescence efficiency of P increases as solute concentration
increases.
(a) Show that this phenomenon is rationalized if there is excitation transfer
with C acting as donor and P acting as acceptor.
(b) Assume that excitation is transferred or decays by the following reactions:

+ hv

(1)

C*

(2)

C*~~C

+ C ~2C
(4) C* + P ~C + P
(5) p* ~P + hv
(3)

C*

(6)

p*

+ P ~2P
+ C~P + C
C* + P ~C + P*

(7)

p*

(8)

p*

(9)

~~P

General References

199

Show that
rpfipc

{I

rp0fipp = {I

+ (k,[C] + (k. + k.)[P])Tc}(1 + y)

+ (k [P] + k8[C])Tp}(1 + y)/(f. + y)
7

where Tp and TC are the radiative lifetimes; y determines the fraction of

radiation absorbed by C and P, y ~ Ep[P]/ECC]; ip0 is the total fluorescence
efficiency of the mixture at infinite dilution; ipc and ipp are the concentrationdependent fluorescence efficiencies; J. is the fraction of excitation transfer,
fo ~ koTCP]/[l

+ (k,[C] + (k. + k.)[P])TC]

(c) When [P] = 0.002 M, ipp/ippo ~ 2; estimate the range of the C-P
interaction.
(d) At infinite dilution rppo/ip0 ~ i. Estimate y and the extinction coefficient
ratio Ep/EC.
(e) Measured fluorescence efficiencies are
[P] x 10' (M):

3.4 4.5

rpp/ ipc:

4.0

5.7

8.0 9.0

ip/rpc:

3.5

3.8

4.2 4.6

5.5
12
5.7

6.6
12
5.1

Assume k7 and k8 are negligible and estimate k9; TC is 1.2 X 10-8 sec. (The
diffusion limit in benzene is ~ 1010 M -1 sec- 1 .)

General References
Photochemistry
P. G. Ashmore, F. S. Dainton, and T. M. Sugden, eds., Photochemistry and Reaction Kinetics (Cambridge: University Press, 1967).
A. Cox and T. J. Kemp, Introductory Photochemistry (London: McGraw-Hill, 1971).
J. P. Simons, Photochemistry and Spectroscopy (London: Wiley-Interscience, 1971).

Energy Transfer
L. Krause, Adv. Chem. Phys. 28, 267 (1975).
D. Secrest, Ann. Rev. Phys. Chem. 24, 379 (1973).
I. W. M. Smith, in Gas Kinetics and Energy Transfer, P. G. Ashmore and R. J. Donovan, eds.
(London: Chemical Society, 1977), pp. I-56.

Chemiluminescence
T. Carrington and J. C. Polanyi, in MTP International Review of Science, Physical Chemistry,
Series I, Vol. 9, J. C. Polanyi, ed. (London: Butterworths, 1972), pp. 135-171.

200

Chap. 6

Photochemistry

Orbital Symmetry Correlations

T. L. Gilchrist and R. C. Storr, Organic Reactions and Orbital Symmetry (Cambridge:
University Press, 1972).
R. G. Pearson, Symmetry Rulesfor Chemical Applications (New York: John Wiley, 1976).

Lasers in Chemistry
M. J. Berry, Ann. Rev. Phys. Chem. 26, 259 (1975).
S. Kimel and S. Speiser, Chem. Revs. 77, 437 (1977).
J. T. Knudtson and E. M. Eyring, Ann. Rev. Phys. Chem. 25, 255 (1974).
c. B. Moore, ed., Chemical and Biochemical Applications of Lasers (New York: Academic
Press, 1974), Vol. 1, Chapters 6 and 7.

Nonstationary State
Mechanisms
7.1. Introduction
In the last two chapters chemical mechanisms were studied under
simplified conditions. An attempt was made to identify each elementary
step individually. Naturally this was not always possible. Competition,
reverse reaction, consecutive reaction, and inhibition all complicated the
analyses. However, by limiting consideration to systems for which the
steady-state assumption is appropriate, reasonably simple portraits of
reaction could be drawn.
Problems of practical interest such as those that arise in combustion,
explosion, metabolism, and manufacturing processes represent a more
formidable challenge. Here the system is far from equilibrium. Various
forms of feedback may operate so that a wide range of behavior can be
found. The most familiar form of feedback leads to explosion. An exothermic reaction evolves heat thus raising the system's temperature. At the
higher temperature the reaction velocity is accelerated, the rate of heat
evolution is also accelerated, and therefore so is the rate of temperature
increase. Unless enough heat can be lost to the surroundings, explosion is
inevitable.
A less familiar form of coupling involves chemical feedback. Imagine
two parallel competing processes. As we have seen, each retards the other
since both deplete the same pool of reactants. If, in addition, an intermediate produced in one pathway is an intermediate reactant in the other,
the two paths are no longer parallel and independent. Instead the inter201

Chap. 7

202

Nonstationary State Mechanisms

mediate's concentration can control whichever of the two pathways is

dominant. In a system far from equilibrium there may be periodic switching
from one pathway's dominance to the other's. As a consequence the concentration of some components can oscillate in homogeneous systems.
Such behavior is particularly evident in biological systems where numerous
periodic processes occur in nature.
The oscillations resulting from coupled competing processes need
not just involve a temporal change of concentration. Diffusion, temperature,
and voltage also couple with chemical reaction resulting in concentration
waves, temperature oscillations, or voltage oscillations. Under special
conditions a system may even be stable in more than one stationary state
and, as a system evolves, a chemical "hysteresis" can be observed.

7.2. Thermal Explosion

In the treatment of chain reactions (Section 5.6) it was clear that under
suitable conditions the system could become unstable. In an exothermic
reaction, the temperature rises thus accelerating the reaction. Positive
feedback causes another increase in temperature. Unless depletion of the
reactants or heat loss to the surroundings is sufficiently rapid, the coupling
of temperature and reaction rate sets the stage for an explosion.
There are three thermal effects: thermal conduction within the reaction
vessel, described by (2.27); heating due to reaction exothermicity; and heat
loss due to interaction with the surroundings maintained at a temperature
To. The local temperature is the solution of the differential equation

-aT =

at

(2
x 17 T + Q -d~) - V
~

- yeT - To)

(7.1)

The second term is the chemical contribution and the last one accounts
for coupling with the surroundings; ~ is the progress variable for chemical
reaction, Q is the enthalpy release per mole of reaction, (1) and y accounts
for the effectiveness of thermal coupling between system and surroundings. (2)
Solution of (7.1) requires knowing how the rate of reaction, d~/dt, depends
upon T. Assume a simple situation and consider a single, irreversible
(1)

(2)

In general more than one chemical process contributes to (7.1) and

~ Qi~i' The resulting equations are extraordinarily complex.
If the system were adiabatic y would be zero.

Q~

is generalized to

Sec. 7.2

Thermal Explosion

rate of heat
evolution or
heat loss

203

r
I

Temperature

Fig. 7.1. Heat evolution curves (1, 2, 3) and heat loss curve (straight line) as a function
of temperature.

bimolecular reaction
A

k
+B~
products

for which the rate of progress is

dt

k(T)(Ao -

~)(Bo - ~)

(7.2)

where Ao and Bo are the initial reactant concentrations. Depending upon

initial conditions (7.1) and (7.2) mayor may not permit an explosion.
Analysis of (7.1) and (7.2) is difficult; however, some limiting cases
are tractable. Ignoring depletion (Ao ?> ~, Bo }> ~) and assuming homogeneity W2 T = 0), combining (7.2) with (7.1) yields
(7.3)
The first (heat evolution) term accounts for chemical heating and the second
(heat loss) term accounts for conduction to the surroundings. Their separate
temperature dependences are shown in Fig. 7.1, where the effect of temperature on Q, C v , and y is not considered. Only the rapid increase of rate
constant is accounted for. Due to activation effects heat evolution increases
exponentially as T increases; heat loss is but a linear function of T. (3)
(3)

For an extended treatment with examples see N. N. Semenov, Some Problems in Chemical Kinetics and Reactivity, translation by M. Boudart (Princeton, N.J.: Princeton
University Press, 1958), Vol. 2, Chapter 8.

204

Chap. 7

Nonstationary State Mechanisms

Consider low initial reactant density (curve 1). At low temperature

heat evolution exceeds heat loss. The temperature increases until Ts is
reached when the two rates balance and there is a steady state. Any fluctuations around T s , possibly caused by local inhomogeneities, are rapidly
damped-for T > Ts the cooling rate exceeds the heating rate while for
T < Ts the reverse is true. In either case the system is stable; the temperature
returns to Ts. Another steady state appears to occur when T = T i ; however,
this is unstable. A fluctuation for which T < Ti leads to rapid cooling
while one for which T> Ti leads to unchecked heating. At Ti the system
is on the brink of exploding and may be driven in that direction by local
temperature fluctuations. Since any process which allows T to exceed T i ,
no matter how briefly, leads to thermal instability the temperature Ti is
an ignition temperature. At this temperature a thermal explosion occurs
even though the walls of the system remain at To; heat dissipation is too
slow to forestall explosion.
As reactant concentration increases, heat evolution at fixed temperature
also increases. Critical conditions finally arise such that explosion is inevitable. At the critical concentration (curve 2) Ts and Ti coalesce; the steady
state is unstable to local temperature fluctuations and ignition occurs.
The external temperature is too high to permit sufficient heat loss to the
surroundings. At yet higher initial concentration (curve 3) no steady state

is possible when the external temperature is To; explosion always occurs.

Uncritical consideration of Fig. 7.1 suggests that any exothermic
reaction has an ignition point; this is a consequence of our simplification
of (7.1) and (7.2). We assumed no significant depletion occurs during
the preignition stages. However, if Ao and/or Eo are too small depletion
keeps the rate of progress (7.2) under control. As a consequence the unchecked temperature increases suggested by (7.3) and illustrated in Fig. 7.1
need not occur.
Coupling between system and surroundings greatly affects the possibility of thermal explosion. With adiabatic walls, explosion is only avoided
if there is rapid depletion. The stronger the diathermal coupling, the greater
the heat loss and the less likelihood there is of an explosion. Thus the size
and shape of the reaction vessel are significant variables; increasing the
surface-to-volume ratio decreases the possibility of ignition.
The qualitative features which control thermal explosion do not depend
upon the specific chemical process taking place. Whether the reaction is a
simple one such as (7.2) or a multistep chain reaction, the principles remain
the same. Chemical heat evolution accelerates the reaction which then
produces more heating.

Sec. 7.3

Population Explosion-Autocatalysis

205

7.3. Population Explosion-Autocatalysis

In Section 5.6 we noted that, for a branching-chain reaction, the population of chain carriers may grow without limit. As a consequence reaction
velocity accelerates independent of chemical heating. At low reactant
concentrations the rate law for carrier formation is, from (5.35),
a>l

(7.4)

Here n chain carriers are formed in the initiation step for which the rate
per carrier is rr. In the branching step (or steps) a carriers are formed for
each one that is consumed; the rate is (a - l)rB[R]. The termination rate
is rT[RV 4 )
To illustrate the general considerations consider the hydrogen-oxygen
system for which the important elementary steps at low pressure appear
to be the following, (5) where S is used for surface and the values in parentheses represent iJH (kJ mol-I):
Initiation:
Branching:

H2+ S

{H +0,
0+ H2

Propagation:
Termination:

HO +

kl
~HS+H
k

~HO+O

(+64)

~HO+H

(- 4)

H2~HOH

{H I S

+ H

(7.5)

(-58)

k5

--" --" stable species

k.

HO+S --" --" stable species

Chain initiation occurs by surface reaction, as in this example, or homogeneously, as in the CO + O 2 chain. (6) Since heterogeneous processes are
very significant in the kinetics of branching chains great care in surface
preparation is required to achieve reproducible results. (7) The propagation
14)

(5)

16)

17)

See N. N. Semenov, Some Problems in Chemical Kinetics and Reactivity, translation

by M. Boudart (Princeton, N.J.: Princeton University Press, 1958), Vol. 2, Chapter 9
for a general discussion of ignition in branching-chain systems.
S. W. Benson, The Foundations of Chemical Kinetics (New York: McGraw-Hili, 1960),
p.454.
A. S. Gordon and R. H. Knipes, J. Phys. Chem. 59, 1160 (1955).
N. N. Semenov, Some Problems in Chemical Kinetics and Reactivity, translation by
M. Boudart (Princeton, N.J.: Princeton University Press, 1958), Vol. 2, Chapter 8.

206

Chap. 7

Nonstationary State Mechanisms

and branching steps of (7.5) are typical; each is bimolecular. In a mixture

of fixed composition we expect that
(7.6)

rB ex Pr exp( - EB/ RT)

where Pr is the reactant pressure and EB an activation energy for branching.

At low pressure termination is always heterogeneous; any probable homogeneous bimolecular reaction of R must, at the very least, propagate the
chain. Termination is diffusion controlled. In Section 2.5 we showed that
the characteristic time to diffuse a distance d in a linear system is ",-,d 2/ D,
where D is the diffusion coefficient. A similar result is obtained in threedimensional systems. (8) Using (2.41), the termination rate is
(7.7)
where P is the total pressure and d is now the diameter of a spherical or a
long cylindrical reaction vessel.
Ignoring the effect of reaction on the various r, (9) (7.4) can be integrated
(7.8)

<

rT, A is negative
and, after sufficient time, [R] reaches the steady-state limit (5.36). However,
if A > 0 there is positive feedback. The system is unstable, branching
dominates, and [R] grows exponentially. There is a population explosion
of chain carriers. In the H2
O 2 system, which is typical, the branching
and propagation sequence, for which the stoichiometry is 3H 2 O 2 =
2H 20
2H, is only slightly exothermic. Here 28 kJ are released per mole
of H 20 while !JHf"'-' -240 kJ mol-I. The bulk of the chemical heating
occurs as a consequence of termination. Since the population explosion
leads to more termination the system heats rapidly and ignition occurs.
From (7.8) the critical condition for explosion is A = 0 or with (7.6)
and (7.7)

It is assumed that initially [R]

O. Now, if (a - l)rB

Prexp(-EB/RT) ex P/2/pd 2
For a buffer-gas pressure of Po we find a critical pressure, PI,

(8)

(9)

PI ex

p/4

exp(En/2RT)/d,

Po

PI ex

p/2

exp(EB/RT)/p od 2,

Po Pr

(7.9)

L. D. Landau and E. M. Lifschitz, Fluid Mechanics (London: Pergamon, 1959), Chapter 6.

This is equivalent to ignoring depletion of reactants and thermal heating.

Sec. 7.3

Population Explosion-Autocatalysis

207

D
Region of
explosion

region of
normal
reaction

C
B

lIT - -...
Fig. 7.2. Explosion limits for a typical branched-chain reaction. The region ABC is known
as the explosion peninsula. The curve ABCD is the boundary between explosive and
bounded reaction rates.

Ignition is expected when P > Pl' Reality is more complicated. The explosion limits characteristic of branching chains are illustrated in Fig. 7.2.
Increasing pressure with fixed temperature and composition leads to
ignition at Pl' The unexpected feature is the region of stability at high
pressure, P2 < P < P3' The reaction velocity at constant temperature and
composition is plotted in Fig. 7.3. For P < PI the rate is very slow, if not
imperceptible. Above P2 the rate is again very slow but here it is measurable,
increasing gradually until P3 is reached. The behavior near PI and P2 is a
significant indicator of a population explosion. A system on the verge of
thermal explosion does not have an imperceptible reaction velocity.

rate of
reaction

/Pz

\
\

pressure

"-

-~

Fig. 7.3. Rate of a branching-chain reaction as a function of pressure for fixed temperature
and composition. The dashed lines illustrate the effect of inert buffer gas on the first and
second explosion limits.

Chap. 7

208

Nonstationary State Mechanisms

Consider PI; its temperature dependence corresponds to (7.9). Increasing the volume-to-surface ratio (increasing d) or increasing buffer-gas
pressure is predicted to lower Pl. Precisely such behavior is found in the
Hz-O z , or other branched-chain systems as indicated in Fig. 7.3. Note
that were the explosion at PI thermal in nature a buffer gas would stabilize
the system as Cv would increase and as a result the heating rate would
drop (Fig. 7.1).
In order for the system to be stable at high pressure, a process that
consumes chain carriers must become significant, otherwise autocatalysis
would remain dominant. Chain-trapping generally involves a three-body step
R

+A +M

-->.

AR

+M

(7.10)

where M is buffer gas and A is reactant. As a result adding inert buffer

gas to a mixture of fixed composition shifts pz to lower values, as shown in
Fig. 7.3. The trapping reaction in the Hz-O z system is(lO)
LlH", -167 kJ mol- I

(7.11)

H0 2 is stabilized by heterogeneous reaction or by combining with chain

carriers.(ll) In general, quenching reduces the carrier formation rate (7.4)
by adding a term of the form -rdR]. Assuming an elementary step like
(7.11) is typical
(7.12)
where EQ is the activation energy of the quenching reaction. (1Z) Modifying
(7.8), the critical condition is

A = -rQ

+ (a -

l)rB - rT = 0

which, if P2 is fixed by the quenching-to-branching balance, yields

P2

+ Po ex: exp[-(EB -

EQ)/RT]

(7.13)

s. W. Benson, The Foundations a/Chemical Kinetcs (New York: McGraw-Hili, 1960),

p.454.
1111 Possible trapping reactions are H + H0 2 ~ H2 + O 2, HO + H0 2 ~ H 20 + O 2;
reactions prejudicial to branching are H + H0 2 ~ 20H, 0 + H0 2 ~ OH + O 2.
All <.;e highly exothermic.
1121 In the CO-O, system a rather different reaction sequence is significant: initiation
CO -r- O. ~ CO. + 0; branching and propagation 0 + CO ~ CO 2*, CO.* + O. ~
CO 2 + 20; termination CO 2 * + S ~ CO 2 , CO, * + M ~ CO 2 + M. Modification of
(7.12) is needed in this case; see A. S. Gordon and R. H. Knipes, J. Phys. Chern. 59,
1160 (1955). The chain carrier CO,* is probably electronically excited CO,; see B. H.
Mahan and R. B. Solo, J. Chern. Phys. 37, 2669 (1962).
1101

Sec. 7.4

Lotka Problem: Chemical Oscillations

209

Since quenching reactions require little activation, EB ~ EQ , and P2 should

increase with increasing T (Fig. 7.2). As already mentioned adding buffer
decreases P2.
No general theory of the third explosion limit has been advanced;
however, it is generally assumed that such explosions are thermal in nature
and reflect the system's inability to dissipate heat. However, in the H 2-0 2
system a new kinetic pathway may become important. As diffusion of
H0 2 is retarded surface reaction is less important. The reaction

may be significant; H0 2 then acts as a propagator rather than a chain

trap. (13)

7.4. The Lotka Problem: Chemical Oscillations

We have just seen that positive feedback can lead to explosive reaction,
either by unchecked heating or unchecked population growth. We now
consider a different example of instability. Consider the hypothetical
mechanism
k.

X+Y~2Y,

(7.14)

which does not describe any known chemical process but has a crude ecological analog. If A is grass, X are deer, Yare wolves, and Z are dead
wolves (or fertilizer) then deer eat grass and reproduce, wolves eat deer
and reproduce, and wolves die. In wilderness management it is well known
that animal populations oscillate rather than settling into a steady state.
As we shall see (7.14) can account for such behavior.
Before analyzing this mechanism consider the related problem, where
reverse reaction is possible,
k.

X+Y~2Y,
k_.

(7.15)

For simplicity assume that [A] has a constant value Ao which may be arranged if, initially, A is in great excess or if the system is open and A is
(13)

N. N. Semenov, Some Problems in Chemical Kinetics and Reactivity, translation by

M. Boudart (Princeton, N. J.: Princeton University Press, 1958), Vol. 2, p. 196;
J. Nicholas, Chemical Kinetics (New York: Halstead, 1976), p. 148.

Chap. 7

210

Nonstationary State Mechanisms

continually replenished. Since reverse reaction is possible (7.15) admits

of a true equilibrium state. The equilibrium constraints are
(7.16)

and the coupled rate equations are

d~]

k1AO[X] - Ll[X]2 - k 2[X][Y]

---;Jt

k2[X][Y] - L 2[Y] - k3[Y]

d~~]

ka[Y] - L3[Z]

d[Y]

+ L2[Y]2

+ L3[Z]

(7.17)

If there are small perturbations about the equilibrium state the system,
as discussed in Section 4.9, relaxes to equilibrium (see Problem 7.4). The
displacement variables [X] - X o , etc., are linear combinations of decaying
exponential functions. Direct negative feedback, due to reverse reactions
like 2X ----'" X + A, keep [X] and [Y] under control no matter what the
initial conditions.
If reverse reaction is forbidden no true equilibrium exists. No direct
feedback controls [X]; instead it is kept in check by reaction with [Y].
From (7.14) or (7.17) the coupled rate equations, maintaining [A] = A o ,
are(14)
d[X]
---;Jt = k1AO[X] - k 2[XHY]
(7.18)
d[Y] = k 2 [X][Y] - k3[Y]
dt
There appears to be a steady state when d[X]/dt
case

d[Y]/dt

0 in which
(7.19)

which are substantially different constraints than (7.16). Defining displacement variables
[X] = $(t)
(14)

+ X o,

[Y]

rJ(t)

+ Yo

(7.20)

There will be a steady increase in Z but its growth is of no immediate interest. The system
(7.18) was first analyzed by A. J. Lotka, J. Am. Chern. Soc. 42,1595 (1920); Proc. Nat.
Acad. Sci. (U.S.A.) 6, 410 (1920).

Sec. 7.4

Latka Problem: Chemical Oscillations

211

and substituting in (7.18) yields, with (7.19),

~ = -k2X01] - k2~1] = -k21](~

+ k2~1] =

ij = k2YO~

k2~(1]

+ Xo)

(7.21)

Yo)

If the displacements are small the quadratic terms can be dropped. Using
the matrix method of Chapter 5 and assuming that ~ and 1] are linear
combinations of decaying exponential functions e-J.t, the decay constants
A, are solutions of the determinantal equation
(7.22)
from which we find

A,2

+k

2 2X OYO

0 or, with (7.19),

(7.23)

The "decay constants" are imaginary! The displacement variables, being

linear combinations of the functions eiwt and e- iwt (or equivalently sin wt
and cos wt), do not describe relaxation to a steady state. Instead a system
perturbed from the steady state undergoes perpetual oscillation. The origin
in the ~-1] plane (i.e., X = X o, y = Yo) is not a steady state at all; rather
it is a singular point which is not accessible. Even if the system were prepared
at the steady state, it could not remain there. Due to the fluctuations which
constantly arise, a system in which initially; = 1] = 0 is unstable. Fluctuations create displacements and oscillations commence.
The analysis of the displacement equation (7.21) has been limited to
small values of ~ and 1]. To generalize it is easier to determine the functional
relationship between the displacement variables ~ and 1] than to find the
separate solutions ~(t) and 1](t). Divide ~ by i;; the result is
(7.24)
which may be rewritten

d~(l

Xo

+ Xo

= -d1](1 _

1]

Yo

Yo

and integrated
(7.25)

Chap. 7

212

Nonstationary State Mechanisms

The equation determines a set of curves that describe the relationship

between ~(t) and 'Y)(t). The curves are characterized by certain properties:
(i) they cannot cross since the slope, d~/d'Y), is uniquely specified at
each point in the ~-'Y) plane [from (7.24)];
(ii) as [X] > and [Y] > 0, ~ > -Xo and 'Y) > - Yo [from (7.20)];
(iii) unbounded increase of ~ or 'Y) is impossible [from (7.25)];
(iv) for [X] > II and [Y] > 0, to each 'Y) there corresponds one value
of ~ > and one value of ~ < 0; similarly there are two 'Y) for
each ~ [from (7.25)];
(v) when ~ > and 'Y) > 0, ~ < and iJ > 0, etc. [from (7.21)].

The general features of such curves are indicated in Fig. 7.4a. They form a
set of nonintersecting closed curves as shown in Fig. 7.4b. The singularity
is the origin in the ~-'Y) plane. Trajectories do not terminate at this point.
Instead a system displaced from the singularity oscillates in perpetuity.
Furthermore the larger the initial perturbation the larger the amplitude of
the oscillations. No relaxation occurs, even for large amplitude perturb ationsY5) What is not apparent from Fig. 7.4 is the oscillation period. However, by expanding ~(t) and 'Y)(t) in Fourier series the oscillation frequency
(7.23) may be shown to be a general feature of the rate law (7.18); w does
not depend upon the perturbation amplitude. (16)
The model mechanism predicts undamped oscillations in the concentrations of intermediates X and Y. The concentrations of reactant A and
product Z do not oscillate; however, their rates of consumption and formation do. Three conditions were needed to generate such behavior:
(i) continuous flow of matter into the system ([A] is constant);
(ii) a mechanism with coupled indirect feedback steps;
(iii) a system far from equilibrium (some steps are irreversible).
The constraints appear to be general. (17) Unlike explosion where feedback
leads to direct acceleration of reactant consumption, oscillatory mechanisms
are more subtle. According to (7.18) increasing [X] when [Y] is small
(15)

(16)

(17)

The approach outlined, considering the trajectories in the !;-'Y} plane instead of !;(t)
and rl(t) individually, has numerous applications. For an introduction to the theory of
trajectory classification see D. A. Sanchez, Ordinary Differential Equations and Stability
Theory (San Francisco: Freeman, 1968).
A. J. Lotka, J. Arn. Chern. Soc. 42,1595 (1920); Proc. Nat. Acad. Sci. (U.S.A.) 6, 410
(1920).
A discussion of the requirements for oscillatory behavior is given by G. Nicolis and J.
Portnow, Chern. Rev. 73,365 (1973); see particularly Sections L D. and lILA. A more
qualitative treatment is given by H. Degn, J. Chern. Ed. 49, 302 (1972).

Sec. 7.4

Latka Problem: Chemical Oscillations

I.I
I
I
I
I
I
I

213

Xo

I
I

L __
X--+

(0)

( b)

Fig. 7.4. (a) General features of trajectories (7.25). (b) Detailed curves showing three trajectories of the concentration displacement variables ~ and 'Y} for different initial displacements.
The steady state is the origin of the ~-'Y} plane. It is inaccessible. [Figure 7.4b adapted from
A. J. Lotka, J. Am. Chem. Soc. 42, 1595 (1920).]

leads to acceleration in the rate of production of X. Finally enough Y has

formed to consume X with sufficient rapidity so that [X] starts to drop and
the first step in (7.14) slows down. As Y is consumed by the third reaction
its concentration drops and [X] can build up again resuming the cycle. The
net effect on [X] is immediate autocatalysis followed by delayed inhibition
by Y. It is delayed feedback that characterizes chemical oscillation.

In a closed system reactant concentration must decrease. Thus perpetual oscillations in the concentration of intermediates are not possible.
However, if initially A is in great excess and conditions (ii) and (iii) are
met, very slowly damped oscillations may be observed. In the remainder of
this chapter we discuss some specific examples. (18) Figure 7.5 illustrates
allowed and forbidden chemical oscillation. Detailed balance forecloses
oscillations about equilibrium, as shown in Fig. 7.5a. Oscillations about
the steady state, as shown in Fig. 7.5b, violate no thermodynamic principle.
While the Lotka mechanism exhibits many features found in chemical
oscillators, it is misleading in one important qualitative respect. According
to Fig. 7.4b the amplitude of the concentration oscillations is determined
by the magnitude of the initial perturbation. Such behavior is apparently
not characteristic of real systems. Rather, independent of the initial values of
(18)

For a catalog of chemical, thermochemical, electrochemical, and biochemical oscillators see G. Nicolis and J. Portnow, Chem. Rev. 73, 365 (1973), Section II.

Chap. 7

214

Nonstationary State Mechanisms

(a)

(b)

Fig. 7.5. (a) Forbidden and (b) allowed oscillations in the relaxation of a concentration
displacement, 6X, as a function of time.

intermediate concentrations, chemical oscillators evolve to periodic behavior

determined by the rate constants of the elementary steps and by the concentrations of the major reactants. This phenomenon, known as limit cycle
behavior, has its analog in the evolution of systems displaced from equilibrium; the size of the initial displacement has no effect on the approach
to equilibrium. The study of limit cycle mechanisms is a complex problem
of trajectory analysis and classification. (19)

7.5. The Belousov-Zhabotinskii Reaction

While concentration oscillations have been observed in a variety of
systems, only two homogeneous chemical oscillators have been unambiguously established. When malonic acid (HMa) is oxidized by Br03 - in
sulfuric acid solution using a Ce(IV)jCe(III) couple as catalyst, Belousov observed that the [Ce(lV)]j [Ce(III)] ratio oscillated periodically. (20)
Further studies by Zhabotinskii demonstrated oscillatory behavior when
Ce(IV)jCe(III) was replaced by Fe(III)jFe(II) or Mn(III)jMn(II) and
(19)

(20)

Two such mechanisms, presumed to describe specific oscillating systems, have been
formulated. For details see P. Glansdorff and I. Prigogine, Thermodynamic Theory
of Structure, Stability, and Fluctuations (New York: Wiley-Interscience, 1971), Chapter
14; I. Prigogine and G. Nicolis, Quart. Rev. Biophys. 4,107 (1971); R. J. Field and R.
M. Noyes, J. Chem. Phys. 60, 1877 (1974).
B. P. Belousov, Sb. Ref Radiats. Med. (Medgiz, Moscow) 1958,145 (1959).

Sec. 7.5

Belousov-Zhabotinskii Reaction

215

Table 7.1. Range of Reactant Concentrations Permitting Concentration Oscillations

in the Belousov-Zhabotinskii Reactiona
Species

Malonic acid

Concentration (M)

0.013-0.5

KBrOa

0.013---{).063

Ce(NH.MNO a).

0.000 1---{).0 1

H.SO.
KBr

0.5-2.5
Trace (;S 2 x to-)

a R. F. Field, E. Karas, and R. M. Noyes, J. Am. Chern. Soc. 94, 8649 (1972).

when malonic acid was replaced by other organic acids. (21) Another example
of oscillation is the 103 -/1 2 catalyzed decomposition of H 20 2 The production of O 2 is not continuous but proceeds in bursts that are correlated with
oscillations in [1 2 ].(22) Since oxygen is produced, considerable care was
required to demonstrate that the oscillations reflect homogeneous processes. (23)
Both the Belousov-Zhabotinskii and the Bray-Liebhafsky systems have
been characterized mechanistically. (24) We consider the Belousov reaction
in detail. Oscillatory behavior is not limited to a narrow range of concentrations. Initial conditions which lead to periodic concentration variation are
given in Table 7.1. The period of oscillation is quite dependent upon initial
concentrations, as suggested by the Lotka result (7.23); it varies from ,....." 15
to ,.....,,200 sec in different experiments. Not only are there oscillations in
the [Ce(IV)]/ [Ce(III)] ratio but [Br-], which is present in trace quantities,
oscillates as well. Typical results are shown in Fig. 7.6; the amplitudes
diminish only very gradually over dozens of cycles. For the experiment il('1) A. M. Zhabotinskii, Dokl. Akad. Nauk. SSSR 157, 392 (1964); Bio/izika 9,306 (1964).
(") W. C. Bray, J. Am. Chern. Soc. 43, 1262 (1921).
(.a) W. C. Bray and H. T. Liebhafsky, J. Am. Chern. Soc. 53, 38 (1931); H. T. Liebhafsky,
J. Am. Chern. Soc. 53, 896, 2074 (1931); H. T. Liebhafsky and L. S. Wu, J. Am. Chern.
Soc. 96, 7180 (1974). Liebhafsky studied the reaction at two widely separated periods
in his career; the papers cited represent only a fraction of his work.
('.) Belousov reaction: R. J. Field, E. Karas, and R. M. Noyes, J. Am. Chern. Soc. 94,
8649 (1972); Bray-Liebhafsky reaction: K. R. Sharma and R. M. Noyes, J. Am. Chern.
Soc. 98, 4345 (1976).

Chap. 7

216

Nonstationary State Mechanisms

0

1000
500
[Ce(lV)]
~ 200 A 1 0 9 - - i 100
[Ce(III)]
50
20
, 10 B

e
L

to

~I~U

2
1 C

Ql

Ql

Cl

SECONDS .....

Fig. 7.6. Periodic behavior of [Br-] and [Ce(IV)]/[Ce(III)] in the Belousov reaction [R. J.
Field, E. Koros, and R. M. Noyes, J. Am. Chern. Soc. 94, 8649 (1972), reprinted by permission of the American Chemical Society].

lustrated in Fig. 7.6 cyclical behavior can be characterized as follows:

(i) [Br-] relatively large and decreasing, [Ce(IV)]j [Ce(III)] ratio
decreasing;
(ii) [Br-] plummets sharply;
(iii) [Br-] relatively small and increasing, [Ce(IV)]j [Ce(III)] ratio
increasing;
(iv) [Br-] grows abruptly.
The overall stoichiometry, in which none of the species [Br-], [Ce(IV)],
or [Ce(III)] appear, is
3Br03 - + 3H+ + 5CH z(COOHh

3BrCH(COOH)z + 4CO z
+ 2HCOOH + 5H zO

(7.26)

The key to understanding the phenomena is knowledge of the chemistry of

Br03 - which, while a powerful oxidizing agent, does not react directly
with organic acids. The brominating agent in (7.26) is Brz while production
of CO z and formic acid is due to attack by Ce(IV) on either HMa or BrMa.
As is so often the case, kinetically important species do not appear in the
overall stoichiometry.
Two separate sequences account for (7.26). Schematically they may be
summarized as
1(+Br-)------>.> BrMa

(A)

Br0 3 - + HMa
Ce(IV)jCe(IlI) catalyzed
'---------.....>.> HCOOH + CO z + BrBr- inhibited

(7.27)
(B)

Sec. 7.5

Belousov-Zhabotinskii Reaction

217

Sequence (A) requires Br- as a reactant; Br- also acts indirectly to inhibit
sequence (B). When Br- is depleted reaction switches from (A) to (B).
However, (B) replenishes Br-, becoming self-inhibitory and reinitiating
(A). Oscillations arise because of the switching.
The individual steps are complex and will only be outlined. (25) Reduction of Br03 - to Br2 proceeds by consecutive removal of oxygen (twoelectron transfer)

+ Br0 + 2H+ -----' HBr0 + HOBr

Br- + HBr0 2 + H+ -----' 2HOBr
Br- + HOBr + H+ -----' Br2 + H 20

(a) Br(b)
(c)

3-

(7.28)

The Br2 reacts with HMa

(d) Br2

+ HMa -----' BrMa + H+ + Br-

The overall process (a)

(A) 2Br-

+ (b) + 3(c) + 3(d)

(7.29)
is sequence (A)

+ Br0 + 3H+ + 3HMa -----' 3BrMa + 3H 20

3-

(7.30)

The slow step in the process is (a). None of the oxybromine intermediates
are thermodynamically plausible reactants in one-electron steps such as
are required to oxidize Ce(III).
A way to reduce Br03 - in one-electron transfers postulates the Br0 2
radical as intermediate,

+ HBr0 2 + H+ -----' 2Br0 + H 0

Ce(lII) + Br0 2 + H + -----' Ce(lV) + HBr0 2

(e) Br0 3 (f)

(7.31 )

+ 2(f) is autocatalytic
+ 3H+ + HBr0 2Ce(lV) + H 20 + 2HBr0

where the net reaction (e)

(C) 2Ce(I1I)

+ Br0

3-

2 -----'

(7.32)
It might appear that (C) would rapidly deplete Br03 -, however, as HBr0 2
disproportionates

(7.33)
(25)

The analysis is due to R. J. Field, E. Karas, and R. M. Noyes, J. Am. Chem. Soc. 94,
8649 (1972), where corroboratory details are given.

218

Chap. 7

Nonstationary State Mechanisms

[HBr0 2] is always small and (C) cannot lead to a population explosion.

Here the slow step is (e). As long as [Br-] is sufficiently large (b) competes
favorably with (e) and the one-electron sequence cannot proceed. However,
when [Br-] drops sufficiently, as it must in sequence (A), the autocatalytic
pathway (C) takes over and [Br-] drops precipitously since (b) continues.
As HOBr is produced in the disproportionation (g) it can react with HMa
via (c) and (d). Thus the net reaction of Ce(III), BrO a-, and HMa is 2(C)
+ (g) + (c) + (d)
(D) 4Ce(III)

+ BrO a- + HMa + 5H+ ---' 4Ce(IV) + BrMa + 3H 20

(7.34)

The individual steps which regenerate Br- and Ce(lII) and complete
the cycle have not been as precisely characterized. However, Ce(IV) can
oxidize HMa or BrMa in a series of rapid one-electron transfers; the net
result is
(E) 6Ce(IV)

+ CH 2 (COOH)2 + 2H 20

(F) 4Ce(IV)

+ BrCH(COOH)2 + 2H 20

---' 6Ce(III)

+ HCOOH

+ 2C0 2 + 6H+

---' 4Ce(III) + HCOOH

+ 2C0 2 + 5H+ + Br-

(7.35)

The set of processes (D), (E), and (F) comprise sequence (B).
Now consider a complete cycle in the oscillatory regime as illustrated
schematically in Fig. 7.7. At a, [Br-] is large; the major pathway is (A)
so [Br-] decreases. The ratio [Ce(lV)]j[Ce(III)] drops as reactions (E) and
(F) continue. At (3, [Br-] has become too small to sustain sequence (A);
reaction switches to sequence (B) and [Br-] plummets due to reaction
(b) reaching point y. Then the [Ce(lV)]j[Ce(III)] ratio grows due to (C)
and, as more Ce(IV) forms it replenishes [Br-] via (F). At <5, [Br-] is large
enough to allow switching back to (A); sequence (B) stops abruptly; the
remaining Ce(lV) continues to oxidize HMa via (F) and [Br-] rises sharply.
The system has returned to a.
Our analysis of the Belousov reaction has just considered the oscillatory
domain. Whether sequence (A) or sequence (B) is followed, initially there
must be an induction period before oscillation commences since periodic
behavior is triggered by replenishing [Br-], which cannot occur until
[BrMa] is large enough so that (F) becomes important. The behavior
illustrated in Fig. 7.6 is consistent with this interpretation. After initiation

Sec. 7.5

Belousov-Zhabotinskii Reaction

I [CeCIV)]
09 [CemnJ

219

(a)

(a)

(ct)

(Ii,a)

~
(tJ. y)

time

,.

Fig. 7.7. Schematic diagram of one cycle in the Belousov reaction.

[Be] drops rapidly and remains small for a considerable time until it
rises abruptly signaling the onset of oscillations.
The Belousov reaction satisfies the prerequisite conditions for oscillation. The consumption of reactants is slow; there are irreversible steps so
the system is far from equilibrium; there is delayed feedback, in this case
both catalytic and inhibitory. The concentrations of intermediates oscillate
around the steady state, which is a singular point that is never attained.
There are no oscillations in the concentrations of reactants or products,
only in their rates of consumption and formation. Evolution seems Characteristic of a limit cycle; the initial concentrations of the intermediates appear
not to affect the limiting amplitudes of the stable concentration oscillations. (26) The effect. of perturbing the oscillations by adding extra Bror Ce(IV) has been studied; after a short lag time the oscillations revert
to their preperturbation behavior, again indicative oflimit cycle behavior.(27)
A simplified mechanistic scheme, based upon the reactions outlined in this
section, has been analyzed using numerical methods such as those outlined
in Section 5.3. The time dependence of the various concentrations calculated from the model is in good agreement with experiment; the numerical
solutions appear consistent with limit cycle behavior.(28)
(26)

(27)

(28)

R. J. Field, E. Koros, and R. M. Noyes, J. Am. Chem. Soc. 94, 8649 (1972).
V. A. Vavilin, A. M. Zhabotinskii, and A. N. Zaikin, in Biological and Biochemical
Oscillators, B. Chance, E. K. Pye, A. K. Ghosh, and B. Hess, eds. (New York: Academic Press, 1973), pp. 71-79.
R. J. Field and R. M. Noyes, J. Chem. Phys. 60, 1877 (1974).

220

Chap. 7

Nonstationary State Mechanisms

7.6. Other Oscillating Systems

As may well be apparent periodicity in chemical systems requires
coupled, competing processes. Without coupling via bromide ion the
Belousov-Zhabotinskii reaction would just provide parallel pathways for
the consumption of bromate ion and malonic acid. The coupling compels
feedback which creates the possibility of switching from one path to the
other. Since the time scale of the processes is comparable switching and
oscillation are observable. In a similar way thermochemical and electrochemical oscillators may be designed. Furthermore spatial, rather than
temporal, periodicity may be observed. We consider a few examples.

7.6.1. Concentration Waves in the Belousov Reaction

Zhabotinskii observed that when the Belousov system was not stirred
the [Ce(lV)]/ [Ce(III)] ratio, instead of oscillating homogeneously, evolved
as traveling waves. (29) A geometry and set of initial concentrations which
dramatize the effect has been developed by Winfree. (30) An explanation
requires considering diffusion. In an unstirred system local fluctuations
ensure that all regions do not reach the end of the induction period simultaneously. Consider a domain in which [Br-] has reached the triggering
concentration, as shown in Fig. 7.8a. According to the model of the previous
section [Br-] rises abruptly and the [Ce(IV)]/ [Ce(III)] ratio is high. Thus
Br- and Ce(IV) tend to flow out of the area and Ce(III) tends to flow
into it. As a consequence sequence (A) is triggered in the neighboring
regions; as [Ce(III)] and [Br-] are dropping in the central area it switches
to sequence (B), as shown in Fig. 7.8b. Diffusive triggering of sequence (A)
propagates outward. Subsequently diffusion and depletion cause a shift
to sequence (B). The interchange continues with resulting concentration
waves. Their wavelength is determined by the diffusion coefficients of the
intermediates in addition to the rate constants and the concentrations of
major reactants. (31)

(29)

(30)

(31)

A. M. Zhabotinskii, Dokl. Akad. Nauk SSSR 157, 392 (1964); Bio/izika 9, 306,
(1964).
A. T. Winfree, Science 175, 634 (1972); Sci. Amer. 230(6), 82 (1974). The latter article
has particularly pretty photographs.
A. M. Zhabotinskii, in Biological and Biochemical Oscillators, B. Chance, E. K. Pye,
A. K. Ghosh, and B. Hess eds. (New York: Academic Press, 1973), pp. 89-95.

Sec. 7.6

Other Oscillating Systems

Ce(lV)

C'OllI0
(a)

221

Br-

(b)

Fig. 7.8. Schematic diagram showing development of concentration waves in the Belousov
reaction. (a) Sequence A is triggered in central domain; diffusion flows are indicated.
(b) At a later time sequence B dominates in central region; rings of concentration variations propagate outward.

7.6.2. Liesegang Bands: Periodic Precipitation

When a concentrated salt solution of, e.g., AgN0 3, is allowed to diffuse
into a sheet of gel impregnated with, e.g., K 2Cr0 4 , the resulting Ag 2Cr0 4
does not precipitate uniformly. Instead it forms in bands (or rings) parallel
to the diffusion front. (32) The phenomenon is quite general; similar structures
are found with many sparingly soluble salts. Numerous explanations have
been proposed. The simplest presumes that some degree of supersaturation
is necessary before precipitation occurs.(33) As the AgN0 3 diffuses into the
gel it forms a diffusion front parallel to the edge of the gel sheet. Crystallization occurs when the leading area is supersaturated with Ag 2Cr0 4
As the Ag2Cr0 4 precipitates, the [Cr0 4 2 -] concentration in that region
drops and Cr0 4 2- diffuses to the growing crystals thus depleting the neighboring area. As AgN0 3 continues to diffuse into the gel it passes through
a region of low [Cr0 4 2 -]; precipitation recommences when a normal area
is reached. The phenomenon then repeats itself leading to alternating
bands.
A somewhat different interpretation postulates that precipitation is
preceded by the formation of colloidal particles of the insoluble material. (34)
These then become charged by preferential adsorption of either anion or
cation which, due to electrostatic interactions, is autocatalytic. The larger
The effect was discovered by R. E. Liesegang, Naturw. Wochschr. 11, 353 (1896). A
voluminous literature exists on the subject; see K. H. Stern, Chern. Revs. 54, 79
(1954).
(33) Originally due to Ostwald, a modern treatment is due to S. Prager, J. Chern. Phys. 25,
279 (1956).
(34) M. Flicker and J. Ross, J. Chern. Phys. 60, 3458 (1974).
(32)

222

Chap. 7

Nonstationary Stale Mechanisms

the colloidal particle the more charge it may carry and the more counterions
it will attract thus accelerating its growth and ultimate precipitation. Then,
due to diffusive coupling, the characteristic rings are formed. The significant
feature of this model is that spatial inhomogeneities form before precipitation. as seems to occur. The supersaturation explanation requires both
to be simultaneous.

7.6.3. The "Beating Mercury Heart": Electrocapillarity

When "-' 1 cm3 of mercury is placed in a watch glass and covered with
a dilute solution of strong acid (e.g., HN03 , HCI, H 2S04) to which a few
crystals of a strong oxidizing agent (e.g., K 2Cr20 7 , Na2S20S, or KMn0 4 )
are added, the solution is complex and a film forms on the surface of the
mercury. Then, if an iron wire just touches the Hg globule, the "heart"
pulsates 2 or 3 times per second. (35) Recent experiments have shown that
the phenomenon is not arcane. (36) The system forms an electrochemical
cell which may be crudely described as
Fe(s) I H+(aq), Cr20 7 2-(aq) I Hg(liq)

(7.36)

with iron being the negative electrode. Iron is not required; similar results
are obtained using aluminum or magnesium. The oxidizing agent may also
be discarded although the "heartbeat" is then less powerful.
When there is Fe-Hg contact the iron is oxidized, Fe ---->. Fe2+ + 2e-,
and solvated Fe2+ diffuses into solution. Electrons are transferred to the
Hg surface where a cathode reaction occurs; as a result negative charge,
due to ions adsorbed on the surface, builds up. As long as there is contact
between the iron and the mercury the surface charge continues to increase;
it is apparently only limited by the rate of ion desorption from the surface.
However, the surface charge, being negative, is repelled by the negatively
charged iron electrode. If the wire does not penetrate the liquid too deeply
the Hg drop can deform sufficiently for electrical contact to break. The
adsorbed negative ions desorb into solution; the liquid resumes its spherical
shape and the circuit is reestablished. Continuous repetition leads to
pulsation.
(35)

This curious phenomenon was discovered by G. Lippman, Ann. Phys. 28 149, 546
(1873).

(36)

S.-W. Lin, J. Keizer, P. A. Rock, and H. Stenschke, Proc. Nat. Acad. Sci. (U.S.A.) 71,
4477 (1974).

Sec. 7.6

Other Oscillating Systems

223

7.6.4. Hydrocarbon Combustion: Thermal Oscillations

Hydrocarbon combustion proceeds via complicated chain mechanisms. (37) A simplified low-temperature scheme is
Initiation:
Propagation:
Branching:
Termination:

+ O 2 ----'- R + H02
ROO
{ R + O
ROO + RH ----'- ROOH + R
----'- RO + HO
{ ROOH
HO + RH ----'- R + H 20
RH

----'-

(7.37)

R +R

H0 2 is unreactive (see Section 7.3) and the chain carriers are R, ROO,
RO, and HO. The effect of buffer gas on energy transfer is kinetically
unimportant since the experimental conditions conform to the high-pressure
regions for unimolecular decomposition and bimolecular recombination.
As temperature is increased an unusual effect is observed; the reaction
rate drops. Thus explosion can be quenched by raising the temperature.
A new pathway that reduces the chain-carrier concentration must become
significant. A plausible candidate is the decomposition of the weakly bound
peroxy radical
ROO ----'- R + O 2
followed by olefin formation, e.g.,
(7.38)
This sequence both reduces branching by forming less ROO and increases
termination by forming more H0 2 At still higher temperatures H0 2
itself becomes reactive and may lead to propagation and branching. Thus
the reaction rate will again rise.
In the temperature domain where reaction rate falls as temperature
increases, thermal oscillations are observed for systems at pressures below
the explosion limit. The temperature of the burning gas is a periodic function of time. (38) The temperature may increase by as much as 200 K in a
(371

(381

See S. w. Benson, The Foundations of Chemical Kinetics (New York: McGraw-Hill,

1960), pp. 479-489 or J. Nicholas, Chemical Kinetics (New York: Halstead, 1976),
pp. 150-157 for details.
For a discussion of propane combustion see P. Gray, J. F. Griffiths, and R. J. Moule,
Faraday Symp. Chem. Soc. 9, 103 (1974).

Chap. 7

224

Nonstationary State Mechanisms

pulse. While it might seem that the phenomenon is an indication of inhomogeneity in the reactor, in fact oscillation is more regular and less
damped in a well-stirred system. In closed reactors only a few pulses are
observed before depletion damps the oscillations; in a continuous-flow
reactor pulsation continues indefinitely. (39)
From (7.1) the temperature-time profile in a homogeneous reactor is

dT
dt

QV
Cv

.!!:I
_ Y (T- T.)
dt
0

In the domain of falling reaction rates, at lower temperatures dTJdt is large

and positive since heat evolution is large and heat loss is small. As a consequence the temperature in the reactor increases abruptly after which the
reaction rate drops sharply. As a result heat loss greatly exceeds heat evolution and the temperature plummets. As long as reactants are not depleted
or in a continuous-flow system the process may repeat itself. There is
obviously an intermediate temperature at which the heat evolution and the
heat loss balance; one might therefore expect a steady state instead of limit
cycle oscillations. However, as in all the other examples considered there is
delayed feedback. The reaction rate does not instantaneously reflect the
ambient temperature; thus the temperature overshoots the steady-state
value before a correction occurs.

7.6.5. The Glycolytic Pathway: A Biochemical Oscillator

Whereas oscillatory phenomena have been, until recently, considered
chemical curiosities, they had been of great interest to biochemists for years.
Metabolic pathways conform to all requirements for the observation of
oscillations: the systems are open-reactants constantly enter across cell
walls and products are removed; they are far from equilibrium; there are
feedback loops to maintain and reassert control even when the cell is
drastically perturbed. Most biochemical oscillators are so complex that a
detailed mechanistic analysis, such as that given for the Belousov reaction,
is not available. The mechanism of glycolytic oscillation in yeast is the
most well established.
The overall yeast metabolism is
glucose -->. -->. -->. ethanol, glycerol, CO 2
(S9)

B. F. Gray and P. G. Felton, Comb. and Flame 23, 295 (1974).

(7.39)

Sec. 7.6

Other Oscillating Systems

225

As well as being of biochemical interest glycolysis is commercially important; it is used in the production of bread, beer, and wine. For a CO'lstant
rate of glucose input, oscillation is observed in the rate of CO 2 production
as well as in the concentrations of many reaction intermediates. (40) There
are multiple feedback loops; the control mechanism is far more elaborate
than in any of the physicochemical oscillators discussed previously. We
shall sketch some of the results to indicate the complexity of even a rather
simple biochemical pathway. (41)
The overall reaction (7.39) involves at least 10 separate intermediates.
The various steps, requiring C-C bond cleavage or dehydrogenation,
require mediation by a minimum of six enzymes working together. Feedback
is affected by variation of pH and of the concentration of the reduced
dinucleotide, NADH. (42) Oscillations in concentration have been found for
at least 16 separate species involved in glycolysis. The oscillators fall into
two categories. In each group the maxima and minima in the concentrations
occur at the same time. However, the two groups oscillate out of phase,
an observation that has been used to provide insight as to some features
of the feedback loop. Furthermore, the character of the oscillations changes
as the rate of glucose input is varied. At very low input level no oscillations
are found. Above a critical input level doubly periodic oscillations of small
amplitude and frequency develop. As the input rate increases further the
multiple periodicities wash out. The amplitude and frequency of the stable
oscillations increase. Finally at very high input levels oscillations are again
damped.
Of these phenomena, three have simple rationalizations. At very low
input levels the biochemical system is not sufficiently displaced from
equilibrium for stable oscillations to form. Increasing the input level
increases the concentration of the fundamental reactant which, in terms
of the Lotka model, requires an increase in frequency from (7.23). Too
great an input rate finally saturates the intermediate enzymes in a fashion
similar to that outlined in Section 5.10. As a result feedback controls are
no longer effective and oscillations damp out.

(40)

(41)

(42)

There is an enormous literature on the subject. For representative work see Biological
and Biochemical Oscillators, B. Chance, E. K. Pye, A. K. Ghosh, and B. Hess, eds. (New
York: Academic Press, 1973), Part III.
For an analysis of many of the features of the metabolic pathway see A: Boiteux and B.
Hess, in Biological and Biochemical Oscillators, B. Chance, E. K. Pye, A. K. Ghosh,
and B. Hess, eds. (New York: Academic Press, 1973), pp. 243-252.
NADH is the reduced form of nicotinamide-adenine-dinucIeotide.

226

Chap. 7

Nonstationary State Mechanisms

7.7. Multiple Stationary States: The NO z-Nz0 4 Reaction

Systems far from equilibrium exhibit unusual kinetic behavior. Feedback may lead to instability and ultimately explosion or to undamped (or
weakly damped) oscillations about an inaccessible steady state. Here we
treat another example of chemical instability-the abrupt switching between
steady states with an attendant chemical hysteresis. (43) The coupled variables
are temperature and the progress variable for a chemical reaction, the same
coupling that creates thermal explosion (Section 7.2) and thermal oscillation
(Section 7.6.4). The difference is that here the chemical reaction is endothermic so explosion is ruled out. Instability is induced by heat flow into
the system. Unlike the examples considered previously, the mathematical
description of this coupled system is simple. The resulting equations may
be solved and detailed theoretical predictions verified.
Consider the qualitative aspects of the N0 2-N 2 0 4 equilibrium; the
reaction
(7.40)
is endothermic (LJH = 58.1 kJ mol-I). When an N0 2-N 20 4 mixture is
irradiated with visible light only the N0 2 absorbs energy which, by various
photochemical relaxation processes, ultimately serves to heat the mixture
thus raising the temperature. At higher temperature the equilibrium is
shifted to favor N0 2 ; thus a greater fraction of the incident radiation may
be absorbed and the temperature is further raised. The feedback is positive. (44)
In terms of concentration variables
(7.41 )
where C would be [N 2 0 4 ] if no N0 2 were present, extension of (7.1) to
account for radiation heating yields
(7.42)
Here 10 is the radiation intensity, E the molar extinction coefficient of N0 2 ,
and I the path length; Q, the heat evolved per mole of N0 2 formed, is now
negative. The rate law for N02-N 20 4 interconversion is, from (7.40) and
(43)

(44)

The phenomenon was predicted on theoretical grounds by A. Nitzan and J. Ross, J.

Chem .. Phys. 59, 241 (1973).
See C. L. Creel and J. Ross, J. Chern. Phys. 65, 3779 (1976).

Sec. 7.7

227

Multiple Stationary Stales

(7.41),

~;

2k+(T)(C -

~/2) - 2L(T)~2

(7.43)

In (7.42) and (7.43) the quantities under experimental control are 10 , To,
and C. Increasing the radiation dose has two direct consequences-radiation
heating and increased heat loss to the surroundings-and one indirect
one-heat absorption as more N02 forms.
In a uniform steady state V 2 T, aT/at, and a~/at are all zero. Since the
parameters in (7.42) and (7.43) are known the equations, which are nonlinearly coupled due to the temperature dependence of the rate constants
k+ and k_, may be solved for ~ (or T) as functions of 10 , To, and C. Typical
results, for the same value of C, are presented in Fig. 7.9. As the surroundings cool from 250 K to 230 K the structure of the solutions changes
dramatically. At higher temperatures there is a unique N02 concentration
for each value of 10 , However, as To decreases a critical point is reached
below which there are three possible values of ~(Io, To). Two represent
stable steady states; one is unstable.
Consider the 230 K curve in detail for a system initially in steady
state A. As 10 increases ~ increases until the marginal stability point B is
reached where d~/dlo --" 00. Further increase of 10 requires an abrupt jump
in ~ to point C if a steady state is to be maintained. Yet higher-power levels
again lead to a gradual increase of ~. Now consider reducing the radiation

20

Percent absorption
(proportional to
[N0 2
10

l)

Power of incident light (watts)

Fig. 7.9. Steady-state solutions of (7.42) and (7.43) for various choices of external temperature To. Note the hysteresis in the 230 K curve. [Adapted from C. L. Creel and J. Ross,
J. Chern. Phys. 65, 3779 (1976).]

228

Chap. 7

Nonstationary State Mechanisms

power of incident light

Fig. 7.10. Comparison of ideal (solid line) and observed (dashed line) hysteresis in the
NO,-N,04 reaction.

intensity. The system traces a path DCEFA since it is only at E that marginal
stability occurs. The net result is a hysteresis loop as 10 is varied. Above
240 K there is no hysteresis; ~ varies continuously with 10 , The set of
curves for different values of To look just like isotherms of a van der Waals
gas (turned on their side) above and below a critical point. Hysteresis
below the critical temperature has its analog in supercooling and superheating. (45)
Experiment confirms the predictions of Fig. 7.9. Above the critical
value To there is no hysteresis. Below To hysteresis is observed; however,
the jump from one branch to the other is not as abrupt as predicted. (46)
A comparison of observed and predicted hysteresis loops is shown in Fig.
7.10. Transition occurs before the marginal stability point is reached.
Under the experimental conditions there is sufficient inhomogeneity to
account for this effect. The temperature of the gas is probably not uniform
throughout the sample so that observation is of a set of systems undergoing
sequential transitions between the branches.
If uniformity were established one would still expect deviations from
the ideal hysteresis loop. Fluctuations about the steady state could induce
transitions from one branch to the other before the marginal stability point
is reached. Such behavior is analogous to nucleation of a supercooled
liquid; freezing then occurs abruptly before the limit of metastability is
reached. Referring to Fig. 7.9, one might wonder whether there could be
an analog to the Maxwell construction in the van der Waals fluid. While
a definitive answer is not yet available, the tentative answer is yes. Since
(45)

(46)

For extension of the thermodynamic analogy see A. Nitzan, P. Ortoleva, J. Deutch,

and J. Ross, J. Chern. Phys. 61, 1056 (1974).
C. L. Creel and J. Ross, J. Chern. Phys. 65, 3779 (1976).

229

Problems

power dissipation in the mixture depends on ~ and T, it is multiply valued

in the domain where there are multiple solutions. After taking proper account for heat flow into (by radiation) and out of (by conduction at the
walls) the system, the state of lower power loss could be found. In this
sense point F is more stable than point E; similarly point C is more stable
than point B. At some intermediate power level the states on each branch
should be equally stable, thus defining a Maxwell-like construction. (47)

Problems
7.1.

(a) From Fig. 7.1 it is clear that Tc is the temperature at which thermal
explosion is inevitable. Use (7.3), assume an Arrhenius form for k(T),
and show that Tc - To = RT'/Ea
(b) Compute Tc for a reaction where Ea = 300 kJ mol- l when the system
is in a thermostatted furnace at (a) 600 K, (b) 800 K, (c) 1000 K.
(c) Why does Tc - To increase as To increases?

7.2.

The seven elementary steps of (7.5) and (7.11) account for kinetics in the
H 2-0 2 system below the third explosion limit.
(a) Show that at the first limit ks = 2k2[02]/O + k./k.[H2D and thus,
as volume-to-surface ratio increases, ks = 2k2[02].
(b) At the second limit show that k 7[M] = 2k2/(1 + k./k.[H 2]) which,
for a large reaction vessel, reduces to k7 [M] = 2k 2.

7.3.

In a stoichiometric mixture of
the second explosion limit is
P2 (atm):
T(K):

H2

and O 2 the temperature dependence of

0.043

0.066

750

775

0.104
800

0.156
825

Estimate E B ; the quenching reaction (7.11) has a small negative activation

energy, EQ ,...., -7 kJ mol-l. To which of the branching reactions does the
computed EB correspond?
7.4.

To demonstrate that chemical oscillations are a far-from-equilibrium

phenomenon consider the mechanism (7.15) with the equilibrium constraints (7.16).
(a) Define displacement variables ~ = [X] - X o , 17 = [Y] - Yo, and
, = [Z] - Zo and use (7.17) to show that, near equilibrium,
~ = -(LlXo + k2YO)~ + k_ 2Y 0 17
iJ = k2YO~ - (L 2 Yo + ka)17 + La'
, = ka17 - La'

(47)

I. Procaccia and J. Ross, J. Chern. Phys. 67, 5558 (1977).

Chap. 7

230

Nonslationary State Mechanisms

(b) Assume Z = IZo], i.e., that product is always drawn off so that C = o.
Use the matrix method of Chapter 5 and show that the decay constants
are solutions of the equation
A' - [LlXo

+ (k-2 + k.)Yo + k.]A + k_lXO(k. + L

Yo)

+ k.k.Yo

= 0

(c) Show that both roots are positive and therefore that the system relaxes
to equilibrium without oscillation. (The roots of the equation A' - BA + C
= 0 satisfy the relation AlA. = C.)

+ 1')'/Yo = const,

7.5.

Show that when ~ and 1') are small (7.25) becomes

i.e., the ~-1') trajectory is an ellipse.

7.6.

Mechanistic details strongly influence the possibility of chemical oscillation.

Consider the autocatalytic mechanism
kl

A~X,

+ Y -..:.,., 2Y,

~'/Xo

ka
Y~Z

[A. J. Lotka, J. Phys. Chem. 14, 271 (1910)].

(a) Assume [A] is constant and formulate the rate laws. Show that in the
steady state Xo = k./k. and Yo = klAo/k .
(b) Define displacement variables ~ == [X] - Xo and 1') == [Y] - Yo and
show that close to the steady state
~ = -k.Yo~ - k.Xo1'),

(c) Show that the decay constants are

A = k.Yo[1 (1 - 4Xo/YO)l/]/2

and thus that the system exhibits damped oscillations as it approaches the
steady state if Yo < 4Xo, i.e., if 4k.' > klk.A o.
(d) Can you give a physical rationalization for the condition governing
the onset of damped oscillation?
7.7.

Spatial inhomogeneities couple with chemical relaxation. One set of possible

consequences is illustrated when the Lotka rate law (7.18) is coupled with
Fick's law (2.25)
[X] = klAo[X] - k.[X][y]

[Y]

= k.[XHY] -

k.[Y]

+ Dx 17'[X]

+ Dy V"[Y]

(a) Show that the linearized rate equations for the displacement functions

W, r) and 1')(1, r) are

(b) Assume a perturbation of wave vector q, displacement functions of the

form ~ = lX(t) cos q rand 1') = f3(t) cos q r, and show that

Problems

231

(c) Now show that the decay constants are

,1.

= (D" + Dy)q2/2 [(D" - D lI )'q'J4 -

w0 2

)l"

where Wo is the Lotka frequency (7.23).

(d) Do inhomogeneities in the Lotka system persist? What is the effect of
increasing q?
7.8.

A proposed model for the Belousov reaction is

k8

k.

B+X~2X+Z,

X+Y~P,

with the identifications X = HBrO., Y = Br-, Z = Ce(IV), and A = B =

BrOa- [R. J. Field and R. M. Noyes, J. Chern. Phys. 60, 1877 (1974)].
(a) Identify the steps in the Belousov system which correspond to reactions
1-5.
(b) Show that an overall stoichiometry is fA + 2B = fP + Q.
(c) From the discussion in Section 7.5 determine the permissible range of
values for the stoichiometric factor f
(d) Formulate the rate laws and show that in the steady state, assuming
[A] and [B] are constant: Zo = kaBoXo/ks, Yo = fkaBoXo/(k1Ao + k.Xo),
and Xo is the solution to the equation k.Yo + k.Xo = (1 + f)kaBo/2.
(e) If Ao = Bo show that Xo = kaBo {b-a-c+ [(a+b)2+ c2+2c(a-b)l/' }/2
where a = kl/k2ka, b = (1 + f)/2k., and c = f/k.k .
(f) Define displacement variables and show that the decay constants,
assuming small deviations from the steady state, are solutions of the cubic
equation
k8BO - k.Yo - 4k.Yo

0=

+ ,1.

-k 2 Y O

k1Ao - k.Xo
-k1Ao - k.Xo + A

kaBo

0
fkG
-k5 + A

(g) The critical condition for onset of oscillation occurs when ,1. = O.
Show that, assuming Ao = B o , the constraint is Xo ~ kaBoa/[k.(a - b)].
7.9.

Consider the effect of inhomogeneity in the N0 2-N.O. system discussed

in Section 7.7. The modified form of (7.43) is
~

= k+(2C -

n-

2k_~'

+ D J72~

(a) Ignore any temperature dependence of C, and 'Y. Define displacement

variables T = T" + T and ~ = ~" + x. Show that if the perturbation from
the steady state is small

+ ~ V', + Qx)V/Cv

(dolx/V

-k+x - 4k_~ssx

where k+'

==

yT

+ D V'x + (2C -

(dk+/dT)Tss, etc.

~ss)k+'T - U:sL'T

Chap. 7

232

Nonstationary State Mechanisms

2
(b) Show that k + , = E. i +lk + ( Tss )/ RTss,
etc.
(c) Define displacements of wave vector q: T

= J(t)

cos q r and x

get)

cos q r; show that J = -aJ+bg+eg and g = cJ-dg, where a =

'Y+ XQ2V/Cv, b = dollC v, c = (2C-~ss)k+!1E/RTs~, d = k+ +4L~ss+
Dq2, with!1E = Ea(+)-E a(-).

(d) Show that the decay constants are A = {(a + d) [(a -d)2+4bc]'/2}/2.
(e) The parameters a,b,c, and d are always positive. However, as Q<O
(the reaction is endothermic), e < O. Are there value of q which permit
either damped or undamped oscillation about the steady state? Is it
possible that }.._ < O? What is the physical significance of a negative }..?

What is the physical significance of a negative A?

7.10. An interesting ecological problem (with chemical analogs) occurs when two
similar predators compete for the same food.
(a) If the predator species Pi have mean lifetimes Ti and reproduce according
to the mechanism Pi
.
[Pd

k
+F~
2P i

1
Ti

[Pd

(i

1, 2) show that the rate laws are

+ ki[Pd[F],

1,2

(b) Now show that

k2 [1\] _ k, [F 2]
[P,J
[P,]

(~

~)

T2

T,

and thus that k2In[P,J - k,ln[P 2] = Kt + const, with K = k , /T 2-k 2/T , .

It is generally possible for both species to persist or will one become extinct?

General

Referel1c~s

Explosion and Ignition

P. G. Ashmore, in Photochemistry and Rearfion Kinetics, P. G. Ashmore, F. S. Dainton, and
T. M. Sugden, eds. (Cambridge: Call1bridge University Press, 1967), pp. 309-386.
N. N. Semenov, Some Problems in Chemical Kinetics and Reactivity, translation by M.
Boudart (Princeton, N.J.: Princeton University Press, 1959), Vol. 2.

Oscillatory and Multiple Steady-State Phenomena

B. F. Gray, in Reaction Kinetics, P. G. Ashmore, ed. (London: Chemical Society, 1975),
Vol. I, pp. 309-386.
R. J. Field and R. M. Noyes, Ace. Chem. Res. 10, 214 (1977).
R. M. Noyes and R. J. Field, Ace. Chern. Res. 10, 273 (1977).
A. Pacault, P. Hanusse, P. De Kepper, C. Vidal, and J. Boissonade, Ace. Chem. Res. 9,
438 (1976).
I. Prigogine, in Fast Reactions and Primary Processes in Chemical Kinetics, S. Claesson, ed.
(New York: Interscience, 1967), pp. 371-382.
J. Ross, Ber. Buns. Gesell. Phys. Chern. 80, 1112 (1976).

8
Single-Collision Chemistry
8.1. Introduction
Even if the rate law and the mechanism of a reaction have been determined there are still many unanswered questions of chemical interest.
The rate law only accounts for the overall phenomenology. The mechanism
provides a model for the basic steps in the reaction. The molecular properties which determine whether or not reaction occurs can only be inferred
from other chemical knowledge. Our previous discussion leads to only two
very general, and hardly informative, conditions for reaction: molecules
must interact (collide) and must have sufficient energy.
There are many other questions which can be posed. How is a theory
for computing a rate constant to be constructed? How may the orientational requirements for a reactive collision (such as the familiar Walden
inversion) be studied directly? Can reaction take place in a glancing collision or must molecules always collide head-on? Does reaction require the
formation of a long-lived complex or can reaction take place as molecules
fly past one another? If molecules have sufficient energy, will they invariably
react? What molecular properties promote catalysis?
These and other questions can only be investigated by isolating each
elementary step and studying the dynamics of the interacting molecules.
For gas-phase reactions, which usually proceed via binary collisions, the
chemistry of single collisions can be studied using molecular beams thus
providing new perspectives. For solution kinetics the problem is more
complicated since the basic reaction step is not the collision of two molecules
but the trapping of the reactants in a cage of solvent. Gas-phase studies
in which a molecule is surrounded by a few solvent molecules provide a
233

Chap. 8

234

Single-Collision Chemistry

way to isolate a solvated molecule from the bulk solvent. Some molecular
beam investigations of this type have been carried out.
The inherent distinction between gas and solution kinetics must be
part of any successful theory of chemical kinetics. In particular the mechanics of gas-phase reaction depend upon forces involving two, or at
most three, molecules at a time. In solution the complicating effects due to
interaction with a number of surrounding solvent molecules must also
be considered.
In this chapter we focus upon the dynamics of the reactive collision
process in the gas phase. To this end we first show how the basic measurable
parameter in collision chemistry, the reaction cross section, may be related
to its thermal counterpart, the rate constant. We then discuss the experimental approach to measurement of reaction cross section as well as a
naive model. The remainder of the chapter is devoted to the interpretation
of a number of experiments in terms of simple dynamical models for reactive
collisions. We make no attempt to develop a theory of reaction cross
sections.

8.2. Reaction Cross Section: Relation to the Rate Constant

To obtain a basis for computing the rate constant for an elementary
step consider the simple bimolecular gas-phase transformation
A

+ BC----'AB + C

The A and BC molecules are constantly colliding with one another, but
not all such collisions can lead to reaction. If A collides with the C end of
BC, AB bond formation is probably much less likely than if collision is
with the B end. If the reaction is endothermic, energy must be supplied
in order for reaction to take place. The necessary energy is supplied by
collision and then redistributed among the reactants. Thus, when molecules
collide, we can expect that there is a probability of reaction (less than I)
which depends upon their relative kinetic energy, their relative orientation,
and possibly other factors which have not yet been considered.
The rate of reaction of A atoms per unit volume can be expressed as

_ dnA _
dt -

number of A-BC COlliSions) ( fraction of SUCh)

per unit volume
collisions
(
e
with relative speed
which lead to
between e and e + de
reaction

foo d
0

(8.1)

Sec. 8.2

Reaction Cross Section: Relation to Rate Constant

235

where nA = NAIV is the concentration of A. The first factor in the integral

states that as more collisions occur, more molecules can react. The second
factor states that each collision need not lead to reaction; there is a reaction
probability, peE), which may vary with E, the relative kinetic energy of the
colliding molecules.o> From (2.21) we know that the total frequency of
A-BC collisions per unit volume is
(8.2)
where CA,Be is the mean relative speed of A and BC, and d is a hard-core
diameter. Naturally, real molecules are not hard spheres and d, as noted
in Chapter 2, is to be interpreted as a phenomenological parameter. There,
the collision cross section was related to transport coefficients. Here, by
introducing suitable modifications we shall be able to define a reaction
cross section.
The same arguments which led to (2.21) can be modified to show that
the collision frequency per unit volume of A and BC molecules with relative
speed between e and e + de is

where ft is the reduced mass. (2) The bracketed quantity is e times the probability that a pair of molecules will collide with a relative speed between
c and c + dc. Integrating (8.3) over all speeds leads immediately to (8.2).
Substituting (8.3) into (8.1) the reaction rate is

Since ifte2 is the relative kinetic energy,

with energy as a variable

E,

) 112

the integral can be rewritten

J'OoO p(E)Ee- dk7' dE ]

(8.4)

which is precisely the form of the rate law for the elementary step being
studied; the bracketed term is the rate constant. Here nd 2 is the collision
(1)
(2)

Molecular energy is denoted as E; molar energy is denoted as E.

For a derivation of (8.3) see W. J. Moore, Physical Chemistry, 4th ed. (Englewood Cliffs,
N. J.: Prentice-Hall, 1972), pp. 150-152.

Chap. 8

236

Single-Collision Chemistry

cross section which measures how close the centers of mass of A and
must pass in order for collision to occur. The factor p() accounts for
fraction of the collisions that lead to reaction. Both nd 2 and p( )
phenomenological parameters. Rather than treating each separately
define a cross section for reaction(3)

Be
the
are
we

and rewrite (8.4)

dnA

[8

- ( f t = nAnBC nf-t(kT)3

1/2

IOO

o Q()e-

dkT

(8.5)

The reaction cross section is equivalent to the collision cross section for
hypothetical spherical molecules which react with each collision; in the
next section we indicate how Q() can be measured directly. The rate
constant is thus identified from (8.5)
k(T) = [

8
nf-t(kT)3

]112 IOO Q()e0

dkT

(8.6)

Since concentrations, in the rate law (8.5), are measured in molecules

cm-3, k(T) has units of cm 3 sec- 1 molecule-I. The task of calculating a
binary rate constant is thus reduced to the problem of measuring the reaction
cross section, Q().
The parameters in the Arrhenius expression (4.42) can be related to
the rate constant determined from the cross section. The activation energy
is defined as the negative slope of In k(T) vs. (RT)-1. This indicates Ea
is to be computed as
dInk
(8.7)
Ea = -R d(1/T)
In general, values of Ea calculated from (8.7) are temperature dependent
(Problem 8.1). There is, in fact, only one functional form of Q( ) that
yields the Arrhenius form (4.42) (Problem 8.17).
(3)

The d that measures the reaction cross section may be substantially different from a d
determined from gas-phase transport experiments. Ion-molecule and alkali metalhalogen reactions often have cross sections substantially larger than what would be
expected from analysis of transport data for the corresponding species. As we have defined Q(E), it is a function of molecular energy; however, experimental data usually
present Q(E), a function of molar energy. At 300 K the mean collision per colliding pair
is ~6 x 10- 21 J (equivalent to 0.04 eV or, in molar units, to 4 kJ mol-I).

Sec. 8.3

Scattering Measurements-Molecular Beams

237

8.3. Scattering Measurements-Molecular Beams

The study of individual molecular collisions is a topic of great current
interest. The experimental methods are variants of the model experiment
outlined in Section 2.2 to illustrate the idea of mean free path. While that
experiment determines a total cross section for loss of reactant, with more
sophisticated apparatus the nature of the product can be determined,
i.e., partial cross sections for elastic, inelastic, and reactive events can be
distinguished. Let us reconsider the reaction
A

+ BC ---' AB + C

To measure the reaction cross section as a function of relative kinetic

energy requires forming beams of A molecules and BC molecules with
specified velocities. The beams are oriented so that they collide; collision
products are monitored as a function of energy and orientation. A representative apparatus is sketched in Fig. 8.1.
Beam experiments are rather difficult to carry out since they require
intense molecular beams of particles with a well-defined velocity; the
technical problems are greatest for neutral species. (4) To form an ionic
beam electric fields are used to accelerate the ions to the desired speed.
To form a beam of neutral molecules various techniques have been used.
The most intense beams are formed when the gas is driven, under high
pressure, through a small nozzle; the gas emerges at supersonic speeds.
The result, on the molecular scale, is similar to running water through a
hose with a nozzle. As the nozzle radius diminishes (within limits) the
stream of water emerges at higher speeds; all regions in the stream move
with the same speed.
In a molecular beam experiment the rate of product formation is
measured as a function of the energy of the products and the direction in
which they are scattered. The immediate problem is to interpret these
data and deduce the reaction cross section. Consider crossed beams such
as those of Fig. 8.1. The beams have fluxes IA and IBC; they collide with a
relative energy to for which the reaction cross section is Q(tO). Reaction
occurs in the small collision volume Veon . As the beam fluxes are kept
constant, the number of BC molecules in the collision volume is also a
constant, NBC. The quantity IAQ( to) is the number of A molecules which
(4)

For a description of the apparatus necessary for production, detection, and analysis
see J. M. Parson and Y. T. Lee, J. Chern. Phys. 56,4658 (1972); Y. T. Lee, J. D. McDonald, P. R. Le Breton, and D. R. Herschbach, Rev. Sci. Insfr. 40,1402 (1969).

238

Chap. 8

..
Colhsion regIon
Source of A

Single-Collision Chemistry
Detector
~

of volume Vcoll

/;/;;

/V

' "\
/::(~~attered
r---L.~~~~i~~li~ 19 :lIision products
Beam of
intensity IA

---""_--'

Fig. 8.1. Schematic diagram of a molecular beam experiment. The detector is movable; it
subtends the solid angle dQ = sin (J dO drp.

react per unit time per Be molecule in the collision volume so that the total
rate of product formation is
d[Prod]
dt

[hQ(E)]NBC

(8.8)

To compute the number of BC molecules in the collision zone requires

adapting the molecular flux calculations of Chapters 1 and 2. From the
perspective of a viewer in the collision volume, A and BC molecules approach one another with the relative speed Cre!. The larger the relative speed,
the fewer the molecules that are in the collision zone at any instant.
In Section 1.5 we showed that, for particles moving in the +x direction,
the flux of molecules with x component of velocity between u and u + du is
G(u) du

N
VuexCu) du

Here the fraction of molecules in the specified velocity domain, the analog
to exCu) du, is 1 since the beam is velocity-selected. The approach speed
is now Crel> not u, and the beam intensity is [BC, not G(u) duo Thus, the
flux is

from which, with (8.8), we obtain

d[Prod]
dt

II

= ~Q(E)[BCh
e re1

(8.9)

Sec. 8.3

Scattering Measurements-Molecular Beams

239

The result is symmetric in A and Be as is mandatory. It is somewhat similar

to a binary rate law with beam flux corresponding to concentration and
the quantity Q(E)/Crel corresponding to the rate constant. The rate constant
provides a measure of reaction probability for molecules at a specific
temperature while the cross section (or equivalently Q(E)/Crel) is a measure
of reaction probability for molecules with specific relative kinetic energy.
The units of Q/c, length X time, do not immediately suggest its similarity
to a rate constant. However, the differences arise because fluxes, not concentrations, appear in the rate law (8.9). The parallel is, of course, implicit in
the expression (8.6). The beam intensities, relative speed, and collision
volume are the experimental parameters. The rate of product formation is
measured from which, with (8.9), Q(E) can be determined.
The detector shown in Fig. 8.1 may be moved to measure the rate
of product formation as a function of position. At an orientation specified
by e and cp the detector subtends a solid angle dQ = sin e de dcp. Generalizing (8.9) to account for. angular variations in the rate of product formation
we have
(8.10)

where {d[Prod]/dt}e.'P is the rate at which product arrives at the detector.

The quantity fJ2Q/oe ocp is the differential cross section for molecules with
collision energy E.
The detector must be able to distinguish different products as well
as determine the amount of each that is formed. A mass spectrometer is
generally used for this task. When coupled with a time-of-flight analyzer
it is possible to determine the distribution of product velocities as well.
Such data provide information about the dynamics of a reactive collision
(Section 8.7).
An even more detailed picture can be constructed if the reactant
molecules can be prepared in specific excited states. For the example
considered, A may be formed in a variety of electronic states and Be may
be formed with a specific amount of electronic, vibrational, and rotational
energy. If i and j denote the internal states of A and Be, respectively, the
reaction cross section is found to be a function of both i and j.
Thus Q(E), as it is used in (8.5) and (8.6), is really a thermal average
since the population of the internal states of the reactants is temperature
dependent. A more complete specification requires defining the reaction
cross section for molecules in particular internal states colliding with a
specified collision energy, Q(E, i, j). The mean cross section in a thermally

240

Chap. 8

Single-Collision Chemistry

equilibrated system is a function of temperature and energy; it is

Q(E; T)

L L fA(i)(T)f~8(T)Q(E, i,j)
i

(8.11 )

where fA(i)(T) is the fraction of A molecules in the internal state i at temperature T. From statistical thermodynamics this fraction is rfJ<i/qA,
where qA is the partition function for the internal degrees of freedom of
A. (5) With this result and its analog for the fraction of BC molecules, the
average cross section can be computed from (8.11); it is this average which
is substituted in (8.6) to calculate a rate constant.
If the colliding molecules can produce a variety of collision products,
there are separate reaction cross sections for each chemical transformation.
In the terminology of the field the individual pathways are called channels.
A channel is open if the total energy available, defined as the energy of the
separated reactants plus the collision energy, is greater than that of the
separated products. Even though a channel is open there may be no detectable product yield if the reaction cross section is sufficiently small. To
each channel there corresponds a rate constant for reaction in a system at
thermal equilibrium to be calculated using (8.11) and (8.6); the activation
energy is found from (8.7).

8.4. Reaction Cross Section: Hard-Sphere Model

Consider a particularly simple model of chemical reaction. The molecules are imagined to be hard spheres of mean diameter (J and the reaction
has an energy barrier of height Eo. For reaction to occur two conditions
must be met: the molecules must collide and they must have sufficient
collision energy to overcome the barrier.
Case A: All collision energy available for surmounting barrier. In these
circumstances the energy dependence of the reaction cross section is particularly simple
E

<

Eo

E> Eo

(8.12)

Even at this level of simplification the model cannot be correct. Not all
the collision energy is available to overcome the barrier. The fraction of
(0)

A good elementary discussion of statistical thermodynamics is given by T. L. Hill,

Introduction to Statistical Thermodynamics (Reading, Mass.: Addison-Wesley, 1960).
See especially Chapter 3.

Sec. 8.4

Hard-Sphere Model

241
relative speed c

,.-- "'
'
, --l.
I

" '--

/
/

T
projection of c along line of
centers

=c

cos

e = c j 1- b2/(T2

Fig. 8.2. Diagram for determining line-of-centers velocity for hard spheres which do not
collide head-on; the impact parameter is b.

energy accessible depends upon the impact parameter, which, as illustrated

in Fig. 8.2, is th-e quantitative measure of the extent to which the collision
is off center. As b increases from 0 to a the character of the collision changes
from head-on to glancing. If b > a no collision takes place.
Case B: Line-ol-centers collision energy available for surmounting
barrier. For the collision geometry of Fig. 8.2 the velocity directed along
the line of centers is c(l - b2 j( 2 )1/2. The energy available for overcoming
the barrier is therefore only

Since t(b) > to for reaction to occur, we can solve for b and find that, for
reactive collisions, b is less than a critical value be(t)

Collisions which occur with impact parameter b have their loci on a cylinder
of radius b. Thus, the collision cross section for an impact parameter
between band b + db is the annular area 2nb db. The fraction of collisions
which occur with this range of impact parameters is therefore 2nb db
divided by the total collision cross section na 2 Since reaction only occurs
if b < bit), the fraction of reactive collisions is

which when combined with (8.1) yields the result

(8.13 )

Chap. 8

242

Single-Collision Chemistry

For E < Eo no reaction occurs; as E increases the cross section increases

reaching its limiting value of n()"2 when E ~ Eo.
Both models for the reactive cross sections are grossly oversimplified.
They depend on substituting a hard-sphere potential for a realistic intermolecular potential. The inadequacy of this replacement is apparent from
Fig. 2.9 which compares particle trajectories based upon the hard-sphere
and Lennard-Jones potentials.
Since the models are mathematically tractable it is instructive to deduce
a form for the rate constant. We only consider the (slightly) more realistic
Case B and combine (8.13) with (8.6) to yield, with f3 = l/kT,
k(T) = (8f33/n f-t)1/2
=

n~f-t

foo a-li <n()"2(l

- EO/E) dE

<0

2n ()"2 exp( -f3Eo)

(8.14)

In SI, the units of k(T) are m 3 molecule- 1 sec-I; to convert to the more
commonly used units liter mole- 1 sec- 1 multiply (8.14) by 103 No. The
activation energy can be computed from (8.7); it is
(8.15)
which, as expected, varies slowly with temperature. The activation and
threshold energies are not the same, which is generally the case. However,
the change in activation energy predicted by this model is ,....,2 kJ mol-1
for a temperature change of 500 K. Experimental determinations of Ea
are rarely both sufficiently accurate and made over a large-enough range of
temperature to render such variation observable.
The preexponential factor in the line-of-centers model is proportional
to T1!2 which is a direct consequence of the cross-section function (8.13).
With a different, but less readily rationalized, model for Q(E), a constant
preexponential factor is predicted (Problem 8.18).
Both Eo and ()" are parameters of the model. It is not possible to determine Eo from the properties of the individual reactants. However, ()" can be
estimated from the molecular diameters of isolated reactants using transport,
spectroscopic, or X-ray diffraction data. Thus, the preexponential factor in the
Arrhenius expression can be computed. From (4.42), (8.14), and (8.15) we find
8e )112
A= ( - - n()"2
nf3f-t

(8.16)

which is compared with experimental results in Table 8.1. As expected,

CI + H.

0.80
1.6
12
0.035
9.4
0.058
400
200
0.35
10
7
14
85
12

Experiment

900
920
1400
480
340
330

47
59
110
47
59
26
65
1070

Calculated
from (8.16)

~
a

~
w

'r~

..,~

:c...

00

Reference

a D. R. Herschbach, H. S. Johnston, K. S. Pitzer, and R. E. Powell, J. Chern. Phys. 25, 736 (1956), where original references are given.
b V. N. Kondratiev, Rate Constants of Gas Phase Reactions, translation by L. J. Hoitschlag (Washington: National Bureau of Standards, 1972),
where original references are given.

CCls + CI.
HCI + CI

H. + N.H.
NH. + NH.
HC] + CCl.
2Br+Ar

3.5
2
8
29
16
138
84
18

CI. + O.
CO + CI.
HI +1

10.5
43.5
132
36
102.5

NO. + O.
N0 2F + F
NO + CO 2
FCIO. + F
2NO + CI 2

Activation energy
(kJ mol-i)

H + CCl.
Br2 + Ar
CI + CCl,

H + 12
H + N.H,
H + N,H,

NO +0 8
N02 + F2
NO. + CO
F. + CIO.
2NOCI
2CIO
COCl + CI

Reaction

Ftequency factor x 10-12

(emS mol~i sec-i)

Table 8.1. Kinetic Parameters jor Some Bimolecular Reactions

244

Chap. 8

Single-Collision Chemistry

the model is poor. Calculated values of A all fall in the fiftyfold range,
3 X 1013-1.5 X 1015 cm3 mole-1 sec-I, in disagreement with experiment, a
result that cannot be ascribed to errors in the estimate of cr. A twofold
error would only change A by a factor of 4 and the actual discrepancies
are usually much larger. In one case (H + N 2H 4 ), two reaction channels
with significantly different A-factors exist; collision theory makes little
distinction between them. In all but one case the calculated value of A
is much larger than that found experimentally, indicating that (8.l3) generally overestimates reaction cross sections.
Some of the defects of the hard-sphere model for the reaction cross
section are apparent. It assumes that all the line-of-centers kinetic energy
can be used for overcoming the threshold. It does not account for the
dependence of Q(E, i, j) on the internal states of the reactants. It does not
consider any effect due to molecular structure or to the details of the collision process. Improvement requires either direct experimental measurement of Q(E, i, j) or equivalently, a calculation which takes into account
the interaction potential between colliding molecules in specific internal states.

8.5. Reaction Cross Section: Ion-Molecule Systems

The hard-sphere model, as described in Chapter 2, only accounted in
a qualitative fashion for the mass, momentum, and energy transport that
underlie the phenomena of diffusion, viscous flow, and heat conduction.
For systems other than monatomic gases, the model was of limited utility.
Thus, its failure to predict reasonable values of the Arrhenius A-factor is
to be expected.
An understanding of reaction cross sections is only possible by considering some aspects of the dynamics of molecular interaction. (6) Naturally
not all molecular collisions lead to reaction. There is always the possibility
(often quite substantial) of inelastic collision. However, we shall only
discuss reactive events.
Simplest to describe are those systems in which there is no barrier to
reaction. Exoergic(7) ion-molecule reactions fall into this category. Due both
(5)

(7)

For an extension of the discussion in this and the next section see R. Wolfgang,
Ace. Chern. Res. 2, 248 (1969).
This term denotes that electronic energy is released in the reaction, i.e., the products are
more stable than the reactants. It is the equivalent, in discussions of chemical dynamics,
to the familiar word exothermic which is used to indicate that heat is evolved in a
reaction at constant temperature.

Sec. 8.5

Ion-Molecule Systems

245

0.5

0.2
0.1
0.05
Q(E) (nm 2 )
0.02
0.01
0.005

/
/
~

r
/J

co+ +

0.002 5 10 20 50 100 200 1000 2000

Ecoll (kJ mol-I)

Fig. 8.3. Reaction cross section, Q(E), for two ion-molecule systems as a function of collision energy. [Data for the Ar+ + D2 reaction from A. J. Masson as cited by M. J. Henchman, in Ion-Molecule Reactions, J. L. Franklin, ed. (New York: Plenum Press, 1972),
p. 180. Data for the C+ + D2 reaction from E. Lindemann, L. C. Frees, R. W. Rozett,
and W. S. Koski, J. Chem. Phys. 56, 1003 (1971).]

to electrostatic forces and to a tendency toward chemical binding, there

are long-range attractive interactions between ions and molecules. Reactants
with zero initial collision energy are accelerated toward each other.
In Fig. 8.3 the energy dependence of Q(E) for the
(8.17)
reaction is shown. (SJ The data indicate that Q(E), for this barrierless reaction, drops steadily as the collision energy of the reactants increases, an
observation that directly contradicts the prediction of the hard-sphere
model (8.13).
To understand the behavior of Q(E) requires considering the effect
of the long-range attractive forces on the reactants' trajectories. This is
(8)

In the remainder of this chapter we conform to the experimental convention and

present data as a function of molar energy, E.

246

Chap. 8

Single-Collision Chemistry

Fig. 8.4. Effect of collision energy and impact parameter on trajectories when there are
long-range attractive forces and no threshold for reaction. (a) Variation of energy at constant impact parameter; (b) variation of impact parameter at constant collision energy.

illustrated in Fig. 8.4. When the impact parameter is large, only reactants
with low collision energy are deflected sufficiently to spiral in toward one
another and react. As the collision energy increases, the relative speed
increases and the trajectory is less curved. As E increases further, there is
less deflection and the ion and molecule no longer pass close enough for
reaction to take place. The qualitative picture is the same as that illustrated
in Fig. 2.9 for interaction via a Lennard-Jones potential; at large impact
parameters significant deflection only occurs for small collision energies.
The effect of varying E at constant b is shown in Fig. 8.4a. There is a critical
energy Ee(b) above which reaction cannot occur; as b increases Ee must
drop. The effect of varying b at constant E is shown in Fig. 8.4b. It is
complementary. At any E there is a critical impact parameter be(E) above
which no reaction can take place. As E increases, be must drop until it
becomes comparable with the "hard-core" distance (j which is determined
by the short-range repulsive forces between the reactant species. At this
and higher energies the repulsive forces are presumably most important
for determining reaction dynamics.
The existence of the critical impact parameter is often discussed in
terms of the so-called centrifugal barrier.(9) As particles spiral in toward
one another, the energy available for radial motion is less than the initial
collision energy. Conservation of angular momentum of two colliding
particles requires that there be nonzero centrifugal kinetic energy (that
part of the kinetic energy which accounts for motion perpendicular to the
line of centers). As the interparticle separation decreases so does the moment
of inertia; the centrifugal kinetic energy increases to conserve the angular
momentum.
For a pair with reduced mass fl, collision energy Eo, impact parameter
(9)

For a quantitative treatment of the relevant collision dynamics see R. E. Weston, Jr.,
and H. A. Schwartz, Chemical Kinetics (Englewood Cliffs, N.J.: Prentice-Hall, 1972),
Chapter 3.

Sec. 8.5

Ion-Molecule Systems

247

b, and interaction potential VCR), conservation of energy requires that,

at any interparticle separaJion R,

(8.18)
Here E ob2jR2 is the centrifugal kinetic energy. The radial speed dRjdt
is therefore governed by an effective potential E ob 2 jR2 + VCR). The first
term (the centrifugal barrier) is always positive and the second term is
negative for R large enough for particles to attract one another. Again
referring to Fig. 8.4b, which illustrates the effect of varying b at fixed Eo,
we note that as b increases, the distance of closest approach Ro (where
dRjdt = 0) also increases. When b is large, particles do not "collide";
instead they bounce off the centrifugal barrier and reaction, involving
approach to within distances comparable to chemical bond lengths, may
not occur. As b decreases, so does the centrifugal barrier; finally particles
may surmount it and penetrate to the repulsive wall. The critical impact
parameter bcCEo) is the maximum value of b for which such penetration
occurs.
For the Ar+ + D2 reaction the cross section is very large at low collision energies. Even for b < be(E), not all collisions are reactive; thus
for a given energy an upper bound to Q(E) is nb e2(E). The maximum
measured value of Q is '"'-'0.5 nm 2, which occurs at the lowest energy,
'"'-' 10 kJ mol- 1 . Thus, at this energy be is greater than 0.4 nm while the hardcore diameter, from Table 2.6, is '"'-'0.3 nm. The observed cross sections
are consistent with a model in which long-range attractive forces dominate
dynamics at low collision energy. Both the data and the model illustrate
an important generalization: the effective range of interaction decreases
as the collision energy increases.

As the collision energy increases further Q(E) continues to drop

steadily. Reaction cross sections of ;S0.03 nm correspond to values of
b ;S 0.1 nm, which is much less than the "hard-core" diameter of 0.3 nm.
Reaction cross sections decrease at high energy for an obvious reason. If
too much energy is transferred to the product it dissociates. Since the
strongest chemical bonds are '"'-' 103 kJ mol -l, at larger collision energies
we may always expect fragmentation to be significant. (10) There is an important corollary. Dissociation only occurs if the collision energy of the
reactants is converted into vibrational energy of the products. Thus the
energy disposition between translation and vibration in the products depends
(10)

Naturally if the products which are formed have lower bond energies, dissociation
occurs at smaller collision energies.

Chap. 8

248

Single-Collision Chemistry

upon the initial energy. We shall investigate this point when we study a
model for the mechanics of reaction dynamics in Chapter 10.
Thus far we have considered the barrierless, exoergic Ar+ + D2
reaction. To illustrate the effect of a potential barrier the reaction cross
section for
is included in Fig. 8.3. The threshold for reaction is ,-...,40 kJ mol-l. The
cross section is zero if the collision energy is less than the threshold value;
it then rises rapidly until the inevitable decrease occurs at high energy.
The low-energy cross sections for reactions with barriers are always significantly less than for those without barriers. At higher energies they are
comparable. At yet higher energies the cross section is determined by the
repulsive, short-range forces which are similar for all molecules.

8.6. Reaction Cross Section: Atom-Molecule Systems

Most gas-phase reactions of chemical interest involve neutral species.
Because of the difficulties of producing monoenergetic beams of neutral
molecules few systematic studies of cross section vs. energy have been
carried out when both reactants are neutral. The available evidence, both
experimental and theoretical, indicates that the general features of Q(E)
are much the same whether or not the reactants are charged. The important
consideration is whether there is a barrier.
Although cross sections have been measured for many reactions, few
have been studied over a range of energies. One for which data are available
is the reaction
CH3I

+ K----" KI + CH 3,

LlE= -IOOkJmol- l

The cross section is plotted in Fig. 8.5. The threshold is small (;5 3 kJ)
and the cross sections large, comparable to those in the low-energy domain
of the Ar+ + D2 system (Fig. 8.3). The studies of the K + CH3I reaction
were not continued to energies where there is a precipitous drop in Q(E).
Even larger low-energy cross sections have been observed. Studies
of reactions of alkali metals with halogens are even more dramatic, as
long as there is no barrier. For the K + Br2 system a cross section of
,-...,2 nm 2 has been measured(ll); assuming each collision to be reactive,
(11)

J. H. Birely, R. R. Herm, K: R. Wilson, and D. R. Herschbach, J. Chern. Phys. 47,

993 (1967).

Sec. 8.6

Atom-Molecule Systems
1.0

0.3

0.1

249

0.03
Q(E)(nm2)
0.01

+ O2

TO

T + O2

+0

~T +0 + 0

0.003
0.001
0.0003

Exchange
Threshold

0.0001
5

10

20

50

Dissociation
Threshold

100 200

500 1000 2000 5000

Ecoll(kJ mol-I)
Fig. 8.5. Reaction cross section, Q(E), for three atom-molecule reactions as a function of
collision energy. [Data for K + CRaI reaction from M. E. Gersh and R. B. Bernstein,
J. Chern. Phys. 55, 4661 (1971). Data for T + D. reactions from M. Karplus, R. N. Porter, and R. D. Sharma, J. Chern. Phys. 45, 3871 (1966).]

reaction occurs at impact parameters of ;;:;0.8 nm, far larger than any
sensible "hard-core" distance. A reasonable chemical explanation of such
enormous cross sections may be based on the low ionization potentials
of alkali metals and the high electron affinities of halogens. Because of
this complementary relationship at distances of ;S 1 nm electron transfer
is highly probable, producing two charged species which are strongly
attracted to each other; reaction is then very likely. The mechanism has
been dubbed "harpooning" because of the following picturesque analogy.
At a fairly large distance the alkali metal "tosses out" an electron which
acts as a harpoon and attaches itself to the halogen; the ionic attraction
then provides the line for the whaler (alkali metal) to haul in the whale
(halogen). (12)
(1')

As is the case in small-boat whaling the distinction between the hauler and the hauled
is fuzzy. In addition, if the whale is powerful enough (large collision energies) it can
break the line and escape. In that event the whale carries the harpoon with it; here the
electron is transferred back (unless the collision energies are huge).

Chap. 8

250

Single-Collision Chemistry

The available experimental data on reactions with a significant threshold

suggest that cross sections are ;S0.05 nm 2, as found in the C+ + D2 system.
A theoretical study of the tritium-deuterium reaction
T

+D

----'-TD

+ D,

11E = -1.6 kJ mol- 1

is available; the results are included in Fig. 8.5. This cross-section function,
when used to compute rate constants via (8.6), led to results in good agreement with experimentally determined Arrhenius parameters. (13) The threshold is ,.....,,30 kJ mol-1 and the qualitative similarity with the C+ + D2
system (Fig. 8.3) is apparent. The decomposition cross section for the
reaction
11E = +437 kJ mol- 1
T + D2 -, T + D + D,
was also computed and is included in Fig. 8.5. Note that it does not show
any significant decrease at high energy, which is hardly surprising. If the
collision energy is large enough fragmentation and/or ionization is the
almost-certain outcome of any collision.
In the examples considered so far collision led to one of three results:
inelastic energy transfer, reaction, or decomposition. In more complicated
systems a bewildering variety of reactions becomes possible at large collision energy.u 4 ) The reactions of hydrogen atoms with propane provide
another good example. (15) These experiments were done using tritium
produced by neutron bombardment of 3He via the nuclear reaction
3He

+ n----'-p + T

The tritium, being radioactive, provided a convenient analytical label.

At low collision energies under conditions where the tritium reacts, only
hydrogen abstraction occurs(16):

Eo;S 50 kJ mol- 1
(13)

(14)

(15)
(16)

(8.19)

M. Karplus, R. N. Porter, and R. D. Sharma, J. Chern. Phys. 43, 3259 (1965); 45,
3871 (1966).
The K + CHaI system shows such behavior at higher collision energies. Possible
reactions have been cataloged by M. E. Gersh and R. B. Bernstein, J. Chern. Phys.
56,6131 (1972). The channel forming K+ + CHaI- followed by dissociation of the negative ion has been observed by A. M. C. Moutinho, J. A. Aten, and J. Los, Chern. Phys.
5, 84 (1974).
A. H. Rosenberg and R. Wolfgang, J. Chern. Phys. 41, 2159 (1964).
For a discussion of hot-atom chemistry see R. Wolfgang, Sci. Arner. 214(1), 82 (1966).

Sec. 8.6

251

Atom-Mdecule Systems

Table 8.2. Relative Rate Constants for the Abstraction, Replacement, and Double
Replacement Reactions in the T + C3HS Systetrt'
Temperature

(K)

Abstraction
(8.19)

Replacement
(8.20)

Double replacement
(8.21)

300

8 x lO- s

7 X 10-16

6x lO- N

600

6 X 10-4

8 X 10-8

I X 10-86

1000

2 X 10-'

1 X 10-'

4x 10-.2

Estimated as proportional to {3 e-P< at threshold.

As the energy range is increased replacement also becomes important:

T

+ CaHs ----' CaH7T + H }

----' C2H5T + CH 3 Eo"""" 100 kJ mol-1
----' CHaT + C 2H 5

(8.20)

At still higher energies double replacement reactions, of which

Eo'" 450 kJ mol- 1

(8.21 )

is one example, occur. The collision energy is then so large that there is
partial fragmentation. The generalization is obvious-as collision energy
increases more reaction channels are open. Reaction (8.21) need not occur
in a single step. If vibration ally excited C 2H 5T or CHaT, formed by replacement as in (8.20), has sufficient vibrational energy, it will undergo secondary
decomposition.
At thermal energies only abstraction is observable. From (8.6) the
quantity PEo exp( - pEo) provides a measure of relative rate constants if
we assume that the shapes of the cross-section functions near threshold
are not too vastly different. Values of PEo exp( -PEo) are given in Table
8.2 for the reactions (8.19)-(8.21). Only at 1000 K might replacement be
detectable; double replacement can never be an important thermal pathway.
Thus collision chemistry permits the study of reactions inaccessible by means
of ordinary thermal kinetic techniques.

252

Chap. 8

Single-Collision Chemistry

8.7. Product Distribution Analysis

By studying the scattering pattern (the velocity and angular distribution) of the products formed in a crossed molecular beam experiment,
much can be deduced about mechanistic details of an elementary reaction.
The effect of varying impact parameter, collision geometry, and internal
states of the reactants can all be investigated.
We shall consider three systems which exemplify distinct types of
collision dynamics. Two are direct reactions; collision, reaction, and
separation take place in a time comparable to molecular vibration periods,
about 10-13 sec. The reaction
Ar+

+ D2 ----' ArD+ + D

which we discussed as a prototype barrierless ion-molecule system, occurs at large impact parameter and proceeds via a direct, forward-peaked
mechanism. The reaction
D

+ Cl 2----' DCl + Cl

is favored by small impact parameters; it proceeds via a rebound mechanism. The third example

o + Br 2 ----' BrO + Br
is an indirect reaction; a Br 20 complex forms and survives for at least a
full rotational period (;:::;5 x 10-12 sec) before decomposing. The scattering
pattern is quite different in each case and provides a distinctive signature.
Molecular beam experiments are performed in a laboratory frame of
reference but the chemically interesting events take place with respect
to the center of mass of the colliding species. In order to interpret the
data, differential cross sections measured in the laboratory (LAB) coordinate system must be transformed to reflect events which took place in
the center-of-mass (CM) coordinate system. To effect this transformation
the invariant motion of the center of mass must be subtracted from the
scattering data obtained in the LAB system. A simple example which
illustrates the difference between LAB and CM kinematics is shown, for an
elastic collision, in Fig. 8.6. In CM the particles always move directly toward
one another before interaction and directly apart afterwards. This condition
is a consequence of momentum conservation in a system with a stationary
center of mass. The interaction causes each particle to be deflected through

Sec. 8.7

Product Distribution Analysis

LAB

--VI

before call ision

{v,

eM

,\2

Vc

253

~
VI

(a)

.,

.....

Uz

\.,,'"
.ul

(b)

Fig. 8.6. Comparison of collision kinematics in LAB and CM coordinate systems. Unprimed
and primed vectors refer to velocities before and after collision. In this example
M, = 2.7M2' (a) In LAB the center of mass, marked., moves with constant velocity
Vc throughout the collision. The collision geometry is for crossed beams at 90. (b) In CM
the center of mass is stationary. Conservation of momentum requires that relative motion
be linear before and after interaction. Each particle is deflected through an angle cpo

an angle rp. In LAB the situation is vastly different. The heavy partide,
which carries most of the center-of-mass momentum, is only slightly
deflected while the trajectory of the lighter particle is more significantly
perturbed. Similar diagrams can be constructed if collisions are inelastic
or reactive. The relationship between LAB and eM is more complicated
but, as long as the internal states of both reactants and products are known,
the correspondence can be made. (17)
A number of experimental limitations introduce ambiguities into the
determination of eM scattering patterns from LAB data. Molecules in
the incident beams have a range of velocities. There is therefore a distribution in initial center-of-mass and relative velocities. Thus a particular
deflection angle in LAB is correlated with a range of deflection angles
in eM. The uncertainty can be reduced by improving velocity selection in
the beams. Another source of ambiguity is that, in general, the scattering
pattern of only one collision product is determined which, unless the internal
energy of the product is measured, is insufficient to precisely characterize
the kinematic problem. A final difficulty is that molecules in the incident
beams are usually produced in a range of rotational states, which further
blurs the construction of the eM scattering pattern from LAB data.

(17)

This problem is analyzed by R. D. Levine and R. B. Bernstein, Molecular Reaction Dy-

namics (Oxford: Clarendon Press, 1974), pp. 60-63, 192-196.

Chap. 8

254

Single-Collision Chemistry

1.0
Relative
Intensity 0.5
O.OL-....::::..--~-...1---===O"'_-4

+5

+10

LAB angle

Ar+

Fig. 8.7. The angular distribution of ArD+ scattered in LAB due to the reaction
Ar+ + D2 ~ ArD+ + D. The Ar+ and D2 are incident at 0" and 90. The collision energy
is 7.8 kJ mol- 1 . [Adapted from Z. Herman, J. Kerstetter, T. Rose, and R. Wolfgang, Disc.
Faraday Soc. 44, 123 (1967).]

8.8. Direct, Forward-Peaked-The Ar+

+ D2

System

The reaction
Ar+

+ D2~ArD+ + D

has been extensively studied using molecular beams. The scattering pattern
has been deduced by monitoring the ionic product ArD+. Measurements
have been carried out for collision energies from 7.8 to 880 kJ mol-I, an
energy range over which the reaction cross section drops by a factor of
about 50 (Fig. 8.3).(18)
The LAB angular distribution of ArD+ at the lowest collision energy
is shown in Fig. 8.7. All the ArD+ is scattered at angles close to the direction
of the incident Ar+ beam, which is expected since the center-of-mass momentum does not differ greatly from the momentum of the heavy incident
(Ar+) or scattered (ArD+) particles. By transforming these data to eM a
portrait of the reaction dynamics can be constructed.
The eM angular distributions of ArD+ determined for collision energies
of 7.8 and 440 kJ mol- 1 are compared in Fig. 8.8. In both cases the product
is predominantly forward scattered, i.e., it is moving in roughly the same
direction as the Ar+ incident beam, which reflects the chemical process
since the scattering pattern refers to eM, not LAB. The results reflect
(18)

Z. Herman, J. Kerstetter, T. Rose, and R. Wolfgang, Disc. Faraday Soc. 44, 123 (1967);
M. Chiang, E. A. Gisiason, B. H. Mahan, C. W. Tsao, and A. S. Werner, J. Chern.
Phys. 52, 2698 (1970).

Sec. 8.8

Direct, Forward-Peaked-Ar+

+ Da

255

System

center of moss

90
I
I

center of moss

I
I

I
I

90
(a)

(b)

Fig. 8.8. Polar diagrams showing angular distribution in CM of ArD+ formed in the
Ar+ + D. reaction at collision energies of (a) 7.8 kJ mol- l and (b) 440 kJ mol-l. The radius vector is proportional to ArD+ intensity. The diagrams are not drawn to the same scale.
[Derived from: (a) Z. Herman, J. Kerstetter, T. Rose, and R. Wolfgang, Disc. Faraday
Soc. 44, 123 (1967); (b) M. Chiang, E. A. Gislason, B. H. Mahan, C. W. Tsao, and A.
S. Werner, J. Chern. Phys. 52, 2698 (1970).)

collisions that occurred for a range of impact parameters since experimental

control of the distance of closest approach for molecules in the crossed
beams is, of course, impossible.
The predominant forward scatter and the fact that it is more pronounced at the higher energy can be interpreted in terms of the spectator
stripping model illustrated in Fig. 8.9a. As the Ar+ flies past the D2 at large

o8

before react ion

after reaction

(a)

before reaction

after reaction

(b)

Fig. 8.9. Model for spectator stripping. (a) Idealized: trajectories are not deflected due to
intermolecular forces. Product ArD+ is scattered at 0. (b) Effect of long-range forces.
Product ArD+ is forward scattered.

256

Chap. 8

Single-Collision Chemistry

impact parameter one D atom is snatched from the molecule while the other
is a passive spectator and continues along its original path with velocity
unaffected. The process is analogous to the parlor trick of snatching a
tablecloth out from underneath a set of dishes. If the cloth is pulled abruptly
the dishes do not move; the cloth goes with the prankster and the dishes
remain behind.
If the stripping model described the reaction precisely the Ar D+ would
appear only at 0, the direction of the incident Ar+. Real trajectories are
curved due to the forces between Ar+ and D 2. At low collision energy the
curvature is more pronounced (Fig. 8.4a) and so the ArD+ is considerably
deflected as shown in Fig. 8.9b. (19) At higher collision energies kinematics
always swamp effects due to longer range, weaker forces, as illustrated in
Fig. 2.9 and emphasized in the discussion of the centrifugal barrier (Section
8.5). The experimental consequence for the Ar+ + D2 system is that it
more closely approximates ideal stripping; as a consequence the forward
scatter is more pronounced.
Stripping, by its very nature, is a large impact parameter process.
At small impact parameters the reactants collide head-on; thus the Ar+
rebounds and is back scattered in the eM coordinate system either taking
a D ~tom with it or not, depending upon whether the collision is reactive
or nonreactive. Thus some back-scattered products are always expected,

as is found experimentally (Fig. 8.8).

By considering both scattering and cross-section data a slightly more
detailed portrait of the reaction can be drawn. At the lower energy (7.8
kJ mol-I) Q is ,....,0.5 nm 2. Such a large value implies substantial reaction
probability at large impact parameter, conditions favorable for forward
scattering. At the higher energy (440 kJ mol-I) Q is only ,....,0.04 nm 2.
Since the scattering indicates that stripping is still the predominant mechanism, most reaction must be due to large impact parameter collisions.
The scattering and cross-section data are consistent if two conditions are
met. The probability of stripping is much smaller at high collision energy
(to account for the lower cross section) and, at all energies, the probability
of rebound reaction is far less than that of stripping (to account for the
forward scatter).

(19)

There is a correspondence in the tablecloth-dish analogy. The less abrupt the tug on the
cloth, the more the dishes are perturbed. The reader is cautioned that expertise at the
trick comes at the expense of many broken dishes.

Sec. 8.9

Recoil-The D

+ ct.

System

257
center of mass

Fig. 8.10. Angular distribution in CM of DCl produced in the D + Cl. reaction. The radius vector is proportional to the intensity of DCI. The intensity at angles less than 60
is very small. [Derived from D. R. Herschbach, Disc. Faraday Soc. 55, 233 (1973).]

8.9. Recoil-The D

+ Cl

System

The angular distribution of products in the reaction

D

+ Cl

---->.

DC!

+ Cl

contrasts sharply with that in the Ar+ + D2 reaction. As shown in Fig. 8.10,
the DCl is predominantly back scattered which indicates that small impact
parameter recoil collisions favor reaction, (20) an observation consistent

with the fact that the reaction cross section is small.

In product analyses both velocity and angular distribution are measured. While knowledge of the angular distribution is usually sufficient to
characterize the reaction as a small or large impact parameter process,
the velocity distribution can be used to infer more details of the reaction.
Figure 8.11 is a contour map representing the angle-velocity flux of DCl
produced in the D + Cl 2 reaction; the collision energy is ,.....,35 kJ mol-I.
At the peak in the distribution the DCl has a velocity of,....., 1600 m sec-I.
Since the contour map refers to CM, the peak in the distribution of Cl
must also occur at roughly the same velocity since Cl and DCl are about
equally massive, and in CM, the center of mass of the reacting system is,
of course, stationary. Thus the total relative kinetic energy of the collision
products at this velocity is ,.....,90 kJ mol-I, nearly three times as large as
(20).

J. D. McDonald, P. R. Le Breton, Y. T. Lee, and D. R. Herschbach, J. Chern. Phys. 56,

769 (1972).

Chap. 8

258

Single-Collision Chemistry

60
/

I---f

1000

m 8- 1

Fig. 8.11. Contour map of the angle-velocity flux distributions for DCl in the D + Cl.
reaction. The radius vector is proportional to the velocity of the Cl. Contours are drawn
at relative intensities of 0.1, 0.5, 0.8, and 1.0. [Adapted from D. R. Herschbach, Disc.
Faraday Soc. 55, 233 (1973).]

the collision energy of the reactants. (21) Referring to Fig. 8.11, most of the
product DCI appears at velocities between 1200 and 2000 m sec- 1 which
corresponds to energies between ,,-,50 and "-'150 kJ mol-I. The extra
kinetic energy acquired by the products derives from the large exoergicity
of the reaction, "-' 190 kJ mol-I. On the average "-'40% of the available
energy appears as product recoil energy; the remainder must reside as
internal energy (vibration, rotation) of DCI.
These observations can be used to determine the preferred reaction
geometry. As the D and Cl 2 approach they are decelerated due to repulsive
forces which create an activation barrier. The collision energy being greater
than the threshold energy, a D-CI bond forms with great energy release.
The products then recoil. In Fig. 8.12 two extreme geometries are shown.
As the D-CI bond forms the CI-CI bond is stretched and broken. Neglecting the mass of the D atom and the initial collision energy (which are
both relatively small), the CI atoms must recoil along the direction of the
original CI-CI bond in order to conserve momentum (the D atom is now
attached to one of the Cl atoms). The perpendicular geometry therefore
implies sideways scattering (peaks near 90) while the collinear geometry
is consistent with back scattering. Thus, the reaction probability is greatest
for collinear (or nearly collinear) collisions. (22)
The final energyinCM is E = (1/2)(MDClV~Cl + MCi V~l). Since momentum conservation requires that MDClVDCl + MClVCl = 0, Vel and thereby E can be calculated
from a measurement of VDCI.
( ) A rationalization, based upon consideration of the molecular orbitals involved in the
rearrangement process, is given by J. D. McDonald, P. R. Le Breton, Y. T. Lee, and D. R.
Herschbach, J. Chern. Phys. 56, 769 (1972).
(21)

Sec. 8.10

Collision Complexes-The 0

+ Bra

System

259

energy releas,e
- - - - upon reaction

4l \

1"'-;:"

cCb <=x=)
I

CI 2

~~

trajectories after
reaction

2:J

DCI

DCI

Fig. 8.12. Effect of collision geometry upon scattering of products in exoergic D

reaction.

8.10. Collision Complexes-The 0

+ Br2

+ Cia

System

The angular distribution of product BrO in the reaction

o + Br2~ BrO + Br
is shown in Fig. 8.13.(23) Unlike the reactions discussed in the two previous
sections the scattering pattern is symmetrical; the distribution is strongly
peaked at both 0 and 180, a result that cannot be understood in terms of
either the stripping or the recoil mechanism. However, if the reactants
form a collision complex a completely different picture emerges as shown
in Fig. 8.14. The reactants collide with an impact parameter, b (Fig.8.14a).
If the collision is "sticky" some of the relative momentum is converted into
angular momentum of the complex (Fig. 8.14b) so that the three atoms
rotate about the center of mass of the system. If the complex survives for
at least one full rotation before decomposing, all scattering directions in
the collision plane are equally probable, which accounts for the forwardbackward symmetry in a CM scattering pattern. The reason that the
measured angular distribution is not uniform can be understood from the
following considerations. Collisions with impact parameter b are cylindrically distributed about the z-axis of Fig. 8.14. Thus the full scattering
pattern is found by superposing the pattern for each collision plane as
indicated in Fig. 8.l4c. The axis of the resulting sphere is oriented along
(23)

D. D. Parish and D. R. Herschbach, J. Am. Chern. Soc. 95,6133 (1973).

Chap. 8

260

Single-Collision Chemistry

900

Fig. 8.13. Angular distribution in eM of BrO produced in the Br 2 + 0 reaction. The radius
vector is proportional to the BrO intensity. [Derived from D. D. Parish and D. R. Herschbach, J. Am. Chern. Soc. 95,6133 (1973).]

x
(0 )
b

Br2

---- 19

7------------

--- - z

-_/..-o

x (axis

of rotation
at complex)

(b)

----~--~~~--+-------~z

(el

iE--+---::' - - - - - - - ,

Fig. 8.14. Schematic diagram showing details in a collision typical of complex formation.
[Adapted from J. P. Toennies, Ber. Bunsenges. 72, 927 (1968).]

Sec. B.11

State Selection Experiments

261

the direction of the relative velocity veClOr. All 0 and 180 product trajectories are superposed since these angles are the poles of the scattering sphere.
At other angles the trajectories are distributed on the parallels of latitude.
Thus as a detector is moved from 0 to 180 in a particular plane, observed
CM intensity is symmetric with its minimum at the equator (the 90 line).
The angular distribution of BrO shown in Fig. 8.13 is precisely what
is expected if complex formation has taken place. It is symmetric and
strongly peaked in both forward and backward directions. Such a pattern
only occurs if the complex survives at least one full rotational period
(;:;:;5 X 1O~12 sec) before decomposing. Since vibrational periods are typically ,....." 1O~13 sec, the Br20 complex is long-lived on a molecular time scale.
In this example vibrationally unexcited Br20 is stable relative to Br2 and O.
As is true of all cases in which complex formation has been found to occur,
there are strong long-range attractive forces (and thus no activation barrier)
which would lead to the formation of a stable molecule, radical, or ion if
the excess energy could be removed by collisional deactivation. Energy
removal is not possible given the conditions of a beam experiment. The
excess energy remains trapped m the vibrational modes of the complex,
being transferred between them until enough is concentrated in one mode
at which point decomposition occurs. Since the overall reaction

o + Br2 ----" Br + BrO

is exoergic by ,.....,,40 kJ mol~I, less of the complex's excess energy is required
to split off Br than to split off O. Presumably only a small fraction of the
complex decomposes to reform the reactants.

8.11. State Selection Experiments

The molecules in an ordinary molecular beam have both rotational
and vibrational energy. By suitable modification of the experiment it is
possible to produce beams of molecules in a particular vibrational state
or with a particular orientation. Therefore one can directly study the effect
of vibrational energy or of molecular orientation on chemical reactivity.
By focusing an HCI chemical laser tuned to the v = 0 ----" I vibrational
transition on a beam of HCI molecules, it is possible to excite all the molecules of the beam into the first vibrational state of HC!. The effect on
reactivity in the
K

HCI ----" KCI

+H

262

Chap. 8

Single-Collision Chemistry

system is dramatic. (24) The reactive cross section for ground-state HCI
molecules is ,,-,0.0015 nm 2, less than 1% of that in the K + CHaI system
(see Section 8.6 and Fig. 8.5) which indicates there is an activation barrier,
estimated to be "-' 10 kJ mol-I. Reaction only occurs at small impact
parameters; a recoil mechanism is most likely. One quantum of vibration,
in this case ,,-,36 kJ mol-t, provides more than enough energy to surmount
the barrier and the cross section increases enormously to ,,-,0.2 nm 2, comparable to that found for the reaction K + CHaI. Now reaction takes place
at much larger impact parameters; the mechanism may well be totally
different.
Methyl iodide is a dipolar molecule; it can therefore be aligned in an
electric field. By passing a beam of CHaI molecules through such a field
a beam of oriented molecules is produced. This technique was used in a
study of the
Rb + CH3I ----'" RbI + CH 3
system. (25) Reaction was about four times more probable when the methyl
iodide molecules in the beam were selected such that the rubidium approached the iodine end of the molecule than if the alignment was such
that the rubidium approached the methyl end. The results are consistent
with the model in that electron transfer from alkali metal to halide is the
primary step.
A similar experiment was performed on the
K

+ CFaI----'" KI + CFa

system. (26) The results were strikingly different. Reaction is most likely
if the molecules in the beam are oriented so that the alkali atom approaches
the CFa end ofthe molecule. Presumably electron transfer is still the primary
step; since the F atom is far more electronegative than the I atom, it is
much more likely to participate in the harpooning step. However, the
C-I bond is weaker than the C-F bond (234 vs. 490 kJ mol-I) and the
further course of reaction seems governed by energy considerations. The
I atom is the one removed which simply requires that the charge in the
K +-CFaI - be rapidly redistributed after the electron transfer has taken place.
(24)

(251

(26)

T. J. Odiorne, P. R. Brooks, and J. V. V. Kasper, J. Chern. Phys. 55, 1980 (1971).

R. J. Beuhler, Jr., and R. B. Bernstein, J. Chern. Phys. 51, 5305 (1969). The experiment is
not designed to form a beam of paralIel dipoles since, at the required field strengths, the
molecules would ionize. Instead a field geometry is constructed to reject all molecules
which are leading with their iodine (or methyl) end.
P. R. Brooks, J. Chern. Phys. 50, 5031 (1969).

263

Problems

Problems
8.1.

Using (8.6) and (8.7) show that

Ea = N o <E 2Q(E/<EQ(E

- 3RT/2

where

8.2.

If the temperature dependence of Q(E) is considered show that

Ea

Ea l1l

RP< aQ(E; T)/an/<EQ(;

where Ea l1l is the activation energy calculated in Problem 8.1.

8.3.

In Section 8.3 we described the measurement of a reaction cross section.

What measurements must be made to determine (a) the elastic cross section,
(b) the inelastic (nonreactive) cross section?

8.4.

A molecular beam source contains Cs(liq) in equilibrium with Cs(v) at 600 K.

The vapor pressure of Cs is 4.5 X 10- 3 atm. The aperture is a circular slit
1 mm in diameter; other apertures maintain the beam as an atomic ribbon.
Calculate:
(a) the total cesium atom flux emerging from the source;
(b) the axial intensity in a plane perpendicular to the source and 0.3 m
from the last aperture;
(c) the axial intensity assuming the geometry of (b) if only atoms with
axial speeds in the range 0.95 ::; v/c mp ::; 1.05 are transmitted (c mp is the
most probable speed).

8.5.

Energy-selected ionic beams are produced by electron bombardment of a

thermal distribution of a species A, A + e- ~ A + + 2e-, followed by
acceleration of A + through a potential drop Vo.
(a) Xe, initially at 300 K, is ionized; a 5-V potential is then applied. What
is the mean ionic energy?
(b) What is the mean velocity in the direction of the applied potential?
(c) What is the standard deviation (i) in the speed, (ii) in the energy?

8.6.

Another way to produce an energy-selected ionic beam is to pulse the

potential for a time, T, that is much less than that needed to accelerate
the ions through the electric field, V/d, where d is the plate separation.
Each ion then receives the same velocity increment.
(a)
(b)
(c)
the
the

What is the velocity increment?

Estimate the maximum pulse duration.
If Xe+, formed from Xe at 300 K, is accelerated via this procedure to
mean velocity of Problem 8.5b, what is the standard deviation (i) in
speed, (ii) in the energy?

Chap. 8

264

Single-Collision Chemistry

8.7.

Total cross section, as determined by measurement of beam attenuation,

was discussed in Chapter 2. In one such experiment an atomic beam is
passed through a reaction chamber of 2-cm length containing gas at 10-7 atm
and 300 K. If the attenuation is 50% what is the total cross section for atomgas interaction?

8.8.

At 500 K the viscosity of H2(g) is 1.26 X 10-5 and that of I2(g) is 2.20 X 10-5
in SI units. Estimate the molecular size and then use collision theory to
obtain a value for the A-factor. The experimental value is ,....,4.3 X 1010
M-l sec-I.

8.9. The cross section for the reaction T

the analytical expressions
Q

_{O,y
B

+ H. ~ TH +
E

0.69

exp( -0.47y1.46),

H is roughly given by

<

Eo

E> Eo

where
y == (E -

Eo)/iJ,

Eo =

4.2 X 10-. 0 J,

iJ =

6.0 X 10-19 J,

B = 0.039 nm'

For a temperature 1000 K plot

(a) the Boltzmann function ({3E)'/'e- fJ.;
(b) the cross section Q(E);
(c) the excitation junction {3e-{J<Q().
Note how each of these has its maximum in quite distinct energy domains.
From (c) it should be clear why threshold and activation energies are not
the same.
8.10. What is your prediction of the scattering pattern for reactive collisions in
the Ar+ + D. system at high collision energies in CM? Would forward or
back scattering predominate? (Ignore fragmentation.)
8.11. Consider the bimolecular reactions of Table 8.1. On the basis of the activation energies and any chemical knowledge at your disposal suggest which
reactions might proceed (a) via complex formation; (b) via a recoil mechanism.
8.12. In Figs. 8.8, 8.10, and 8.13, CM angular distributions are given for a number
of reactions. The polar diagrams plot normalized differential cross sections
a'QfaO acp as a function of 0 for fixed cp. As the detector subtends a solid
angle dQ = sin 0 dO dcp, the differential cross section overemphasizes scattering for 0,...., 0 and 0,...., 180. For each example estimate the relative
value of a'Q/ao acp at 15 intervals. Then plot a normalized scattering
pattern as a function of 0 by numerically computing
sin 0

2"

dcp(a 2 Q/ao acp)

Problems

265
Table 8.3

a'Q/ao alP

a2Q/ao alP

(nrn 2 rad-')

(deg)

(nrn" rad- 2 )

(deg)

(i)

(ii)

(i)

(ii)

0.0048

0.16

105

0.060

15

0.15

120

0.049

0.0054

30

0.14

0.0015

135

0.047

0.0061

45

0.13

0.0021

150

0.047

0.0066

60

0.10

0.0026

165

0.044

0.0068

75

0.079

0.0033

180

0.043

0.0074

90

0.071

0.0040

8.13. For reactive scattering of Na and (i) Br., (ii) CHsI at about the same collision energy the variations of differential cross sections with scattering
angles (CM) are [data derived from J. H. Birely, E. A. Entemann, R. R.
Herm, and K. R. Wilson, J. Chem. Phys. 51, 5461 (1969)] as given in
Table 8.3.

(a) Calculate the reaction cross section for each case.

(b) What are likely mechanisms for each reaction?
8.14. Consider a reactive collision between molecules 1 and 2 with initial masses
m i , m. and velocities v" v. The products l' and 2' have masses mi" ma'
and the internal energy change is LIE [c = Vi - V., M = mi + m.].
(a) Show that the relative speed after collision is c' = [(flC' - 2L1E)/fl'J"2;
fl and fl' are reduced masses before and after collision, respectively.
(b) Assume a deflection X during reaction in CM, Show that the relative
velocity after collision is c' = C'(Du cos X + D.L sin X); D. is a unit vector
parallel to c, and D.L a unit vector in the collision plane perpendicular to c.
(c) Show that the product velocities in LAB are

(d) Show that the deflection of l' relative to an incident beam of 1 in

LAB is

where q = (m;c' - m2c)/ M. If q is small show that cos 0 = l-q" sin" W/2Vi"'
where cos W = (Vi q)/I Vi II q I

266

Chap. 8

Single-Collision Chemistry

8.15. In idealized spectator stripping the atom left behind continues with its
velocity unaltered. The reaction Ar+ + D. ~ ArD+ + D is 145 kJ mol- i
exoergic.
(a) Use conservation of momentum and show that, in ideal stripping, the
relative translational energy after reaction is E' = 1OEo/21, where Eo is
the collision energy.
(b) Since excess collision energy must appear as internal energy of ArD+,
at what collision energy does the reactive cross section for stripping go to
zero? The binding energy for ArD+ is 370 kJ mol-i.
8.16. A "steric factor" is often used to account for discrepancies between observed A-factors and those computed using the hard-sphere model (S.16).
The steric factor / accommodates orientational requirements in reaction
so that the A-factor is A = Ao/with Ao given by (S.16).
(a) What are the steric factors for the reactions of Table 8.1?
(b) Comment, if you can, on possible reasons for differences between /
in the following pairs of reactions:
(i)

+ N.H, ~ NH. + NH.

+ N.H, ~ H. + N.H.
+ CCI, ~ CCI. + HCI

CI

+ CCI 4

H
H

(ii)

+ 12
CI + H2

(iii) H

CCI.

CI 2

+I
~ HCI + CI
~

HI

8.17. Show that the cross section

10

<

Eo

10 :::::: Eo

yields an Arrhenius form for keT)o

S.IS. Plot the line-of-centers cross section (S.13) and the cross section of Problem
8.17, and compare with the data presented in (a) Fig. 8.3, (b) Fig. S.5.
S.19. Sketch the energy dependence of the reaction cross section for
(a) D+
(b) D-

+ H. ~ DH + H+

+ H. ~ DH + H-

Do not forget the possible effect of a potential barrier.

8.20. Show that (S.IS) implies that the distance of closest approach of hard
spheres is
b,
b>a
Ro= {
a,
b<a

General References

267

General References
J. L. Kinsey, in MTP International Review of Science, Physical Chemistry, Ser. I, Vol. 9,
J. C. Polanyi, ed. (London: Butterworths, 1972), pp. 173-212.
K. J. Laidler, Theories of Chemical Reaction Rates (New York: McGraw-Hili, 1969),
pp. 182-204.
R. D. Levine and R. B. Bernstein, Molecular Reaction Dynamics (Oxford: Clarendon
Press, 1974), Chapter 6.
.
J. Ross, ed., Molecular Beams, Advances in Chemical Physics, Vol. 10 (New York: Interscience, 1966).

Ion-Molecule Reactions
P. F. Knewstubb, Mass Spectrometry and lon-Molecule Reactions (Cambridge: University
Press, 1969).

Elastic Scattering
R. J. Cross, Jr., Acc. Chern. Res. 8, 225 (1975).

Inelastic Scattering
G. A. Fish and F. F. Crim, Acc. Chern. Res. 10,73 (1977).
I. W. M. Smith, Acc. Chern. Res. 9, 161 (1976).
J. P. Toennies, Ann. Rev. Phys. Chern. 27, 225 (1976).

Reactive Scattering
D. R. Herschbach, Disc. Faraday Soc. 55, 223 (1973).
D. A. Micha, Acc. Chern. Res. 6, 138 (1973).

State-to-State Experiments
J. M. Farrar and Y. T. Lee, Ann. Rev. Phys. Chern. 25, 357 (1974).

Theories of Reaction Rates

9.1. Introduction
A theoretical model for chemical reaction might take one of two forms.
H could be designed to interpret molecular beam experiments which would
necessitate detailed consideration of the possible outcomes of molecular
collisions; it could yield, e.g., the reaction cross section (Figs. 8.3 and 8.5)
and the product angular distribution (Figs. 8.7, 8.8, 8.10, 8.11, and 8.13).
Such an undertaking would be exceptionally ambitious. It would essentially
require theoretical predictions of all details of the reaction process.
A much more limited objective is to try to calculate the chemical rate
constant. Rather than predicting the results of molecular collisions, this
approach is designed to determine the Arrhenius parameters Ea and A.
Naturally if the reaction cross section had been computed, it could be
used to compute the rate constant. On the other hand, in a reaction at
constant temperature the probability of high-energy collisions is small.
Thus, the reaction pathways that are statistically significant are those with
energy barriers near the threshold for reaction. If these chemically significant
pathways can be characterized it is possible to apply statistical mechanics
and estimate the thermal rate constant.
In either case, it is necessary to know something about the way in
which reactants interact with each other. In our discussion of reaction
dynamics we have seen that large reaction cross sections suggest that
there is long-range attraction between the reactants. Conversely, systems
in which there is a barrier to reaction have small reaction cross sections.
In this chapter we shall consider how some knowledge of the interparticle
interaction can be used to construct theories of the chemical rate constant.
269

270

Chap. 9

Theories of Reaction Rates

9.2. Potential-Energy Surfaces

In Section 8.2 we found that the reaction cross section provides the
link between the rate constant and collision dynamics. However, Q(E, i,j)
is just a function of the collision energy and the internal states of the reactants. The outcome of a single collision also depends upon the impact
parameter, as demonstrated for the simple hard-sphere model of reaction
cross section discussed in Section 8.4. The probability of a specific result
pCb, E, i, j) is dependent on all these variables. In Section 8.4 we showed
that the relative importance of a collision with impact parameter between
band b + db is proportional to the annular area 27tb db. The reaction cross
section is found by multiplying the weighting factor for a particular b by
the probability of a specific result and integrating over all values of the
impact parameter
Q(E, i, j)

f~ (27tb db)p(b, E, i, j)

(9.1)

The reaction probability measures the outcome of all possible collisions

with specified initial values of b, E, i, and j. It is particularly simple for the
hard-sphere model of Section 8.4, where
pCb, E)

{~:

> Eo and b < be (E)

otherwise

and be (E) = 0'(1 - Eo/)1/2. Substituting this expression into (9.1) leads
immediately to the result for the reaction cross section obtained previously
(8.13). For more realistic models the determination of the reaction probability is a formidable problem which requires, as a preliminary step, a
knowledge of the interaction potential (the potential-energy surface) of the
reactants as a function of the nuclear coordinates. Such a calculation
determines the electronic energy of the interacting molecules for all nuclear
configurations of interest. Once the potential-energy surface is known
the problem of computing the reaction probability can be tackled. This
requires applying time-dependent quantum mechanics to treat the problem
of nuclear motion in a known potential. The task is exceptionally difficult
and is one that has not been accomplished, even for the simplest reactions,
at the present time. (1)
(1)

Recent calculations on the H + H2 ~ Ha + H exchange reaction come close to carrying out this program; see A. Kupperman and G. C. Schatz, J. Chem. Phys. 62, 2502
(1975); A. B. Elkowitz and R. E. Wyatt, J. Chem. Phys. 62, 2504 (1975).

Sec. 9.2

Potential-Energy Sur/aces

271

The calculation of a potential-energy surface is itself not easy. To

describe the surface for the exchange reaction
A

+ BC----"AB + C

(9.2)

three parameters are needed, e.g., the distances A-B and B-C and the
A-B-C angle. (2) Correspondingly the surface is four-dimensional. If
one of these variables is held fixed, a three-dimensional projection can be
constructed. A series of such diagrams permits visualization of the whole
potential-energy surface.
Reactions that proceed via an intermediate which is more energetic
than either reactants or products, as in Fig. 4.8, have qualitatively similar
potential-energy surfaces. A typical slice at fixed A-B-C angle is shown in
perspective in Fig. 9.1 and as a contour map in Fig. 9.2. When rAB (the
A-B distance) is large, the interaction potential is that of the isolated BC
molecule; the point R represents the classical ground state of the reactants.
Similarly when rBC is large the potential is that of the AB molecule; P is
the classical ground state for products. Large values of both rAB and rBC
correspond to total dissociation into A + B + C; this region of the surface
is the plateau D.
Reaction is represented by motion from R to P on the surface. As
reaction occurs the system starts in the reactant valley, travels over a pass,
and descends into the product valley. Reaction only occurs if the relative
energy of the interacting molecules is sufficient to surmount the pass. For
thermal reactions the system must move along relatively low-energy paths
because, for surfaces like that of Fig. 9.1, the pass height is usually high
enough so that collisions which occur with enough relative energy are
highly improbable. Reaction thus samples the high-energy tail of the Boltzmann distribution and only collisions with energy close to the pass energy
are sufficiently probable to contribute significantly to reaction. (3) The
minimum energy path is shown in both Figs. 9.1 and 9.2. The section through
this path, known as the reaction profile, is shown in Fig. 9.3. It is similar
to the schematic diagram of the activation process shown in Fig. 4.8;
now, however, progress along the path is correlated with specific changes
in geometry. Reactants with collision energy less than the energy of the pass
(2)

(3)

Many other ways of specifying the ABC triangle are possible but this is the most convenient choice.
If the system strays too far from the pass the collision energy must increase in order for
reaction to be possible. At 300 K the fraction of possibly reactive collisions decreases
tenfold with each 6 kJ mol- 1 increase in the barrier.

Chap. 9

272

Theories of Reaction Rates

Potential
Energy

Fig. 9.1. Variation of potential energy with A-B and B-C distances for an ABC system in
which the A-B-C angle is held fixed. [Adapted from K. J. Laidler, Chemical Kinetics (New
York: McGraw-HilI, 1965).)

Activated state (pass)

(A+B+CI

(A+BC) R

Fig. 9.2. Contour diagram for the potential energy of the ABC system of Fig. 9.1. The
dashed line shows the minimum energy path which is followed by a reaction under thermal
conditions.

Sec. 9.2

Potential-Energy Sur/aces

273
Activated complex
ABC

Potential
Energy

A+BC
dx

AB+C

Reaction coordinate

Fig. 9.3. Section through the minimum energy path of Fig. 9.2. There is a barrier to reaction
in either direction.

do not react. (4) In thermal chemical reaction, only trajectories close to the
reaction profile are important; all others require too much collision energy.
Thus most reaction paths pass close to the pass on the potential-energy
surface. Because of the chemical importance of the relatively low-energy
region of the ABC aggregate system, it is specially designated. The pass,
which is the peak of the reaction profile, is known as the transition state
or the activated complex. It bears, of course, no relation to the long-lived
collision complexes discussed in Section 8.10.
Changing the A-B-C angle affects both the potential-energy surface
and the reaction profile but only in those regions where the three particles
are close together. The far valleys and the plateau which describe the
isolated molecules and the separated atoms are not changed. Thus all
reaction profiles are alike at start and finish but their intermediate structure
differs. The pass energy depends upon the angle A-B-C as is illustrated
in Fig. 9.4, where the reaction profile for the exchange reaction

is plotted for different H-H-H angles. The lowest pass height in this system
occurs in the linear configuration. The activated complex or transition state
is identified as the configuration for which the ABC complex has the
minimum pass energy. It is important to recognize that the most probable
reaction paths pass through configurations close to that of the activated
(4)

This is not strictly correct. Molecules follow the laws of quantum, not classical, mechanics. As such they have zero point energy; the system does not precisely follow the potential-energy trough which slightly alters the collision energy necessary for surmounting the pass. It is also possible, but not likely, for reactants with collision energy somewhat less than the pass energy to tunnel through the barrier. These quantum effects
alter the threshold energy which is usually slightly less than the pass energy.

Chap. 9

274

Theories of Reaction Rates

Potential Energy
(kJ mol-I)

-0.05

r(nm) 0.05

Fig. 9.4. Reaction profiles for the Ha system as a function of HHH angle. The distance r
is the separation of the incoming or leaving H atom from the pass on the potential-energy
surface. Energies are measured with respect to the H + H, system. [Adapted from I. Shavitt, R. M. Stevens, F. L. Minn, and M. Karplus, J. Chern. Phys. 48, 2700 (1968).]

complex. Thus knowledge of the molecular structure of the region at the

top of the pass is most important for understanding many aspects of
chemical reactivity.
For the model potential-energy surface we have discussed, the lowest
energy intermediate (the transition state) is at higher energy than either
products or reactants. This is the most common surface; however, other
possibilities also occur. Minimum energy paths for reactions in which
there is no barrier are illustrated in Fig. 9.5. These are typical of exoergic
reactions with large cross sections. The minimum energy path of Fig.
9.5a describes a reaction such as Ar+ + D 2 , discussed in Chapter 8. The
(b)

(0 )

>.

>.

0>

0>

Q)

Q)

'0

:;:
c

:;:
c

0
c..

c..

Q)

Q)

'0

Reaction Coordinate

Reaction Coordinate

Fig. 9.5. Minimum energy path for systems without barriers. (a) Typical profile for reaction with shallow well when complex formation does not occur. (b) Typical profile for
reaction with deep well where intermediate complex may form.

Sec. 9.2

Potential-Energy Surfaces

275

well is too shallow to permit the formation of a collision complex. The

minimum energy path of Fig. 9.5b has a deep well; complex formation is
probable, e.g., the 0 + Br2 system. All low-energy paths are now potentially
reactive; there is no longer a favored reaction trajectory and the idea of
an activated complex is less useful. In these examples the pass energy and
the reactant energy are the same. However, this does not imply that the
pass geometry is the reactant geometry. In terms of Fig. 9.2 we may imagine
a constant energy reactant trough which defines the reaction coordinate
as A and BC approach one another. Trajectories that deviate significantly
from this coordinate must overcome energy barriers. Thus, while there is
no activated complex, there is still a favored intermediate configuration
which plays a similar role.
In general, reactions involve more than three atoms, so a complete
potential-energy surface depends on more variables. To describe the exchange reaction
K + CH3! ----' KI + CH 3
a total of 12 coordinates is needed. Even if a complete calculation could
be done, interpreting such a .surface would present enormous problems.
By considering only reaction channels in which the CH 3 group can be treated
as a single particle, the problem is reduced to that of a simple triatomic
exchange reaction.
The actual construction of a potential-energy surface is a difficult
quantum mechanical problem. It requires calculating the energy of a manyelectron system as a function of nuclear positions. For the similar, but
simpler, problem of computing the known vibrational and rotational
properties of molecules, comparisons of theory and experiment suggest
that approximate theories are not too accurate and accurate theories are
too difficult to be generally useful. A potential-energy surface calculation
involves more electrons (the electrons of all the reactants must be considered
simultaneously) and many geometries. It is not surprising that most such
calculations are both approximate and of limited accuracy. (5)
(5)

Reactions that have been specifically studied include: H + H. ~ H. + H, B. Liu, J.

Chern. Phys. 58, 1925 (1973); K + CH31 ~ KI + CH 3 , L. M. Raff and M. Karplus,
J. Chern. Phys. 44,1212 (1966); H. + He+ ~ HeH+ + H, P. J. Brown and E. F. Hayes,
J. Chern. Phys. 55, 922 (1971); K + Br. ~ KBr + Br, D. R. Herschbach, in Molecular
Bearns, J. Ross, ed. (New York: Interscience, 1966), p. 376; F + H2 ~ H + HF, C. F.
Bender, P. K. Pearson, S. V. O'Neil, and H. F. Schaefer III, J. Chern. Phys. 56, 4626
(1972); Li + F. ~ LiF + F, G. G. Balint-Kurti, Mol. Phys. 25, 393 (1973); H. + 12 ~
2HI, L. M. Raff, L. Stivers, R. N. Porter, D. L. Thompson, and L. B. Sims, J. Chern.
Phys. 52, 3449 (1970).

Chap. 9

276

Theories of Reaction Rates

9.3. Activated Complex Theory

From the viewpoint of a potential-energy surface molecules on one
side of the pass are reactants; those on the other side are products. In
regions close to the pass such a distinction is hard to maintain since the
molecules form an aggregate of strongly interacting particles. Reaction
dynamics can be described as molecules traversing the pass from reactant
side to product side, suggesting that the rate of a chemical reaction may be
calculated by determining the number of systems per second which cross
the pass in the direction reactant to product, an observation that is the basis
for the theory of chemical reaction developed by Eyring. (6)
Consider the bimolecular gas-phase reaction
A

+B

---->.

products

(9.3)

with a reaction profile like that of Fig. 9.3. Motion along the reaction
coordinate is specified by a position x and a reaction momentum p. By
P > 0 we mean that the reactants are moving toward each other; as they
approach the pass they may be considered to form an aggregate or a composite system (sometimes described as an AB complex). The transition from
independent reactants to aggregate system cannot be recognized to have
occurred at any particular point along the reaction coordinate. At small
separations (;50.5 nm) interaction is strong and the composite description
is appropriate. At large separations (;;:; 10 nm) interaction is weak and
describing the reactants as independent species is more reasonable.
The basic assumption of the theory is to consider motion along the
reaction coordinate separately from other motions on the potential-energy
surface. In addition to its translational, rotational, and vibrational modes
the AB aggregate has a reaction mode which describes the chemical transformation in terms of progress along the reaction coordinate. The larger
the value of the reaction momentum p, the more collision energy is available
for surmounting the pass. If the energy in the reaction mode is large enough,
the molecules traverse the pass and react; if not, they are reflected. (7)
Far away from the pass, the number of A-B composite systems, C(p, x),
with reaction coordinate between x and x + dx and reaction momentum
(6)

(7)

The development which folIows is adapted from H. Eyring, J. Walter, and G. E. Kimball, Quantum Chemistry (New York: John Wiley, 1944), pp. 299-311. For a totally
different development see K. J. Laidler, Theories of Chemical Reaction Rates (New York:
McGraw-HilI, 1969), pp. 45-53.
Again this is not quite true since the problem is one of quantum, not classical, mechanics. See footnote (4) in this chapter.

Sec. 9.3

Activated Complex Theory

277

between p and p + dp, can be estimated if we assume that local equilibrium

is established along the reaction coordinate. Naturally the reactant population is being continuously depleted as reaction proceeds so the hypothesis
is not strictly true; it is, however, a plausible first approximation. The rate
at which systems approach the pass can be used to compute the reaction
rate. This rate is the reciprocal of the time required for all systems in the
interval dx to pass a specific point such as Xo on the reaction profile, Fig. 9.3.
Since the systems are distributed in the interval dx, a translation dx is needed
in order for all the systems to pass Xo. Each system in the specified momentum range has the same value of p; thus the time required is
f1 dx/p

where f1 is the reduced mass for motion along the reaction coordinate,
f1 = mAmB/(mA
mB)' The total number of systems approaching the

pass per second is

p

C(P,x)-df1 x

(9.4)

If the reactants A and B were truly in equilibrium with an A-B complex,

the statistical mechanical theory of chemical equilibrium in gases could
be applied.(S) The number of such systems NAB would be
(9.5)
where qA' qB, and qAB are partition functions for A, B, and A-B complex;
NA and NB are the number of A and B molecules. The energy of the complex

is assumed to be separable because we have differentiated motion along the

reaction coordinate from all other modes. Thus, far from the pass, where
the system is in the flat portion of the reaction profile Fig. 9.3, the energy
in the reaction mode is kinetic energy only, p2/2f1. The energy residing in
all modes except the reaction coordinate is n and the energy of the composite is n + p2/2f1. The zero of energy for the composite is the same as
that of the separated reactants. Motion along the reaction coordinate
requires traversing a pass with a classical barrier to reaction t. The total
energy in all the other modes is measured from an energy zero set at the
top of the pass. From equilibrium statistical thermodynamics the number
of composites found with reaction coordinate between x and x + dx,
(8)

T. L. Hill, Introduction to Statistical Thermodynamics (Reading, Mass.: AddisonWesley, 1960), Chapters 4, 8, and 9. A summary of some important results of statistical thermodynamics is included in Appendix A.

278

Chap. 9

reaction momentum between p and p

Cn(p, X) dp dx

NAB
(
qAB exp -

+ dp,

Theories of Reaction Rates

and

En

En)
(
kT exp -

is(9)

p2) dp dx
2flkT - h -

(9.6)

Combining (9.6) with (9.5) and summing over all degrees of freedom
except the reaction coordinate and momentum we find

C(p, x) dp dx

NANB:I:
(
qAqB qAB exp -

p2) dp dx
2flkT - h -

(9.7)

where qiB is a partition function for energy disposition in all modes of

the complex except the reaction mode

qiB

~ exp ( - :~ )

(9.8)

To compute the population of the complex requires integrating (9.7)

over p and x; this is ambiguous since the precise range of values of x that
can be considered to correspond to composite systems is unknown. Fortunately the quantity of interest is the rate at which systems approach the
pass. This can be determined from the behavior in the reactant valley as
long as collisions with a third particle do not perturb the approach to the
pass.(lO) From (9.4) and (9.7) the rate of approach is

~)

;
p
( - - - - dp
NANB~-exp

qAqB

flh

(9.9)

2flkT

which is independent of the reaction coordinate. Only systems with p > 0

are moving toward the pass and possible reaction. Only a fraction of these
descend into the product valley. Introducing a transmission coefficient,
x(p), to account for partial transmission, the reaction rate is found by
multiplying (9.9) by x(p) and integrating over all p > 0

dNA
qiB
---=NAN
Bdt
qAqB

foo --x(p)exp
P
(P2-) dp
0

flh

2flkT

See Appendix A for a justification of this expression. The factor of h- 1 occurs because
the formulation is semiclassical. It is against the rules of quantum mechanics to treat
reaction coordinate and momentum separately.
(10) This is an excellent approximation in dilute gases. At 1 atm and 500 K the mean free
path is ~50 nm, much larger than the range of values of the reaction coordinate that
can be considered to correspond to the composite. At very low pressure the approximations used to derive (9.7) are no longer reasonable; under such conditions the concentration of composites is sensitive to the rate at which they traverse the pass. In
addition, depletion is significant and (9.5) is no longer a good approximation.
(9)

Sec. 9.3

Activated Complex Theory

279

and the second-order rate constant can be identified as

(9.10)
If chemical reaction were a problem in classical mechanics the momentum dependence of the transmission coefficient would only reflect whether
there was sufficient energy to surmount the pass. For a pass energy Et
we have
p < pt = (2,uEt)I/2
(9.11)
x(p) = {~:
P > pt
The problem of computing x(p) is actually a quantum mechanical one.
The momentum dependence is determined by the shape of the reaction
profile. Theoretical models show that x(p) is small but not zero for p < pt
and that for p > pt it is oscillatory, but generally increasing, until the
momentum is large enough so that fragmentation is possible. The qualitative
behavior of x(p) for momenta small enough so that fragmentation is not
a problem is shown in Fig. 9.6.(11)
While the quantum and classical forms for x(p) are strikingly different,
the Boltzmann factor in the integrand of (9.10) ensures that the oscillatory
structure of the quantum form of x(p) only slightly influences the rate
constant. In Fig. 9.7 the quantities px(p) exp( -p2j2,ukT) are plotted for
the two forms of x(p). In both cases roughly the same range of momenta
contribute significantly to the integral. Thus (9.10) is approximately

(9.12)
where <x) is a mean transmission coefficient. Its value, determined by
comparing the computed and experimental A-factors, is generally between
0.1 and 0.9; in practice it is often assumed to be ,..".,0.5. As formulated,
k 2 (T) has units of molecules- l sec-I. By introducing appropriate factors
of V and No it can be converted into concentration units, cm 3 mol- l sec-I.
In terms of the molar partition function per unit volume,
(9.13)
(11)

This problem is treated by H. Eyring, J. Walter, and G. E. Kimball, Quantum Chemistry

(New York: John Wiley, 1944), pp. 311-326.

Chap. 9

280

x(p)

.- /

I
I
i
I
I
I
I
p'f =

Theories of Reaction Rates

I
I
I

\...1

htJE'

vi

--

reaction momentum

Fig. 9.6. Behavior of transmission coefficient ,.;(p) as a function of reaction momentum for
classical and quantum mechanical models of reaction; a rectangular barrier is assumed.

the rate constant is

(9.14)

which is the famous Eyring equation. The rate constant can be calculated
from the properties of the potential-energy surface by identifying the reaction coordinate. The pass height determines the Eyring activation energy
E~, here defined as the classical barrier to reaction; as such it does not include
the zero point vibrational energy of the activated complex. The shape of
the reaction profile determines ,,(p) and thus <,,); the remainder of the
surface near the pass determines Q1B'

Transmission
Function
(Arbitrary
units)

classical
quantum

reaction momentum

Fig. 9.7. Transmission function, p,.;(p) exp( - p2 /2{1kT), as a function of reaction momentum
for classical and quantum models of ,.;(p) plotted in Fig. 9.6. Note that oscillations in
,.;(p) are damped by the Boltzmann factor; E+ /kT is much greater than 1 corresponding to a
significant activation barrier.

Sec. 9.4

Model Calculations of k.(T)

281

As was the case in the hard-sphere model for reaction rates (Section
8.4), c t is not easily computed and thus a complete calculation of k 2 (T)
is not generally possible. To eliminate c t and determine an expression for
the A-factor we write (9.14) as
k 2(T) = B(T) exp( -ctjkT)

Using (8.7) the Arrhenius activation energy is

and the A-factor is

A

B(T)e B

(9.15)

Again both A and Ea are temperature dependent but the variation is usually
too small to be detected experimentally.
The theory is obviously quite approximate even if <u), c t , and Qi.B
could all be computed. We began by assuming an equilibrium population
of A-B composite systems at points along the reaction profile far from the
pass; however, the problem is intrinsically nonequilibrium in nature. We
assumed that motion along the reaction coordinate was not coupled to
other degrees of freedom of the A-B composite, which is highly unlikely,
if only because the moments of inertia of the aggregate vary as the reaction
coordinate changes. We assumed the one-dimensional reaction coordinate
could be uniquely assigned even though molecular beam experiments
indicate that reaction dynamics may be much more complex. Even with
these limitations the theory is still a considerable improvement over the
hard-sphere model of Section 8.4. Its great achievement is that, through
consideration of the partition function Qh, plausible assumptions as to
the nature of the A-B interaction can be introduced and used to calculate,
in an admittedly approximate manner, either k 2 (T) or, more commonly,
the A-factor.

9.4. Model Calculations of k 2 (T)

The activated complex theory provides a link which connects the potential-energy surface with the chemical rate constant. To apply the theory
requires knowing the structure of the potential-energy surface near the
saddle point. As discussed in Section 9.2 accurate determination of the
surface is a difficult problem. Instead of computing the properties of the

282

Chap. 9

Theories of Reaction Rates

surface, the properties of the activated complex are usually estimated. These
estimates then provide the data used to determine k2(T).
Here we treat two specific examples. First we consider the hard-sphere
model for chemical reaction and show that activated complex theory leads
to a k 2(T) identical with that calculated using kinetic theory arguments in
Section 8.4. We then consider the molecular example
H

+1

2 ----'

HI

+I

and, by making reasonable assumptions about the structure of the activated

complex, show that the theory provides a means for discriminating between
two possible geometries for the activated complex.
In addition to constructing a model for the potential-energy surface
or postulating a structure for the activated complex, the reaction coordinate
must be identified. Then all other degrees of freedom in the A-B complex
can be described in terms of the translational, rotational, and vibrational
modes of the complex. If these are assumed to be independent, the computation of k 2 (T) requires calculating QiB which, from statistical thermodynamics, is
(9.16)
The q's are partition functions for the various degrees of freedom and
is the electronic degeneracy factor for the complex. (12) According to
our assumptions only motion along the reaction coordinate leads to chemical
change. Other degrees of freedom are presumed unaffected by reaction.
Thus the geometry and vibrational frequencies determined at the saddle
point can be used to characterize energy disposition in the nonreactive
modes of the complex. It is this feature of the saddle point which endows
the activated complex with special importance. (13) The partition functions
for the various degrees of freedom may then be computed using statistical
thermodynamics. (14)

ge

(12)

(13)

(14)

We assume that only one electronic state of the complex is important in thermal reactions.
It might seem that this assumption of separability implies that knowledge of the saddle
point structure can be used to characterize the reactants and products. If each degree
of freedom were strictly independent then precise description of the saddle point
structure would specify the molecular properties of both reactants and products. Since
separability is only approximate, no such extrapolation is possible. In fact, the reaction
coordinate evolves from a molecular vibration in a manner unspecified by knowledge
of the saddle point region.
T. L. Hill, Introduction to Statistical Thermodynamics (Reading, Mass.: AddisonWesley, 1960), Chapters 4, 8, and 9; or see Appendix A.

Sec. 9.4

Model Calculations of k.(T)

283

The hard-sphere approximation of Section 8.4 is the simplest model

for reaction. The molecules A and B are treated as spheres which must
approach within a distance a to react. Only the translational motion of
A and B is considered, and from (9.13) and (9.A6) of Appendix A
(9.17)
with a similar expression for QB. The complex has only translational and
rotational degrees of freedom. Relative motion, which would correspond
to vibration if AB formed a bound molecule, is the reaction mode. Using
(9.16) and (9.A6) we find
(9.18)
where M = mA + mB. The moment of inertia of the activated complex,
I = fUJ 2 , is determined by a, the distance of closest approach; fl is the
reduced mass mAmB/M. Combining (9.14), (9.17), and (9.18) we find
(9.19)
which, but for the factor of No, is identical with (8.14); <x) is exactly
one in this model since collision theory is a classical mechanical model
for chemical reaction.
As a specific molecular example consider the reaction
H

+ I2~ HI + I

(9.20)

In the temperature range 300-700 K the Arrhenius parameters, given in

Table 8.1, are A ~ 2x 1014 cm 3 mol- 1 sec-1 and Ea ~ 2 kJ mol- 1. The
A-factor, which is about 20% of that estimated using collision theory,
may be estimated using (9.14)-(9.16) if reasonable assumptions about
the structure of the activated complex of HI2 are made. Both geometry
and vibrational frequencies must be estimated.
At the pass both H-I and I-I distances are somewhat larger than
in the corresponding stable molecules. Actual calculations of potentialenergy surfaces(15) suggest that this expansion amounts to 20-30%; it
affects not only the moments of inertia but also the vibrational frequencies
of the HI2 complex. The parameters required to describe molecular properties are: g, the electronic degeneracy; I, the moment of inertia (for a
115)

See references in footnote (5) in this chapter.

Chap. 9

284

Theories of Reaction Rates

linear molecule); v, the vibrational frequencies; and (1, the symmetry number. If HI2 is assumed to be linear with interatomic distances 30% greater
than those in HI or 12, the following set of molecular parameters can be
obtained(16) :
H: g=2
12 : g = 1, (1 = 2, 1 = 7.5 X 10-45 kg m 2, v = 6.45 X 1012 sec-1

H1 2: g=2, (1=1, 1= 13.0 X 10-45 kg m 2, v=5.3xI0 13, 1.45 X 1013,

1.45 X 1013 sec-1
Both Hand HI2 have electronic degeneracies of 2 since each has a single
unpaired electron; 12 , being a symmetric diatomic molecule, has a symmetry number of 2. The diatomic molecule 12 has two rotational and one
vibrational degree of freedom. The linear complex HI2 has two rotational
and four vibrational modes; because one possible vibrational mode corresponds to the reaction coordinate, three are true vibrational degrees
of freedom. From (9.14) the preexponential factor B(T) is
B(T) =

<x>

kT
h

Q~I.

QHQI.

which may be evaluated using (9.16) and the statistical mechanical expressions for the various partition functions, (17)

(16)

(11)

With a 30% bond expansion, the bond lengths in HI. are /(H-I) = 0.21 nm and /(1-1) =
0.35 nm. Calculation of the vibrational frequencies requires estimating the force constants
in the activated complex. In the series HF, HCI, HBr, and HI, the quantity [(frequency)
x (bond length)] is essentially constant. Assuming this to be generally applicable, the
vibrational frequency of HI (6.94 x 1013 sec- 1 ) leads to the estimate given for the stretching frequency of HI.. The bending frequency in HI. cannot be estimated from the
properties of HI. However, assuming that HI and HTe bonds are similar, a bending <L
force constant, k~, can be determined from the spectroscopic properties of H.Te (bend-.'
ing frequency 2.58 x 1013 sec-l, bond length 0.165 nm). Using this value for k~, the.
bending frequency for both linear and bent HI. may be calculated. Formulas.. relating
force constants to frequencies are given by G. Herzberg, Infrared and Raman Spectra
(Princeton, N.J.: Van Nestrand, 1945), pp. 168-175.
T. L. Hill, Introduction to Statistical Thermodynamics (Reading, Mass.: AddisonWesley, 1960), Chapters 4, 8, and 9; or see Appendix A.

Sec. 9.4

Model Calculations of k.(T)

285

Table 9.1. Arrhenius A-Factor for the Reaction H + I.

Using Activated Complex Theory"
A

HI

+I

Calculated

10- 14 (em" mol- 1 sec-I)

HII angle, degrees

180
90
a

250K

500K

750 K

0.055 (0.034)

0.055 (0.028)

0.073 (0.042)

3.04

2.73

2.95

(2.70)

(2.13)

(2.13)

Two possible geometries for the activated complex are considered. At each geometry the
first value is based upon a 30% bond expansion; the values in parentheses are based upon
no bond expansion. The experimental value for A in the temperature range 300-700 K is
2 x )014 em" mol- 1 sec-I.

Introducing the molecular parameters into this expression yields

B(T)

2.05 X 10
<,,> ---=T=1/=214

(l (l -

e-310IT)
e-2560IT)

(9.21)

from which A can be calculated using (9.15). A similar calculation can

be carried out assuming that the activated complex has a bent geometry.
If an HII angle of 90 is assumed, the following set of estimated molecular
parameters is obtained for the HI2 complex:
g

= 2,

= 1,

ABC

= 3.63 X 10-135 kg 3 m 6 ,

= 5.3 X 1013 , 1.45 X 1013 sec- 1

Here ABC is the product of the three principal moments of inertia. (18)
The difference is that one vibrational degree of freedom has been altered
to a rotational degree of freedom. By using statistical thermodynamics a
result analogous to (9.21) is found. The results of calculations assuming
linear and bent configurations for the activated complex are given in
Table 9.1 for three temperatures assuming both 30% and no bond expansion. A
of 1 was assumed so that the calculation should overestimate
the A-factor if the properties postulated for the HI2 complex are reasonable.
Some interesting points emerge. The calculation is insensitive to the
amount of bond expansion assumed. The temperature dependence of A
for the two geometries is quite small, which is a general feature so that
any temperature in the experimental range can be used in these calculations.

<,,>

(18)

The procedure for calculating the principal moments of inertia is given by I. N. Levine,
Molecular Spectroscopy (New York: John Wiley, 1975), pp. 197-203.

Chap. 9

286

Theories of Reaction Rates

Table 9.2. Comparison of A-Factors Computed Using Activated Complex Theory

and Collision Theory with Experimental Resultsa

Reaction
Experiment

+ 0 ~ NO. + O.
NO. + F. ~ NO.F + F
NO. + CO~ NO + CO.
Fa + CIO. ~ FCIO. + F
2NOCI ~ 2NO + Ci.
2CI0 ~ CI. + O.
COCI + CI ~ CO + CI.
NO

0.80
1.6
12

Collision
theory

Activated
complex theory

47

0.44

59

0.12

110

6.0

0.Q35

47

0.082

9.4

59

0.44

0.058

26

0.010

65

1.8

400

D. R. Herschbach, H. S. Johnston, K. S. Pitzer, and R. E. Powell, J. Chern. Phys. 25, 736 (1956).

The A-factors for the two geometries are very different. The bent activated
complex leads to a reasonable value; the linear complex leads to a gross
underestimate. In the bent geometry the calculated A-factor is 1.5 times
the experimental value, which is remarkable considering the approximate
nature of the calculation. In the linear geometry the calculated value is
30-40 times too small. Thus we can argue that this structure for the activated complex is very unlikely. (19) The calculation discriminates against
one geometry. It does not prove the other to be correct. Nonetheless given
a choice between a linear or nearly linear activated complex and one that
is strongly bent, there is reason to opt for the latter.
Similar calculations have been carried out to estimate the A-factor
for various reactions. The results, for some of the examples of Table 8.1,
are given in Table 9.2. The calculations did not account for bond-length
increases in the activated complex. Since expansion affects both the moments
of inertia and the vibrational frequencies in the same way, recalculation
would increase the predictions of the theory by factors of 3-5. Even without
this correction the predictions of activated complex theory are generally
within a factor of 10 of the experimental value. Only when the experimental
(19)

It is interesting that the product distribution in molecular beam studies of the H + I.

system also suggests a bent geometry for the activated complex. See D. R. Herschbach,
Disc. Faraday Soc. 55, 233 (1973).

Sec. 9.5

Kinetic Isotope Effect

287

A is very large are there gross discrepancies. However, there are other
difficulties. According to the theory as developed in Section 9.3, the mean
transmission coefficient is less than 1. Thus activated complex theory should
always overestimate k 2 (T), and therefore the A-factor as well. Even recognizing that the method of calculation used in making the predictions of
Table 9.2 leads to underestimates, the differences between calculation and
experiment are too large to be attributable to neglect of bond expansion in
the complex. Whether the differences are due to weaknesses in the theory
leading to (9.14) or to errors in the proposed structure of the activated
complex is unclear. In any case activated complex theory provides another
tool, albeit one which must be laced with a liberal dose of chemical intuition, to be used in studying a chemical reaction. It must be emphasized
that agreement between calculated and observed values of A does not
"prove" anything. As with alI other kinetic arguments disagreement indicates that the postulated mechanism or structure for the activated complex
is incorrect. A case in point is the H2 + 12 reaction. The A-factor calculated
assuming the Bodenstein mechanism

is 4 X 1013 cm 3 mol -1 sec-I, half the experimental value. (20) In spite of the
satisfactory agreement, there is other kinetic evidence that the reaction
does not proceed via the bimolecular mechanism.

9.5. The Kinetic Isotope Effect

Statistical thermodynamics can be used to explain the effect of isotopic
substitution on equilibrium constants. Similarly the statistical theory of
reaction rates provides a basis for understanding isotope effects in rate
constants. Consider a prototype atom-molecule reaction
A

+ BC-----"AB + C

where B can exist in a number of isotopic forms. Then, from (9.14), the
ratio of the rate constants for reaction involving different isotopes Band
B' is
Q(BC)
(9.22)
Q(B'C)
(20)

S. W. Benson, The Foundations of Chemical Kinetics (New York: McGraw-Hili, 1960),

pp. 285-286; recalculation based upon the quoted data leads to the value given here.

288

Chap. 9

Theories of Reaction Rates

The partition functions are determined by nuclear motion on the potentialenergy surfaces for the BC molecule or the ABC complex, and the activation
energy is the potential step required to traverse the lowest energy path on
the ABC surface. The phenomenon is a direct manifestation of quantum
mechanics. In the classical limit (massive species) the ratio is 1.
Simplification of (9.22) is based on the fact that the potential-energy
surface is the ground state for electronic motion in a charge distribution
determined by the nuclear configuration. Since isotopes have the same
nuclear charge, the potential-energy surface is unaffected by isotopic substitution. As a consequence the classical barrier height is isotope independent
and (9.22) becomes
k'
k

Q(BC)
Q(B'C)

<x')
<x)

(9.23)

Furthermore the shape of the surface in the reactant valley and at the
saddle point is not changed. The force constants for nuclear motion in
either region are invariants. Knowledge of the moments of inertia and
vibrational frequencies of one isotopic species allows computation of the
same quantities for any other. For a diatomic molecule like BCthe results
are particularly simple
I' _ (

)2 _

T - --;;- -

mB,(mB
mB(mB'

+ me)
+ me)

Similar, but more complicated, expressions can be found relating the properties of any molecular species.
Further simplification of (9.23) can be carried out in different ways.
In one, the quantities Q+ and Q are determined by a calculation similar
to the one used in Section 9.4 to compute the A-factor for the H + 12
reaction, leading to an expression involving masses, moments of inertia,
and vibrational frequencies. The other method depends upon the solution
of the vibrational secular equation in the reactant valley and at the saddle
point. (21) The result, for a linear transition state, is
k'

<x')

T - <x)
(21)

v'+

vr

r(BC)
r(B'C)

r 1(AB'C) [ r 2(AB'C)
r 1(ABC)
r 2 (ABC)

]2

(9.24)

Detailed analyses are given by J. Bigeleisen and M. Wolfsberg, Adv. Chem. Phys. 1,
15 (1968); L. Melander, Isotope Effects on Reaction Rates (New York: Ronald Press,
1960), Chapter 2.

Sec. 9.5

289

Kinetic Isotope Effect

where

The frequencies VI and V 2 correspond, respectively, to the symmetric stretch

and the doubly degenerate bend of the activated complex. Motion along
the reaction coordinate replaces the asymmetric stretch. The quantity vt,
which is a computational device without physical meaning as a frequency,
is determined by the curvature of the reaction profile at the peak. For a
profile like that of Fig. 9.3 it is
vt

=--

2n

(-k
)1/2
-fl

where k is the curvature 0) and fl is the reduced mass for motion along
the reaction coordinate. Clearly vt, while having the dimensions of frequency, has no relation to the rate at which systems cross the pass; that
quantity is determined by (9.9).
After the potential-energy surface near the saddle point has been
constructed and the properties of the activated complex determined, all
factors in (9.24) except for the transmission ratio </(')/</() may be immediately computed. The transmission ratio may be calculated if highly
simplified models of the reaction profile are used. However, in most instances little error is introduced by assuming that </(')/</() = 1.(22) The
exceptions are for reactions in which isotopes of hydrogen are intimately
involved in the reaction process. While both the transmission ratio and the
partition function ratios of (9.24) approach 1 for massive reactants, the
transmission ratio tends to 1 more rapidly. Therefore its influence on the
isotope effect is relatively less significant, except at low temperature.
A system for which the effect of isotopic substitution has been thoroughly investigated is the reaction
Cl

+ H2 ----' HCl + H

The rate law for the rate of disappearance of reactants is

_ d[iH jH]
dt

k.[C1][iH jH]
~J

'

i, j = 1,2,3

Experimental values for the isotope effect knlk ij are plotted in Fig. 9.8.
Values computed from (9.24) are included in the figure. The parameters
(22)

R. E. Weston, Jr., Science 158, 332 (1967).

Chap. 9

290

Theories of Reaction Rates

50~
20

kll

10

kij

.... .. ..
3.0

IH3H
IH2H

4.0

Fig. 9.8. Kinetic isotope effect for the reaction H. + CI ~ HCI + H. Solid lines calculated
using (9.24). [Adapted from R. E. Weston, Jr., Science 158, 332 (1967), where original references are given.]

required to evaluate (9.24) were calculated using a semiempirical potentialenergy surface; variation of the transmission ratio was taken into account.
Note that the kinetic isotope effect can be quite large; 3H 2 at the lower
temperature studied (273 K) reacts 35 times slower than IH 2 The calculated
values are in fair agreement with the data. The trends are correctly predicted
and numerical discrepancies are not large. It appears that deviations become
more severe at lower temperatures. Whether these are due to the weaknesses
of transition-state theory itself, the model used for the potential-energy
surface, or the way in which the transmission ratio has been estimated is
unclear.
Isotopic substitution in the Cl + H2 system leads to a "normal"
isotope effect, i.e., [k(light)]/ [k(heavy)] > 1. In other systems this ratio
is less than I-the so-called "inverse" isotope effect. The isotope effect is
determined by the difference between the shape of the potential-energy
surface at the reactant configuration and at the pass. Ignoring vibrational
excitation, the ratio (9.24) reflects differences in zero point energies (Problem
9.15). In the example considered the zero point energy of the activated
complex is less than that of the reactants-the activated complex is "floppy."

Sec. 9.6

Unimolecular Reaction-General Features

291

If the activated complex were "constrained," as is the case when the isotopic
reactants are atoms, the opposite would be true; an "inverse" isotope effect
is expected. (23)

9.6. Unimolecular Reaction-General Features

In Chapter 5 we examined the phenomenology of unimolecular reactions and their pressure-dependent order of reaction. The simple mechanistic model (5.18) predicted the rate law

_ d[A]
dt

k* [A]

(9.25)

with k*, a concentration-dependent "first-order" rate constant given by

(9.26)

where koo is the high-pressure unimolecular rate constant and kd/k_ is

the ratio of the rate of reaction of activated molecules to their rate of
deactivation. To provide a theoretical model, we must recognize that
activated molecules are formed with a distribution of energies and that
the rate constant for decomposition to products must be energy dependent,
as follows from a rudimentary molecular picture of the decomposition
process. As an example consider the reaction

For products to form, the ethyl-carbonyl C-C bond must stretch and the
H atom on the carbonyl group must approach the ethyl group. Only if
sufficient internal energy is concentrated in a "critical mode" that leads
to reaction can decomposition occur. The total internal energy of a molecule
is distributed among all its internal modes. Only rarely is it concentrated
in a single vibrational mode. Thus, if the internal energy E is just greater
than the threshold energy Eo, few molecules will be collisionally activated
to a state for which sufficient energy is in the critical mode; the decomposition probability is thus small. As E increases, the likelihood of finding an
energy Eo in the critical mode also increases thus enhancing the decomposition probability.
(23)

K. J. Laidler, Chemical Kinetics, 2nd ed. (New York: McGraw-HilI, 1965), p. 97.

292

Chap. 9

Theories of Reaction Rates

These ideas on energy transfer can be formalized by considering the

rate of formation and the rate of destruction of excited molecules with
energies between E and E dE. Excited molecules are formed via collisions.
If n is the concentration of unexcited A molecules, the total collision rate
as computed in (2.21) is yn2j2, where y, assuming the molecules to be hard
spheres of diameter d, is(24)

= 4nd 2 (kTjnm)1!2

(9.27)

If we assume that reaction does not significantly affect energy distribution

among the low-energy states of the A molecule, the collisions between
unexcited molecules produce molecules with an energy distribution typical

of a gas at equilibrium. The approximation, that reaction does not affect

the partitioning among low-energy states, is similar to the one used in
estimating the number of composite systems in the development of activated
complex theory (Section 9.3). If the equilibrium probability density is
P(E), the total rate of production of excited molecules with energies in the
range E to E dE is

(9.28)
A factor of 2 is introduced because in the collision process either of the
A molecules might be excited.
While the population of low-lying states identified with A is not
altered by reaction, the population of the states identified with A*, i.e.,
those with sufficient energy to react, is affected. The total concentration
of A* molecules with energy between E and E dE is n~(E)P(E) dE, where
HE) is a depletion factor accounting for reaction. Since A *-A* collisions
are highly unlikely, they can be ignored. The total collision rate of A *
molecules of the specified energy is yn[n~(E)P(E) dE). Since the overwhelming
majority of A-A * collisions lead to deactivation, the total rate of destruction
of A * molecules of energy E is

(9.29)
(24)

We ignore any relation between collision frequency and the internal energy of the
products of the collision. Furthermore, the effect of the intermolecular potential is
ignored. To lift these restrictions a more general formulation of the problem, based
upon the "master equation" must be constructed. The master equation focuses upon
the transition probabilities botween molecular states. For a discussion and a general
review of unimolecular reactions see D. C. Tardy and B. S. Rabinovitch, Chern. Rev.
77, 369 (1977).

Sec. 9.7

Unimolecular Reaction-RRKM Theory

293

The second term accounts for the effect of reaction; k(E) is the reaction
rate constant for molecules of energy E. Once a steady state is established
(9.28) and (9.29) are equal and we find
1

(9.30)

+ k(E)/yfJp

where the ideal gas law, n = fJp, has been used. At high pressure there is
little depletion of excited states due to reaction. At low pressure the effect
is important; molecules react too rapidly for the excited-state population
to be maintained by collisional activation. At low energy, where k(E) = 0,
there is also no depletion.
The total rate of product formation is found by integrating the second
term in (9.29) over all energies greater than Eo, the threshold for reaction

so that, using (9.25) and (9.30) the unimolecular rate constant k* is

(9.31)
which has as its high- and low-pressure limits
(9.32)

Note that at low pressure the rate constant is determined by the fraction
of activated molecules. If k(E) is assumed to be constant for E > Eo, (9.31)
reduces to the Lindemann expression (9.26).

9.7. Unimolecular Reaction-RRKM Theory

To apply (9.31) requires estimating the energy dependence of the excited-state population peE) and the decomposition rate constant k(E).
Progress to this end, based upon treating molecular vibration as harmonic, was made by Hinshelwood, (25) Rice and Ramsperger, (26) and Kas(2S)
(26)

C. N. Hinshelwood, Proc. Roy. Soc. (London) AI13, 230 (1926).

O. K. Rice and H. C. Ramsperger, J. Am. Chern. Soc. 49,1617 (1927); SO, 612 (1928).

Chap. 9

294

Theories of Reaction Rates

sel. (27) The modern picture, based, as is activated complex theory, (28) upon
identifying the reaction coordinate (the critical coordinate in unimolecular
vocabulary), proposes the following three-step process to describe unimolecular decomposition
A+A~A*+A

A*

At ----'" products

In the first step the molecule is collision ally excited. There is then intramolecular energy transfer as A* undergoes various conformational changes.
Among the states of A * is a subset that describes the critical conformation
A t, defined so that there is an even chance that molecules in the configuration At react. (29) The distinction between the isoenergetic A * and At is
apparent in considering the reaction cyclopropane ----'" propene. Unless
sufficient energy resides in a C-C vibrational mode a cyclopropane molecule
does not isomerize, no matter how energized it is. As energy is redistributed
the molecule evolves into the critical configuration from which it may
react or revert back to states of the activated molecule.
The RRKM theory proposes a model to describe energy transfer among
molecules belonging to the classes A* and At. It assumes the following:
(i) In collisional processes that form A * molecules of energy , molecules are randomly distributed among all degenerate states at that energy
(strong collision assumption).
(ii) Vibrational energy redistribution within states of A * is rapid with
respect to the rate of unimolecular reaction (random access assumption).

Strong collision means that there is no preference for populating a subset

of states of A * in the excitation process. It is the antithesis of laser activation
(discussed in Section 6.8) in which particular vibrational-rotational states
are populated. Random access means that molecules produced in particular
states of A * are rapidly transformed into states of At; there are no states
that are weakly coupled. (30) Since rotational-vibrational (R-V) energy
(27)
(28)

(29)

(30)

L. S. Kassel, J. Phys. Chern. 32, 225, 1065 (1928).

R. A. Marcus and O. K. Rice, J. Phys. Chern. 55,894 (1951); R. A. Marcus, J. Chern.
Phys. 20, 359 (1952).
The form of RRKM theory presented here is adapted from W. L. Rase, in Modern
Theoretical Chemistry, W. R. Miller, ed. (New York: Plenum Press, 1976), Vol. 2,
Part B, Chapter 3.
Non-RRKM behavior is discussed by W. L. Rase, in Modern Theoretical Chemistry,
W. R. Miller, ed. (New York: Plenum Press, 1976), Vol. 2, Part B, Chapter 3; and
D. L. Bunker, Theory of Elementary Gas Reaction Rates (Oxford: Pergamon Press,
1966), Chapter 3.

Sec. 9.7

Unimolecular Reaction-RRKM Theory

295
Total En ergy

r t .L

T- . .

r
-i r-

v t

v
_
o -

rO'' tiOI """ lJ

required to
attain the critical
configuration

J.
Fig. 9.9. Diagram for energy disposition in RRKM theory.

transfer is slow, only V-V transfer can significantly affect unimolecular

reaction. (31)
The energetics of the species A * and At are illustrated in Fig. 9.9.
For all states considered the energy, E, is greater than a threshold value,
Eo, required to form molecules in the states of At. The energy of At available
to drive reaction is only Et = E - Eo - E/, since due to conservation of
angular momentum, some of the internal energy of the molecule cannot
be converted to vibrational energy. Of the total useful energy, a portion,
Eo t, is concentrated in a critical coordinate (the analog to the reaction
coordinate of activated complex theory).(32)
The molecular states that correspond to either A * or At are highly

excited ones; as a result they are nearly continuous. To compute P( E) and

k(E) it is convenient to reformulate the idea of degeneracy in terms of a
density of states, N(E). The total number of states (the degeneracy) in the
energy range E to E + dE is N(E) dE; N(E) is the number of states per unit
energy. With this definition, the equilibrium probability for a molecule
to be in a state of energy E, peE) dE, is
(9.33)
(31)

R-V transfer requires a change of angular momentum, which is not possible in an

isolated molecule unless it undergoes a pure rotational transition. Typical rotational

lifetimes at very low pressure are ~1O-6 sec, much l.)nger than the decomposition
time.
(32) Note that lOr and lOr t are not necessarily equal. Although angular momentum is conserved
as A* evolves to At, the rotational energy may change since the molecular structure is
not invariant. The separation of rotational and vibrational degrees of freedom is only
approximate in both configurations.

296

Chap. 9

Theories of Reaction Rates

where qi is the partition function for the internal degrees of freedom of the
molecule.
To compute k(E) consider the basic RRKM assumptions. They imply
no preference for populating states of the same energy. Thus the states of
A * and At are populated in proportion to their degeneracies. At an energy
E and a specified angular momentum the ratio [At]J[A*] is determined
statistically
[A t] _ g/N(Et)
[A*] - grN(Ev)
where N(Et) is the density of vibrational states of At molecules with vibrational energy Et and likewise N(Ev) for A * molecules with vibrational energy
Ev. The rotational degeneracies gr t and gr are equal since angular momentum
is conserved; as a result no rotational transition can occur as the molecule
evolves from A * to At. The density of states N(Et) is further decomposed
since the critical mode is presumed to be distinguishable from all other
vibrational modes of At,
N(Et) =

L g(E/)N:(E t -

E/ = E/)

.~~o

Here Nc(E/) is the density of states of the critical mode at energy } and
g( Ev t) is the degeneracy of the other vibrational modes at energy Ev t. This
decomposition, treating the critical coordinate as continuous and all other
vibrational modes as discrete, is analogous to distinguishing a specific
reaction coordinate in activated complex theory. Defining an energydependent frequency of passage from A t to products, Y(Ec t), the expression
for the reaction rate constant of A * molecules is
(9.34)

The factor of ! accounts for the fact that half the At molecules revert
to A * molecules. The bracketed expression is the probability that a molecule
with total energy E and vibrational energy Ev has an energy Ec t concentrated
in the critical coordinate.
The product Y(E/)Nc(Ec t) can be estimated from uncertainty principle
arguments. A density of states Nc(E/) implies a separation between levels
--.-1/Nc(E/) and a corresponding energy uncertainty -;5IJNcCE/). Since
(lifetime) X (energy uncertainty) ,....., h

Sec. 9.7

Unimolecular Reaction-RRKM Theory

297

we find that(33)

The lifetime is twice the reciprocal of the passage frequency V(E/). (34) Then

and with (9.34)135)

(9.35)
In this form k(E) depends upon En Ev , and E/ since, from Fig. 9.9,
Et = E - Eo - E/ = Ev - Eo + (Er - E/). It is qualitatively correct; as
E -" Eo, Et decreases and fewer vibrational states of At are accessible. As
a result k(E) -" O.
Calculations based upon (9.35), (9.33), and (9.31) require detailed
models to determine the density of states in the critical configuration. The
problems are similar to those which arise in the application of activated
complex theory, but possibly more difficult. Useful expressions for k*
are found by assuming rotational and vibrational degrees of freedom are
separable. The elaboration of (9.33) is then
(9.36)
where N(Ev) and qv are the density of states and partition function for vibration, respectively, and likewise N(Er) and qr for rotation. Substituting this
and (9.35) into (9.31) yields
k*

_1_
hqrqv

foo fOO L:!~o g(E/)N(Er ) exp[ -,B(Er + Ev)] dE, dEv

0

+ k(E)/Y,Bp

(9.37)

Since the sum over vibrational states of At is zero unless Er + Ev = E > EO,
the range of integration has been formally extended. It is generally more
(33)

Careful analysis is required to demonstrate that h, and not Ii, appears here. O. K. Rice,
J. Phys. Chern. 65, 1588 (1961).

(34)
(35)

The factor of 2 reflects the even chance of proceeding from A + to products.

A different argument, based upon treating motion along the critical coordinate as
equivalent to translational motion in a one-dimensional box, yields the same result.
See W. L. Rase, in Modern Theoretical Chemistry, W. R. Miller, ed. (New York: Plenum Press, 1976), Vol. 2, Part B, Chapter 3.

298

Chap . .9

Theories of Reaction Rates

convenient to express k* in terms of the properties of the critical configuration. Since the rotational degeneracy of both A * and A t are the same we
find

If we assume Eo is the same for all states in the critical configuration then,
incorporating the energy constraints of Fig. 9.9, k* is

The high-pressure limit determined from (9.38) is

The integrals are evaluated in Appendix B and the result is

(9.39)
where qv t is the vibrational partition function for all noncritical vibrational
modes of At. The result is identical to that found by applying activated
complex theory and assuming a quasi-equilibrium population of critical
molecules At. At high pressure, where depletion is not significant, the two
approaches must be equivalent. At low pressure the correspondence cannot
hold.
The low-pressure limit is most easily deduced directly from (9.31)
and (9.33); it is

which is precisely the result of collision theory since the bracketed expression
is the fraction of molecules with E > Eo. To relate ko to the properties of
the critical configuration is more difficult and can only be done approximately. Various prescriptions have been suggested. (36)

(36)

See W. L. Rase, in Modern Theoretical Chemistry, W. R. Miller, ed. (New York: Plenum Press, 1976), Vol. 2, Part B, Chapter 3.

Sec. 9.8

Critique of RRKM Theory

299

9.B. Critique of RRKM Theory-Isomerization of Methyl

Isocyanide
The RRKM theory was designed to account for the pressure dependence of the unimolecular rate constant. The theory has been successfully
applied to many reactions, e.g., cyclopropane
propene, (37) C2H 5
C 2H 4 + H, (38) and CH 3NC
CH 3 CN. (39) Of these the latter is a particularly
interesting example. The theory accounts for the experimental results
over a range of temperature yet there is evidence that energy transfer within
the CH 3NC molecule is not described by the RRKM hypotheses. The falloff
curves calculated using (9.38) provide a test of RRKM theory. The falloff
ratios in isomerization reactions are fairly insensitive to the choice of critical
mode or to moderate variation of the structure and the vibrational frequencies in the critical configuration as long as the molecular model reproduces the high-pressure Arrhenius A-factor.(40) Thus the specifics of the
dynamical model of the isomerization process are not too important, which
limits the rigor of the test.
Making reasonable estimates of the molecular parameters and adjusting
d to make theory and experiment coincide when k*/kC<) = 0.1 yields the
results of Fig. 9.10. The corresponding values of d are given in Table 9.3.
The agreement between theory and experiment is gratifying. An indication
of the theory's consistency is the behavior of d. Its temperature variation
is small and its value is very close to the hard-sphere diameter determined
from viscosity measurements, 0.45 nm.
Both observations suggest that the observed agreement is not fortuitous
since d, in RRKM theory, establishes the cross section for collisions in
which there is substantial energy transfer. In any reasonable theory this
cross section must be close to that which determines the transport coefficients
and must not be very temperature sensitive.
The structure of the critical configuration was tested by means of
isomerization studies on a series of isocyanides: CD3NC, CH 2DNC,
C2H5NC, and C 2D 5NC. (41) With the same critical configuration as was
-->.

-->.

-->.

(37)
(38)

(39)
(40)
(41)

M. C. Lin and K. J. Laidler, Trans. Faraday Soc. 64, 927 (1968).

J. V. Michael and G. N. Suess, J. Chem. Phys. 58, 2807 (1973).
F. W. Schneider and B. S. Rabinovitch, J. Am. Chem. Soc. 84, 4215 (1962).
G. M. Weider and R. A. Marcus, J. Chem. Phys. 37, 1835 (1962).
F. W. Schneider and B. S. Rabinovitch, J. Am. Chem. Soc. 85, 2365 (1963); B. S. Rabinovitch, P. W. Gilderson, and F. W. Schneider, J. Am. Chem. Soc. 87, 158 (1965);
K. M. Maloney and B. S. Rabinovitch, J. Phys. Chem. 73,1652 (1969); K. M. Maloney,
s. P. Pavlou, and B. S. Rabinovitch, J. Phys. Chem. 73, 2756 (1969).

300

Chap. 9

Theories of Reaction Rates

log,o p (otm)
Fig. 9.10. Pressure dependence of the unimolecular rate constant for CHaNC isomerization.
For clarity the 533 K curve is displaced one log p unit to the left and the 473 K curve is
displaced one log p unit to the right. Solid lines are calculated values based upon RRKM
theory, adjusted to coincide with experiment when k* /koo = 0.1. [Adapted from F. W.
Schneider and B. S. Rabinovitch, J. Arn. Chern. Soc. 84, 4215 (1962).1

Table 9.3. Temperature Dependence of Collision Diameter as Required by RRKM

Theorya
T(K)

d (nm)

473
504

0.43
0.47

533

0.44

a The experimental value is 0.45 nm.

used to correlate the CHaNC data, good agreement between theoretical

and experimental falloff ratios was found in all cases.
There remain unresolved questions. Trajectory calculations, based
upon classical treatment of intramolecular energy transfer, suggest that
the random access hypothesis does not apply to CH aNC.(42) It appears that
(42)

D. L. Bunker, and W. L. Hase, J. Chern. Phys. 59, 4621 (1973); H. L. Harris and D. L.
Bunker, Chern. Phys. Lett. 11,433 (1971).

Sec. 9.8

Critique of RRKM Theory

301

there may not be facile energy transfer between states of the critical configuration and some of the states of the activated molecule. How this is
to be interpreted in light of the agreement depicted in Fig. 9.10 is, at present,
unclear. Possibly a quantum treatment of energy flow would establish
CHaNC as obeying the RRKM assumptions. Alternatively, it may be that
the RRKM approach is not sufficiently sensitive to its own assumptions,
at least for the CHaNC isomerization reaction. The fact that the falloff
ratio, for isomerization reaction, is not too dependent on the detailed
structure of the critical configuration(4a) gives this idea some credibility.
If the RRKM assumption of random access is invalid, different models
must be formulated in order to obtain an expression for k(c), if (9.31)
is to be used. Slater(44) investigated the other extreme model and assumed
no vibrational energy transfer between normal modes of a molecule. (45).
This approach is too restrictive. Recent work attempts to steer a middle
(and more difficult) course. (46) The dynamics of vibrational energy transfer
are incorporated in the theory. Effects attributable to vibrational excitation
of modes both strongly and weakly coupled to the critical coordinate can
then be exhibited.
Experimental tests of the RRKM hypotheses have also been developed.
By means of chemical activation molecules are formed with sufficient
energy to undergo unimolecular processes directly. Typical activation
processes are H-atom addition to olefins, radical recombination, and CH 2
addition or insertion. Molecules formed in this way have non-Boltzmann
energy distributions; their rate of decomposition can be used to test the
random access hypothesis introduced previously. A rather different formulation for the rate constant is needed. (47) The decomposition of C 2H 5, formed
by chemical activation, can be accounted for using the same parameters
for the critical configuration that were used to fit the thermal decomposition
(43)
(44)

(45)

(46)

(47)

G. M. Weider and R. A. Marcus, J. Chem. Phys. 37, 1835 (1962).

N. B. Slater, Theory of Unimolecular Reactions (Ithaca, N. Y.: Cornell University Press,
1959).
Such a model assumes, in effect, that molecular vibration is adequately treated as harmonic. Theories of crystals, which might be viewed as enormously large molecules, have
shown conclusively that anharmonicity is needed to properly account for nonequilibrium behavior; see 1. Prigogine, Nonequilibrium Statistical Mechanics (New York:
Interscience, 1962), Chapter 2.
W. M. Gelbart, S. A. Rice, and K. F. Freed, J. Chem. Phys. 52,5718 (1970); 57,4699
(1972).
W. L. Hase, in Modern Theoretical Chemistry, W. H. Miller, ed. (New York: Plenum
Press, 1976), Vol. 2, Part B, Chapter 3.

302

Chap. 9

Theories of Reaction Rates

data. (48) More often, thermal and chemical activation experiments cannot
be carried out on the same molecule. Instead, experiments are performed
on a homologous series for which a single critical configuration is suitable
throughout. The most extensive kinetic studies, involving chemiactivated
radicals formed by H-atom addition to olefins, can be rationalized using the
RRKM approach. (49)

9.9. Thermodynamic Analogy

In addition to theoretical estimates of the rate constant based on postulated structures for the activated complex, the statistical theory of rate
constants can be used to correlate kinetic trends in homologous series of
reactions. For this purpose we construct a "thermodynamic" interpretation
of the rate constant.
The reaction rate, in terms of activated complex theory, is the number
of A-B composite systems that traverse the pass per second. We computed
it by treating separately two aspects of the problem. First local equilibrium
was assumed and the number of A-B complexes calculated. Then progress
along the reaction coordinate was considered and the rate of barrier crossing
determined. From (9.10)-(9.12) this frequency (sec- 1 ) can be identified
as kT<"')lh. Assuming <"') = 1, the reaction rate is
(9.40)
where NiB is to be interpreted as the number of activated complexes with
energy greater than the pass energy; these complexes are presumed to
differ from excited reactants or products by having molecular configurations
which closely resemble the structure of the activated complex. Naturally
this identification is empirical. Transforming to concentration units (9.40)
becomes

from which the rate constant may be identified as

(9.41 )
(48)
(49)

J. V. Michael and G. N. Suess, J. Chern. Phys. 58, 2807 (1973).

C. W. Larson and B. S. Rabinovitch, J. Chern. Phys. 52, 5181 (1970).

Sec. 9.9

Thermodynamic Analogy

303

The ratio niBjnAnB has the form of an equilibrium constant for formation
of activated complex. For systems in which the activity of the various
species equals their concentration (dilute gases, dilute solutions of nonelectrolytes) (9.41) defines a totally empirical function, the activation equilibrium constant Kt(T),
(9.42)
Continuing in this vein, the corresponding enthalpy, entropy, and Gibbs
free energy of activation are

and the rate constant is(50)

k 2 (T)

kT
-h- exp( -!JGotjRT)

kT
-h- exp(!JSotjR) exp( -!JHotjRT)

(9.43)
These quasi-thermodynamic parameters can be related to the Arrhenius
activation energy by means of (8.7); using (9.42) the result is
Ea

dIn Kt
RT + RP --;:::::dT

As the equilibrium constant Kt is based on concentrations

dIn Kt
dT
and
Ea

RT + !JUot

!JHot

+ RT(l

!JHot

+ RT -

- !Jngt)

!J(PV)t
(9.44)

where !Jngt is the change in the number of moles of gas in the activation
process. For gas-phase binary reactions !Jng t = -1.

(50)

Units must be treated carefully in (9.43). The left-hand side has dimensions of a secondorder rate constant while the right-hand side appears to have dimension (-1. A choice
of standard state is implicit in the definition of the free energy of activation, L1 Go t ..
For details see A. C. Norris, J. Chern. Ed. 48, 797 (1971).

Chap. 9

304

Theories of Reaction Rates

9.10. Gas-Phase Reactions

The thermodynamic analogy is not to be used as a method for establishing absolute values for the various activation parameters; these are of
limited significance since they derive from an equilibrium thermodynamic
interpretation of intrinsically nonequilibrium properties. Furthermore, to
accept such numbers uncritically ascribes a measure of definiteness to the
activated complex which is unwarranted. On the other hand, trends and
similarities may be useful in helping to characterize reaction mechanism.
In Table 9.4 the values of LlSot and LlHot calculated from (9.43) and (9.44)
are given for a number of gas-phase reactions. For the bimolecular reactions
the value of LlSot depends upon the choice of standard state; for rate constants in units of cm3 mol- 1 sec- 1 the natural standard state is a concentration of 1 mol cm- 3
Table 9.4. Activation Parameters for Some Gas-Phase Reactions
Reaction

iJHot (kJ)

Reference

C.H5Cl ~ C.H. + HC!

232

+11.7

C.H.Cl. ~ C.H.C! + HC!

220

+16.5

Cyclopropane ~ Propene

270

+50.0

Cyclobutene ~ trans-Butadiene

133

+15.6

24

-20.4

CH. + C.H. ~ CaH7

3.3

+26.5

-8.3

+19.4

7.9

-80.1

CHa + C.H. ~ CH. + C.H5

35.1

-49.7

CHa + C.H. ~ CH. + C2Ha

33.5

-50.1

Isobutene + HC! ~ t-C.HoC!

92.5

-70.9

H + C.H. ~ C.H5
CHa + CHa ~ C2H.
CHa + CO ~ CHaCO

a Standard state for LlSo" for binary reactions is a concentration of 1 mol em-a.

o S. W. Benson and H. E. O'Neal, Kinetic Data on Gas Phase Unimolecular Reactions (Washington: National Bureau of Standards, 1970).
C J. A. Kerr and M. J. Parsonage, Evaluated Kinetic Data on Gas Phase Addition Reactions
(London: Butterworths, 1972).
d V. N. Kondratiev, Rate Constants 0/ Gas Phase Reactions, translation by L. J. Hoitschlag
(Washington: National Bureau of Standards, 1972).
S. W. Benson, The Foundations o/Chemical Kinetics (New York: McGraw-Hill, 1960).

Sec. 9.10

Gas-Phase Reactions

305

The activation entropies for the four unimolecular reactions are

positive and can be interpreted if we assume that the molecule, in its activated state, is less ordered than in the ground state. Disordering occurs
because a bond must be weakened or broken for reaction to be possible.
The cyclopropane isomerization is a particularly good example. As a C-C
bond breaks the molecule passes through an intermediate, possibly a
diradical,

/\
The intermediate has greater rotational and vibrational freedom than
either reactant or product. Thus it is not surprising that LlSo+> LlSrxn ;
the values are 50 and 28 J K -1, respectively.
The numerical values of LlSo+ for binary reactions have no particular
significance since they depend critically on the choice of reference state.
It is no longer meaningful to suggest that LlSo+> 0 indicates an activated
complex that is less ordered than the reactants. On the other hand, trends
are important. The more negative the value of LlSo+, the more severe are
the orientational requirements for reaction to occur, i.e., the pass on the
potential-energy surface is narrow. The more positive the value of LlSo+,
the fewer are the orientational constraints corresponding to a much broader
pass on the potential-energy surface. One possible explanation for positive
LlSo+is that, in the activated complex, the reactants are very loosely bound
leading to a large moment of inertia; this is assumed to be the case for
CRa dimerization.
The trends in the other bimolecular examples are reasonably understandable. There are fewer orientational constraints in R-atom addition
to ethylene than in methyl radical addition; the values of LlSo+are in accord
with this interpretation. The hydrogen-transfer reactions of methyl with
ethane and ethylene have the same entropy of activation, indicative of
similar orientational requirements. The addition of CR 3 to CO has a large
negative LI So +, reflecting the severe geometric constraints necessary for this
reaction to be possible.
It is the Arrhenius A-factor which is related to LlSo+and thus to mechanistic models for the structure of the activated complex. The value of
LlHo+ is determined by the barrier to reaction and it cannot be simply
related to a model for the reaction intermediate.
Alternatively, the thermodynamic analogy can be used to provide
estimates of rate constants. In equilibrium thermodynamics one uses

306

Chap. 9

Theories of Reaction Rates

tabulations of bond enthalpies and entropies to estimate iJHo and iJSo

for compounds for which thermodynamic measurements are unavailable.
Similar, but less reliable, methods have been devised for estimating iJHot
and iJSot . While they require numerous assumptions about the structure
of the activated complex, these are really not much different in spirit from.
guessing the mechanical properties of the activated complex. (51)

9.11. Solution Reactions

The activation entropy for solution reactions can be interpreted in
similar fashion. The solvent exchange reactions

have been studied with X = Cl-, Br-, and NCS-. They are interesting
because, with each ligand, two types of exchange kinetics are found. (52)
Four of the water molecules exchange slowly; iJSot is between 0 and 12
J K-i. The fifth exchanges much more rapidly; iJSot is between 57 and
90 J K -1. These observations may be related to a possible exchange mechanism.
The complex has the octahedral structure shown in Fig. 9.11 a. Exchange
of water may be imagined as occurring through a variety of intermediate
structures; the most extreme examples are shown in Figs. 9.11b and c.
In the first case an extra water molecule is bound before another molecule
leaves; the activated complex is seven coordinate. In the second case a
water molecule is lost to solvent in the activation process; the complex is
five coordinate. The difference between the two intermediates is striking.
Two water molecules bound to the seven-coordinate complex are free to
mingle with the solvent if the complex is five coordinate. The seven-coordinate complex is thus the more ordered and correspondingly should have
a lower value of iJSot. As four of the water molecules exchange with iJSot
< 12 J K-1 we can propose that, for cis exchange, a seven-coordinate
intermediate forms. For trans exchange, on the other hand, a water molecule
leaves before another one is attached; the intermediate is five coordinate.
(51)

(52)

For a discussion of this approach see S. W. Benson, Thermochemical Kinetics (New

York: Wiley-Interscience, 1976).
T. W. Swaddle, Coord. Chern. Rev. 14,217 (1974); D. E. Bracken and H. W. Baldwin,
Inorg. Chern. 13, 1325 (1974).

Sec. 9.12

Primary Salt Effect

307
(c)

(b)

(a)

Fig. 9.11. Structures of the (a) octahedral CrS 6 X complex ion; (b) seven-coordinate intermediate; (c) five-coordinate intermediate. The solvent S is water.

9.12. Primary Salt Effect

In the process of identifying the quasi-thermodynamic activation
parameters in Section 9.9 we assumed that activity could be equated with
concentration. While reasonable for the examples discussed in Sections
9.10 and 9.11 it is obviously inappropriate if we wish to consider ionic
reactions. From (9.41) the rate constant is related to the ratio nh/nAnB
which can no longer be set equal to an equilibrium constant. It is the
activities, not the concentrations, which are related to the equilibrium
constant
(9.45)
Since a

yn, where y is the activity coefficient, (9.45) can be rewritten

t
nAB
nAnB

Kt(T) YAYB

yiB

which can be substituted in (9.41) with the result

(9.46)
The quantity ko(T) is the rate constant in very dilute solution where all
the activity coefficients are unity. The formula (9.46) is obviously applicable
to any system but the activity corrections are most dramatic for ionic
reactions.
In ionic solution the ion activity coefficients are strongly charge dependent; from Debye-Htickel theory we have
(9.47)

Chap. 9

308

Theories of Reaction Rates

where ZA is the ionic charge and I is the ionic strength of the solution
1= L Zi2Ci. In water at 25C the constant B is 1.0. Since the activated
complex in the reaction

+ ~ (AB)zA+ZB -" 'products

has a total charge ZA + ZB we may now use (9.46) to determine the effect
AZA

BZB

of ionic strength on the rate constant. From (9.46) and (9.47) we have
log k 2(T)

log koCT) -

tBI1!2[zA2

+ ZB 2 -

(ZA

+ ZB)2]
(9.48)

Thus increasing the ionic strength speeds up reactions between similarly

charged ions and slows down those between ions of opposite charge, a
point already mentioned in our discussion of diffusion-controlled reactions

VII
-0.6~-----L------~----~~

0.1

0.2

If (M1/2)

0.3

Fig. 9.12. Effect of ionic strength on rates of some ionic reactions:

(1) Co(NH3)5Br2+ + Hg2+ + H 20 ~ [Co(NH3)5H20j3+ + HgBr+;
(II) S20.'- + 1- ~ 13- + 2S0.'-;
(III) [O,N-N-CO,Etj- + OH- ~ N 20 + C03'- + EtOH;
(IV) Sucrose + OH- ~ inversion;
(V) H 20, + H+ + Br- ~ H,O + ! Br,;
(VI) Co(NH.)5Br'+ + OH- ~ Co(NH.)50H"+ + Br(VII) Fe 2+ + Co(C.0 4 )a"- ~ Fe3+ + CO(C 20 4 ).4-.

Some reactions have not been balanced. [Adapted from S. W. Benson, The Foundations of
Chemical Kinetics (New York: McGraw-Hill, 1960), p. 525, where original references are
given.]

Sec. 9.13

Diffusion-Limited Kinetics

309

in Section 5.7. Some data illustrating the effect of ionic strength are plotted
in Fig. 9.12. The expression (9.48) accounts for the observations at ionic
strengths where Debye-Hiickel theory is reliable.(53)
The effect is readily understood qualitatively. Increasing ionic strength
always reduces the magnitude of the electrostatic effects since the ions
become more shielded by their ion atmospheres. Thus repulsion between
ions of like sign is reduced allowing them to approach one another more
easily; the result is an increase in k2(T). Attraction between ions of opposite
sign is also reduced; they approach one another less readily and k 2(T) falls.

9.13. Diffusion-Limited Kinetics-Debye Theory

Activated complex theory and the RRKM approach to unimolecular
decomposition emphasize the specific conformational rearrangements that
are required for reaction to take place. In gas-phase reactions such considerations must be dominant. However, in solutions, reactants that have
diffused together cannot simply collide and separate. They remain constrained by the surrounding solvent molecules which must be displaced
if the particles are to move apart. The reactants remain trapped, typically
for 10-11 sec, and while so constrained undergo "-' 100 collisions with one
another. At the same concentrations species in both gas and solution phase
have the same collision frequency. The difference is that in a gas all collisions are independent while in a solution, two particles, having encountered
one another, make multiple collisions before separating. If the chemical
rearrangement is facile it is the encounter frequency, governed by diffusion,
which sets an upper bound to the bimolecular reaction rate.
To compute the maximum rate of a bimolecular chemical reaction in
solution assume that, after the molecules have diffused together, reaction
takes place with each encounter. (54) In order to simplify the mathematics
we center our attention, and the coordinate system, on a representative
molecule of type B. Molecules of type A diffuse through the solution and,
occasionally, encounter the B molecule. Details of molecular structure are
suppressed; Both A and B are spherical; the distance of closest approach
(53)

(54)

Ionic strength can affect reactions of neutral species. An example is the reaction of
CO(aq) with hemoglobin [L. H. Parkhurst and Q. H. Gibson, J. Bioi. Chem. 242, 5762
(1967)]. Here changing ionic strength greatly alters the activity of the dissolved CO
thus changing the rate constant.
The treatment in this section is modeled after that given by I. Amdur and G. G.
Hammes, Chemical Kinetics (New York: McGraw-Hill, 1966), pp. 59-64.

Chap. 9

310

Theories of Reaction Rates

(0)

( b)

Fig. 9.13. Model for AB relative diffusion through solvent. (a) A and B separated by solvent
sheath. (b) A and B have diffused into contact when reaction occurs; the contact
distance is d.

is d. Since reaction takes place at each encounter, the concentration of A

molecules at the surface of a B molecule is zero; at large distances from the
B molecule the molecular concentration of A is that of the bulk solution,
nAG. In terms of r, the AB distance, the boundary conditions are
r

{d,

(9.49)

00,

Since reaction occurs whenever r = d, the rate of reaction of A with a

representative B molecule is the total flux of A at the surface of a sphere
of radius d, as indicated in Fig. 9.13.
To continue, generalize the ionic flux equation (3.15) to consider
effects due to any potential

Here UA is the potential energy of interaction between A and B molecules.

The net diffusion coefficient, D, is DA + DB' which accounts for centering
the coordinate system on a B molecule. As the molecules and the intermolecular potential are chosen spherically symmetric, J A is orientation
independent and directed along r, the A-B radius vector. Thus we have

dnA

-D ( - dr

dUA )
+ {lnA-dr

(9.50)

The total flux across a surface of radius r is

(9.51)

Sec. 9.13

lit

Diffusion-Limited Kinetics

311

Since the system is in a steady state, h is constant; it is the total flux of

A molecules directed radially toward the representative B molecule. Combining (9.50) and (9.51) we find that
d

h = 4nr2D exp( -(3UA ) dr

[nA

exp({3UA )] = const

(9.52)

If D is assumed concentration independent, (9.52) can be manipulated

into a form that provides a useful expression for fA' Multiply both sides
of the equation by exp( -(3UA)/r2 and integrate over r; the result is

(9.53)
where we have used the boundary conditions (9.49) and the fact that the
A-B interaction UA(r) must be zero at large A-B separation. Since h
represents the rate at which A molecules reach the surface of the representative B molecule, it is the rate of reaction of A per B molecule. Multiplying
by nBo to account for the population of B we find the rate law,

where DA

+ DB

has been substituted for D, and

(9.54)

from which the rate constant can be found. Expressing it in the conventional units M-l sec- 1 yields the rate constant for formation of an encounter
complex,
(9.55)
with d and the D's measured in SI units.
This treatment, an elaboration by Debye of an approach introduced
by Smoluchowski, (55) is based upon a succession of rather gross assumptions.
The equations which describe bulk diffusion were assumed to remain accurate even at distances where A-B interaction is significant. We then im(55)

M. V. Smoluchowski, Z. Physik. Chem. 92, 129 (1917); P. Debye, Trans. Electrochem.

Soc. 82, 265 (1942).

Chap. 9

312

Theories of Reaction Rates

posed a steady state and assumed the D's to be concentration independent.

Using a more realistic approach only slightly alters the results, except for
reactions with very short half-times.
The same approach may be used to calculate the maximum rate constant
for dissociation of the encounter complex. The only difference is the choice
of boundary conditions. To replace (9.49) we have

r=

00,

nAB =

d,

nAB

IjLl V, LI V

= 4nd j3
3

(9.56)

since at infinite A-B separation the complex does not exist and at contact
there is one encounter complex in a volume "-'4nd 3j3. The arguments which
led to (9.53) can be modified to yield

from which the maximum dissociation rate constant is

(9.57)
since

lAB

is the fiux per encounter complex, dln[AB]/dt.(56) Combining

(9.55) and (9.57) yields an equilibrium constant for the formation of

the complex
(9.58)
in units of M-l. This expression, which is valid for any nonchemically
bound complex, was first derived to treat ion-pair formation in liquids. (57)
The limitation to spherical molecules is a more severe restriction; it
may be relaxed by introducing a steric factor which reduces the factor 4n
appearing in (9.52), (9.53), and (9.55). Rough quantitative evaluation of
steric factors have been made. (58) It might seem that such factors are inconsistent with the idea of an encounter followed by multiple collisions.
However, nonspherical molecules must rotate while they are trapped in the
encounter complex if they are to assume a reactive configuration (assuming
that in the initial moments of the encounter the configuration was non(5')

(57)
(58)

The first derivation of (9.57) was given by M. Eigen, Z. Phys. Chern. NFl, 176 (1954).
R. M. Fuoss, J. Am. Chern. Soc. 80, 5059 (1958).
R. M. Noyes, Prog. React. Kinetics 1, 129 (1961).

Sec. 9.13

Diffusion-Limited Kinetics

313

reactive). Since molecules remain together for,..."" 10-11 sec there is time for
,..."" 10 rotations, which would seem to allow a sampling of many orientations.
However, if the reactants are fairly strongly associated with the surrounding
solvent, any such rotation would be significantly hindered, and the reaction
rate correspondingly less than the limiting value for spherical molecules.
The most common application of (9.55) and (9.57) is to the problem
of ionic reactions in solution. If the solution is very dilute the AB potential
energy is
(9.59)
where E is the dielectric constant of the bulk solvent. For this simple potential (9.54) can be integrated with the result
(9.60)
The value of kdiff can then be estimated using (9.55). From the NernstEinstein equation (3.13) and the data of Table 3.1 we find that ion diffusion
coefficients at infinite dilution are ,...",,0.6-2 X 10-9 m 2 sec- 1 at 25C (except
for H+ and OH- which are 9.3 X 10-9 and 5.3 X 10-9, respectively). Typical
encounter distances are ,...",,0.5 nm and in water at 25C E"""" 80. Values of
kdiff computed on the basis of these assumptions are plotted in Fig. 9.14

II

10
/

..Q

.....-

--

"'"Q
0>

..... ",,11

o //
o/if

/0/
0
/
/

(/)

::ii!

'"

.4'

J>" "

",-6

Reaction Type
Experimental
Data

Calculated
Values

H+ + B

------

0 A+B

/I

---

o A + OH-

H++OW

-2

-4

ZAZe

Fig. 9.14. Comparison of experimental and calculated values of k dilf for diffusion-limited
reactions. Values calculated using (9.55).

Chap. 9

314

Theories of Reaction Rates

as a function of ZAZB; for comparison, the experimental rate constants

quoted in Table 5.3 for ions with the same charge product are also indicated.
The agreement between theory and experiment is excellent, which suggests
the model represents accurately the major features which account for the
magnitude of the rate constants. The same approach allows estimation of
kdiss from (9.57). For uncharged species in water at 25C the result is
k diss ' " 3.1 X 1010 sec-I. At concentrations ;;:; 10-4 M where the ion-atmosphere affects conductance and equilibrium properties, it also affects kdiff
and k diss . The potential energy (9.59) must be modified to account for the
shielding in which case (9.54) can only be integrated numerically. The net
result is to somewhat reduce the effects attributable to ionic charge. (59)
The calculation of kdiff and k diss was based on two assumptions:
(i) chemical reaction of the encounter complex is instantaneous;
(ii) the system is in a steady state.
Neither assumption is universally valid. As a result, a measured rate constant, even if of the order of magnitude expected for kdiff' cannot be directly ide~tified as k diff .
.
If the rate constant for reaction of the encounter complex is k2' a
steady-state treatment, presuming the mechanism
A

+ B ~ (A
kd!ff

kdlss

k.

B)~products

yields the rate law

d[A]
dt

so that unless k2 ~ k diss the measured value of the apparent rate conStant
does not depend on kdiff alone. While k2 is hard to estimate, the related
quantity k' = k2kdifr/kdiss can be given a physical interpretation. It is the
intrinsic rate constant for the A + B reaction, i.e., the rate constant which
would be observed if the reactants were placed in contact. (60) As such k'
(59)

(60)

It is worth noting that the method of calculation must be significantly altered for

more concentrated solutions to insure that the values of k dlfr found by letting A diffuse
toward a stationary B or vice versa be the same. See L. Bass and W. J. Greenhalgh,
Trans. Faraday Soc. 62, 715 (1966); A. M. Watts, Trans. Faraday Soc. 62, 2219, 3189
(1966).
R. M. Noyes, Prog. React. Kinetics 1,129 (1961); T. R. Waite, J. Chern. Phys. 28,103
(1958); 32, 21 (1960).

Sec. 9.13

Diffusion-Limited Kinetics

315

is not necessarily limited by the encounter frequency (in solution) or the

collision frequency (in a gas). In terms of k', the apparent rate constant is
(9.61)

In general, given the uncertainty in the estimate of kdiff' it is hard to disentangle chemical and diffusive contributions to k app . (61)
A steady state is not established instantaneously. After A and B form
an encounter complex and react, the mean concentrations of unreacted
A and B in the vicinity are below their steady state values. Diffusion of
A and B is required to reestablish the steady state. As a result there is an
induction period before steady-state behavior is observed. At shorter times
the binary rate constant, k app , is time dependent, and greater than its
steady-state value, (62)
kapp(t) =

k'

k'kdiff
{
kdiff
1

+ 1 + kdifrlk'

(9.62)

Choosing the same values for d and D that were used in estimating kdiff
we find that, as long as the reaction is diffusion limited (k' > k diff), ",-,10- 7
sec is required for kapp to differ by less than I %from its steady-state value.
After "'-' 10-9 sec the difference is "'-' 10%.
The effect is significant for reactions with short half-times. In an
ordinary diffusion-limited reaction under conditions where the species
B is in great excess, the half-time is Ij(kdifdBD. If k diff "" 1010 M-1 sec-I,
a steady-state treatment of the kinetics will suffice as long as [B] ;:S 0.1 M. (63)
If [B] is larger, the effect of transient behavior cannot be ignored. Another
instance where transients are important is in the kinetics of fluorescence
quenching. Here, the intrinsic lifetimes of the photo excited species are
typically "'-' 10-9 sec. The photostationary state is not established and
effects attributable to the time dependence of kapp can be observed.
For the practicing kineticist this is usually not an issue since the major significance of
the diffusion limit is that derived binary rate constants cannot exceed kdl!!. The
individual contributions to kapp have been distinguished in an analysis of the kinetics
of iodine-atom recombination in CCI, [R. M. Noyes, J. Am. Chern. Soc. 86, 4529
(1964)].
(62) R. M. Noyes, Prog. React. Kinetics 1,129 (1961); T. R. Waite, J. Chern. Phys. 28,103
(1958); 32, 21 (1960).
(63) The measured rate constants will be "'" 10% too large, but efforts to obtain greater accuracy are usually unwarranted.
(61)

Chap. 9

316

Theories of Reaction Rates

Appendix A. Statistical Thermodynamics

This section is meant only as a compilation of some of the important
results of statistical thermodynamics. No proof will be given. (64) The
fundamental quantity which relates the mechanical properties of molecules
to the thermodynamic properties of a dilute gas is the molecular partition
function q(T)
q(T) =

~ exp ( - :~)

(9.AI)

where En are the energy levels of the molecule of interest. The thermodynamic consequence is that q(T) determines the molar chemical potential
via the relation
(9.A2)
fl = -RTln q(T)/N
Thus, if a chemical equilibrium is established in the reaction
A+
we know that
flA

B~AB

+ flB =

flAB

or using (9.A2) and taking antilogarithms

(9.A3)

so that knowledge of the partition functions determines the equilibrium

constant.
Another important result is that statistical thermodynamics allows
calculation of the fraction of molecules, In' in a particular energy level.
The expression is
J, = exp( -En/kT)
(9.A4)
n
q
If this is summed over all energy levels it reduces to

'LIn =
n

which is of course required.

(M)

A fuller discussion can be found in texts on this subject, e.g., T. L. Hill, Introduction to
Statistical Thermodynamics (Reading, Mass.: Addison-Wesley, 1960), Chapters 4,
8, and 9.

Appendix A

317

Application of (9.AI) and (9.A4) requires knowledge of the molecular

energy levels. A good approximation at low energy is to assume that the
various degrees of freedom are independent and that the energy of a molecule can be decomposed into electronic, vibrational, rotational, and translational contributions,

The consequence is that the partition function (9.Al) may be factored into
a product of contributions from each degree of freedom
(9.A5)
where each q is now calculated by an expression like (9.AI); the energy
levels now refer to those corresponding to a particular degree of freedom.
When expressions for the energy levels, deduced from quantum mechanics, are introduced into (9.Al) the following results are obtained:
_ ( 2nmkT )3/2
qt h2
V

qr

Sn 2 IkT
ah 2
Sn2(ABC)1!2(2nkT)3/2

(linear molecules)
(nonlinear molecules)

ah 3

=
v

qe

11
i

exp( -hvi/2kT)

[1 - exp( -hVi/kT)]

II [2 .smh (hVi)]-l
2kT

rigid
rotors
harmonic
oscillators
(9.A6)

In these expressions V is the volume of the system; m is the molecular mass;

I is the moment of inertia (linear molecules); ABC is the product of the
principal moments of inertia (nonlinear molecules); Vi are the vibrational

frequencies; g is the electronic degeneracy; and h is Planck's constant. The

quantity a is the symmetry number which takes into account the number
of ways a molecule can be rotated to exchange identical nuclei; a equals
2 and 3 for 12 and NH 3 , respectively.
The expression for qr is not accurate when used to determine the
properties of isotopes of H2 at temperatures below 1000 K. However,
the errors are less than 10% if T > 250 K. The approximate form for qr
is thus adequate at moderate temperatures when used to estimate qt since
the uncertainties of activated complex theory surely exceed 10%.

Chap. 9

318

Theories of Reaction Rates

In some instances it is convenient to use classical rather than quantum

mechanics. The analog to (9.AI) is
q =

drp dx
f --he-</kT

(9.A7)

for one classical degree of freedom. An integration over position (x) and
momentum (p) replaces the sum over energy levels. The energy 10 is given
by the classical expression
10 =

f~ + Vex)

the sum of kinetic and potential energies. The factor h is introduced so that
classical and quantum mechanics correspond in the limit where they are
known to lead to the same predictions; if it were not included the classical
partition function (9.A7) would also not be dimensionless.

Appendix B. Evaluation of kXJ in RRKM Theory

To determine k"", requires evaluating the integrals

f~ N(E,t) exp( -/1/) dE,t

(9.BI)
(9.B2)

Since N(E,t) dE/ is just the rotational degeneracy (9.BI) is equivalent to

L g(E,t) exp( -{1,t) =

(9.B3)

q/

<t
r

the rotational partition function of the critical configuration.

To evaluate (9.B2) note that L~i=o g(Ev t) is a discontinuous function
of lOt,

o <10/ <
101 t

where g / and

tit

<

Ev t

Elt

<

102 t,

etc.

are the degeneracy and energy of the discrete vibrational

Problems

319

states of A t. Hence (9.B2) is

L g/ roo

J ,.t

'i

exp( -(JEt) dE t

7f ~ g/ exp( -(JE/) =

(9.B4)

kTq/

'i

i.e., kT times the partition function for noncritical vibrational modes of

the critical configuration.

Problems
9.1.

Show that the moment of inertia of a linear triatomic molecule, XYZ, is

where the XY distance is R I

of the molecule.
9.2.

In the temperature range

the YZ distance is R" and M is the mass

~4()()

CI

K, the A-factor for the reaction

+ H. ~ HCI + H

is 1.2 X 10 13 cm 3 mole-'sec-'. Calculate the A-factor for this reaction assuming a linear transition state with bonds 300/0 longer than in H2 or HC!.
Assume that the vibrational frequencies in the activated complex are sufficiently large so that none are excited. The bond lengths in H2 and HCI are
0.0742 nm and 0.1275 nm, respectively. The electronic degeneracies are gel
= 6, gH = 1, gt = 2. (Estimation of the vibrational frequencies of HHCl
is not required.)
9.3.

For the reaction of Problem 9.2 compute the A-factor assuming the CIHH
angle to be 90. For an XYZ molecule with such geometry the moment of
inertia product is

+ I")(lllI,,

ABC

(Ill

III

mx(my

122

where

- I;,)

+ mz)R,'/ M
mz(mx + my)R.'/M

I., = mxmzR,R./ M

Chap. 9

320

Theories of Reaction Rates

Does the Eyring theory allow discrimination between the two extreme
geometries for the CI + H. reaction?
9.4.

Recalculate the A-factors in Problems 9.2 and 9.3 assuming no bond expansion in the activated complex. Does this alter your choice for the more
probable geometry?

9.5. A calculation of the A-factor for the H. + I. reaction using Bodenstein's

mechanism leads to agreement with the data. Show that either of Sullivan's
postulated mechanisms (Problem 4.1) leads to the identical result if the
geometry of the activated complex is the same in each case. (No numerical
calculation is needed.)
9.6.

For the reaction

NO

+ NO.CI -' NO. + NOCI

the Arrhenius parameters are Ea = 28.9 kJ mol- I and A = 0.83 X 1012

emS mol- I sec-I.
(a) Compute LJH+, LJS+, and K+ at 350 K.
(b) Assuming PNO = PN02Cl = 0.1 atm, compute p+ and nt at 350 K.
9.7.

(a) Using the data of Table 9.4 determine n+ in the isobutene + HCI
reaction at 400 K; the pressure of each of the reactants is 1 atm. To what
molecular concentration does your value of n+ correspond?
(b) How do you define the activated complex? Interpret the result of part
(a) within this context.

9.8.

LJSo+is much larger for the reaction cyclopropane -' propene than for the
reaction cyclobutene -' trans-butadiene. Why?

9.9.

Look up So for HCI(g), C.H,(g), and C.H 5CI(g) at 298 K. Use these values
and the data of Table 9.4 to compute LJSo+for the reaction

Compare your value with that for the addition of HCI to isobutene. Comment on the difference.
9.10. (a) Extend the thermodynamic analogy of Section 9.9 to define an activation
volume LJ Vo+ in terms of the pressure dependence of the rate constant.
(b) The pressure dependence of the rate constant for the solvent exchange
reaction

was measured in 0.01 M HCIO, at 70.5C [So B. Tong and T. W. Swaddle,

Problems

321

Inorg. Chern. 13, 1538 (1974)]:

k (sec-I) x 105 :
p (kbar):

3.82

4.01

4.56

4.95

5.79,5.76

6.00

0.001

0.059

1.027

2.02

3.31

4.00

What is Ll Vol?
(c) Would you expect any correlation between Ll Vol and LlSot? Explain.
9.11. Stokes' law permits an estimate to be made of ionic radii. Combine this
with the Nernst-Einstein relation (3.13) and the expression for the diffusion
limited rate constant to show that

and that

Use these results with the data of Table 3.1 to compute kdtrr and d for the
reaction, at 25C,
H+ + OH- ~H20
At this temperature 'YJ = 0.89 X IO- a kg m- I sec-I and
mental value of kdtrr is 1.4 X 1011 M -I sec-I.

E =

80. The experi-

9.12. Carrv out the same calculation described in Problem 9.11 to estimate d
and kdtrr for the reactions

+ OH- ~ NH,OH
+ HCO a- ~ COa 2- + H,O
+ HCO a- ~CO, + H 20

NH,+
OHH+

Compare your results with the experimental data quoted in Table 5.3.
9.13. The general theory of diffusion-limited rate constants may be combined
with the theory of the primary salt effect to estimate the effect of ionic
strength on k dW Assume mean diffusion coefficients 1.3 x 10- 9 m' sec I
and encounter distances of 0.5 nm for ions in water at 25C.
(a) Use (9.55), (9.59), and (9.60) to show that, when I

0,

(b) Combine this with (9.48) and estimate k,(T) for the reaction
A ZA

+B

ZB

~ product

at ionic strengths 0, 10-' and 10-' for values of

2, 3.

ZA

= 1,2,3,

ZB

= 1,

Chap. 9

322

Theories of Reaction Rates

(c) From the formula of (b), determine the value of I at which kdlt! is
the same if the ionic charge product ZAZB is +q or -q.
(d) What does the result of (c) suggest about the range of validity of the
general formula deduced in (b)?
9.14. The RRKM model may be treated phenomenologically as

where the rate determining step is the decomposition of A+.

(a) What is the decomposition rate -d[A]/dt, in a steady state?
(b) What are the high- and low-pressure limits?
(c) Is this phenomenological approach any improvement over Lindemann's
treatment? Why?
9.15. Show that, if vibrational excitation is ignored, (9.24) reduces to
<u') )1'+
-k' = - - - exp [h
- - (Llv' - Llv) ]

<u)

2kT

v+

where Llv = v,+(ABC) + 2v.+(ABC) - v(BC); thus the isotope effect,

. under these limiting conditions, reflects differences in zero point energy.
9.16. At 273 K the isotope effect,
(1)
(2)

CI
CI

k,/k.,

is 35 for the reactions:

+ H2 ~ HCI + H
+ T2 ~TCI + T

Determine the isotope effect for the reverse reactions. The vibrational
frequencies are 1.32 x 1014 sec- 1 for H. and 8.97 x 1013 sec- 1 for HCI. Those
for T. and TCI can be estimated since

and the force constant is determined by the shape of the potential surface.
9.17. Demonstrate the equation for

k d188 ,

(9.57).

9.18. Estimate kd188 as a function of ZAZB for ionic complexes in water at 25C.
Assume D A + DB = 2.6 X 10-9 mO sec- 1 and d = 0.5 nm.
9.19. What is

Keq

for ion-pair formation in the reaction

Na+

+ Ac- =

(Na+ ... Ac-)

General References

323

Assume d = 0.4 nm, 0.5 nm, and 0.6 nm. At what separation is Keq (a)
maximum, (b) minimum? Can you ascribe physical significance to these
results?
9.20. Singlet-singlet excitation transfer, as discussed in Section 6.5.2, takes place
at distances of 6-7 nm. Use data from Problem 9.13 and estimate the
diffusion limit for this process in water. (The largest values observed are
~ 3 x 10 11 M -1 sec- 1 .) Is the process diffusion-limited?
9.21. (a) Use data from Problems 9.13 and 9.20 to estimate the time required to
establish a steady state when studying singlet-singlet excitation transfer.
(b) Would transient behavior complicate studies made using pulsed-laser
flash photolysis?
9.22. As the intrinsic rate constant varies, the time required to establish a steady
state does also. For reactions for which the measured rate constant is less
than ! the diffusion limit, under what conditions, if any, does non-steadystate behavior complicate the interpretation of an experiment?

General References
I. Amdur and G. G. Hammes, Chemical Kinetics (New York: McGraw-Hill, 1966), Chap-

ters 2 and 4.
H. S. Johnston, Gas Phase Reaction Rate Theory (New York: Ronald Press, 1966).
K. J. Laidler, Theories of Chemical Reaction Rates (New York: McGraw-Hill, 1969),
Chapters 2, 3, and 6.

Potential-Energy Surfaces
D. G. Truhlar and R. E. Wyatt, Adv. Chem. Phys. 36, 141 (1977).

Activated Complex Theory

K. J. Laidler and A. Tweedale, Adv. Chem. Phys. 21,113 (1971).
W. H. Miller, Acc. Chem. Res. 9, 306 (1976).

Models for Calculation of Rate Constants

D. L. Bunker, Acc. Chem. Res. 7, 195 (1974).

10

A Dynamical Model
for Chemical Reaction
10.1. Introduction
If the reaction cross section, Q(E, i, j), is known, the rate constant for
the corresponding chemical reaction can be calculated from (8.6) and (8.11).
To interpret the details of molecular beam experiments even more information is needed. The hard-sphere model was obviously far too naive to give
reliable estimates of the reaction cross section. Improvement, by considering
the dynamics of reaction across a realistic potential-energy surface such as
that described in Section 9.2, is a formidable quantum mechanical problem.
As already mentioned it has not been solved, except for low-energy H + H2
chemical reaction.
In this chapter we shall try to clarify this chemically significant problem
by studying the dynamics of the collision process A + BC in extremely
simplified terms'!!) We shall not attempt to calculate Q(E, i, j). Instead we
shall see how changing various molecular parameters influences reaction
probability and affects energy transfer. The effect of varying the collision
energy, the reactant mass, and the shape of the reaction profile will be
investigated.
The results of our analysis will show that slight variations in the nature
of the ABC potential-energy surface, in the collision energy of the reactants,
or in the initial vibrational energy of the BC molecule can have dramatic
effects on the collision dynamics. In our model we shall emphasize collinear
(1)

The treatment is adapted from B. H. Mahan, J. Chern. Ed. 51, 308, 377 (1974).

325

326

Chap. 10

Dynamical Model for Chemical Reaction

collisions and ignore effects due to rotational motion. The model, while
highly restrictive, nonetheless illustrates many aspects of reaction dynamics.
The fact that so many features of chemical reaction can be reproduced
by a simple model emphasizes the point that chemical reaction is not a
mysterious problem of rearrangement and recombination. It is, on the
other hand, directly determined by the type of collision taking place and
the nature of the interparticle forces.

fO.2. Kinetic Energy of a Triatomic System

In order to describe the dynamics of the triatomic system A + BC
a convenient coordinate system must be constructed. Six degrees of freedom
are required to completely describe the relative motion of the three particles.
These may be chosen in many ways. The most familiar set of coordinates
is that which we used in describing the ABC potential-energy surface in
Section 9.2: the AB and BC distances, the A-B-C angle, and three angular
coordinates to specify the orientation of the ABC system in space. Such a
coordinate set is shown in Fig. 10.1, where the vectors r AB and rBC provide
all the information required to specify the relative motion and the rotational
motion of the three particles. While a natural set of coordinates, it is not
convenient for calculation. The discussion is greatly simplified if we describe
the system in terms of BC relative motion and the relative motion of A
with respect to the center of mass of BC; these coordinates are also shown
in Fig. 10.1.
Complete analysis of the dynamics of collision in the A + BC system
is extremely difficult even if A, B, and C are atoms. To avoid the complications introduced by molecular rotation we limit consideration to collinear configurations such as those shown in Fig. 10.2; then only A-B

Fig. 10.1. Coordinates for atom-diatom system.

Sec. 10.2

Kinetic Energy of a Triatomic System

327

X -+-----01

Fig. 10.2. Coordinates for collinear A-BC system.

collisions are possible. Many features of reaction dynamics can be illustrated even if the model is simplified further and A, B, and C are assumed
to obey the laws of classical mechanics.
For the collinear system the total kinetic energy of relative motion is
the sum of the vibrational kinetic energy of the BC diatom and the kinetic
energy of A relative to the center of mass of BC Letting mA, mB, and mc
denote the atomic masses yields
(10. I)
Here mBmC/(mB -+- me) is the reduced mass for BC vibration and
mA(mB -+- me)/(mA -+- mB -+- me) the reduced mass for A-BC relative

motion; Y and A are the corresponding velocities. The kinetic energy is

then particularly simple. The chemically more intuitive coordinates rAB
and rBC may be expressed in terms of X and Y; using Fig. 10.2 we find
rAB

(10.2)

X - yY

with y == mC/(mB -+- me). A further coordinate transformation yields an

even simpler expression than (10.1). Choosing

x = X and y = Y/n
with n = [mA(mB -+- mc)2/mBmcM]1/2, where M
kinetic energy becomes

(10.3)
= mA

-+-

mE

-+- mc, the

(10.4)

In addition to its obvious interpretation as the relative kinetic energy of the

three collinear particles, (10.4) may also be interpreted as the kinetic energy
of a single particle of mass mA(mB -+- mc)/M moving on a planar surface.

Chap. 10

328

Dynamical Model for Chemical Reaction

rAB = constant

&....11......._ " " - - - ' ' - -........_

......-

rBe

= constant

x
Fig. 10.3. Mapping of molecular coordinate system into Cartesian (x, y) system. Lines
of constant r AB and rBC intersect at an angle f3 = tan- 1(m BM/mAmc)1/2.

The relation between x and y and the chemically interesting coordinates

and rBe may be found from (10.2) and (10.3),

rAB

(10.5)
Lines of constant rBe run parallel to the x axis and lines of constant r AB
have a slope of (Y1'))-l in the x-y coordinate system; the angle between
these lines is determined by the atomic masses
(10.6)
The mapping is shown in Fig. 10.3.

10.3. Model for Inelastic Collision

To utilize the representation we have just presented it is necessary to
consider the interaction potential (the potential-energy surface) of the linear
ABC system and determine the corresponding potential energy for motion
of the equivalent single particle in the x-y plane. The simplest problem of
chemical interest is a model for energy transfer in inelastic collisions, which
was the mechanism we proposed to understand unimolecular decomposition and bimolecular radical recombination reactions. Assuming the
square-well, hard-sphere potentials of Fig. 10.4, the corresponding potential-energy surface in the x-y plane is given by Fig. 10.5. The regions where
V= +
are configurations in which rAB and/or rBe are less than their

Sec. 10.3

329

Model for Inelastic Collision

00

rAB

Fig. lOA. Interaction potentials for hard-sphere square-well oscillator model.

minimum allowed values. The trough where V = 0 corresponds to the

well in the Be potential, and when V = D the diatom is dissociated.
The simplest example is the somewhat nonphysical trajectory shown
in Fig. 10.5. Here the initial motion is parallel to the x axis; y is constant
and from (10.5) rBe is constant. In dynamical terms the A atom is approaching a Be diatom which has no vibrational energy. The representative
particle collides elastically with wall I (an AB collision) and is reflected;
the new trajectory makes an angle n - 2{3 with the initial one. Reflection
then occurs at wall 2 (a Be collision). Atthis step the representative particle
is either reflected or transmitted depending upon whether the Be diatom
has acquired sufficient vibrational energy to overcome the dissociation
barrier. If transmitted, the y component of kinetic energy decreases in
order to conserve energy in crossing the step. The y component of velocity

is reduced, accounting for the deflection which is analogous to the phe-

V=

+ 00

rr-2t!

V=D

refraction
/ //(diatom dissociates)
~
step

v=o

f3

v=+oo

reflection
(diatom remains bound)

wall 2

Fig. 10.5. Potential-energy surface and trajectory of representative particle for collinear
hard-sphere A interacting with square-well oscillator Be.

Chap. 10

330

Dynamical Model for Chemical Reaction

nomenon of refraction of light. Thus we shall designate the deflection of

the particle as it enters a region of different potential as refraction.
Since, from Figs. 10.1 and 10.2, X represents the distance between A
and the center of mass of BC, the relative velocity of A and BC is, from
(10.3), x = vx , which, in terms of a trajectory, is just the projection of the
velocity vector along the x axis. The finai relative velocity of A and BC is
Vx

= v cos(n -

2~)=

-v cos

2~

where v is the magnitude of the initial velocity vector. Using (10.3) and
(10.1), the kinetic energy of relative A-BC motion after collision is
E'

mA(mB

+ me)

2M

v 2 cos 2

2~

Eo cos 2 2~

where Eo is the initial kinetic energy of the ABC system, which, in this
example, is the initial relative A-BC kinetic energy. If BC remains bound
the rest of the energy resides in the vibrational mode of BC, Ev':
Ev' = LIE = Eo - E' = Eo sin 2

2~

Dissociation occurs if Ev' > D. The fractional energy transfer is a function

of particle mass only; using (10.6) we find
(10.7)
If A is lighter than B the relative energy transfer is small
LIE
Eo

4mAmO

+ me)

--~-----:---'---:-

mB(mB

while if A is heavy the energy transfer is independent of A

LIE

If; ~

4mBme
(mB
me)2 '

The mass effect may be roughly correlated with the activation (deactivation)
efficiency of buffer gases in unimolecular decompositions (bimolecular
recombinations ).
As an example, catalysis efficiency in the isomerization of cyclopropane
to propene can be interpreted, in part, in terms of the mass of the buffer-gas
molecules. To activate CaH6' energy is most likely transferred first to a
C-H bond by collision of a buffer-gas molecule with the H atom; the extra

Sec. 10.3

Model for 1nelastic Collision

331

Table 10.1. Energy Transfer Efficiency to an 1solated HC Bond by Buffer Gases

of Various Mass Assuming the Model of Section 10.3
Buffer mass
(molecular analog)

Efficiency computed
from (10.7)

Relative efficiency in
cyclo-C 3 H. isomerization
from Table 5.1

4 (He)

0.77

0.060

40 (Ar)

0.35

0.053

2 (H 2 )

0.95

0.24

28 (N 2 )

0.39

0.060

energy can then be partitioned among other vibrational modes and eventually weaken a C-C bond. From (10.7) the relative energy transfer may be
estimated by considering the head-on collision of a buffer gas A with the
CH moiety so that B = I and C = 12. The ratio iJEj Eo, computed for
buffer gases of mass 4, 40, 2, and 28 (He, Ar, H 2 , N 2 ), is tabulated in
Table 10.1; the results are in qualitative agreement with the experimental
activation efficiencies of Table 5.1. A buffer of mass 4 is more efficient
than one of mass 40; one of mass 2 is better than one of mass 28. Considering the approximate nature of the model being used-effects due to
multiple collisions, molecular structure, molecular rotation, and intermolecular forces having all been ignored-no better agreement could be
expected.
The dynamics of the hard-core, square-well problem is particularly
simple for the initial conditions just considered. Sometimes multiple collisions occur as shown in Fig. 1O.6a. The representative particle collides
with wall I more than once, i.e., A hits B, then B hits C, and finally Band
A collide again before atom and diatom separate. If, as in Fig. 1O.6a,
BC is initially vibrationless, multiple collisions can only occur when f3 < 60;
as mB decreases these become more common and are predominant if
f3 < 50. Such repeated collisions significantly increase the energy transfer;
instead of (10.7) the inelasticity is

iJE
Eo

sin 2 2nf3

(10.8)

where n is the number of AB collisions that occurred.

If the BC diatom initially has vibrational energy (as it must) the incoming path of the representative particle is no longer parallel to the x

332

Chap. 10

Dynamical Model for Chemical Reaction

Fig. 10.6. Possible trajectories with potential-energy surface of Fig. 10.5. (a) Multiple AB
encounter. (b) Be has initial vibrational energy. (c) Be has initial vibrational energy and
multiple AB collisions occur. (d) Be has zero final vibrational energy.

axis. Two such trajectories are shown in Fig. 1O.6b and c. The energytransfer calculation is more complicated but, on the average, less energy
is transferred to the vibrational mode. The possibility of multiple collisions
is much greater. For some trajectories, such as the one shown in Fig. 1O.6d,
collision leads to vibrational deactivation, which affects bimolecular recombination kinetics.
In bimolecular recombinations the vibrational energy is much larger
than the relative translational energy. Under such circumstances (10.8)
also describes the fractional energy transfer from vibration to translation
(see Problem 10.2). Let us consider the reaction
Br

+ Br + M ~ Br2 + M

In highly vibrationally excited molecules the quantum mechanical aspects

of particle motion are less significant and the details of the interparticle
potential also less important. Just as with high-temperature gas-phase
transport, discussed in Chapter 2, the repulsive part of the potential has
the most significant effect on reaction dynamics. A study of the recombination kinetics of Br2 with the buffer gases He, Ne, Ar, and Kr shows the
steady trend predicted by (1O.8)-the more massive the rare gas atom,
the more effective it is at stabilizing Br 2. (2)
(2)

J. K. K. Ip and G. Burns, J. Chern. Phys. 51, 3414 (1969).

Sec. lOA

Effects Due to More Complex Potential-Energy Surfaces

333

Mass effects do not always predominate. In the vibrational deexcitation

of eO(v= I -'- v=O), He is far more effective than Ne in direct contradiction of (10.8). (3) Here the model potential of Fig. 10.5 is far from appropriate; the details of the interaction are significant and vibrational quantization is very important.

10.4. Effects Due to More Complex Potential-Energy Surfaces

The model may be altered to describe the effect of atom-diatom interactions other than hard-sphere repulsions. Assume A is attracted equally
by both Band e. This may be incorporated by adding a well to the potential
trough and assuming the force acts between A and the center of mass of
Be. The modified potential-energy surface is shown in Fig. 1O.7a. The wall
is at constant x, which, from (10.3) and Fig. 10.2, is equivalent to a consta!1t
distance between A and the center of mass of Be. For this trajectory, the
representative particle accelerates when it enters the well, the total kinetic
energy becoming E - .1, where .1 < o. After it strikes wall I its y component of kinetic energy can be shown to be (E - .1) sin 2 2{J by analogy
with the previous calculation. As it leaves the well only the x component
of kinetic energy is affected. Thus the net inelasticity is increased (.1 < 0)
(10.9)

If the interaction were repulsive (.1 > 0), the inelasticity would decrease;
if E t , the relative A-Be translational energy, is less than .1, the particle
cannot overcome the barrier and is reflected at the discontinuity.
If A is preferentially attracted to B, assume the attraction is to B alone
and modify the potential surface by introducing a step at constant r AB
as shown in Fig. 1O.7b. For the initial conditions shown the particle is
refracted upon entering the well, reflected at wall I, and refracted again
upon leaving the well. Geometric considerations show that the well has
not altered the energy transfer. The inelasticity is still given by (10.7) and
(10.8).
The other limiting case occurs when A is attracted to e, which can
be modeled by incorporating a potential well with a discontinuity at a
constant separation r AC as shown in Fig. 1O.7c. The inelasticity calculations
(3)

R. C. Millikan, f. Chem. Phys. 38, 2855 (1963).

Chap. 10

334

AE

= (E- il)sin2

2/3

Dynamical Model for Chemical Reaction

AE

AE ) (E - ill sin 2 211

= E sin 2

AE

211

=0

Fig. 10.7. Potential-energy surface features which complicate trajectory analysis; .1 is the
depth of the potential well. (a) Inelasticity altered due to potential step perpendicular to
x axis. (b) If A only attracted to B, .1 Ev is not affected. (c) A is attracted to C. (d) If potential
well is deep enough temporary trapping occurs.

for the trajectory shown are no longer simple. There is, however, an increase
in energy transfer when the interaction is attractive.
The atom-diatom attraction not only affects energy transfer but it
also introduces the possibility of "sticky" collisions. A possible trajectory
is shown in Fig. 1O.7d. After reflection at wall 1 the representative particle
attempts to recross the step but its x component of kinetic energy is not
sufficient. Here the kinetic energy of the particle is E - L1 while it is in the
well; the condition for crossing the step on the first try is
(E - L1) cos 2 2f3 > I L1 I

(10.10)

If (10.10) is not satisfied the particle is reflected and thus temporarily

trapped. As diagrammed in Fig. 1O.7d the ABC complex undergoes a
complete AB vibration; the second AB collision converts enough energy
into relative kinetic energy of A and BC so that dissociation occurs. Given
different initial conditions the complex may survive many AB vibrations
before dissociating.
Qualitatively, trapping takes place because of the attractive force
between atom and diatom and because there is an internal degree of freedom (the BC vibrational mode). Some of the relative kinetic energy resides
temporarily in the internal degree of freedom. By increasing the well depth,
increasing the number of vibrational degrees of freedom, or decreasing the
relative kinetic energy, the chances of a sticky collision are favored and
longer-lived complexes will be formed. This correlates with another feature

Sec. 10.5

Model for Thermoneutral Chemical Reaction

335

of bimolecular recombination kinetics as presented in Section 5.5. In atomic

recombination long-lived complexes cannot form and a third particle is
required immediately to take away the excess kinetic energy. Methyl radicals
form a long-lived complex; only at low pressures is the time between collisions long enough to allow dissociation instead of collisional deactivation.
The phenyl radical complex has such a long lifetime that experiments have
never been run at a pressure so low that dissociation is competitive with
collisional deactivation.

10.5. Model for Thermoneutral Chemical Reaction

The model presented to study inelastic energy transfer for a collinear
atom-diatom system can be extended to treat the exchange reaction
A

+ Be ---" AB + C

The simplest potential-energy surface, shown in Fig. 10.8, is one where

AB and BC are assumed to be bound by identical square-well potentials
and where the potential energy remains the same whether two or three
particles are interacting. It is a primitive model for a thermoneutral exchange
reaction. The potential-energy surface for linear H3 (the portion corresponding to the regions in which all three particles in H3 are interacting is
shown in Fig. 10.9) has a significant barrier in the trough when all interatomic distances are small. However, if the relative kinetic energy is large,
the details of the trough are less significant than its general shape. Thus
the model surface is an aid to understanding qualitative features of highenergy atom-molecule collision processes (the restriction to collinear trajectories is still a significant limitation).
A+B+C

v =0
Fig. 10.8. A square-trough potential-energy surface for the reaction
A +BC~AB + c.

V=OD
x

A+BC

336

Chap. 10

Dynamical Model for Chemical Reaction

Fig. 10.9. A portion of the potential-energy surface for the exchange reaction H + H2 ~
H2 + H assuming collinear geometry. The energies are in units of kJ mol-I. [Adapted
from R. E. Weston, Jr., J. Chern. Phys. 31, 892 (1959).]

As an example consider f3 = 60, which corresponds to an unlimited

number of mass ratios, among them A:B:C = 1:1:1, 1:2:6, 2:3:5, etc.
Different combinations can be deduced from (10.6). From Fig. 1O.10a
we see that, since both reactant and product channels have the same width,
all horizontal trajectories (zero vibrational energy) are reactive; the products
formed have zero vibrational energy. The result is peculiar to the choice
f3 = 60. For other f3 this trajectory is not always reactive; in the example
of Fig. 1O.10b, large values of y lead to inelastic collisions.
If the reactants have vibrational energy, the initial trajectory is no
longer horizontal. An angle I () I determines the partitioning of energy
between vibration and translation; it is defined as
tan ()

y/x

where velocities are measured at the standard reference line perpendicular

to the x axis shown in Fig. 1O.lOc. We then find that
E t = E cos 2 (),

(10.11)

where E is the total kinetic energy of the system; E t and Ev are the initial
relative A-BC kinetic energy and BC vibrational energy, respectively.
As long as f3 > 60 there are only three types of reactive trajectories, those
shown in Figs. 1O.lOc,d, and e. Since both y and x are negative at the
reference line in Fig. 1O.lOc, () is positive. In Fig. 10.1 Od () is negative,
since y > 0 and x < 0 at the reference line. In Fig. lO.lOe () is positive for
the two trajectories shown.

Sec. 10.5

(c)

f3

Model for Thermoneutral Chemical Reaction

337

tl'=tl- Iel
cptf3tf3' = IT

f3' = f3 tiel
+f3 +f3' = TT

e' t cp = f3

e' t cp = tl

reference line

e ttl = e'

Fig. 10.10. Effects of initial vibrational energy and phase on product formation and energy
distribution. The diagrams at the top show trajectories with no initial vibrational energy
for (a) f3 = 60, (b) f3 > 60 surface. (c) Construction for deducing 8' for collision sequence
where A hits B, then B hits C. (d) Effect of phase on reactivity. (e) Construction for finding
8 when collision sequence is B hits C, then A hits B. (f) Trajectories with 8 = 30 on f3 = 60
surface.

338

Chap. 10

Dynamical Model for Chemical Reaction

Table 10.2. Summary of Reaction Possibilities on the Potential-Energy Surface

of Figure 10.8, with f3 = 60, as a Function of Vibrational Energy of Reactants,
as Characterized by 1 (J 1 [Equation (10.11)]
1 (J I,

Extent of reaction

degrees

0
0-30
30
30-60
60-90

1 (J' I,

Total
Partial
None
Partial
Total

degrees

1 (J 1
1 (J 1

120 120 -

1 (J 1
1 (J 1

t (3 112 cot 1 (J 1 + 1)2

t .(3 112 cot (J + 1)2
1

The {J = 60 case is particularly simple. As () increases from 0 to 30

an increasing fraction of trajectories lead to inelastic collisions. Paths
with () < 0 and large values of y at the reference line are the first to be lost
from the product channel (Fig. 1O.lOd); for -30 < () < -16.1 there
are no reactive trajectories. For () > 0 reaction is not impeded until () =
16.1; the number of reactive paths then decreases until, when () = 30,
all paths are blocked (Fig. 10.1Of). As 1 () 1 increases reaction is again
possible; for 1 () 1 > 60 all paths are reactive. Thus reactivity depends
not only on the ratio Ev/Et (as determined by 1() I) but also upon the
vibrational phase of the reactants (as determined by the sign of () and the
value of y at the reference line). The various possibilities are summarized
in Table 10.2.
The angle 1 ()' 1 characterizes energy disposition in the product channel,
Ev' = E sin2 ()'. For each of the trajectories in Figs. 1O.lOc,d, and e, ()'
is a different function of (). Noting that the sum of the angles of a triangle
is n radians, we find that when 0 < () < {J/2 (Fig. 10.lOc)

{J'

= {J - 1() I,

rp

+ {J + {J' =

n,

rp+O'={J

(10.12)

and hence
()' =

3{J -

1 () 1 -

(0

< () < (J/2)

(10.13)

The case -n/2 + {J < () < 0 (Fig. 1O.10d) differs only in that {J'
{J + 1 () 1 and thus
()' =

3{J

1 () 1 -

(-n/2

+ (J < () < 0)

(10.14)

For all other values of () (Fig. 1O.IOe) we obtain

()' =

n - {J -

1 () 1

({J12 < () < n12, -n12 < () < -n12

+ (J)

(10.15)

Sec. 10.6

Exoergic Reactions

339

The case {J = 60 is again the simplest. Energy disposition in the products

depends on the relative energy but not on the phase of the reactants. From
the value of 1 ()' 1 the ratio Ev' / Ev may be calculated; these are included in
Table 10.2. In the range 30 < 1 () 1 < 60 the products have more vibrational energy than the reactants and the AB formed dissociates if Ev' > D.
Changing initial conditions affect both reactivity and energy disposition. For fixed E v , increasing E t corresponds to decreasing 1() I. For
{J = 60 we find from (10.11) and Table 10.2 that when E t < E,,/3 all
trajectories are reactive. For E t > E,,/3 the reaction probability drops until,
when E t = 3Ev , no reaction occurs. With a further increase of E t , the
reaction probability rises again. For other choices of {J product energy
disposition as well as reactivity depend upon both phase and 1() I.

10.6. Exoergic Reactions

In order to discuss anything other than thermoneutral reactions, the
potential-energy surface must be modified. The easiest way to represent an
exoergic reaction is to introduce a single step in the potential energy trough.
The location and orientation of the step can be varied to model different
situations. The simplest case arises if the step is located in the product
channel as shown in Fig. 10.11 a; as drawn it corresponds to a repulsion
between C and the center of mass of AB. Trajectories which were reactive
for the potential surface of Fig. 10.8 remain that way. When the representative particle crosses the step it is refracted. Only the component of velocity
perpendicular to the step is affected. The excess energy appears as relative
translational energy of the products
Ee'

E cos 2 ()'

Ll,

E,,'

E sin 2 ()'

(10.16)

where ()' is the angle characterizing energy disposition in the product

channel for the simple surface of Fig. 10.8. When the exoergicity is due to
BC repulsion the step is parallel to the x axis (Fig. 10.11 b) and the excess
energy affects both translation and vibration.
The effect of a step in the reactant channel is quite different. Figures
lO.llc and d illustrate dynamics on potential surfaces where the exoergicity
is due to attraction between A and the center of mass of BC. In Fig. 1O.lIc
the relationship of initial phase to reactivity is shown by considering two
trajectories for which the reactants have zero initial vibrational energy.
For the path with large y the representative particle is reflected back toward

Chap. 10

340

Dynamical Model for Chemical Reaction

v=o

v=o

v=o

v=o

v
Fig. 10.11. Effects of exoergicity (Ll < 0) on product formation and energy distribution.
(a) Step in exit channel corresponding to repulsion between C and center of mass of AB.
Final translational energy altered by step. (b) Step in exit channel due to BC repulsion. Both
translational and vibrational energy of products are affected. (c) Step in entrance channel
due to attraction between A and center of mass of BC. Large y trajectory which would have
led to inelastic collision becomes reactive if E is small. (d) Step in entrance channel. Collision between A and vibrationally excited BC leads to inelastic energy transfer instead of
reaction because of the exoergicity.

the reagent channel; if the reaction were thermo neutral (,1 = 0) the trajectory would be nonreactive. From Fig. 10.5 we see that 'JI = :rr - 2fJ. Since
the total kinetic energy in the region to the left of the step is E - ,1, where
,1 < 0, the x component of the kinetic energy is
(E - ,1) cos 2

'JI

(E - ,1) cos 2 2fJ

Thus if
(E - ,1) cos 2 2fJ

<

-,1

or E

<

-,1 tan 2 2fJ

(10.17)

the representative particle is reflected when it returns to the step. In this

example the representative particle enters the product channel after reflection at the step. Note that if the initial kinetic energy is too large reaction
does not occur. The exoergicity permits low-energy trajectories which would

Sec. 10.6

Exoergic Reactions

341

otherwise be nonreactive to enter the product channel. In addition, energy

disposition in the products is affected. For the small y path, ()', from (10.13),
is 3(J - n; similarly for large y, when reaction occurs, ()' is (J. Since the
total kinetic energy is E - Ll the final vibrational energy is
E'
v

= {

(E - Ll) sin 2 (J,

(E - Ll) sin 2 3(J,

large y
small y

(10.18)

Of the total exoergicity - Ll, only a fraction appears as product vibrational

energy, the amount depending upon y as well as (J. The large I () I trajectory
of Fig. 10.11 d corresponds to an initial state for which Ev > E t . Were
Ll = 0, this trajectory would lead to the product channel; however, the
new reaction dynamics requires that it return to the potential step and
attempt to cross. Since the x component of kinetic energy is greater after
the AB collision than before, the representative particle recrosses the step;
the result is inelastic collision. Of the two effects the enhanced low-energy
reactivity is particularly significant chemically. It correlates with the observation that exoergic reactions have large low-energy cross sections. The fact
that reaction probability drops with an increase in E t may account for
negative Arrhenius activation energies in some exothermic reactions.
The disposition of the excess energy differs markedly depending upon
the placement of the step in the potential-energy surface. When the exoergicity is due to a step in the product channel (termed "late" or "repulsive"
energy release) as shown in Figs. 1O.11a and b, refraction at the step leads
to a major portion of the exoergicity appearing as relative translational
energy. The situation that occurs when the step is in the reactant channel
(termed "early" or "attractive" energy release) is quite different. From
Fig. 10.llc it is clear that much of the exoergicity has appeared as vibrational
energy in the product Be. The results, distinguishing attractive and repulsive
potential-energy surfaces, are similar to those obtained using far more
sophisticated models. (4) In addition to the shape of the surface, the mass
ratios of the interacting particles also affect energy disposal, as reflected
by the way in which Ev', from (10.16) or (10.18), depends upon (J. Naturally
the distinction-attractive or repulsive-is an oversimplification. The
step need not occur primarily in one channel or the other. It may appear
closer to the midpoint of the path in which case the energy release is termed
"mixed," indicating that neither translational nor vibrational energy
transfer predominates.
(4)

A good account of more realistic treatments of reaction dynamics is given by J. C.

Polanyi, Acc. Chern. Res. 5, 161 (1972).

Chap. 10

342

Dynamical Model for Chemical Reaction

10.7. Endoergic Reactions

Endoergic reactions can be treated by introducing potential steps in
which LI > O. Analysis is simplest when the step is placed in the reactant
channel (Fig. 1O.12a). If E t = E cos 2 () < LI, the A atom is reflected at the
step and reaction is forbidden. In this example, representing repulsion
between A and the center of mass of Be, only the relative translational
energy is useful for surmounting the barrier. If the particle has sufficient
energy to cross the barrier it is refracted. After passing the barrier the total
kinetic energy is E1 = E - LI; the vibrational energy is unaltered, thus

and
sin ()1

[Ej(E - LI)J 1/2 sin ()

(10.19)

The further course of reaction is determined by the angle ()1' Any trajectory

(0)

v=o
reflection
if Et ( Ll

nonreactive path due to endoerllicity

Fig. 10.12. Effect of endoergicity (.d > 0) on product formation and energy distribution.
(a) Step in entrance channel. Reaction only occurs if E t = E cos 2 0 >.d. (b) Step in entrance
channel. Effect of endoergicity in modifying trajectories to decrease reactivity. (c) Step in
product channel. Reaction only occurs if E t ' = E cos 2 0' > .1.

Sec. 10.7

Endoergic Reactions

343

which leads back to the step is necessarily nonreactive. Since 1 ()l 1 > 1 () I,
initial trajectories corresponding to small vibrational energy (small 1() I)
may, after refraction, be redirected along a nonreactive path as can be
seen in Fig. 10. 12b. Thus not only does the barrier prevent reaction if
E t < LI, it also reduces the reaction probability (when Ev is small) for
many trajectories if E t > LI.
A barrier in the product channel (Fig. 10.12c) can only be surmounted
if Ee' = E cos 2 ()' > LI. In this example the endoergicity is due to attraction
between C and the center of mass of AB. Here a small amount of vibrational
energy is far more effective in promoting reaction than is increasing the
translational energy. As long as fJ - nl2 < () < 2fJ - n12, (10.13) or
(10.14) are valid for all fJ and

()' = 3fJ - () - n
so that
Ee'

E cos2(3fJ - ())

E(cos

3fJ cos () + sin 3fJ sin ())2

(10.20)

As usual Ee' depends on the sign of (). If fJ > n13, Ee' is larger for () positive.
Using (10.11) and choosing the phase which maximizes E/, (10.20) becomes
E/

= (EF2 1 cos 3fJ 1 + Evl/2 1 sin 3fJ 1)2

(10.21)

If the initial total energy is E it is completely available for surmounting the

barrier only if
E
_v = tan 2 3fJ

Et

Thus, unless fJ = 60, a small amount of vibrational energy is required

in order to most efficiently utilize the available translational energy. The
effect is clearly demonstrated in the reaction(5)
He

+ H 2+--->.HeH+ + H,

LIE = 77 kJ mol-1

where, near threshold, vibrational energy in H2 + is many times more effective

in promoting reaction than is relative translational energy. The relative
reaction probability as a function of vibrational state and total energy is
tabulated in Table 10.3. At the lowest total energy, which is only slightly
above threshold, the effectiveness of vibrational excitation is dramatic.
One quantum of vibration is 27.5 kJ mol-I, an excitation equivalent to
(5)

W. A. Chupka and M. E. Russell, J. Chern. Phys. 49,5426 (1968).

Chap. 10

344

Dynamical Model for Chemical Reaction

Table 10.3. Relative Probability for the Reaction H 2+ + He ~ HeH+ + H as a

Function of E, the Total Energy (Translational and Vibrational), and of the
Vibrational Quantum Number, v, of the H2 +a
E (kJ mol-i)
v

o
2

96

192

288

385

0.06

0.10

0.13

0.17

0.49

0.35

0.31

0.25

1.95

0.93

0.55

0.34

1.70

0.99

0.56

2.35

1.22

0.68

2.49

1.70

0.89

a W. A. Chupka, in Ion Molecule Reactions, J. L. Franklin, ed. (New York: Plenum Press, 1972),
Vol. 1, p. 73.

increasing the Ev/E t ratio from 0 to 0.4; it leads to an eightfold increase

in reaction probability. A second quantum means Evl E t increases to 1.3;
the reaction probability increases by another factor of 4. Also interesting

is the effect of increasing energy at constant vibrational excitation. In the

ground vibrational state, reaction probability increases slightly as the
total energy increases. In all other vibrational states the effect of increasing
the total energy is counterproductive. Accurate quantum mechanical
calculations have corroborated the model for the potential-energy surface;
the 77 kJ mol- 1 barrier is almost entirely in the product channel. (6)
A similar situation, in which vibrational excitation enhanced reactivity,
has been mentioned in a different context. Laser activation in the decomposition of D 3 BPF 3 seemed to lead to much more efficient excitation
(Section 6.8). Of course for such a complex molecule no potential surface
has been calculated. No correlation with a product channel barrier is
presently possible.
Combinations of wells and barriers may be used to model more
complex potential-energy surfaces. In the examples we considered both
reactivity and energy disposition were significantly affected by the position
and nature of a single well or barrier. Further structure in the potential
surface, such as placing steps in both reactant and product channels,
(O)

P. J. Brown and E. F. Hayes, J. Chern. Phys. 55, 922 (1971).

Sec. 10.8

Noncollinear Geometries

345

greatly complicates the dynamics. (7) Since even for the simpler examples
slight changes in initial conditions may alter the outcome of interaction,
it is clear how difficult it is to deduce a potential-energy surface from the
results of molecular beam experiments.
For the much simpler problem of an intermolecular potential between
argon atoms, discussed in Section 2.8, we saw that only by combining data
from different types of experiments could a reliable potential be constructed. In that case the potential was a function of one variable only,
the argon-argon distance. For the atom-molecule system the potentialenergy surface is a function of the three independently variable atomic
distances. To vary the parameters of such a potential in order to reproduce
molecular beam scattering data is a herculean task; some efforts toward
this goal are discussed in Section 10.9. To carry out a similar program for
reactions involving more complicated molecules is presently beyond consideration. At present reactive scattering data are used to establish the
general features of a potential-energy surface, not the precise details.

10.8. Noncollinear Geometries

The model, based as it is on the assumption of collinear collision
geometry, is a gross oversimplification. Even if the Be diatom had no
rotational angular momentum, real A-Be trajectories occur with nonzero
impact parameter. Two such possibilities are illustrated in Fig. 10.13.
After the first AB collision takes place, some of the collision energy must
be converted to rotational energy, either of the Be diatom or, if a single
collision leads to reaction, of the AB diatom. The consequences are particularly important if multiple collisions are predicted for the collinear
geometry.
An example is the isotope effect in the reaction of Kr+ with HD at
high collision energy. (8) The facts are as follows:
(i) At very low collision energies the cross section ratio Q(KrH+)/
Q(KrD+) is slightly greater than 1; the ratio peaks at Ecoll '"" 100 kJ mol- 1
and then drops steadily. For Ecoll > 180 kJ mol- 1 the ratio is less than 1;
(7)

(8)

A model, classical trajectory calculation for collinear H + H. ~ H. + H, using a

realistic potential function, was set up as an experiment in the application of programming skill and of Runge-Kutta numerical integration methods [J. M. White,
Physical Chemistry Laboratory Experiments (Englewood Cliffs, N.J.: Prentice-Hall,
1975), pp. 301-326].
S. Chivalak and P. M. Hierl, J. Chern. Phys. 67, 4654 (1977).

346

Chap. 10

Dynamical Model for Chemical Reaction

0-

~B

(1)

(i)

'ej .....

(i i)

(iii)

e:e

(II)

.e
(j)

%
-e

(iv)

A"B

(ii)

(ii i)

Fig. 10.13. Effect ofnoncollinearity on CM reaction dynamics for a thermoneutral reaction

for particles A, B, and C of mass 86, 1, and 2, respectively-a model for Kr+ + HD
dynamics. Length of vectors indicates velocity. Particle A is so heavy that it is nearly
stationary; direction of motion is indicated. Case I: (i) initial trajectory; (ii) instant
after AB collision; (iii) instant after AC collision; (iv) instant after AC distance reaches
its maximum. Note that an AC diatom has formed even though initial collision was
with B. The AC diatom has vibrational and rotational energy. Case II: (i) initial trajectory;
(ii) instant after AB collision; (iii) BC remains bound-no reaction occurs. The BC
diatom has vibrational and rotational energy.

KrD+ is then the predominant product. The effect is illustrated in Fig.

10.14.
(ii) KrH+ is forward scattered at all energies studied; the cross section
decreases by a factor of 20 over the energy range considered (8 kJ mol- l
Ecoll
300 kJ mol-l).
(iii) At low collision energies KrD+ is forward scattered; as the energy
increases it is back scattered. The cross section drops by ,.....,4 as the collision energy increases from 8 to 300 kJ mol-l.

::s

::s

Consider high collision energies. Here the details of the potential

surface are less significant. Our model may be applicable since experiment
suggests that Ecoll is substantially larger than any barrier to reaction. (9)
Furthermore, the exoergicity ('"""-'30 kJ mol-l) is small. The high collision
energy data seem to suggest that KrD+, as it is back scattered, is formed
(9)

K. R. Ryan and I. G. Graham, J. Chern. Phys. 59, 4260 (1973); P. F. Fennelly, J. D.

Payzant, R. S. Hemsworth, and D. K. Bohme, J. Chern. Phys. 60, 5115 (1974).

Sec. 10.8

Noncollinear Geometries

347

3r-------------------------------~

o
2

0 0

100

200

300

Relative Energy (kJ mol-I)

Fig. 10.14. Kinetic isotope effect in the Kr+

+ HD system

[So Chivalak and P. M. Hierl,

J. Chern. 67, 4654 (1977)].

mainly in a recoil (small impact parameter) process. At the same energies

KrH+ is forward scattered, which is suggestive of a large impact parameter
process.
Now consider small impact parameter processes and assume that
reactive trajectories are collinear, or nearly so. Ignore exoergicity and
model the reaction with an atom-diatom potential such as that of Fig.
10.10. Two cases arise, corresponding to the geometries Kr+HD and
Kr+DH with fJ = 34.8 and 55.2, respectively. If the diatom is presumed
to have no initial vibrational energy, trajectory analysis indicates that for
the two values of fJ all collisions in both geometries are reactive (Problem
10.10) thus suggesting that equal amounts of KrD+ and KrH+ are formed.
The difference between the two paths is that Kr+ collides with the H end
of the DH molecule twice before KrH + can separate. Only a single collision
between Kr+ and the D end of the DH molecule is needed to form KrD+.
At nonzero impact parameter rotation greatly changes the predictions
of the model. Because KrD+ can form in a single collision, collisions with
the D end of the molecule still lead to KrD+, which is formed with some
rotational energy. The consequences of induced rotation upon collisions
of Kr+ with the H end of the molecule are strikingly different. Because
the H atom is not transferred on the first collision, the HD molecule starts
to rotate as shown in Fig. 10.13. Depending upon the initial conditions
the Kr+ might separate or it might take part in another collision with
either the H or the D end of the molecule. As collision energy or impact
parameter increase, more angular momentum is imparted so the Kr+ ion's
second collision is more likely to be with the D end of the molecule. (10)
(10)

If the impact parameter is larger than the hard-core diameter the model is no longer
remotely applicable. The arguments presented are useful at smaller impact parameters.

Chap. 10

348

Dynamical Model for Chemical Reaction

As a consequence the amount of KrD+ formed in recoil processes is enhanced; the amount of KrH+ is decreased.
The qualitative features of the high-energy cross-section ratio and
scattering patterns can now be understood. Smaller impact parameters
favor formation of KrD+ because of rotation and the requirement of multiple collision for formation of KrH+. The major recoil product is KrD+,
as reflected in its scattering pattern. Recoil processes produce little KrH +;
only large impact parameter processes, which lead to forward scatter,
are effective. The data reflect these considerations. The enhanced cross
section for KrD+ is also understandable. As collision energy increases
large impact parameters are less likely to lead to reaction (Sections 8.5-8.9).
Thus KrH+ formation becomes increasingly improbable. Smaller impact
parameters, which are now more significant, favor KrD+ formation. The
result is an enhancement of Q(KrD+), as observed.
Problem 10.3 illustrates that for collinear trajectories in which the
diatom has no initial vibrational energy (0 = 0), multiple collision is
mandatory if fJ < nj4. According to (10.6) such values of fJ are possible
only if mB < me. At nonzero values of the impact parameter b, since
reaction and separation cannot occur, the Be diatom starts to rotate after
collision of A with the B end of the molecule. The larger the collision energy
or the larger the value of b,cl1l the greater the angular velocity. The second
collision is more likely to involve the heavy end of the molecule. Sufficient
energy transfer is now possible; separation may occur without further
collisions. In general, therefore, the collinear approximation overestimates
effects due to multiple collisions. At nonzero impact parameters more than
two collisions are quite unlikely unless the potential surface favors complex
formation.

10.9. Realistic Scattering Calculations

The procedure we have outlined, a classical mechanical trajectory
analysis based upon a chosen potential-energy surface, has been applied
in a far more sophisticated fashion to a number of triatomic systems. The
restriction to collinear collision geometries has been lifted and more realistic
potential-energy surfaces have been used. We consider two examples, the
reactions H + D2
HD + D and Ar+ + D2
ArD+ + D, to demon->.

(11)

->.

There are limits on the range of b for which these considerations apply. See footnote
(10) in this chapter.

Sec. 10.9

en
f-

Realistic Scattering Calculations

349

15
Trajectory Calculation

:l

>-

Experimental Results

~10

a:
!:::
co
a:

~
.J

t9

en
0

J:

-20

20

40

60

LABORATORY SCATTERING ANGLE

Fig. 10.15. In-plane angular distribution ofHD formed in the reaction H + D2 ~ HD + D.

The trajectory calculation is by P. Brumer and M. Karplus, J. Chern. Phys. 54, 4955 (1971);
the experimental data are from J. Geddes, H. F. Krause, and W. L. Fite, J. Chern. Phys.
52, 3296 (1970). The calculated curve is normalized to agree with experiment at + 36.
[From P. Brumer and M. Karplus, J. Chern. Phys. 54, 4955 (1971), reprinted by permission, American Institute of Physics.]

strate the accuracy with which scattering patterns can be calculated. (12)
In particular, we shall find that good results can sometimes be obtained
using most approximate model potentials.

10.9.1. D+H2 ----'HD+H

Classical trajectory calculations for the hydrogen exchange reaction
have been used to compute the angular distribution of the HD produced. (13)
The potential-energy surface was a realistic one. (14) Theory and experiment
are compared in Fig. 10.15 for low-energy collisions. Agreement is quite
good; the calculation reproduces the general shape of the experimental
scattering curve. The results presumably provide a test of the potential1121

(13)
(14)

Another interesting system for which scattering patterns have been calculated and compared with experiment is the reaction of H+ with D 2. Depending upon energy, various
channels are open. The possibilities are: D+ + HD (E> 4 kJ mol-I); H + D2+'
D + HD+ (E> 180 kJ mol-I); H + D+ + D, H+ + D + D (E> 440 kJ mol-I)
[J. R. Krenos, R. K. Preston, R. Wolfgang, and J. C. Tully, J. Chern. Phys. 60, 1634
(1974)].
P. Brumer and M. Karplus, J. Chern. Phys. 54, 4955 (1971).
R. N. Porter and M. Karplus, J. Chern. Phys. 40, 1105 (1964).

350

Chap. 10

Dynamical Model for Chemical Reaction

EXPERIMENT

KINEMATIC

N Dire<: tlon

D~

MODEL

Direction

- - - _ I ....- - - - - -..............- - - _....

, --CII!

Fig. 10.16. Comparison of calculated and observed angle-velocity flux distribution for the
reaction Ar+ + D. ~ ArD+ + D at a collision energy of 260 kJ mol-I. Calculated values
from T. F. George and R. J. Suplinskas, J. Chern. Phys. 54,1037 (1971); experimental data
from M. Chiang, E. A. Gislason, B. H. Mahan, C. W. Tsao, and A. S. Werner, J. Chern.
Phys;52, 2698 (1970). [From Z. Herman and R. Wolfgang, in lon-Molecule Reactions, J.
L. Franklin, ed. (New York: Plenum Press, 1972), Vol. 2, p. 580, reprinted by permission.]

energy surface since the collision energies are only somewhat above threshold.

Contour maps for the angle-velocity flux of ArD+ produced in this

reaction have been computed on the basis of a highly approximate potentialellergy surface. (15) The model treats Ar+ as a hard sphere and D2 as a pair
of hard spheres nearly in contact. The relative motion prior to Ar+-D2
collision is governed by the ion-induced-dipole potential between Ar+
and D 2. During collision the ion and each atom are treated as hard spheres.
An ArD+ ion is presumed to have formed if the relative translational energy
of the nuclei is less than the bond dissociation energy. As the products
(15)

T. F. George and R. J. Suplinskas, J. Chern. Phys. 54, 1037 (1971).

Problems

351

separate motion is again controlled by an ion-induced-dipole potential.

The model potential is essentially the same one we have treated in this
chapter, and is extended to incorporate the long-range ion-induced-dipole
interaction.
A comparison of calculation and experiment, for a collision energy
of 260 kJ mol-I, is presented in Fig. 10.16. The agreement is remarkable,
considering the crudity of the model. Presumably this may be partly rationalized since the collision energy is high, a condition under which the detailed
structure of the potential-energy surface is less important. In any case,
it is clear from this example that agreement of the sort illustrated provides
little basis for assessing the accuracy of a potential-energy surface. (16)
The problem is similar to that of establishing an intermolecular potential
from limited data, first discussed in Chapter 2.

Problems
Problems 10.1-10.4 are designed to exhibit some general features of nonreactive
collisions assuming the potential-energy surface of Fig. 10.5.
10.1.

Show that for fJ = n/2n all collinear collisions are elastic (no interconversion of translational and vibrational energy).

10.2.

Consider a collinear collision for which the initial partitioning of the

energy is
Ev = Eo sin 2 (j
E t = Eo cos 2 (j,

> 0 if P/x > 0 when the first AB collision occurs.

(a) Show that, if the trajectory leads to m AB collisions, the vibrational
energy after interaction is

(j

Ev = Eo sin2 (j',

(b) Show that iJEv/Ev

(j'

= n

+ (j -

2mfJ

1 - sin 2(j' /sin 2(j and that, if Ev:> E t ,

(c) Compute iJEv/Eo in Br2 recombination when the buffer gas is He,Ne,Ar,
Kr. (The ratio of the recombination rate constants is 1 : 1.4 : 2.2 : 2.9
at ......,300 K.)
(16)

A detailed critique of the model has been given by M. Henchman, in Ion-Molecule

Reactions, J. L. Franklin, ed. (New York: Plenum Press, 1972), Vol. 1, pp. 206---208.

352
10.3.

Chap. 10

Dynamical Model for Chemical Reaction

(a) Show that the only possible values of f) are -n12 < f) < f3.
(b) For f3 > nl4 show that only one AB collision is possible if f) < 3f3 - n.
(c) For f3 > n/4 show that there are two AB collisions if 2f3 - nl2 < f) < f3.
(d) Energy disposition is determined by I f) I, not f). For I f) I < f3 show
that the fraction of collisions with f) = -I f) I is HI + tan I f) I cot f3].
(e) For 3f3 - n < f) < 2f3 - nl2 there can be 1 or 2 AB collisions. Show
that, of the trajectories with f) in this range, a fraction 2/[1- cot f3 tan(f3 -f))]
leads to only a single collision.

lOA. Consider systems for which f3 = 50 and 70.

(a) In each system what are the possible values of I f)' I if I f) I = 0, 15,
30, 45, 60, 75?
(b) What fraction of the trajectories with the If) I's of (a) lead to a particular I f)' I?
(c) From your results deduce the qualitative features that encourage
vibrational excitation. What is the effect of the mass of the buffer gas?
What is the effect of increasing the vibrational energy in the molecule?
10.5.

Consider a potential surface such as that of Fig. 1O.7d, where there is a

potential well.
(a) Show that, for a path where there is no initial vibrational energy,
the collision is elastic if the path only involves one reflection at the step.
(b) Now show that, even if there is initial vibrational energy, a collision
involving only one reflection at the step is elastic.
(c) What is the condition under which there can be more than one reflection at the step, i.e., formation of a long-lived collision complex?

In Problems 10.6-10.8 do not consider multiple collisions.

10.6.

(a) For the potential surface of Fig. 1O.l1b calculate E t ' for the trajectory
shown as a function of Eo, Ll, and f3.
(b) Construct a potential-energy surface for which the exothermicity is
due to A-B attraction and determine E t ' for a trajectory where f) = 0
(Ev = 0).

10.7. Assume I Ll I?> Eo and determine Et'/l Ll I and Ev'/l Ll I for the two "repulsive" surfaces of Figs. 1O.lla and b; for the two "attractive" surfaces
of Fig. lO.11c and the one constructed in Problem 10.6b. Note the qualitative differences.
10.8.

(a) Construct a potential surface for "mixed" energy release by placing

the step so that it connects the corners of the surface of Fig. 10.10.
(b) Determine E t ' for an initially vibrationless system.
(c) When I Ll I?> Eo compare your results for Ev'/l Ll I with that of Problem 10.7.

353

General References

10.9.

Chemiluminescence data are used to determine whether an exoergic

potential surface is "attractive," "mixed," or "repulsive." Refer to the
data of Table 6.3 and determine the following:
(a) Which reactions do not involve multiple collisions.
(b) Of these systems, which have surfaces that are (i) relatively more
attractive, (ii) relatively more repulsive.

10.10. Consider the surface of Fig. 10.10 with an initial trajectory where () = O.
Modify the analysis used in Problem 10.2 and show that:
(i) all paths are reactive if n/(2n + 1) 2': f3 2': n/(2n
(ii) no paths are reactive if f3 = n/(2n), n 2': 1.

+ 1.5),

n 2': 1;

General References
P. J. Kuntz, in Modern Theoretical Chemistry, W. H. Miller, ed. (New York: Plenum Press,
1976), Vol. 2, Part B, Chapter 2.
R. D. Levine and R. B. Bernstein, Molecular Reaction Dynamics (Oxford: Clarendon Press,
1974), Chapter 6.
B. H. Mahan, Ace. Chern. Res. 8, 55 (1975).
R. N. Porter, Ann. Rev. Phys. Chern. 25, 317 (1974).

Physical Constants
Quantity

Symbol

Value

Avogadro's number

No

6.02252 X 10 23 mol- 1

Boltzmann's constant

1.38054 X 10- 23 J

Electron charge

1.60210 X 10-19 C

Electron mass

me

9.1091 X 10-31 kg

Faraday's number

9.64868 X 104

C mol-1

Gas constant

8.3143

Planck's constant

6.6256 X 10-34 J sec

Velocity of light

Co

2.997925 X 108 m sec- 1

355

K-1

K-1

mol-1

Index
Where new material or new concepts are introduced in problems, the page numbers are italic.
Absorption of radiation
coefficient of, 166
quantum mechanical expression for,
167
measurement of intensity, 166
nonstationary states, 173
Acridine fluorescence lifetime, 178
Actinometry, 166-167
Activated complex, 273
and bond expansion, 283
and critical configuration, 294
and pass energy, 273
structure estimation, 284
structure in HI" 283-285
Activated complex theory
critique of assumptions, 281
and depletion, 278
and kinetic isotope effect, 287-291
and local equilibrium, 277
model calculations of rate constants
discrimination between reaction
geometries, 285, 287
hard-sphere model, 282-283
identification of reaction coordinate,
282
in reaction of Hand 12, 283-285
tabulation, 286
and separability of degrees of freedom,
277
and three-body collisions, 278
Activated molecules
and buffer gases, 126
and collision energy, 125
in unimolecular reaction, 124, 294

Activation energy (see Arrhenius activation

energy, Enthalpy of activation)
Activity coefficient
from activated complex theory, 307309
and rate constant, 75, 103
5'-Adenosine monophosphate, reaction
with Rh2(OAc). 2H20, 96-98
Adsorption
chemisorption (see Chemisorption)
effect of temperature, 150
in electro capillarity , 222
and heterogeneous catalysis, 148-149
physical, 1.50
Affinity, 76
Angle-velocity flux patterns, 258
Ar +, beam reaction with 2H2
angle-velocity flux distribution calculations, 350-351
angular distribution of products,
254-256
energy dependence of cross section,
245-248
and spectator stripping, 255-256
Argon intermolecular potential, 43
Arrhenius activation energy
and bond energies, 132-133
definition, 98-99
and energy of reaction, 100-101
and enthalpy of activation, 303
and enthalpy of reaction, 100-101
and hard-sphere model, 242
in laser-activated reactions, 190
molecular interpretation, 12, 99

357

Index

358
Arrhenius activation energy (cont.)
negative values
in radical recombination reaction,
130

and reaction dynamics, 341

and potential-energy surface, 341
from reaction cross section, 236
temperature dependence of, 99-100, 109
and threshold energy, 242, 281
in unimolecular reaction, 124
Arrhenius A-factor
from activated complex theory, 281
and bond expansion, 285
tabulation, 286
temperature dependence, 285
definition, 99-100
from hard-sphere model, 242
tabulation, 243
Atom-molecule scattering, 248-252
and harpooning model, 249
Autocatalysis, 87, 205-209
in Belousov-Zhabotinskii reaction, 218
Autoinhibition, 87
in Belousov-Zhabotinskii reaction, 218
BCb
isotopic separation of, 192-193

reaction with H2, 190-191

Beer-Lambert law, 81, 166
Belousov-Zhabotinskii reaction, 214-219
and concentration waves, 220
induction period, 218
mechanism of, 216-218
switching control, 218
Benzophenone
excitation transfer, 197
phosphorescence lifetime, 195
Bipyridine, reaction with Fe(I1I), 160
Bjerrum faults, 62
Bohr frequency rule, 167
Bond energy, in estimation of Ea, 132-133
Br2
beam reaction with Na, 265
beam reaction with 0, 252, 259-261
angular distribution of products, 259
complex formation in, 261
reaction with H2, 72-73, 77, 131-133
Branching-chain reactions (see Chain
reactions)
Bray-Liebhafsky reaction, 215

C, beam reaction with 2H2, 248

Carbonyl addition reactions, 137, 139, 159

Catalysis
by enzymes, 145-147
competitive inhibition, 147
noncompetitive inhibition, 147, 161
uncompetitive inhibition, 147
heterogeneous, 147-152
and adsorption, 148
analogy to enzyme catalysis, 148-149
and chemisorption, 151
and diffusion, 147-148
homogeneous, 142-144
simple mechanism, 144
Catalysts, 142-144
poisoning, 143
and surface chemistry, 151-152
and surface reconstruction, 152
CCIF3, laser-induced decomposition, 192,
198

CCbF, laser-induced decomposition, 192,

198

Centrifugal barrier, 246-247

CH4, reaction with Clep), 99-100, 109
C2H4, oxidation to CH3CHO, 143-144
C2H6, pyrolysis of, 158
Chain branching
and chain traps, 208
rate of, 206
Chain carriers, 132, 205
in hydrocarbon combustion, 223
Chain initiation, 205
Chain reactions, 131-135
apparent activation energy of, 133
branching chains, 134-135, 205-209
and breakdown of the steady state, 134
and population explosions, 134, 206
and thermal oscillations, 223
inhibition of, 132
mechanisms of H:z-Br2 reaction, 131-133
and thermal explosion, 133, 202-204
Chain termination, 132
via chain traps, 208
via heating, 206, 223
and surface effects, 134, 206
CH3CHO, pyrolysis of, 159
Chemical activation, 301
Chemical hysteresis, 226-230
Chemical oscillations (see Oscillatory
reaction)
Chemiluminescence, 184-185
and energy partioning, 184-185
spectrum as analytical tool, 193
Chemisorption
and catalysis, 151

Index
Chemisorption (cont.)
energetics of, 150
site specificity, 151-152
surface reconstruction, 152
surface specificity, 151
CH3I
beam reaction with K, 248-249
beam reaction with Na, 265
l-Chloroanthracene, excitation transfer to
perylene, 198
CH3NC, isomerization and RRKM theory,
299-302
Clep), reaction with CH4, 99-100, 109
Ch
beam reaction with 2H, 252, 257-258
angle-velocity flux distribution, 258
angular distribution of products, 257
collision geometry of, 258
reaction with H 2, 133
CIO-, reaction with r, 77, 160
CO
laser, 189, 195
reaction with 02, 184, 208
C02 laser, 188, 195
Collision complexes
angular distribution of products, 259
lifetime, 259
Collision dynamics (see also Elastic,
Inelastic, Reactive collision
dynamics)
with collinear geometries, 325-345
with noncollinear geometries, 345-348
Collision energy
and centrifugal barrier, 246-247
and reactivity in beam reactions, 246254
and thermal reaction, 271
Collision frequency, 20-24
as function of relative speed, 235
and mean free path, 20
total, 24
Collision probability, 18-19
Competition methods, 84 (see also Quenching, photochemical)
in kinetics of cyclohexane rearrangement,
86-87
limitations of, 178-179
Concentration waves, 220
Concerted rotation, and orbital symmetry
correlations, 186-187
[Co(NH3)s]202 4+, reaction with V(II), 160
Consecutive reactions, arbitrary order,
120-124

359
Consecutive reactions, arbitrary order
(cont.)

general solution, 122-123

closed forms, 122
numerical methods, 122
and preequilibrium, 120-122
and the steady state, 120-122
Consecutive reactions, first-order (see
First-order reactions)
Continuous flow, 82
Conversion per flash, and progress variable,
198
Cr(VI)
reaction of HCr04- with Fe(III), 109
reaction with r, 107-108
Cr(H20)SX 2+, solvent exchange kinetics, 306
Critical coordinate, 291, 294, 295
sensitivity of choice of, 299
Cross-diffusion, 54
Crystal radius, 58
Cu(I), reaction with Fe (III), 160
Current density, 50, 54
Cyclohexane rearrangement kinetics, 86-87
Cyclopropane isomerization, 124-126
effect of buffer gas, 126
entropy of activation, 303
pressure dependence of rate constant, 125
Debye-Hiickel theory
and electrophoretic effect, 63-64
and primary salt effect, 309
and relaxation field, 64-65
Debye theory of diffusion-limited reactions,
309-317
Decay constant
first-order
concentration dependence, 88, 94
and rate constants, 85, 153-156
and relaxation methods, 90, 96-98
non-Arrhenius temperature dependence,
120
non-first-order, from data analysis,
88-89
Depletion
in diffusion-limited kinetics, 315
in unimolecular reaction, 292-293
Detailed balance
and cyclic processes, 79
and Maxwell-Boltzmann velocity distribution, 5
and microscopic reversibility, 79
Deuterium (see 2H, IH2H, 2H2)
Diathermal coupling, 204

360
Differential cross section, 239
in beam reaction of Na and Br2, 265
Diffusion (see also Transport equations)
and branching-chain reactions, 206
as competition method, 84
and electrocapillarity, 222
and heterogeneous catalysis, 147-148
and oscillating reactions, 220
and periodic precipitation, 221
Diffusion coefficient, 24
of dilute hard-sphere gases, 32-38
of electrolytes, 56
and equivalent conductance, 53-55
and Walden's rule, 35, 59-60
Diffusion-limited rate constant
Debye theory, 309-317
for dilute electrolytes, 313-314
and transient effects, 314-315
and dielectric constant, 137, 313
and ionic charge, 136, 313
and ionic strength, 137, 314
mechanistic interpretation, 137, 139
and primary salt effect, 321
tabulation of, 138
and temperature, 137
and viscosity, 136
Dipole-dipole interaction, 180
Dissociation field effect (see Wien effects)
Dymond-Alder potential, 42-43
Eadie transformation, 146
Effusion of gases
and isotopic enrichment, 15
and mean free path, 30
Knudsen flow, 30
and mean speed, 14
and molecular flux, 13-14
Elastic collision dynamics
and collision energy, 41
and impact parameter, 41
and intermolecular potential, 41
and microscopic reversibility, 79
simplified model, collinear geometries
equivalent single particle, 327
kinetic energy, 326
trajectories, 40-41
and velocity equilibration, 6
and walls, 7
Electric dipole transition moment, 167
Electrocapillarity, 222
Electrochemical potential, 55

Index
Electrolyte diffusion coefficient, 56
Electronic energy
and radiationless conversion, 173-174
tabulation of, 196
Electrophoretic field
in high electric fields, 66
for moderately dilute electrolytes, 63
Encounter complex, 136
Encounters
distinction from collisions, 136
rate of, 136, 309-311
Energy
of activation (see Arrhenius activation
energy)
of reaction, 100
Enthalpy
of activation, 303
estimation methods, 306
tabulations, 304
of reaction, 10 1
of solvation, 49
Entropy of activation, 303
estimation methods, 306
interpretation of
for gaseous reactions, 304-306
for solution reactions, 306
tabulation, 304
Enzyme catalysis (see Catalysis)
Equivalent conductance (see also Ion
equivalent conductance)
apparent
additivity rule, 51
concentration dependence, 51-52
concentration dependence, 52, 62-64
and electrophoretic field, 62-63
and ion atmosphere, 62-64
and relaxation field, 62, 64
Error function, 11, 21, 28
Eu(II), reaction with Fe (III), 94-96
pH dependence, 140-141
Excimers, 181
Excitation function, 264
Excitation transfer
in benzophenone, 197
diffusion limitations, 181
and excimer formation, 181
and molecular complexity, 180
and photodissociation, 181
and spin, 180
Explosions, 123
population, 205-209

Index
Explosions (cont.)
population (cont.)
explosion limits, 207
and surface effects, 205
thermal, 133, 202-204
diathermal coupling, 204
and heat evolution, 203
and surface effects, 204
Fast reaction methods, 83-85 (see also
Competition methods, Flash
photolysis, Rapid mixing methods,
Temperature jump)
Fe(ln
reaction of 4,7-(CH3h-1,IO-phenanthioline complex with IrCI/-, 106-107
reaction with Hg(II), 109
Fe(III)
reaction with bipyridine, 160
reaction with Cu(I), 160
reaction with Eu(II), 94-96
pH dependence of, 140-141
reaction with HCr04-, 109
reaction with V(III), 141-142
Feedback mechanisms
and branching chains, 206
and e1ectrocapillarity, 222
and glycolysis, 225
and thermal explosion, 202
and thermochemical oscillations, 223
Fick's laws, 25, 71
first law, 25
molecular justification, 33-34
second law, 27
First-order reactions
analysis of consecutive reactions, 114120, 153-156
and formation of trace intermediates,
1I5-1I6, 1I8
and steady-state hypothesis, II 6
and time-scale separation, 115, 117
Flames, 185
Flash photolysis
formation of trace intermediates, 83
nanosecond, 84
picosecond, 85
Flooding (see Isolation methods)
Fluctuations
and chemical hysteresis, 228
and nucleation, 228
and chemical oscillations, 211, 220
and thermal explosions, 204

361
Fluorescence, 170-172
quenching in diffusion-limited reaction,
315
vibrational, 185
Fourier's law, 24, 25, 71
molecular justification, 33
Franck-Condon principle
and dissociation, 174
and isomerization, 177
and shape of absorption band, 168
Friction, molecular, 57
Fumarate, reaction to I-malate, 161
r-function, 8
Gibbs free energy of activation, 303
Glutamate, reaction with pyridoxal phosphate, 104-105
Glycolysis, 224-225
IH
A-factor for reaction with h, 283-285
product distribution in beam reaction
with 2H2, 348-349
2H, beam reaction with Ch (see Ch)
3H
beam reaction with 2H 2, 249-250
reactions with propane, 250-251
H2
reaction with BCh, 190-191
reaction with Br2, 72-73, 77, 131-133
reaction with Ch, 133
reaction with lz, 72-73, 103
reaction with 02, 205-209
IH2H, isotope effect in beam reaction with
Kr+, 345-348
2H2
beam reaction with Ar+ (see Ar)
beam reaction with C, 248
beam reaction with 3H, 249-250
product distribution in beam reaction
with IH, 348-349
H/, beam reaction with He, 343-344
H3
potential-energy surface, 336
reaction profile, 274
Halogen hydrolysis, 110
Hanes transformation, 146
Harpooning model, 249, 262
2H3BPF 3, laser-induced decomposition,
189-190
H2C03, dissociation equilibrium, 67

362

Index

He, beam reaction with H/, 343-344

Heat conduction (see also Transport
equations)
and explosions, 202
and hysteresis, 226
and thermal oscillations, 224
Hemoglobin, conversion to oxyhemoglobin,
93
Heterogeneous catalysis (see Catalysis)
HF laser, 189, 195
Hg(II)
reaction with Fe(II), 109
reaction with V(I1I), 160
Hg(T2), as excitation transfer donor, 181
H 20 2, reaction with S20/-, 103-104
Homogeneous catalysis, 142-144
Hot-atom chemistry, 250
Hydrocarbon combustion
mechanism, 223
and thermochemical oscillations, 223-224
I, atomic recombination reaction, 130
and buffer gases, 130
1reaction with CIO-, 77, 160
reaction with h, 162
12
photodissociation in solution, 139
reaction with H2, 72-73, 103
reaction with r, 162
Ignition temperature
and population explosions, 206
and thermal explosions, 204
Inelastic collision dynamics
and radical recombination reactions,

127-128

simplified model, collinear geometries,

328-333
mass effects, 330-331
and multiple collisions, 331
and translational-vibrational energy
transfer, 330
and trapping, 334
and vibrational-translational energy
transfer, 331-332
Inhibition
in chain reactions, 132
in oscillatory reactions, 213
in parallel reactic ...s, 141-142
in thermochemical oscillators, 223
Integrated rate laws
first-order, 85-87
interpretation of decay constant, 86

Integrated rate laws (cont.)

general-order, irreversible, 88
second-order, reversible, 87-88
Intermediates, trace
in catalysis, 142
in chain reactions, 132
in first-order sequential reactions, 115
in flash photolysis, 83
identification methods, 80
in oxidation-reduction reactions, 142
photoexcited molecules, 172
Intermolecular potential
of argon, 43
and elastic collision dynamics, 41
experimental determination of, 42-43
Lennard-Jones form, 40-42
quantum mechanical calculation of,
43-44
Internal conversion, 170, 173
Interstellar chemistry, 127
Intersystem crossing, 170, 173
Ionic atmosphere, and equivalent conductance, 63-64
Ion equivalent conductance (see also
Equivalent conductance)
and diffusion coefficient, 54
and electrolyte diffusion, 56
at infinite dilution, 52
tabulation of, 53
and ionic radius, 57-58
Ionic radius
and crystal radius, 58
and ion equivalent conductance, 58
and proton mobility, 61-62
tabulation of, 58
and Walden's rule, 59-60
Ionic strength, effect on rate constant,
137, 307, 309, 314
Ion-molecule reactions
and energy barrier, 248
and long-range forces, 245-247
Ion pairs
in electric fields, 66
equilibrium constant for formation of,
312
IrCI/-, reaction with Fe(4,7-(CH3)2-1,10phenanthioline), 106-107
Ir(NH3)5H203+, rate of water exchange,
320

Irreversible thermodynamics
and reaction rate laws, 76
reciprocal relations, 26
and transport, 26
Isolation methods, 87-89

Index

363

Isotope separation
via effusion, 15
laser-controlled, 192-195
effect of pressure, 194
effect of vibrational anhannonicity,
194

Jablonski diagram, 170

Jacobian method, 23, 44
K, beam reaction with CH3I, 248-249
Ketene photolysis, 181
Kinetic isotope effect
in beam reaction of Kr+ and H2, 345-348
general theory, 287-289
inverse effect, 290
and potential-energy surface, 290
Kinetic theory of transport
for dilute hard-sphere gases
comparison with experiment, 36-39
results of exact theory, 35-36
simplified theory, 30-34
Sutherland modification, 38
and effusion, 30
and energy flux, 34, 46-47
and local equilibrium, 29-30
and molecular flux, 31-32
King-Altman rules, 146
Knudsen flow, 30, 47
Kr+, beam reaction with iH 2 H, 345-348
Langmuir adsorption isothenn, 148-149
Laser activation
enhancement of reactivity, 190
and wavelength, 191
in isotope separation, 192-195
and multiple-photon absorption, 189, 194
and pressure, 190-19 I
and selective bond breaking, 188-192
and state selection, 26 I
Lasers
and isotope separation, 192-195
and ultrafast flashphotolysis, 84-85
and vibrational activation, 188-192, 261
Lennard-Jones potential, 40-42
estimation of parameters, 42
and temperature dependent molecular
diameters, 46
Liesegang bands, 221-222
Lifetime, intrinsic
experimental detennination of, 169
for vibrational fluorescence, 184
Limit cycles, 214, 219

Linear least-squares method, IOI-I02

in analysis of rate law data, 93-98
for determining activation energy, 99
and random error hypothesis, 101
Lineweaver-Burk transfonnation, 146
Line-width measurements, 84
in cyclohexane rearrangement, 87
Local equilibrium postulate, 29, 277
Lotka mechanism (see Oscillatory reaction)
Low-energy electron diffraction (LEED),
150, 152

Mass spectrometry, 239

Master equation, 292
Matrix isolation methods, 183
Maxwell-Boltzmann velocity distribution
alternative forms, IO
derivations, 2-6
mean energy, 8
mean speed, 8, II-12
for molecular pairs, 23
and reaction probability, 12
Mean free path
and collision frequency, 20, 22-24
of molecules in a beam, 18-20
speed dependence of, 19, 22
tabulation of, 39
Mechanism, as distinct from rate law, 7273

"Mercury heart," 222

Mesitylene isomerization, 183
Metastable excited states, 175-176
Michaelis-Menton mechanisms, 145-147
and series-parallel reactions, 146
Microscopic reversibility
and time-reversal invariance, 79-80
in magnetic fields, 79
and transition probabilities, 79
Mobility, ionic, 58
anomalous, H+ and OH-, 61-62
and Bjerrum faults, 62
and proton-jump model, 61
Molecular beams
attenuation in a gas, 18-19
product detection, 239
state selection
by field alignment, 262
by laser activation, 261-262
velocity selection, 237
Molecular diameter
apparent temperature dependence, 38
from critical point data, 37
and hard-sphere rate constant, 242
and Lennard-Jones potential, 40-42

364
Molecular diameter (cont.)
and reaction cross section, 236
in RRKM theory, 299-300
from transport data, 38-39
Molecular flux
and determination of pressure, 7
and effusion, 13-14
in molecular beams, 238
speed dependence in nonequilibrium
systems, 46
and transport, 31-32
Molecularity of reaction, 77
Molecular orbitals
in collision reactions, 258
and orbital symmetry correlations,
186-187
Molecular symmetry, and selection rules,
168
Molecular trajectories (see Elastic, Inelastic,
Reactive collision dynamics)
Multiple-photon absorption
and Arrhenius activation energy, 190
in laser-induced isotope separation, 194
and reactivity, 189
Multiple steady states, 122
analogy to van der Waals isotherms,
228-229
and chemical hysteresis, 226-230
and marginal stability, 227
Na
beam reaction with Br2, 265
beam reaction with CH3I, 265
Naphthalene, as excitation transfer donor,
181
Nernst-Einstein relation, 53-56
NO, as catalyst in reaction of 02 and S02,
142
N02, photolysis, 182
N205
decomposition of, 86
reaction with 03, 144, 161
Noncrossing rule, 187
Nonstationary states, 173
0, beam reaction with Br2 (see Br2)

O2

reaction with CO, 184, 208

reaction with H2, 205-209
reaction with S02, 143
03
decomposition of, 120-122
reaction with N20 5, 144, 161

Index
Ohm's law, 25, 50, 71
deviations from, 65-67
Onsager conductance formula, 64-65, 68
Onsager reciprocal relations, 26
Orbital symmetry correlation, 185-188
Oscillatory reaction, 122
concentration waves in, 220
conditions for, 212
damping in closed systems, 213, 215
and detailed balance, 213
and distance from equilibrium, 212-213
and electrocapillarity, 222
and glycolysis, 224-225
and limit cycles, 214
and periodic precipitation, 221
and thermal oscillations, 223
Ostwald dilution law, 67
Periodic precipitation, 221-222
Perturbation methods (see Flash photolysis,
Relaxation methods)
Perylene, excitation transfer from
l-chloroanthracene, 198
Phase trajectory analysis, 212-214
Phosphorescence, 170-171
Photodissociation, 174-176
and chemiluminescence, 184
and excitation transfer, 181
Photoisomerization, 177-178
and orbital symmetry correlation,185188
Photolysis, 181-183
Photoproducts, methods of characterization, 168
Physical adsorption
energetics, 150
of rare gases, 150
and specificity, 150
Piperazine, reaction with pyridine-4aldehyde, 92-93
Poiseuille's law, 25, 71
molecular justification, 32
Polarography, 84
Potential-energy surfaces, 270-275
dimensionality, 271
for H3, 336
and isotopic substitution, 288
in many-atom systems, 275
models
for inelastic collisions, 329-334
for reactive collisions, 335, 340, 342,
348-350
and reaction profile, 271
saddle-point structure, 277, 282

Index

365

Predissociation, 175-176, 181

Preequilibrium hypothesis
in first-order sequential reactions,
117-118

in general sequential reactions, 120-121

Preexponential factor (see Arrhenius
A-factor)
Pressure, relation to velocity distribution, 6
Pressure jump, 83
Primary salt effect, 307-309
and diffusion-limited kinetics, 321
Principle of mass action
and progress variable, 74
and rate law, 73-76
are reverse rate laws, 76, 109
Product distribution analysis, 252
in beam reaction of Ar+ and 2H2, 254-256
in beam reaction of IH and 2H2, 349-350
in beam reaction of 2H and Cb, 257-259
angle-velocity flux patterns, 258
in beam reaction of Kr+ and IH2H,
345-348

in beam reaction of 0 and Br2, 259

Progress variable, 74-76
and affinity, 76
in laser-activated reaction, 198
and relaxation methods, 91, 123
in single-step reactions, 74
Propane, reactions with 3H, 250-251
Proton-jump model for diffusion, 61-62
Pyrene excimer formation, 181
Pyridine-4-aldehyde, reaction with
piperazine, 92-93
Pyridoxal phosphate, reaction with
glutamate, 104-105
Quantum efficiency
for fluorescence, 172
relative, 178
Quantum yield, 166
for fluorescence, 171
Quasi-continuum, 194
Quenching, photochemical
as competition method, 84, 178-179
limitations of, 178-179
Radiationless conversion, 170-171
and energy differences, 174
and molecular complexity, 174
and nonstationary states, 173
Radical recombination reaction
and buffer g~ses, 129-130
energy conservation in, 127

Radical recombination reaction (cont.)

and molecular complexity, 128
negative activation energies in, 130
and relation to unimolecular reaction,
130-131

simplified collision dynamics for, 330,

332

stabilization via emission of radiation,

127

and three-body processes, 127, 129

Rapid mixing methods
continuous flow, 82
and quenching, 81
stopped flow, 82
Rate constant
from activated complex theory, 2812"1'>7

tabulation of, 286

apparent concentration dependence of,
94-96

apparent pressure dependence of

in radical recombination reactions,
129-130

in unimolecular reactions, 125

and equilibrium constant, 75
dependence on activity coefficients,
75, 103

for excitation transfer, 180

hard-sphere model for, 240-241
critique of, 244
tabulation of, 243
ionic strength dependence of, 307-309
pH dependence of, 140-142
pressure dependence of, 72, 74
and rate law, 72
and reaction cross section, 234-236,
239,250

temperature dependence of, 72, 98-100

thermodynamic analogy, 302-303
for unimolecular reaction, 291-293
energy dependence of, 292
RRKM theory, 297-298
upper bounds to
in bimolecular gas-phase reaction,
123

in bimolecular solution-phase reaction,

123, 136-137,309-317 (see also
Diffusion-limited rate constant)
for decomposition in solution, 312
in termolecular gas-phase reaction,
123, 163

wavelength dependence of, 168

Rate law, 72-73
and data analysis, 93-98

366
Rate law (cont.)
elementary, and principle of mass action,
73-76
experimental methods for establishing,
80-93
and linear irreversible thermodynamics,
76
and reaction complexity, 77
Reaction channels, 240
and Arrhenius A-factor, 244
energy dependence, 250-251
and fragmentation, 250
Reaction coordinate (see also Reaction
profile)
identification of, 282
as separable degree of freedom, 276-277
in unimolecular reaction, 291, 294-295
Reaction cross section
for atom-molecule reaction, 248-252
and collision energy, 245, 247, 249
and energy barrier, 245, 248-249
hard-sphere model for, 240-242
and internal states, 239, 261
for ion-molecule reaction, 245-248
and rate constant, 234-236, 239
as a thermal average, 240
Reaction order, 77
relation to decay constants, 114-120
Reaction probability, 235, 270
Reaction profile, 271
in barrierless reactions, 274-275
for H3 system, 274
and thermal reaction, 273
Reactive collision dynamics
effects due to noncollinearity, 345-348
multiple collisions, 345
model for collinear geometry
endoergic reaction, 342-345
energy partitioning, 338
exoergic reaction, 339-342
mass effects, 337, 341
thermoneutral reaction, 335-338
vibrational energy effects, 337
realistic model calculations, 348-351
Relaxation field
at high electric fields, 66
in moderately dilute electrolytes, 64-65
Relaxation methods, 83-84
decay constant in, 90
for coupled reactions, 91, 123
experimental limitations of, 91-92
in in vivo experiments, 93

Index
Relaxation time (see Decay constant)
Repulsive excited states, 175-176, 181
Resonance radiation, 170-171
Response time, 81
Rh2(OAc). 2H20, reaction with 5'adenosine monophosphate, 96-98
Rotational-vibrational energy transfer, 295
RRKM theory, 293-298
critique of, 299-302
random access assumption, 294, 300-301
rate constants from, 297-298
rotational-vibrational energy transfer
in, 195
strong collision assumption, 294
Runge-Kutta integration methods, 122
Scattering, reactive
complex formation, 252, 259-261
lifetime for, 259, 261
and long-range forces, 261
direct, forward-peaked, 252, 254-256
energy dependence, 254
and spectator stripping model, 255-256
measurement methods, 237-239
mUltiple channels, 244
product analysis, 239
rebound, 252, 257-258
and collision energy, 257
and exoergicity, 257
size of cross section, 257
transformation to center-of-mass coordinates, 252-253
accuracy limits, 253
velocity analysis, 239
Scavengers, in isotopic separation, 191
Second sound, 29
Selection rules for radiation
and molecular structure, 168
and spin, 167
vibronic effects, 168
Sensitization (see Excitation transfer)
SF6, isotopic separation of, 192, 194
Shock tubes, 82-83
Smog, photochemical, 182
S02, reaction with 02, 143
s20l-, reaction with H202, 103-104
Solvent cage
and encounter complexes, 136
and unimolecular reaction, 135, 139
Specific conductivity, 50-51
Spectator stripping model, 255-256
and collision energy, 256

Index
Spectator stripping model (cont.)
and impact parameter, 256
and intermolecular forces, 256
Spin multiplicity
and excitation transfer, 180
and selection rules, 167
Statistical thermodynamics, 316-318
Steady-state hypothesis
analysis of, in coupled first-order reactions, 116-120
in catalysis, 144-145
in chain reaction, 132-134
and branching, 134
and fluctuations, 204
in coupled arbitrary-order reactions,
120-122
in diffusion-limited reactions, 3 I I
transient effects, 314-315
in photochemistry, 172
in systems far from equilibrium, 210
steady state as a singular point, 2 I I, 2 I 9
and time-scale separation, 116-117, 120
Steric factor
in diffusion-limited reactions, 312
in hard-sphere model of reaction, 266
Stern-Volmer plot, 178
Stoichiometry
and mechanism, 72-73
and rate law, 72
Stokes' law
and ion equivalent conductance, 57-60
and stick/ slip boundary conditions, 57
Stopped flow, 82
Sucrose, inversion reaction, 161
Surface chemistry, 148-152
LEED studies, 150, 152
surface specificity, 150
Temperature jump, 83
in carbonyl addition reaction, 92-93
in reaction of Rh2(OAc). 2H20 and
5'-AMP, 96-98
Thermal conductivity, 25
of dilute hard-sphere gases, 32-38
and internal energy transport, 36, 46
and kinetic energy transport, 34, 46
Thermal diffusion, 26
Thermal oscillations, 223-224
Threshold energy
and Arrhenius activation energy, 242, 281
and pass energy, 273
Time-reversal invariance, 79-80

367
Time-scale separation
in arbitrary multistep processes, 120
in consecutive first-order reactions, 115
Trace intermediates (see Intermediates,
trace)
Trajectory analysis
for hard-sphere potential, 40-4 I
for Lennard-Jones potential, 40-41
in RRKM theory, 300
Transition state (see Activated complex)
Transmission coefficient, 278
in classical mechanics, 279-280, 282
isotope effect on, 289
in quantum mechanics, 279-280
Transmission function, 280
Transport coefficients
diffusion coefficient, 24
for dilute hard-sphere gas
interdependence, 35
qualitative properties, 34-35
tabulation, 36-38
temperatnre dependence, 37-39
theory of, 32-35
and reaction cross section, 236
and RRKM theory, 299-300
shear viscosity, 25
specific conductivity, 50-51
thermal conductivity, 24
Transport equations
force-flux relations
current flow, 50
diffusive flow, 24-25
heat flow, 24
viscous flow, 25
molecular justification, 29-34
partial differential equations
formulation, 26-27
solution of boundary value problems,
28,45
Tritium reactions (see 3H)
Tunneling, 273
Turnover number, 145
UF6, isotopic separation, 192, 195
Ultrasonic methods, 83
Unimolecular reaction-gases
and chemical activation, 30 I
energy dependence of rate constant,
291-293
phenomenological theory, 124-126
relation to radical recombination, 130-I 31
RRKM theory (see RRKM theory)

Index

368
U nimolecular reaction-gases (cont.)
simplified dynamical model for, 330-331
Slater theory, 301
Unimolecular reaction-solution
and diffusion, 135
and solvent activation, 135
and solvent cage, 135
V(II), reaction with [Co(NH3)5]20z'+' 160
V(III)
reaction with Fe(III), 141-142
reaction with Hg(II), 160
van der Waals forces, in adsorption, 150
Velocity distribution
in nonequilibrium systems, 29
positional dependence, 30, 40
specification precision limits, 2, 14
Vibrational energy, and reactivity, 190,261,
343-344
Vibrational relaxation, 169-171
Vibrational stabilization, 127
Vibrational-translational energy transfer
and chemiluminescence, 184
and laser activation, 189
and molecular complexity, 170
and pressure, 191
pro bability per collision, 170
and reactivity, 337, 340-341

Vibrational-vibrational energy transfer,

170
and laser activation, 191
Viscosity, shear, 25
of dilute hard-sphere gases, 32-38
and Walden's rule, 35, 59-60
Viscous flow, 25
Voltage jump, 83
Volume
of activation, 320
of solvation, 49

Wacker process, 143-144

Wien effects
and dissociation of H2C03, 67
and electrode kinetics, 67
and equilibrium, 66-67
for strong electrolytes, 66
for weak electrolytes, 66-67
Walden's rule
and diffusivity, 35
and equivalent conductance, 59-60
Woodward-Hoffman rules (see Orbital
symmetry correlation)
Yeast metabolism, 224-225