Physical Chemistry 3: — Chemical Kinetics — - Christian-Albrechts ...

Physical Chemistry 3:
— Chemical Kinetics —
F. Temps
Institute of Physical Chemistry
Christian-Albrechts-University Kiel
Olshausenstr. 40
D-24098 Kiel, Germany
Email: temps@phc.uni.kiel.de
URL: http://www.uni-kiel.de/phc/temps
Summer Term 2013
This scriptum contains notes for the lecture “Physical Chemistry 3:
Chemical Kinetics” (MNF-chem0405) in the summer term 2013 at the
Institute of Physical Chemistry of Christian-Albrechts-University (CAU)
Kiel. This module is the last of the three semester lecture course
in Physical Chemistry for B.Sc. students of Chemistry and Business
Chemistry at CAU Kiel. It includes 3 hours weekly (3 SWS) for lectures
and 1 hour weekly (1 SWS) for an exercise class.
c° F. Temps (2004 — 2013) Last update: June 22, 2013

ii
Contents
Preface
ix
Disclaimer
xi
Organisational matters
xii
1 Introduction 1
1.1 Types of chemical reactions 1
1.2 Time scales for chemical reactions 2
1.3 Historical events 5
1.4 References 7
2 Formal kinetics 8
2.1 Definitions and conventions 8
2.1.1 The rate of a chemical reaction 8
2.1.2 Rate laws and rate coefficients 9
2.1.3 The order of a reaction 10
2.1.4 The molecularity of a reaction 11
2.1.5 Mechanisms of complex reactions 12
2.2 Kinetics of irreversible first-order reactions 16
2.3 Kinetics of reversible first-order reactions (relaxation processes) 20
2.4 Kinetics of second-order reactions 24
2.4.1 A + A → products 24
2.4.2 A + B → products 25
2.4.3 Pseudo first-order reactions 26
2.4.4 Comparison of first-order and second-order reactions 27
2.5 Kinetics of third-order reactions 28
2.6 Kinetics of simple composite reactions 29
2.6.1 Consecutive first-order reactions 29
2.6.2 Reactions with pre-equilibrium 32
2.6.3 Parallel reactions 33
2.6.4 Simultaneous first- and second-order reactions* 33
2.6.5 Competing second-order reactions* 34

iii
2.7 Temperature dependence of rate coefficients 35
2.7.1 Arrhenius equation 35
2.7.2 Deviations from Arrhenius behavior 37
2.7.3 Examples for non-Arrhenius behavior 38
2.8 References 40
3 Kinetics of complex reaction systems 41
3.1 Determination of the order of a reaction 41
3.2 Application of the steady-state assumption 43
3.2.1 The reaction H 2 + Br 2 → 2 HBr 43
3.2.2 Further examples 45
3.3 Microscopic reversibility and detailed balancing 47
3.4 Generalized first-order kinetics* 48
3.4.1 Matrix method* 48
3.4.2 Laplace transforms* 53
3.5 Numerical integration 57
3.5.1 Taylor series expansions 58
3.5.2 Euler method 58
3.5.3 Runge-Kutta methods 59
3.5.4 Implicit methods (Gear): 61
3.5.5 Extrapolation methods (Bulirsch-Stoer) 62
3.5.6 Example: Consecutive first-order reactions 62
3.6 Oscillating reactions* 66
3.6.1 Autocatalysis 66
3.6.2 Chemical oscillations 67
3.7 References 75
4 Experimental methods 76
4.1 End product analysis 76
4.2 Modulation techniques 77
4.3 The discharge flow (DF) technique 78
4.4 Flash photolysis (FP) 82
4.5 Shock tube studies of high temperature kinetics 85
4.6 Stopped flow studies 86

iv
4.7 NMR spectroscopy 87
4.8 Relaxation methods 88
4.9 Femtosecond spectroscopy 89
4.10 References 93
5 Collision theory 94
5.1 Hard sphere collision theory 95
5.1.1 The gas kinetic collision frequency 95
5.1.2 Allowance for a threshold energy for reaction 99
5.1.3 Allowance for steric hindrance 100
5.2 Kinetic gas theory 101
5.2.1 The Maxwell-Boltzmann speed distributions of gas molecules in 1D and 3D 101
5.2.2 The rate of wall collisions in 3D 106
5.3 Transport processes in gases 110
5.3.1 The general transport equation 110
5.3.2 Diffusion 112
5.3.3 Heat conduction 113
5.3.4 Viscosity 114
5.4 Advanced collision theory 115
5.4.1 Fundamental equation for reactive scattering in crossed molecular beams 116
5.4.2 Crossed molecular beam experiments 117
5.4.3 From differential cross sections () to reaction rate constants ( ) 120
5.4.4 Laplace transforms in chemistry 121
5.4.5 Bimolecular rate constants from collision theory 123
5.4.6 Long-range intermolecular interactions 132
6 Potential energy surfaces for chemical reactions 138
6.1 Diatomic molecules 138
6.2 Polyatomic molecules 139
6.3 Trajectory Calculations 143
7 Transition state theory 146
7.1 Foundations of transition state theory 146
7.1.1 Basic assumptions of TST 147

v
7.1.2 Formal kinetic model 149
7.1.3 The fundamental equation for ( ) 149
7.1.4 Tolman’s interpretation of the Arrhenius activation energy 153
7.2 Applications of transition state theory 154
7.2.1 The molecular partition functions 154
7.2.2 Example 1: Reactions of two atoms A + B → A···B ‡ → AB 156
7.2.3 Example 2: The reaction F + H 2 → F···H···H ‡ → FH + H 157
7.2.4 Example 3: Deviations from Arrhenius behavior (T dependence of ( ))* 159
7.3 Thermodynamic interpretation of transition state theory 160
7.3.1 Fundamental equation of thermodynamic transition state theory 160
7.3.2 Applications of thermodynamic transition state theory 164
7.3.3 Kinetic isotope effects 166
7.3.4 Gibbs free enthalpy correlations 168
7.3.5 Pressure dependence of 168
7.4 Transition state spectroscopy 170
7.5 References 174
8 Unimolecular reactions 175
8.1 Experimental observations 175
8.2 Lindemann mechanism 178
8.3 Generalized Lindemann-Hinshelwood mechanism 182
8.3.1 Master equation 182
8.3.2 Equilibrium state populations 184
8.3.3 The density of vibrational states () 185
8.3.4 Unimolecular reaction rate constant in the low pressure regime 188
8.3.5 Unimolecular reaction rate constant in the high pressure regime 190
8.4 The specific unimolecular reaction rate constants () 191
8.4.1 Rice-Ramsperger-Kassel (RRK) theory 191
8.4.2 Rice-Ramsperger-Kassel-Marcus (RRKM) theory 193
8.4.3 The SACM and VRRKM model 198
8.4.4 Experimental results 199
8.5 Collisional energy transfer* 200
8.6 Recombination reactions 201

vi
8.7 References 202
9 Explosions 203
9.1 Chain reactions and chain explosions 203
9.1.1 Chain reactions without branching 203
9.1.2 Chain reactions with branching 204
9.2 Heat explosions 207
9.2.1 Semenov theory for “stirred reactors” ( =const) 207
9.2.2 Frank—Kamenetskii theory for “unstirred reactors” 211
9.2.3 Explosion limits 212
9.2.4 Growth curves and catastrophy theory 212
10 Catalysis 214
10.1 Kinetics of enzyme catalyzed reactions (homogeneous catalysis) 214
10.1.1 The Michaelis-Menten mechanism 215
10.1.2 Inhibition of enzymes: 217
10.2 Kinetics of heterogeneous reactions (surface reactions) 218
10.2.1 Reaction steps in heterogeneous catalysis 218
10.2.2 Physisorption and chemisorption 218
10.2.3 The Langmuir adsorption isotherm 219
10.2.4 Surface Diffusion 222
10.2.5 Types of heterogeneous reactions 223
11 Reactions in solution 227
11.1 Qualitative model of liquid phase reactions 227
11.2 Diffusion-controlled reactions 230
11.2.1 Experimental observations 230
11.2.2 Derivation of 230
11.3 Activation controlled reactions 234
11.4 Electron transfer reactions (Marcus theory) 235
11.5 Reactions of ions in solutions 238
11.5.1 Effect of the ionic strength of the solution 238
11.5.2 Kinetic salt effect 238
11.6 References 240

vii
12 Photochemical reactions 241
12.1 Basic radiative processes in a two-level system; Einstein coefficients 241
12.2 Fluorescence quenching (Stern-Volmer equation) 242
12.3 The radiative lifetime 0 and the fluorescence quantum yield Φ 243
12.4 Radiationless processes in photoexcited molecules 244
12.4.1 The Born-Oppenheimer approximation and its breakdown 244
12.4.2 Avoided crossings of two potential curves of a diatomic molecule 248
12.4.3 Quantum mechanical treatment of a coupled 2×2 system 250
12.4.4 Radiationless processes in polyatomic molecules: Conical intersections 252
13 Atmospheric chemistry* 256
14 Combustion chemistry* 257
15 Astrochemistry* 258
16 Energy transfer processes* 259
17 Reaction dynamics* 260
18 Electrochemical kinetics* 261
A Useful physicochemical constants 262
B The Marquardt-Levenberg non-linear least-squares fitting algorithm 263
C Solution of inhomogeneous differential equations 267
C.1 General method 267
C.2 Application to two consecutive first-order reactions 268
C.3 Application to parallel and consecutive first-order reactions 270
D Matrix methods 273
D.1 Definition 273
D.2 Special matrices and matrix operations 274
D.3 Determinants 277
D.4 Coordinate transformations 278
D.5 Systems of linear equations 280
D.6 Eigenvalue equations 282

viii
E Laplace transforms 286
F The Gamma function Γ () 288
G Ergebnisse der statistischen Thermodynamik 290
G.1 Boltzmann-Verteilung 291
G.2 Mittelwerte 293
G.3 Anwendung auf ein Zweiniveau-System 295
G.4 Mikrokanonische und makrokanonische Ensembles 298
G.5 Statistische Interpretation der thermodynamischen Zustandsgrößen 300
G.6 Chemisches Gleichgewicht 301
G.7 Zustandssumme für elektronische Zustände 303
G.8 Zustandssumme für die Schwingungsbewegung 304
G.9 Zustandssumme für die Rotationsbewegung 306
G.10 Zustandssumme für die Translationsbewegung 307

ix
Preface
This scriptum contains lecture notes for the module “Physical Chemistry 3: Chemical
Kinetics” (chem0405), the last part of the 3 semester course in Physical Chemistry for
B.Sc. students of Chemistry and Business Chemistry at CAU Kiel. It is assumed (but
not formally required) that students have previously taken courses on chemical thermodynamics
(“Physical Chemistry 1: Chemical Equilibrium” or equivalent) and elementary
quantum mechanics (“Physical Chemistry 2: Structure of Matter” or equivalent) as
well as “Mathematics for Chemistry” (parts 1 and 2, or equivalent). Module chem0405
includes 3 hours weekly (3 SWS) for lectures and 1 SWS of exercises (56 contact hours
combined). Students who started their studies at CAU Kiel should normally be in their
4th semester.
The scriptum gives a summary of the material covered in the scheduled lectures to
allow students to repeat the material more economically. It covers basic material that
all chemistry students should learn irrespective of their possible inclination towards inorganic,
organic or physical chemistry, but goes well beyond the standard Physical Chemistry
textbooks used in the PC-1 and PC-2 courses. This is done in recognition of the
established research focus at the Institute of Physical Chemistry at CAU to enable students
to pursue their B.Sc. thesis project in Physical Chemistry. Some more specialized
sections have been marked by asterisks and may be omitted on first reading towards the
B.Sc. degree. Useful additional reference material is given in the Appendix.
A brief lecture scriptum can never replace a textbook. For additional review, students
are strongly encouraged (and at places required) to consult the recommended books
on Chemical Kinetics beyond the standard Physical Chemistry textbooks. The book
by Logan 1 is a comprehensive overview of the field of chemical kinetics and a nice
introduction; it also has the advantage that it is written in German. The book by Pilling
and Seakins 2 gives a good introduction especially to gas phase kinetics. The book by
Houston 3 includes more information on the dynamics of chemical reactions. The book
by Barrante 4 is highly recommended for everyone who wants to brush-up some math
skills. A highly recommended web site for looking up mathematical definitions and
recipies is the MathWorld online encyclopedia (http://mathworld.wolfram.com).
The computer has become indispensable in modern research. This applies especially
to Chemical Kinetics, where only computers can solve (by numerical integration) the
coupled nonlinear partial differential equation systems that describe important complex
reaction systems (e.g., related to combustion, the atmosphere or climate change). The
integration of differential equations, a major task in chemical kinetics, can be performed
using Wolfram Alpha (http://www.wolframalpha.com). More complicated problems can
be solved with versatile computer algebra systems, e.g. (in order of increasing power)
Derive, 5 MuPad, MathCad, 6 Maple, Mathematica 7 . A graphics program (e.g., Origin, 8
1 S. R. Logan, Grundlagen der Chemischen Kinetik, Wiley-VCH, Weinheim, 1996.
2 M. J. Pilling, P. W. Seakins, Reaction Kinetics, Oxford University Press, Oxford, 1995.
3 P. L. Houston, Chemical Kinetics and Reaction Dynamics, McGraw-Hill, 2001.
4 J. R. Barrante, Applied Mathematics for Physical Chemistry, Prentice Hall, 2003 ($ 47.11).
5 Derive is available as shareware without charge.
6 MathCad is installed in the PC lab.
7 Mathematica has been used by the author for some of the lecture material.
8 Student versions of Origin are available for ≈ 35 − 120 (depending on run time).

x
or QtiPlot) may be needed for data evaluation and presentation. 9
As practically all original scientific literature is published in English (British and American;
in fact many chemists at some time in their professional lifes have to master the
subtle differences between both), this scriptum is written in English to introduce students
to the predominating language of science at an early stage. Students should also
note that the working language of some of the research groups (including that of the
author) in the Chemistry Department at CAU is English.
Finally, I would like to encourage all students to ask questions when they appear, whether
during the lecture or in the exercise classes!
Kiel, June 22, 2013,
F. Temps
9 Excel also allows you to plot simple functions and data, but is not sufficient for serious scientific
applications.

xi
Disclaimer
Use of this text is entirely at the reader’s risk. Some sections are incomplete, some
equations still have to be checked, some figures need to be redrawn, references are still
missing, etc. No warranties in any form whatever are given by the author.

xii
Organisational matters
• List of participants
• Lecture schedule
• Exams and grades
— Homework assignments
— Short questions
— Written final exam (“Klausur”)
I
Table 1: List of participants.
Chemistry (B.Sc.) 82
Business Chemistry (B.Sc.) 25
Biochemistry (B.Sc.)
Chemistry (2x-B.Sc.)
Physics (M.Sc.) 1
Other ) 1
Total 109
) Not registered for exercise classes and final written exam (“Klausur”).
I
Figure 1: Physical chemistry courses at CAU Kiel.

xiii
I
Figure 2: Organisational matters.
I
Figure 3: Requirements and grading.

xiv
I
Figure 4: Lecture contents.
I
Figure 5: Table of contents (continued). a
a Topics marked with asterisks will be omitted in this course for lack of time. They are the subject of
specialized lectures for advanced students.

xv
I
Figure 6: Recommended textbooks on chemical kinetics.
References to original publications and other recommended literature for further reading
willbegivenattheendofeachchapter.

1
1. Introduction
1.1 Types of chemical reactions
• Homogeneous reactions:
— Reactions in the gas phase, e.g.,
H 2 +Cl 2 → 2HCl
— Reactions in the liquid phase, e.g., H + -transfer, reduction/oxidation reactions,
hydrolysis, . . . (most of organic and inorganic chemistry)
— Reactions in the solid phase (usually very slow, except for special systems
and special conditions; not covered in this lecture).
• Heterogeneous reactions:
— Reactions at the gas-solid interface, e.g.,
C()+12O 2
() → CO()
— Reactions at the gas-liquid interface, e.g., heterogeneous catalysis, atmospheric
reactions on aerosols (e.g., Cl release into the gas phase from
polar stratospheric clouds in antarctic spring)
ClONO 2 ()+HCl() → Cl 2 ()+HNO 3 ()
— Reactions at the liquid-solid boundary, e.g., solvation, heterogeneous catalysis,
electrochemical reactions (electrode kinetics), . . .
• Irreversible reactions, e.g.,
H 2 +12O 2 → H 2 O
• Reversible reactions, e.g.,
H 2 +I 2 À 2HI
• Composite reactions, e.g.,
CH 4 +2O 2 → CO 2 +2H 2 O
• Elementary reactions, e.g.,
OH + H 2 → H 2 O+H
CH 3 NC → CH 3 CN

1.2 Time scales for chemical reactions 2
1.2 Time scales for chemical reactions
A chemical reaction requires some “contact” between the respective reaction partners.
The “time between contacts” thus determines the upper limit for the rate of a reaction.
I
Reactions in the gas phase:
• Free mean pathlength À molecular dimension (diameter) .
• Reaction rate is determined by time between collisions.
• Time between two collisions @ :
≈ 500 m s (1.1)
y
y
≈ 10 −7 m (1.2)
∆ ≈ ≈ 2 × 10 −10 s (1.3)
• Gas phase reactions take place on the nanosecond time scale:
I @ ≤ 1 ∆ =5× 109 s −1 (1.4)
I
Reactions in the liquid phase:
• Molecules are permanently in contact with each other.
• Typical distance between neighboring molecules (upper limit allowing for solvation
shell):
≈ 2 × 10 −8 10 −7 cm (1.5)
• Maximal reaction rate is determined by the rate of diffusion of the reacting molecules
in and out of the solvent cage.
I
Reactions in the solid phase:
• Atoms and molecules localized on fixed lattice positions.
• Reaction rate is determined by rate of diffusion (“hopping”) of the atoms and
molecules via vacancies (unoccupied lattice positions, “Fehlstellen”) or interstitial
sites (“Zwischengitterplätze”).
• Hopping from one lattice position to another usually requires a corresponding high
activation energy.
• Solid state reactions are usually very slow.

1.2 Time scales for chemical reactions 3
I
Time scale of molecular vibrations:
• The frequencies of the vibrational motions of the atoms in a molecule with respect
to each other determine an upper limit for the rate of a unimolecular reaction.
• Uncertainty principle (Fourier relation between frequency and time domain):
y
With = = ˜:
∆ ∆ ≥ ~ (1.6)
∆ ∆ ≥ 1
2
1
∆ ≈
2 ∆˜
• Application to a C-C stretching vibration:
y
y
(1.7)
(1.8)
˜(C-C) ≈ 1000 cm −1 (1.9)
= = ˜ =3× 10 13 s −1 (1.10)
= −1 =333 × 10 −14 s ≈ 33 fs (1.11)
• Application to slow torsional motions of large molecular groups in solution:
˜ ≈ 100 cm −1 (1.12)
y
∆ ≈ 330 fs (1.13)
I Femtochemistry: We can observe ultrafast chemical reactions directly nowadays by
following the motions of atoms or groups of atoms in the course of a reaction in real
time by femtosecond spectroscopy and femtosecond diffraction techniques. This new
area of “femtochemistry” has grown enormously in importance in the last decade, since
the award of the Nobel prize in chemistry to A.H. Zewail in 1999.
1 femtosecond =1fs=10 −15 s (1.14)
I Timescale of e − -jump processes: Only electron jumps are faster than vibrational
motions (by factor of roughly ).
• Estimated time for − -jump to another orbital due to absorption/emission of a
photon:
∆ ≈ 10 −18 s (1.15)
There are indications that electron correlation may slow this time to some
10 −17 s. 10
10 The group of F. Krausz (MPI for Quantum Optics, Munich) recently used attosecond metrology
based on an interferometric method to measure a delay of 21 ± 5as in the photo-induced emission
of electrons from the 2 orbitals of neon atoms with respect to emission from the 2 orbital by a
100 eV light pulse (Schultze et al, Science 328, 1658 (2010)).

1.2 Time scales for chemical reactions 4
• Other fast electron transfer processes:
— Autoionization of a doubly excited state, by one electron falling to a lower
orbital, while the other is excited into the ionization continuum.
• But “chemical” intra- and intermolecular − -transfer reactions are again slower:
— The electron jump itself is very fast (≈ 10 −18 s,asabove).
— However, in the Franck-Condon limit, the reactions usually require substantial
molecular rearrangements due to the structural differences between the
reactants and products (⇒ Marcus theory, Section 11.4). The net reaction
rate is thus again limited by the time scale of vibrational motion.
I
Conclusions:
• The time scales of chemical reactions span the enormous range of some 30 orders
of magnitude, from 1fs=10 −15 s up to 10 9 y=10 16 s!
• In view of this huge span, a factor-to-two experimental uncertainty in an experimental
measurement or a theoretical prediction of a rate constant really means
rather little.
• The enormous range of reaction rates, their complex dependencies on temperature,
pressure, and species concentrations, and the question of the underlying
reaction mechanisms that determine the reaction rates, make the field of chemical
reaction kinetics and dynamics very attractive and rewarding!

1.3 Historical events 5
1.3 Historical events
I Sucrose inversion reaction (Wilhelmy, 1850): Kinetic investigation of the H + catalyzed
hydrolysis of sucrose by monitoring the change as function of time of the optical
activity of the solution from reactant (sucrose) to products (glucose and fructose). 11
I
Table 1.1: Optical activity parameters for the sucrose inversion reaction.
C 12 H 22 O 11 +H 2 O −→ C 6 H 12 O 6 + C 6 H 12 O 6
sucrose glucose fructose
(1 M) ) +665 ◦ — +520 ◦ −920 ◦
(1 M) ) Ø =+665 ◦ Ø = −200 ◦
) (1 M) =specific angle of rotation of a 1 M solution at (Na-D).
I
Figure 1.1: Observed concentration-time profile of sucrose.
11 This experiment is carried out in the Physical Chemistry Lab Course 1 at CAU (“Grundpraktikum”).

1.3 Historical events 6
I
Empirical rate law:
ln
0
= − (1.16)
Differentiation gives
or
y Empirical rate law:
ln
− ln 0
= − ()=− (1.17)
| {z }
=0
1
= − (1.18)
= − (1.19)
Reaction rate :
≡−
(1.20)
Notice that even though there is one molecule of H 2 O taking part in the reaction, H2 O
does not appear in the rate law, because it is practically constant in the dilute solution.
I
Figure 1.2: Milestones in chemical kinetics.

1.3 Historical events 7
1.4 References
Arrhenius 1889 S. A. Arrhenius, Über die Reaktionsgeschwindigkeit bei der Inversion
von Rohrzucker durch Säuren, Z.Physik.Chem.4, 226 (1889).
Bunker 1966 D. L. Bunker, Theory of Elementary Gas Reaction Rates, Pergamon,
Oxford, 1966.
Glasstone 1941 S. Glasstone, K. Laidler, H. Eyring, TheTheoryofRateProcesses,
Mc Graw-Hill, New York, 1941.
Homann 1975 K. H. Homann, Reaktionskinetik, Grundzüge der Physikalischen
Chemie Bd. IV, R. Haase (Ed.), Steinkopf, Darmstadt, 1975.
Houston 1996 P. L. Houston, Chemical Kinetics and Reaction Dynamics, McGraw-
Hill, Boston, 2001.
Johnston 1966 H. S. Johnston, Gas Phase Reaction Rate Theory, Ronald,NewYork,
1966.
Kassel 1932 L. S. Kassel, The Kinetics of Homogeneous Gas Reactions, Chem.Catalog,
New York, 1932.
Logan 1996 S. R. Logan, Grundlagen der Chemischen Kinetik, Wiley-VCH, Weinheim,
1996.
Pilling 1995 M. J. Pilling, P. W. Seakins, Reaction Kinetics, Oxford University Press,
Oxford, 1995.
Smith 1980 I. W. M. Smith, Kinetics and Dynamics of Elementary Gas Reactions,
Butterworths, London, 1980.
Steinfeld 1989 J. I. Steinfeld, J. S. Francisco, W. L. Haase, Chemical Kinetics and
Dynamics, Prentice Hall, Englewood Cliffs, 1989.

1.4 References 8
2. Formal kinetics
2.1 Definitions and conventions
2.1.1 The rate of a chemical reaction
I
Rate of a simple reaction of type A + B → C+D:
≡− A
= − B
=+ C
=+ D
= − [A]
= − [B]
=+ [C]
=+ [D]
(2.1)
(2.2)
We will usually write the concentrations of a species A by using square brackets, i.e.,
[A] and understand that [A] is time dependent, i.e., [A ()].
I Definition of the rate of reaction: 12 Using the standard convention for the signs
of the stoichiometric coefficients of the reactants (−1) and products (+1), we write a
generic chemical reaction as
1 B 1 + 2 B 2 + → B + +1 B +1 + (2.3)
I
The progress of the reaction from reactants to products can be described using the
reaction progress coordinate :
=
(2.4)
The time dependence leads to the definition of the reaction rate.
Definition 2.1: The reaction rate is defined as
≡ 1
(2.5)
For the above reaction, therefore,
= − 1 1
| 1 | = − 1 2
| 2 | = =+ 1
| | =+ 1 +1
| +1 |
= (2.6)
I Units: 13
• SI unit for the reaction rate:
[] = mol m 3 s (2.7)
12 Reaction rate = Reaktionsgeschwindigkeit. Von manchen Schulen wird auch der Begriff “Reaktionsrate”
verwendet. Die bevorzugte “einzig wahre korrekte Übersetzung” wird von der jeweils anderen
Seite sehr ernst genommen, allerdings gilt in Kiel in dieser Hinsicht Waffenstillstand.
13 For general information on the SI system see (Mills 1988) or (Homann 1995).

2.1 Definitions and conventions 9
• Convenient units for gas phase reactions:
— molar units:
[] = mol cm 3 s (2.8)
— molecular units: 14 [] =molecules cm 3 s ⇒ cm −3 s −1 (2.9)
— We prefer to use molar units for rate constants, because rate constants are
inherently quantities that are “averages” of some other quantity, e.g., a molecular
cross section weighted by the relevant statistical distribution function
(usually the Boltzmann distribution). Rate constants are thus properties of
an ensemble of molecules, they are not original molecular quantities.
• Convenient units for liquid phase reactions:
[] = mol ls=M s (2.10)
2.1.2 Rate laws and rate coefficients
I Therateequation(ratelaw): Experiments show that the reaction rate generally
depends on the concentrations of the reactants. The functional dependence is expressed
by the rate equation (also called rate law), 15 e.g.
for reaction 2.3.
= − 1 [B 1 ]
=
| 1 |
(2.11)
= ([B 1 ] [B 2 ] [B ] [B +1 ] ) (2.12)
Theratelawsofcomplexreactions(i.e.,the ([B 1 ] , [B 2 ] , , [B ] , [B +1 ] , ))
are often rather complicated; only rarely (as in the case of elementary reactions) are
they simple and intuitive.
Often (but not always), ([B 1 ] , [B 2 ] , , [B ] , [B +1 ] , ) can be written as a
product of a rate coefficient and some power series of the concentrations, for
example
= − 1 [B 1 ]
= [B 1 ] [B 2 ] ... (2.13)
| 1 |
Note that the exponents . . . are not usually equal to 1 2 ...
I Therateconstant(ratecoefficient): The rate constant (also called rate coefficient)
16 usually depends on temperature ( = ( )) and sometimes also on pressure
( = ()), but not on the concentrations ( 6= ()).
14 IUPAC has decided that “molecules” is not a unit; see (Mills 1989) or (Homann 1995).
15 Rate law = Geschwindigkeits- oder Zeitgesetz.
16 Rate constant = Geschwindigkeitskonstante. Es wird manchmal auch die Übersetzung “Ratenkonstante”
verwendet (aber bitte nicht “Ratekonstante”).

2.1 Definitions and conventions 10
I Rate laws for first-order and second-order reactions: Simple rate laws describe
first-order and second-order reactions:
• Example of a first-order reaction: CH 3 NC → CH 3 CN
= − [CH 3NC]
=+ [CH 3CN]
= [CH 3 NC] (2.14)
• Example of a second-order reaction: C 4 H 9 +C 4 H 9 → C 8 H 18
= − 1 2
[C 4 H 9 ]
=+ [C 8H 18 ]
= [C 4 H 9 ] 2 (2.15)
I Question: How can we determine reaction rates and rate constants ?
I
Figure 2.1: The easiest — and always applicable — method for determining the reaction
rate and rate coefficient is the tangent method. To determine , yousimplytake
the slope of a plot of () at time ; then follows via Eq. 2.13.
We’ll learn about better methods for determining for elementary reactions (first-order
and second-order reactions) further below.
2.1.3 The order of a reaction
I
Definition 2.2: The order of a reaction is defined as the sum of the exponents in the
rate law (Eq. 2.13).

2.1 Definitions and conventions 11
I Example: Considering a generic rate law, like
− [A] = | | [A] [B] (2.16)
the total order of the reaction is = + . We also say, the reaction is -th order in
[A] and -th order in B, ...
The order of a reaction may be determined by the tangent method (Fig. 2.1): Its effects
on the reaction rate are observable by increasing (e.g., doubling) the concentrations of
the reaction partners one by one. Better methods will be explored later (⇒ homework
assignments).
I
Caveats:
• The order of a reaction is, in general, an empirical quantity; it has to be determined
by an experimental measurement.
• The order of a complex reaction cannot be determined by simply looking at the
overall reaction.
• There are reactions with integer order, like the first-order or second-order reactions
above (Eqs. 2.14 and 2.15). However, the order can also be fractional, as for the
following reactions:
(1) CH 3 CHO → CH 4 +CO:
[CH 4 ]
= [CH 3 CHO] 32 (2.17)
(2) H 2 +Br 2 → 2HBr:
[HBr]
= 0 [H 2 ][Br 2 ] 12
1+ 00 [HBr] [Br 2 ]
(2.18)
• A negative order in a concentration means that the respective species acts as an
inhibitor.
• A positive order in a product concentration means that the reaction rate is enhanced
by autocatalysis.
2.1.4 The molecularity of a reaction
I
Definition 2.3: The molecularity of a reaction is the number of molecules in an elementary
reaction step at the microscopic (molecular) level.

2.1 Definitions and conventions 12
I
There are only three possible cases for the molecularity (one example for each
case):
• monomolecular (= unimolecular) reactions:
CH 3 NC → CH 3 CN (2.19)
• bimolecular reactions:
O+H 2 → OH + H (2.20)
• termolecular reactions:
H+O 2 +M→ HO 2 +M (2.21)
• Events where more than three species come together in the right configuration
and react are extremely unlikely and do not play a role.
I Caveat: The order of a reaction can be integer or non-integer. However, the molecularity
can be only 1 or 2, or at the most 3.
2.1.5 Mechanisms of complex reactions
A stoichiometric formula rarely describes the reactive events happening at the microscopic,
molecular level. It is, for example, completely impossible that the combustion
of 1 -heptane molecule with 11 molecules of O 2
-C 7 H 16 +11O 2 → 7CO 2 +8H 2 O (2.22)
in an internal combustion engine occurs in a single collision and gives 15 product molecules.
Instead, the net reaction is the result of a complex sequence of a very large
number (thousands, in this case) of unimolecular, bimolecular, and termolecular elementary
reactions.
The identification of the individual elementary reactions, the determination of their rates
as a function of and , andtheverification of the ensuing reaction mechanism is one
of the main tasks in the field of chemical kinetics.
I
Examples for complex reaction systems:
• H 2 /O 2 system (Fig. 2.2),
• Combustion of CH 4 (Fig. 2.3),
• NO formation and removal in flames (Fig. 2.4).

2.1 Definitions and conventions 13
I
Figure 2.2: Reactions in the H 2 /O 2 system.
I Rate equations for the H 2 /O 2 system: Assuming that we know the reaction mechanism
(i.e., all elementary reactions contributing), we can write down the rate equations
for all species. Doing so, we obtain a set of coupled non-linear differential equations
(DE’s) which can be solved usually only numerically, if we want to describe the
net reaction:
[H]
[OH]
[O]
= − 1 [H] [O 2 ]+ 2 [O] [H 2 ]+ 3 [OH] [H 2 ] − 4 [H] [O 2 ][M] (2.23)
− 5 [H] [HO 2 ] − 6 [H] [HO 2 ]+
= 1 [H] [O 2 ]+ 2 [O] [H 2 ] − 3 [OH] [H 2 ]+2 6 [H] [HO 2 ] (2.24)
− 7 [OH] [HO 2 ] − 2 8 [OH] 2 [M]
+ (2.25)
= (2.26)

2.1 Definitions and conventions 14
I Figure 2.3: Mechanism of CH 4 combustion (Warnatz 1996).
I Figure 2.4: NO in combustion processes (Warnatz 1996).

2.1 Definitions and conventions 15
I
How we will proceed:
• In the remainder of this Chapter, we shall learn how to analytically solve the rate
equations for elementary reactions and for simple composite reactions.
• In Chapter 3, we shall then consider different methods for solving coupled nonlinear
differential equations for complex reaction systems.
— These methods are well established for stationary reaction systems.
— However, the modeling of instationary reaction systems (such as ignition
processes, explosions and detonations, 3D or 4D models of atmospheric
chemistry and climate evolution, . . . ) continues to be a tremendously challenging
task.

2.2 Kinetics of irreversible first-order reactions 16
2.2 Kinetics of irreversible first-order reactions
A → B (2.27)
I Solution of the rate equations for A and B:
− [A]
=+ [B]
= [A] (2.28)
Mass balance:
[A]
+ [B]
=0 (2.29)
Integration of the rate equation (Eq. 2.28):
Z
− [A] = (2.30)
[A]
Z
Z
ln [A] = −= − (2.31)
y
ln [A] = −+ (2.32)
Initial value condition at =0:
[A ( =0)]=[A] 0
(2.33)
y
Solution for [A ()]:
y
Solution for [B ()]:
y
=ln[A] 0
(2.34)
ln [A] = −+ln[A] 0
(2.35)
[A ()] = [A] 0
− (2.36)
[B ()] = [A] 0
− [A ()] (2.37)
¡
[B ()] = [A] ¢ 0 1 −
−
(2.38)

2.2 Kinetics of irreversible first-order reactions 17
I
Figure 2.5: Kinetics of first-order reactions.
I Time constant: The rate coefficient has the dimension of an inverse time. Thus,
= 1
(2.39)
hasdimensionoftimeandisthetimeafterwhich[A] has dropped to 1
its initial value:
[A] =
=[A] 0
− = [A] 0
≈ [A] 0
2718
(=368 %) of
(2.40)
I Half lifetime: The half lifetime 12 is the time after which [A] has dropped to 1 2 of
its initial value (Fig. 2.5):
y
£
A
¡
= 12
¢¤
=[A]0 − 12
= [A] 0
2
− 12
= 1 2
(2.41)
(2.42)
y
12 =ln2 (2.43)
12 = ln 2
(2.44)

2.2 Kinetics of irreversible first-order reactions 18
I
Conclusion:
Thehalflifetimeforafirst-order reactions is independent of the initial concentration!
I Experimental determination of k:
(1) Plot of ln [A] vs. gives a straight line with slope = − (see Fig. 2.5):
ln ([A] [A] 0
)=− (2.45)
Advantage: We do not need to know the absolute concentration of [A].
(2) Numerical fit of to the exponential decay curve using the Marquardt-Levenberg
non-linear least-squares fitting algorithm (Fig. 2.6; see Appendix B for details).
This method is useful in particular for analyzing experimental decay curves with
a constant background term. Again, we do not need to know the absolute concentration
of [A].
Advantages: Can be applied in presence of a constant background signal (noise,
weak DC signal). Can also be applied for analyzing data from pump-probe experiments,
when the duration of the pump is not short to the decay (deconvolution).
I Figure 2.6: Exponential decay curve of a particular vibration-rotation state of
the CH 3 O radical resulting from its unimolecular dissociation reaction according to
CH 3 O → H+H 2 CO (Dertinger 1995). The small box is the output box from a fit
using the software package ORIGIN.

2.2 Kinetics of irreversible first-order reactions 19
I
Figure 2.7: Excited-state relaxation dynamics of the adenine DNA dinucleotide after
UV excitation.
I Note (and warning): Anybody who wants to do least-squares fitting is urged to
consult appropriate literature before doing the fitting! The book by Bevington 17 is
considered required reading for any physical chemistry graduate student.
As the saying goes: With three parameters, you can draw an elephant. With a fourth
parameter, you can make him walk.
17 P. R. Bevington, D. K. Robinson, Data Reduction and Error Analysis for the Physical Sciences,
McGraw-Hill, Boston, 1992.

2.3 Kinetics of reversible first-order reactions (relaxation processes) 20
2.3 Kinetics of reversible first-order reactions (relaxation processes)
A
1 À
−1
B (2.46)
I
Rate equation:
[A]
= − 1 [A] + −1 [B] (2.47)
I
Equilibrium at t →∞:
[A]
y
= − [B]
= − 1 [A] + −1 [B] = 0 (2.48)
[B] ∞
[A] ∞
= 1
−1
= (2.49)
• Therateconstantforthereversereaction −1 can be calculated from 1 and .
• Or the equilibrium constant can be determined from measurements of the
forward and reverse rate constants. From ,wecanthendetermineimportant
thermochemical quantities, for example for reactive free radicals, via
= ( B ª )( C ª )
=exp
∙− ∆ ¸
ª
(2.50)
( A ª )
• Important note:
— In the frequently encountered case that the number of species on the reactant
and product side of a reaction differ, one has to be very careful with the
units of and .
Consider, for example, a gas phase reaction of the type
In this case, has the dimension of mol cm 3 :
B C
= 1 [B] [C]
= =
−1 [A]
0 has the dimension of bar:
A → B+C (2.51)
A
= 1 B C
A
= 0
(2.52)
0 = B C
(2.53)
A
But the thermodynamic equilibrium constant is dimensionless:
= ( ∙
B ª )( C ª )
= 0
( A ª ) =exp − ∆ ¸
ª
(2.54)
ª
— General relation between 0 ,and :
= 0 × ¡ ª¢ −Σ ( ) = × ¡ ª¢ Σ ( )
(2.55)

2.3 Kinetics of reversible first-order reactions (relaxation processes) 21
I
Solution of the rate equations for [A] and [B]:
Returning to the time dependence of the reaction
A
1 À
−1
B (2.56)
we now solve the rate equation
[A]
= − 1 [A] + −1 [B] (2.57)
• Mass balance:
[B] = [A] 0
− [A] (2.58)
• Integration:
[A]
= − 1 [A] + −1 [B] (2.59)
= − 1 [A] + −1 ([A] 0
− [A]) (2.60)
= − ( 1 + −1 )[A]+ −1 [A] 0
(2.61)
y
[A]
( 1 + −1 )[A]− −1 [A] 0
= − (2.62)
• Solution by substitution:
=( 1 + −1 )[A]− −1 [A] 0
(2.63)
y
y
[A] =( 1 + −1 ) (2.64)
[A] =
( 1 + −1 )
(2.65)
= −
( 1 + −1 )
(2.66)
= − ( 1 + −1 ) (2.67)
ln = − ( 1 + −1 ) + (2.68)
• Initial value condition at =0:
y
y
= 0 (2.69)
=ln 0 (2.70)
0
=exp[− ( 1 + −1 ) ] (2.71)

2.3 Kinetics of reversible first-order reactions (relaxation processes) 22
I Solution for [A(t)] (Fig. 2.8):
or
( 1 + −1 )[A]− −1 [A] 0
( 1 + −1 )[A] 0
− −1 [A] 0
=exp[− ( 1 + −1 ) ] (2.72)
−1
[A] − [A]
1 + 0
−1
[A] 0
−
=exp[− ( 1 + −1 ) ] (2.73)
−1
[A]
1 + 0
−1
This expression is not so easy to memorize, but we may recast it in a simple way:
Using
1
= [B] ∞
= [A] 0 − [A] ∞
(2.74)
−1 [A] ∞
[A] ∞
which gives
[A] ∞
=
−1
[A]
1 + 0
(2.75)
−1
we obtain
[A ()] − [A] ∞
=exp[− ( 1 + −1 ) ] (2.76)
[A] 0
− [A] ∞
With ∆ [A]
=[A()]−[A] ∞
and ∆ [A] 0
=[A] 0
−[A] ∞
, we write this result in compact
form as
∆ [A]
∆ [A] 0
=exp[− ( 1 + −1 ) ] (2.77)
I
Figure 2.8: Kinetics of reversible first-order reactions.

2.3 Kinetics of reversible first-order reactions (relaxation processes) 23
I
Relaxation processes:
• The transition from a deviation from equilibrium into the equilibrium is called
relaxation.
• ( 1 + −1 ) −1 has the dimension of time and is called the relaxation time,
=( 1 + −1 ) −1 (2.78)
• The temporal evolution of relaxation processes from a nonequilibrium to the equilibtium
state is determined by the sum of the respective forward and reverse rate
constants!

2.4 Kinetics of second-order reactions 24
2.4 Kinetics of second-order reactions
2.4.1 A + A → products
A+A→ P (2.79)
I
Solution of the rate equation:
Integration:
y
Initial value condition at =0:
− 1 [A]
= [A] 2 (2.80)
2
[A]
2
= −2 (2.81)
[A]
− 1 = −2 + (2.82)
[A]
[A ( =0)]=[A] 0
(2.83)
y
Solution for [A ()] (Fig. 2.15):
= − 1
[A] 0
(2.84)
1
[A] = 1 +2 (2.85)
[A] 0
I Experimental determination of k:
(1) Graphical method:
1
[A] = 1
[A] 0
+2 (2.86)
⇒ Plot of [A] −1 vs. gives a straight line with slope =2 (Fig. 2.9).
⇒ We do need to know the absolute concentration of [A]!
(2) We can again use a nonlinear least squares fitting method (Appendix B).

2.4 Kinetics of second-order reactions 25
I
Figure 2.9: Kinetics of second-order reactions.
I
Half lifetime:
y
y
y
£
A
¡
12
¢¤
=
[A] 0
2
(2.87)
2
[A] 0
= 1
[A] 0
+2 12 (2.88)
12 =
1
2 [A] 0
(2.89)
The half lifetime for a second-order reactions depends on the initial concentration!
2.4.2 A + B → products
A+B→ P (2.90)

2.4 Kinetics of second-order reactions 26
I
Solution of the rate equation:
− [A]
= − [B]
= [A] [B] (2.91)
(1) Substitution:
=([A] 0
− [A]
)=([B] 0
− [B]
) (2.92)
[A] = [A] 0
− (2.93)
y
= ([A] 0 − )([B] 0
− ) (2.94)
(2) Integration using partial fractions (skipped here):
(3) General solution:
[B] 0
[A]
[A] 0
[B]
=exp([A] 0
− [B] 0
) (2.95)
2.4.3 Pseudo first-order reactions
In experimental studies of second-order reactions of the type A+B→ C, onelikesto
use a high excess of one of the reactions partners (e.g., [B] À [A]) sothat[B] ≈ const
Under this condition, the reaction is said to be pseudo first-order, because
[B] = 0 (2.96)
y Solution of [A ()]:
[A] = [A] 0
−0 =[A] 0
−[B] (2.97)
I
Experimental investigation of pseudo first-order reactions:
(1) Measurements of the pseudo first-order decay of [A] vs. at different concentrations
of [B]: Plotofln [A] vs. give pseudo first-order rate constant 0 = [B]
for each value of [B].
(2) Plot of 0 = [B] vs. [B] gives straight line with slope .
⇒ We do not need the absolute concentration of [A], only the concentration of [B] is
needed.

2.4 Kinetics of second-order reactions 27
I
Figure 2.10: Kinetics of pseudo first-order reactions.
2.4.4 Comparison of first-order and second-order reactions
Consider a first-order and a second-order reaction with the same initial concentrations
and the same initial rate at = 0. As shown in Fig. 2.11 below, the rate of the
second-order reaction rapidly slows down as time increases.
I
Figure 2.11: Comparison of first-order and second-order reactions.

2.5 Kinetics of third-order reactions 28
2.5 Kinetics of third-order reactions
y
A+B+C→ D (2.98)
− [A]
= [A] [B] [C] (2.99)
I Pseudo second-order combination reactions: In combination reactions of two
species A + A → A 2 or A + B → AB (for example, the recombination of two free
radicals), one of the species (bath gas M) is usually present in large excess ([M] À [A],
[B]), so that the kinetics becomes pseudo second-order, giving the two cases:
•
•
A+A+M→ A 2 +M (2.100)
A+B+M→ AB + M (2.101)
M plays the role of an inert collision partner which removes the excess energy from the
newly formed A 2 or AB molecules.

2.6 Kinetics of simple composite reactions 29
2.6 Kinetics of simple composite reactions
2.6.1 Consecutive first-order reactions
A −→ 1
B (2.102)
B −→ 2
C (2.103)
I
Coupled differential equations:
[A]
= − 1 [A] (2.104)
[B]
=+ 1 [A] − 2 [B] (2.105)
[C]
=+ 2 [B] (2.106)
In this case, since the DE’s are linear, there is an exact solution which we will examine
first. We will also look at an approximate solution which can be obtained with the
quasi steady-state approximation and at methods to obtain numerical solutions, which
we have to use for complex non-linear inhomogeneous DE systems.
I
Mass balance:
[A] + [B] + [C] = [A] 0
(2.107)
I
Initial conditions:
[A ( =0)] = [A] 0
(2.108)
[B ( =0)] = 0 (2.109)
[C ( =0)] = 0 (2.110)
I
Exact solution:
• First-order decay of [A]:
y
[A]
= − 1 [A] (2.111)
[A] = [A] 0
− 1
(2.112)
• Inhomogeneous DE for [B]:
[B]
=+ 1 [A] − 2 [B] (2.113)

2.6 Kinetics of simple composite reactions 30
• Solution for [B ()] 18 :
⎧
1 [A] ⎪⎨ 0
¡
− 1 − ¢ − 2
if 1 6= 2
[B ()] =
2 − 1
(2.114)
⎪⎩
1 [A] 0
− 1
if 1 = 2
• Reaction products by mass balance:
[C ()] = [A] 0
− [A ()] − [B ()] = (2.115)
I
Figure 2.12: Kinetics of consecutive first-order reactions (case I).
18 ThesolutionisdetailedinAppendixC.

2.6 Kinetics of simple composite reactions 31
I
Figure 2.13: Kinetics of consecutive first-order reactions (case II).
I Caveat: As seen, the general solution for [B] is the difference of two exponentials,
[B] = 1 [A] 0
2 − 1
¡
− 1 − − 2 ¢ , (2.116)
One exponential describes the rise, the other the fall of [B]. It is sometimes implicitly,
but wrongly assumed that the rise time corresponds to 1 and the decay time to 2 .
The truth is that we cannot tell just from the shape of the concentration-time profile
whether 1 or 2 correspond to the rise or to the fall of [B].
I Quasi steady-state approximation: Under the condition that
2 À 1 (2.117)
we can find a simple solution for the DE’s. Under this condition, after a short initial
induction time (see Fig. 2.13), the change of [B] is very small compared to that of
[A]. Therefore,wehaveapproximately
[B]
≈ 0 (2.118)
This important approximation is known as the (quasi)steady-state approximation. 19
19 Deutsch: Quasistationaritätsannahme.

2.6 Kinetics of simple composite reactions 32
I Quasi steady-state solution for [B (t)]: For we have
y
y
[B]
= 1 [A] − 2 [B] ≈ 0 (2.119)
1 [A] ≈ 2 [B] (2.120)
[B]
= 1
2
[A] = 1
2
[A] 0
− 1
(2.121)
For , [C ()] is therefore given by
y
[C]
= 2 [B] ≈ 1 [A] (2.122)
[C] = [A] 0
¡
1 −
− 1 ¢ (2.123)
2.6.2 Reactions with pre-equilibrium
A
1 À
−1
B (2.124)
B 2 → C (2.125)
I
Equilibrium between [A] and [B]:
1 −1 À 2 (2.126)
y
[B]
[A] = 1
=
−1
(2.127)
[B] = [A] (2.128)
I
Time dependence of [C]:
[C]
= 2 [B] = 2 [A] (2.129)

2.6 Kinetics of simple composite reactions 33
2.6.3 Parallel reactions
A −→ 1
B (2.130)
A −→ 2
C (2.131)
A −→ 3
D (2.132)
y
y
and
[A]
= − ( 1 + 2 + 3 )[A] (2.133)
[A] = [A] 0
−( 1+ 2 + 3 )
(2.134)
[B] =
[C] =
[D] =
1 [A] 0
¡ ¢ 1 −
−( 1 + 2 + 3 )
( 1 + 2 + 3 )
(2.135)
2 [A] 0
¡ ¢ 1 −
−( 1 + 2 + 3 )
( 1 + 2 + 3 )
(2.136)
3 [A] 0
¡ ¢ 1 −
−( 1 + 2 + 3 )
( 1 + 2 + 3 )
(2.137)
Note that the decay of [A] is determined by the total removal rate constant =
1 + 2 + 3
2.6.4 Simultaneous first- and second-order reactions*
A −→ 1
C (2.138)
A+A−→ 2
D (2.139)
y
y
[A]
= − 1 [A] − 2 2 [A] 2
(2.140)
µ
1
[A] = −2 2 22
+ + 1
exp (− 1 )
1 1 [A] 0
(2.141)
I Approximation for k 1 t ¿ 1: Power series expansion of the exponential gives
1
[A] ≈ 1 µ
1
+ +2 2 × (2.142)
[A] 0
[A] 0

2.6 Kinetics of simple composite reactions 34
2.6.5 Competing second-order reactions*
A+A−→ 1
A 2 (2.143)
A+B−→ 2
P 1 (2.144)
B −→ 3
P 2 (2.145)
I
Solution for the case [A] À [B]:
(1)
[A]
= −2 1 [A] 2 − 2 [A] [B] (2.146)
≈−2 1 [A] 2 (2.147)
y
or
1
[A] − 1
[A] 0
=2 1 (2.148)
[A] = ¡ [A] 0 −1 +2 1 ¢ −1
(2.149)
=
[A] 0
1+2 1 [A] 0
(2.150)
(2)
[B]
= − 2 [A] [B] − 3 [B] (2.151)
This is another inhomogeneous first-order DE, for which the solution can be found
as outlined in Appendix C, giving
[B] = [B] 0
× (1 + 2 1 [A] 0
) − 22 1
× exp (− 3 ) (2.152)
I
Example:
1
CH 3 +CH 3 −→ C 2 H 6 (2.153)
CH 3 +OH−→ 2
CH 2 +H 2 O (2.154)
OH −→ 3
P 2 (2.155)
In the experimental study of reaction 2.154 (Deters 1998), the rate coefficient for the
reaction ( 2 according to Eq. 2.152) was fitted to measured OH concentration-time
profiles in the presence of high concentrations of CH 3 using the Marquardt-Levenberg
nonlinear least squares fitting algorithm (Press 1992). The rate coefficients 1 and
3 were determined by independent measurements. The absolute CH 3 concentrations,
which are needed for the evaluation of 2 according to Eq. 2.152, were obtained using
a quantitative titration reaction.

2.7 Temperature dependence of rate coefficients 35
2.7 Temperature dependence of rate coefficients
2.7.1 Arrhenius equation
I Arrhenius: 20 Svante Arrhenius (Arrhenius 1889) found the following empirical relation
describing the temperature dependence of the reaction rate constant:
ln
(1 ) = −
(2.156)
I
Definition 2.4: is called the Arrhenius activation energy of the reaction:
= − ln
(1 )
(2.157)
I
It is very important to keep in mind the following points:
• Eq. 2.156 is nothing else but a simple and very convenient way to describe the
empirical temperature dependence of reaction rate constants!
• Eq. 2.156 says that, within experimental error, a plot of ln vs. 1 should give
a straight line. We will see later, however, that this is rarely exactly true. The
study and the understanding of non-Arrhenius behavior is in fact an important
topicinmodernkinetics.
• as determined by Eqs. 2.156 and 2.157 is therefore a purely empirical quantity!
• We will use Eq. 2.157 simply as the definition and recipy for determining from
experimental data and to relate theoretical models to the experimental data!
I The rationale behind the Arrhenius equation (2.156): The rationale for the
Arrhenius equation is that molecules need additional energy so that chemical bonds can
be broken, atoms can rearrange, and new bonds can be formed (Fig. 2.14).
Van’t Hoff’s equation:
∆ = 2 ln
= 2 ln −→
= 2 ln −→ ←−
− 2 ln ←−
(2.158)
(2.159)
= −→ − ←− (2.160)
Thus we can write
ln
=
2 (2.161)
20 Arrhenius is generally regarded as one of the founders of physical chemistry (together with Faraday,
van’t Hoff, andOstwald).

2.7 Temperature dependence of rate coefficients 36
or 21
ln
(1 ) = −
(2.164)
I
Figure 2.14: Derivation of the Arrhenius equation.
I Experimental determination of E (see Fig. 2.14): Aplotofln vs. 1 gives
a straight line with slope −
I Integration of Eq. 2.156: Assuming that is independent of ,Eq.2.156canbe
integrated:
ln =
(2.165)
2
y
ln = −
+ (2.166)
y
( )= −
(2.167)
Equation 2.167 is used to represent ( ) over a limited -range using the two parameters
and :
• = Arrhenius activation energy (measure for the energy barrier of the reaction)
• = pre-exponential factor (frequency factor, collision frequency).
21
y
(1 )
= − 1
2 (2.162)
= − 2 (1 ) (2.163)

2.7 Temperature dependence of rate coefficients 37
I
I
It is important to realize, however, that the value of E is not equal to the
true height of the potential energy barrier for the reaction!
As said above, is an empirical quantity! Its relation to the true threshold energy 0 ,
which is the minimum energy above which reaction may occur, will be a main subject
in the theory of bimolecular and unimolecular reactions (Sections 5 — 8). There is also
a statistical interpretation of (see Section 8).
Figure 2.15: The reaction HCO → H + CO (G. Friedrichs).
2.7.2 Deviations from Arrhenius behavior
In general, is dependent on temperature, i.e., = ( ). For gas phase reactions,
for example, there is at least the √ dependence of the hard sphere collision frequency.
The -dependence of is often small compared to so that within experimental
error the dependence of ( ) is determined by .
If the -dependence of is not small compared to ,wewillhavetouseamore
general expression for ( ), as explained in the following.
I Generalized three-parameter expression for k (T ): To account for deviations
from Arrhenius because of ( ), one conveniently employs the three parameter expression
( )= − 0
(2.168)
with and 0 now being -independent quantities.
Range of values for specific bimolecular reactions (justified by theory; see later):
− 15 ≤ ≤ +3 (2.169)

2.7 Temperature dependence of rate coefficients 38
I
I
Question: What is the relation of Eq. 2.168 with the Arrhenius equation?
Answer: We have to evaluate the Arrhenius parameters from Eq. 2.168 according to
the definitions of and :
(1) :
=+ 2 ln
(2.170)
=+ 2 µ
ln + ln −
0
(2.171)
y
= 0 + (2.172)
Onlyinthecasethat ¿ 0 can we neglect the term compared to 0 !
(2) : From the expression for ,wehave
y
0 = − (2.173)
( )= −( − )
= −
= −
(2.174)
(2.175)
(2.176)
Comparing coefficients, we find
= (2.177)
2.7.3 Examples for non-Arrhenius behavior
Complex forming bimolecular reactions may show extreme non-Arrhenius behavior. An
extreme case is the reaction OH + CO → H+CO 2 , which incidentally is the single
most important reaction in hydrocarbon combustion next to H+O 2 → OH + O.

2.7 Temperature dependence of rate coefficients 39
I Figure 2.16: The reaction OH + CO → H+CO 2 (Fulle 1996, Kudla 1992).
Another case is encountered if tunneling plays a role (for example, H + transfer reactions
at low temperatures). In this case, ( ) tends to a plateau value as → 0 (Eisenberger
1991).

2.7 Temperature dependence of rate coefficients 40
2.8 References
Arrhenius 1889 S. A. Arrhenius, Über die Reaktionsgeschwindigkeit bei der Inversion
von Rohrzucker durch Säuren, Z.Physik.Chem.4, 226 (1889).
Bevington 1992 P. R. Bevington, D. K. Robinson, Data Reduction and Error Analysis
for the Physical Sciences, McGraw-Hill, Boston, 1992.
Dertinger 1995 S. Dertinger, A. Geers, J. Kappert, F. Temps, J. W. Wiebrecht,
Rotation-Vibration State Resolved Unimolecular Dynamics of Highly Vibrationally
Excited CH 3 O( 2 ): III. State Specific Dissociation Rates from Spectroscopic Line
Profiles and Time Resolved Measurements, Faraday Discuss. Roy. Soc. 102, 31
(1995).
Deters 1998 R. Deters, M. Otting, H. Gg. Wagner, F. Temps, B. László, S. Dóbé,
T. Bérces, A Direct Investigation of the Reaction CH 3 + OH at = 298K:
Overall Rate Coefficient and CH 2 Formation, Ber. Bunsenges. Phys. Chem. 102,
58 (1998).
Eisenberger 1991 H. Eisenberger, B. Nickel, A. A. Ruth, W. Al-Soufi, K.H.Grellmann,
M. Novo, Keto-Enol Tautomerization of 2-(2 0 -Hydroxyphenyl)benzoxazole
and 2-(2 0 -Hydroxy-4 0 -methylphenyl)benzoxazole in the Triplet State: Hydrogen
Tunneling and Isotope Effects. 2. Dual Phosphorescence Studies, J. Phys. Chem.
95, 10509 (1991).
Fulle 1996 D. Fulle, H. F. Hamann, H. Hippler, J. Troe, High Pressure Range of Addition
Reactions of HO. II. Temperature and Pressure Dependence of the Reaction
HO + CO À HOCO → H+CO 2 ,J.Chem.Phys.105, 983 (1996).
Homann 1995 K. H. Homann, SI-Einheiten, VCH,Weinheim,1995.
Kudla 1992 K. Kudla, A. G. Koures, L. B. Harding, G. C. Schatz, A quasiclassical
trajectory study of OH rotational excitation in OH + CO collisions using ab initio
potential surfaces, J. Chem. Phys. 96, 7465 (1992).
Mills 1988 I. M. Mills et al., Units and Symbols in Physical Chemistry, Blackwell,
Oxford, 1988.
Press 1992 W. H. Press, S. A. Teukolsky, W. T. Vetterling, B. P. Flannery, Numerical
Recipes in Fortran, Cambridge University Press, Cambridge, 1992. Versions are
also available for C and Pascal.
Warnatz 1996 J. Warnatz, U. Maas, R. W. Dibble, Combustion. Physical and Chemical
Fundamentals, Modeling and Simulation, Experiments, Pollutant Formation,
Springer, Heidelberg, 1996.

2.8 References 41
3. Kinetics of complex reaction systems
Real reaction systems are usually much more complicated:
• Net reaction mechanisms may consist of ≈ 100 - 10 000 elementary reactions.
• No analytical solutions for the rate laws.
• In favorable cases, one may be able to use certain approximations to simplify the
reaction systems, like
— apply steady-state approximation where possible,
— take into account reactions in equilibrium,
— take into account microscopic reversibility (detailed balancing).
• For more acurate descriptions, however, one usually needs numerical solutions
for the rate equations. Using modern computers, this is not a problem even for
10 000 reactions, as long as the transport processes are not too complicated
(as, for example, 1D reaction systems such premixed flames or 1D models of atmospheric
chemistry). However, numerical solutions are still extremely challenging
for instationary 3D reaction systems, such as
— 3D models of atmospheric chemistry including daily/annual variations and
couplings to the ocean,
— ignition processes (internal combustion engines).
3.1 Determination of the order of a reaction
Any reaction mechanism which we may postulate has to explain the observed reaction
order. Thus, the determination of the reaction order is a central topic.
We explore some general methods for determining the reaction order by considering an
example:
= − [A] = [A] (3.1)
I First-order and second-order reactions: Test whether plots of ln vs. or 1 vs.
, give the straight lines expected for first-order or second-order reactions, respectively.
I Log-log plot of reaction rate vs. concentration: We see from Eq. 3.1 that a
log-log plot of vs. [A] gives a straight line with slope :
lg ∝ lg [A] (3.2)

3.1 Determination of the order of a reaction 42
I Half lifetime method (for n 6= 1): 22
y
Z
− [A] = [A] (3.3)
Z
[A] − [A] = − (3.4)
y
−
1
( − 1) [A]−(−1) = −+ (3.5)
Initial value condition at =0:
[A ( =0)]=[A] 0
(3.6)
y
y
Half lifetime:
y
y
y
= − 1
( − 1) [A]−(−1) 0
(3.7)
1
[A] −1 − 1
[A] −1
0
2 −1
[A] −1
0
=( − 1) (3.8)
£
A
¡
12
¢¤
=
1
2 [A] 0
(3.9)
− 1
[A] −1
0
12 = 2−1 − 1
( − 1)
=( − 1) 12 (3.10)
1
[A] −1
0
(3.11)
lg 12 ∝−( − 1) lg [A] 0
(3.12)
y A log-log plot of 12 vs. [A] 0
gives a straight line with slope −( − 1).
22 We leave aside a usually occuring stoichiometric factor ν here, since we can always define νk = k 0
as rate constant k 0 .

3.2 Application of the steady-state assumption 43
3.2 Application of the steady-state assumption
The steady state assumption allows us to reduce a large reaction mechanism to a smaller
one by eliminating the respective intermediate species.
3.2.1 The reaction H 2 + Br 2 → 2 HBr
The reaction between H 2 and Br 2 (Bodenstein et al.; Christiansen, Polanyi & Herzfeld)
runs as a thermal reaction or as a photochemically induced reaction (initiated by light).
In the latter case, it is a chain reaction with a quantum yield of
Φ ≈ 10 6 (3.13)
I Definition 3.1: Quantum yield Φ:
Φ =
number of product molecules
number of absorbed photons
(3.14)
I
Reaction mechanism (elementary reactions) describing the H 2 -Br 2 system:
Br 2 → Br + Br chain initiation by thermal dissociation (1)
Br 2 + → Br + Br chain initiation by photolysis () (1 0 )
Br + H 2 → HBr + H chain propagation (2)
H+Br 2 → HBr + Br chain propagation (3)
H+HBr → H 2 +Br inhibition (4)
Br + HBr 6→ Br 2 +H endothermic (∆ ª =+170kJ mol) (6)
y reaction (6) can be neglected
Br + Br
+M → Br 2 chain termination (5)
(3.15)

3.2 Application of the steady-state assumption 44
I Rates equations: With the steady state approximations (i.e., [X] ≈ 0) for[Br]
and [H], weobtain 23
[H]
[Br]
[HBr]
= + 2 [Br] [H 2 ] − 3 [H] [Br 2 ] − 4 [H] [HBr] ≈ 0 (3.16)
y 2 [Br] [H 2 ] − 4 [H] [HBr] = + 3 [H] [Br 2 ] (3.17)
2 [Br] [H 2 ]
y [H]
=
3 [Br 2 ]+ 4 [HBr]
(3.18)
(3.19)
= +2 1 [Br 2 ] − 2 [Br] [H 2 ]+ 3 [H] [Br 2 ]+ 4 [H] [HBr]
| {z }
=−
[H]
≈0
−2 5 [Br] 2 (3.20)
= +2 1 [Br 2 ] − 2 5 [Br] 2 ≈ 0 (3.21)
µ 12 1
y [Br]
= [Br 2 ]
(3.22)
5
µ 12 1
2 [H 2 ]
y [H]
= [Br 2 ] ×
(3.23)
5 3 [Br 2 ]+ 4 [HBr]
= 2 [Br] [H 2 ]+ 3 [H] [Br 2 ] − 4 [H] [HBr] (3.24)
Using 2 [Br] [H 2 ] − 4 [H] [HBr] = + 3 [H] [Br 2 ] (Eq. 3.17) and substituting [H]
(Eq.
3.18), the last line simplifies to
[HBr]
=2 3 [H] [Br 2 ] (3.25)
µ 12 1
2 [H 2 ]
=2 3 [Br 2 ] × [Br 2 ] ×
5 3 [Br 2 ]+ 4 [HBr]
(3.26)
= 2 2 ( 1 5 ) 12 [H 2 ][Br 2 ] 12
1+( 4 3 )[HBr] [Br 2 ]
(3.27)
=
0 [H 2 ][Br 2 ] 12
1+ 00 [HBr] [Br 2 ]
(3.28)
I
Result:
with
[HBr]
= 0 [H 2 ][Br 2 ] 12
1+ 00 [HBr] [Br 2 ]
(3.29)
0 =2 2 ( 1 5 ) 12 and 00 = 4 3 (3.30)
I Conclusion: We see that HBr acts as an inhibitor of the total reaction.
23 For the photolysis induced reaction, 1 has to be replaced by 0 1 ,where is the light intensity and
0 1
the corresponding photolysis rate constants (related to the absorption coefficient).

3.2 Application of the steady-state assumption 45
3.2.2 Further examples
I
I
Exercise 3.1: The thermal decomposition of acetaldehyde according to the overall
reaction
CH 3 CHO → CH 4 +CO (3.31)
is known to proceed via the following elementary steps (Rice-Herzfeld mechanism):
CH 3 CHO → CH 3 +HCO (1)
CH 3 +CH 3 CHO → CH 4 +CH 3 CO (2)
(3.32)
CH 3 CO → CH 3 +CO (3)
CH 3 +CH 3 → C 2 H 6 (4)
Derive an expression for the production rate of [CH 4 ]. (Hint: Use the steady-state
approximations for [CH 3 CO] and [CH 3 ] and neglect the HCO for the derivation of the
expression.) ¤
A more complete mechanism takes into account the thermal decomposition of the HCO
radicals according to HCO → H + CO (5) followed by the reaction H + CH 3 CHO →
H 2 +CH 3 CO (6).
Solution 3.1: With the steady state approximations for [CH 3 CO] and [CH 3 ],weobtain
[CH 3 CO]
= 2 [CH 3 ][CH 3 CHO] − 3 [CH 3 CO] ≈ 0 (3.33)
y [CH 3 CO]
= 2 [CH 3 ][CH 3 CHO]
3
(3.34)
and
y
[CH 3 ]
= 1 [CH 3 CHO] − 2 [CH 3 ][CH 3 CHO] (3.35)
+ 3 [CH 3 CO] − 2 4 [CH 3 ] 2 (3.36)
= 1 [CH 3 CHO] − 2 [CH 3 ][CH 3 CHO] (3.37)
+ 3 2 [CH 3 ][CH 3 CHO]
− 2 4 [CH 3 ] 2
3
(3.38)
= 1 [CH 3 CHO] − 2 4 [CH 3 ] 2 ≈ 0 (3.39)
µ 12 1 [CH 3 CHO]
y [CH 3 ]
=
(3.40)
2 4
[CH 4 ]
[CH 4 ]
= 2 [CH 3 ][CH 3 CHO]
= 2
µ
1
2 4
[CH 3 CHO] 32 (3.41)
¥
I Exercise 3.2: The thermal decomposition of ethane gives mainly H 2 +C 2 H 4 . In
addition, small amounts of CH 4 are formed. The decay of the C 2 H 6 concentration can
be described by the reaction steps
C 2 H 6 → CH 3 +CH 3 (1)
CH 3 +C 2 H 6 → CH 4 +C 2 H 5 (2)
C 2 H 5 → C 2 H 4 +H (3)
(3.42)
H+C 2 H 6 → C 2 H 5 +H 2 (4)
H+C 2 H 5 → C 2 H 6 (5)

3.2 Application of the steady-state assumption 46
Using the steady-state approximations for the radicals [CH 3 ], [C 2 H 5 ],and[H], derivea
rate law describing the overall decay of [C 2 H 6 ]. Note: From the C-H bond dissociation
energies in C 2 H 6 and C 2 H 5 ,itisknownthat 3 À 1 . ¤
I Solution 3.2:
− [C 2H 6 ]
=
" µ
3
2
2 1 + 1
4 + 12
#
1 3 4
[C 2 H 6 ] (3.43)
5
¥

3.3 Microscopic reversibility and detailed balancing 47
3.3 Microscopic reversibility and detailed balancing
The principle of microscopic reversibility is a consequence of the time reversal
symmetry of classical and quantum mechanics: If all atomic momenta of a molecule
that has reached some product state from an initial state are reversed, the molecule will
return to its initial state via thesamepath.
A macroscopic manifestation of the principle of microscopic reversibility is detailed
balancing which states that
(1) in equilibrium, the forward and reverse rates of a process are equal, i.e., for a
1
reaction A 1 À A 2 ,
−1
and
1 [A 1 ]= −1 [A 2 ] (3.44)
(2) if molecules A 1 and A 2 are in equilibrium and A 2 and A 3 are in equilibrium, both
with respect to each other, there is also equilibrium between A 1 and A 3 :
A 1
1
À
−1
A 2
2
À
−2
A 3 (3.45)
y
A 1
3
À
−3
A 3 (3.46)
Thus, we can write
[A 2 ]
= 1
= 12
[A 1 ]
−1
(3.47)
[A 3 ]
= 2
= 23
[A 2 ]
−2
(3.48)
[A 3 ]
= 3
= 13
[A 1 ]
−3
(3.49)
= 12 23 = 1 2
−1 −2
(3.50)
Detailed balancing allows us to simplify complex reaction systems by eliminating − 1
out of a sequence of reversible reactions that are in equilibrium.

3.4 Generalized first-order kinetics* 48
3.4 Generalized first-order kinetics*
3.4.1 Matrix method*
For coupled complex reaction systems with only first-order reactions, the solution of the
rate equations can be reduced to an eigenvalue problem (Jost 1974). The solution of
the rate equations, i.e., the calculation of the time-dependent concentration profiles,
requires finding the eigenvalues and eigenvectors of the rate constant matrix of the
system. 24
Eigenvalue problems are readily solved in practice using numerical methods (Press 1992).
We consider an example which we solved using conventional methods in Section 2.3.
I
Example:
The rate equations
A 1
1
À
2
A 2 (3.51)
[A 1 ]
= − 1 [A 1 ]+ 2 [A 2 ] (3.52)
[A 2 ]
=+ 1 [A 1 ] − 2 [A 2 ] (3.53)
canbewritteninamorecompactformasamatrix equation:
⎛ ⎞
[A 1 ] µ µ
⎜
⎝ ⎟ −1 +
[A 2 ]
⎠ =
2 [A1 ]
(3.54)
+ 1 − 2 [A 2 ]
With the concentration vector
and the rate constant matrix
K =
A =
µ
[A1 ]
[A 2 ]
µ
−1 + 2
+ 1 − 2
(3.55)
(3.56)
this becomes
·
A = K · A (3.57)
I
Generalized matrix rate equation for coupled first-order reaction systems:
·
A = K · A (3.58)
24 A brief overview on matrices and determinants is given in Appendix D.

3.4 Generalized first-order kinetics* 49
I Ansatz: The ansatz for solving Eq. 3.58 assumes that we can express A in terms of
another vector B, 25 A = P · B (3.59)
where P · B means the dot product, using a rotation matrix P which is defined so that
it diagonalizes K via the eigenvalue equation K·P = P · Λ , which by multiplication
from the left with P −1 yields
P −1 · K · P = Λ (3.60)
The eigenvalue matrix Λ is a diagonal matrix of the form
⎛ ⎞
1 0 0
Λ = ⎝ 0 2 0
0 0 . ⎠ (3.61)
..
As defined by Eq. 3.60, its elements on the diagonal { 1 2 } are the eigenvalues of
the rate constant matrix. ThematrixP is called the eigenvector matrix associated
with K. The columns of P are the respective eigenvectors
associated with the
respective eigenvalues .
I Solution: Inserting the ansatz
A = P · B (3.62)
into our matrix equation 26 ·
A = K · A (3.63)
we have
or
(P · B)
Muliplication of this equation from the left by P −1 gives
= K · P · B (3.64)
P · ·
B = K · P · B (3.65)
P −1 · P · ·
B = P −1 · K · P · B (3.66)
or
·
B = Λ · B (3.67)
This DE can be immediately integrated since Λ is diagonal. The solution is
B = Λ B 0 (3.68)
where Λ is a diagonal matrix with elements © 1 2 ª , and Λ B 0 means
element-wise multiplication. The result gives the time dependence of B, i.e., B()
starting from the initial values B 0 .
25 The components of B will be seen to be linear combinations of the components of the concentration
vector A. B is immediately obtained once we have the matrix of eigenvectors P (see below).
26 The dot above a concentration vector indicates differentiation by .

3.4 Generalized first-order kinetics* 50
Having B(), we can transform back into the original concentration basis by using
B = P −1 · A (3.69)
and
B 0 = P −1 · A 0 (3.70)
InsertingintooursolutionforB (Eq. 3.68), we obtain
P −1 · A = Λ P −1 · A 0 (3.71)
The final step is now a multiplication from the left with P which yields the solution for
Eq. 3.58
A = P · Λ P −1 · A 0 (3.72)
I Note: This may look complicated to someone who is not used to it. However, once
we have set up K, which is straightforward, everything is done by the computer: The
computer computes the
• eigenvalues of the rate constant matrix K,
• the eigenvector matrix P oftherateconstantmatrixK,
• all necessary inverse matrices, etc.,
• and does all the matrix multiplications.
All you usually have to do is to program the K matrix.
I
Application to our example (solution “by hand”):
• Evaluation of the eigenvalue matrix Λ by solution of the eigenvalue equation
det (K − Λ · I) =0 (3.73)
with I as unity matrix:
y
¯ − 1 − + 2
+ 1 − 2 − ¯ =0 (3.74)
(− 1 − ) · (− 2 − ) − ( 1 · 2 )=0 (3.75)
1 2 + 1 + 2 + 2 − 1 2 =0 (3.76)
( +( 1 + 2 )) = 0 (3.77)
y
1 =0 (3.78)
2 = − ( 1 + 2 ) (3.79)
y
Λ =
µ
0 0
0 − ( 1 + 2 )
(3.80)

3.4 Generalized first-order kinetics* 51
• Conclusions:
(1) The largest (negative) eigenvalue ( 2 in this case) determines the shortest
time scale and thus the short-time evolution of the reactant concentrations!
(2) The smallest (negative) eigenvalue ( 1 in this case) determines the longest
time scale of the reaction, i.e., the evolution of the reactant concentrations
at long times. The smallest (negative) eigenvalue thus determines the net
reaction rate!
(3) Note that in the example the concentrations at long times are constant
(equilibrium!) and therefore 1 =0.
• Evaluation of the eigenvectors p 1 und p 2 by inserting the eigenvalues { 1 2 }
one by one into
K · P = Λ · P (3.81)
yields, for each and the respective p , a linear equation of the type
K · p = · p (3.82)
y
y
(K − ) · p =0 (3.83)
(− 1 − ) 1 + 2 2 =0 (3.84)
1 1 +(− 2 − ) 2 =0 (3.85)
• Inserting 1 =0:
− 1 1 + 2 2 =0 (3.86)
1 1 + − 2 2 =0 (3.87)
y
y
1 1 = 2 2 (3.88)
1 = 2 · (3.89)
2 = 1 · (3.90)
y
• Inserting 2 = − ( 1 + 2 ):
p 1 =
µ
2 ·
1 ·
(3.91)
(− 1 + 1 + 2 ) 1 + 2 2 =0 (3.92)
1 1 +(− 2 + 1 + 2 ) 2 =0 (3.93)
y
2 1 + 2 2 =0 (3.94)
1 1 + 1 2 =0 (3.95)

3.4 Generalized first-order kinetics* 52
y
y
1 = − 2 (3.96)
1 = (3.97)
2 = − (3.98)
y
µ
p 2 =
−
(3.99)
• Eigenvector matrix (= rotation matrix P):
µ
2
P =
1 −
(3.100)
• The arbitrary factor is determined by the initial conditions to be =1,sothat
µ
2 1
P =
(3.101)
−1
1
• Solution for A ():
A = P · Λ P −1 · A 0 (3.102)
I
Solution using symbolic algebra programs:
• Rate constant matrix K:
K =
µ
−1 + 2
+ 1 − 2
(3.103)
• eigenvalues(K) ={− 1 − 2 0} :
µ
− (1 +
Λ =
2 )0
0 0
(3.104)
• eigenvectors(K) =
½µ
−1
1
¾
(Ã 1
!)
↔− 1 − 2
2
1 ↔ 0:
1
⎛
P = ⎝ −1 ⎞
2
1
⎠ (3.105)
+1 1
• inverse(P) :
⎛
− ⎞
1 2
P −1 ⎜
= ⎝
1 + 2 1 + 2
⎟
1 1
⎠ (3.106)
1 + 2 1 + 2

3.4 Generalized first-order kinetics* 53
• B 0 = P −1 A 0 :
• Solution for B ():
• Solution for A ():
B 0 = P −1 A 0 = P −1 ·
µ
10
0
⎛
⎜
= ⎝
− 10
1
1 + 2
10
1
1 + 2
⎞
⎟
⎠ (3.107)
B = Λ B 0 (3.108)
µ ⎛ ⎞
1
−( 1 + 2
=
) −
0 ⎜ 10
0 1 ⎝
1 + 2
⎟
1
⎠ (3.109)
10
1 + 2
=
⎛
⎜
⎝
− 10
1
1 + 2
× −( 1+ 2 )
10
1
1 + 2
× 1
A = P · Λ P −1 · A 0 = P · Λ P −1 ·
µ µ
−( 1 + 2
= P ·
) 0
P
0 1
−1 10
·
0
= P ·
= P ·
=
=
⎜
⎝
⎞
⎟
⎠ (3.110)
µ
10
µ ⎛ 1
−( 1 + 2 ) −
0 ⎜ 10
0 1 ⎝
1 + 2
1
10
1 +
⎛ µ
⎞ 2
−( 1+ 2 )
1
− 10
1 + 2
1
µ
10
1
1 + 2
⎛
2
−( 1+ 2 )
10 + 10 1
⎜ 1 + 2 1 + 2
⎝ 1
−( 1+ 2 )
10 − 10 1
1 + 2
⎛
2 + 1 −( 1+ 2 )
10
⎜ 1 + 2
⎝ 1 − −( 1+ 2 )
10 1
1 + 2
1 +
⎞ 2
0
⎞
(3.111)
(3.112)
⎟
⎠ (3.113)
⎟
⎠ (3.114)
⎞
⎟
⎠ (3.115)
⎟
⎠ (3.116)
This result is identical to the result for [A ()] obtained in Section 2.3.
3.4.2 Laplace transforms*
The concept of Laplace and inverse Laplace transforms is extremely useful in chemical
kinetics because

3.4 Generalized first-order kinetics* 54
(1) Laplace transforms provide a convenient method for solving differential equations,
(2) Laplace transforms form the connection between microscopic molecular properties
and statistically (Boltzmann) averaged quantities.
I
Definition 3.2: The Laplace transform L [ ()] of a function () is defined as the
integral
Z ∞
() =L [ ()] = () − (3.117)
where
0
• is a real variable,
• () is a real function of the variable with the property () =0for 0,
• is a complex variable,
• () =L [ ()] is a function of the variable .
I
Definition 3.3: The inverse Laplace transform L −1 [ ()] of the function () is
defined as the integral
() =L −1 [ ()] = 1 Z
2
+∞
−∞
() (3.118)
where
• is an arbitrary real constant.
I Relation between f (t) and F (p): Since L −1 [ ()] recovers the original function
(), the pair of functions () and () is said to form a Laplace pair:
L −1 [ ()] = L −1 [L [ ()]] = () (3.119)
and
L [ ()] = L £ L −1 [ ()] ¤ = () (3.120)
A list of Laplace and inverse Laplace transforms of some functions can be found in
Appendix E (Table E.1).

3.4 Generalized first-order kinetics* 55
I Solution of differential equations using Laplace transforms: Consider the
Laplace transform of the derivative () of a function (): 27
L [() ] =
Z ∞
We can solve this equation for () via
0
()
= () −¯¯∞
− (3.121)
Z
0 − ∞
= − ( =0)+
0
Z ∞
0
() ¡ −¢ (3.122)
() − (3.123)
= − 0 + () (3.124)
() = L [() ]+ 0
and recover the solution () by inverse Laplace transformation,
() =L −1 ∙ L [() ]+0
¸
(3.125)
(3.126)
I Example: Application to two consecutive first-order reactions. As an example,
we consider the DE’s for two consecutive first-order reactions
A 1
→ B 2
→ C (3.127)
We already solved this problem previously (in Section 2.6.1) using conventional methods,
and repeat the solution now using the Laplace transform method.
Towards these ends, in this example, we shall denote the concentrations [A ()] and
[B ()] as () and () and their Laplace transforms as () and (). Therespective
initial concentrations shall be [A] 0
= 0 and [B] 0
= 0 =0. Thus we have:
• DE for the concentration ():
= 1 0 − 1 − 2 (3.128)
• Laplace transform: () can be found using Eq. 3.124:
L [] =− 0 + () (3.129)
Taking the Laplace transform of the r.h.s. of Eq. 3.128: 28
L [] =L £ 1 0 − 1 ¤ − L [ 2 ()] (3.130)
= 1 0 L £ − 1 ¤ − 2 L [ ()] = 1 0
+ 1
− 2 () (3.131)
= − 0 + () (Eq. 3.129) (3.132)
27 We used in line 2: integration by parts; in line 3: the differivative ( − ) = − − ;inline4:
the definition () =L [ ()] and the abbreviation ( =0)= 0 .
28 We used L [ ]= 1 from Table E.1 and L [ ()] = ().
−

3.4 Generalized first-order kinetics* 56
Since we have 0 =0from the initial condition, this becomes
()+ 2 () = 1 0
+ 1
(3.133)
• Solution for ():
() = 1 0
1
+ 1
1
+ 2
(3.134)
• Inverse Laplace transform: 29
() =L [ ()] = 1 0
2 − 1
¡
− 1 − − 2 ¢ (3.135)
which is the same as obtained in Section 2.6.1.
∙
¸
29 FromTableE.1,weseethatL −1 1
1 ¡
= − ¢ .
( − )( − ) ( − )

3.5 Numerical integration 57
3.5 Numerical integration
I Initial value problems: For complex reaction systems, we have to integrate the coupled
nonlinear DE systems numerically using the computer. We start with the known
initial values of the concentrations at time 0 and the associated DE’s ( · ) and want
to compute the values of at time 0 . In numerics, these problems are known as
initial value problems. With additional conditions, for example for the spatial distributions
of the concentration (fixed values at ( 0 0 0 1 )), they become boundary
value problems.
In solving these problems, we generally have to deal with systems of stiff nonlinear
coupled (ordinary) differential equations (DEsystemsaresaidtobestiff, ifthe
rate constants (more precisely: the eigenvalues) vary by many orders of magnitude;
under these conditons, the changes of the dependable variables differ by many orders of
magnitude).
Thus we have to solve
• equations (kinetic equations + transport processes), with
• variables (concentrations), and the given
• initial values and boundary values.
I
Applications of numerical integration methods:
• Combustion chemistry:
— Stationary combustion (for example diffusion flames),
— Instationary combustion (ignition processes, turbulent combustion) - first
successful attempts have been made,
— Problem: For all but the simplest fuels, we have to deal with hundreds of
species and thousands of elementary reactions.
• Atmospheric chemistry:
— 2D “box” models ( ( )) are straightforward.
— 4D models are required (time, latitude, longitude, height)!
— Huge problems, when it comes to
∗ feedback with oceans, etc.?
∗ do we know all reactions?
∗ heterogeneous reactions?
• Physical organic chemistry:
— Elucidation of reaction mechanisms.
• Macromolecular chemistry (including industrial macromolecular synthesis):
— Prediction of chain length and other propertiesofpolymersandco-polymers.

3.5 Numerical integration 58
I
Survey of methods of numerical integration methods:
• explicit methods (e.g., Runge-Kutta),
• implicit methods (e.g., Gear),
• extrapolation methods (e.g., Richardson, Bulirsch-Stoer, Deuflhard).
In the following, we consider the basics of the most convenient methods.
3.5.1 Taylor series expansions
The Taylor expansion is the basis for all numerical integration methods. We start from
the initial value of ( 0 ) at some time 0
( 0 )= 0 (3.136)
and want to determine the value ( 0 + ) at time 0 + . UsingtheDEfor
we expand in a Taylor series around 0 :
Recursion formula:
·
() =( ) (3.137)
( 0 + ) = ( 0 )+ · · ( 0 0 )+ 2
2! · ··
( 0 0 )+ (3.138)
= ( 0 )+ · ( 0 0 )+ 2
2! · ·
( 0 0 )+ (3.139)
+1 = + · ( )+ 2
2! · ·
( )+ (3.140)
The first-order term ( ) is given by the rate equation at , the second-order
term ( ·
) (which is taken into account for example by the 2 order Runge-Kutta
method) and higher order terms (⇒ higher order Runge-Kutta methods) have to be
evaluated numerically by finite differences.
3.5.2 Euler method
The Euler method is based on a 1 order Taylor expansion. If the step size is small
enough, the higher order terms can be neglected:
( 0 + ) = ( 0 )+ · ·
( 0 )+O( 2 ) (3.141)
Initial value problem:
·
= ( ) (3.142)

3.5 Numerical integration 59
with
( 0 )= 0 (3.143)
Result by substituting ·
( 0 )= ( 0 ):
( 0 + ) = ( 0 )+( 0 )+O ¡ 2¢ (3.144)
Recursion formula (see Fig. 3.1):
+1 = + (3.145)
+1 = + ( ) (3.146)
Problem: Large errors for practical step sizes.
I
Figure 3.1: Euler method.
3.5.3 Runge-Kutta methods
I
2 order Runge-Kutta method:
Initial value problem:
with
·
= ( ) (3.147)
( 0 )= 0 (3.148)

3.5 Numerical integration 60
2 order Runge-Kutta ansatz = Taylor expansion to 2 order:
( 0 + ) = ( 0 )+ · ( 0 )+ 2 ··
2! ( 0 )+O( 3 ) (3.149)
Ã
≈ ( 0 )+ ·
·
( 0 )+ 2 ( 0 + ) − ·
!
( 0 )
+ O( 3 ) (3.150)
2
= ( 0 )+ ³ ·(0 + )+ ·( 0 )´
2 | {z }
=2× slope of () at midpoint
+ O( 3 ) (3.151)
Recursion formula:
+1 = + (3.152)
+1 = + 2 + O( 3 ) (3.153)
1 = ( ) (3.154)
2
µ
=
+ 2 + 1
2
(3.155)
I
Figure 3.2: Illustration of the 2 order Runge-Kutta method.

3.5 Numerical integration 61
I 4 order Runge-Kutta method: In practice, one frequently uses the 4 order
Runge-Kutta method.
I
Figure 3.3: Illustration of the 4 order Runge-Kutta method.
3.5.4 Implicit methods (Gear):
The Runge-Kutta method becomes unstable for stiff DE systems. In this case, it is
preferable to employ implicit (also called iterative or backward) integration methods.
The starting point is the implicit form of the Euler method,
( 0 + ) = ( 0 )+ · ·
( 0 + )+O( 2 ) (3.156)
For the initial value problem ·
= ( ) with ( 0 )= 0 this leads to the recursion
formula:
+1 = + (3.157)
+1 = + ( +1 +1 ) (3.158)
Using the so-called predictor-corrector methods, +1 is predicted in a first step by
starting from using one of the explicit methods described above. The obtained trial
value (0)
+1 is then corrected by using a backward integration scheme (Gear 1971a, Gear
1971a, Press 1992) to obtain a better estimate +1. (1) Then, and (1)
+1 are used for
a new prediction-corrrection cycle, giving +1. (2) The procedure is repeated until the
difference between successive iterations falls below some tolerance limit .
The predictor-corrector method of Gear has been the method of choice for a long time
(Gear 1971a, Gear 1971a).

3.5 Numerical integration 62
3.5.5 Extrapolation methods (Bulirsch-Stoer)
The idea behind extrapolation methods is that the final answer for +1 is approximated
by an analytical function in the step size (Richardson extrapolation). The best strategy
is to use a rational extrapolating function as the ansatz, and not a polynomial function
(Bulirsch-Stoer). This function is determined and fitted to the problem of interest by
performing calculations with various values of . The extrapolating function is then
used for extrapolating from to +1 .
The optimal strategy for stiff DE systems is to follow the Bulirsch-Stoer method (Stoer
1980) in the form implemented by Deuflhard (Bader 1983, Deuflhard 1983, Deuflhard
1985). For Fortran subroutines, see (Press 1992).
I
Figure 3.4: The principle of extrapolation methods.
3.5.6 Example: Consecutive first-order reactions
As an example, we solve the DE’s for two consecutive first-order reactions
A 1
−→ B 2
−→ C (3.159)
(a)exactly,usingsymbolicalgebrasoftwarelikeMathCad,Maple,Mathematica,or
MuPad (the math engine built-in into SWP, the program used to write this script) and
(b) using numeric integration.

3.5 Numerical integration 63
I
Exact solution using symbolic algebra software:
(1) To set-up the initial value problem for the program, 30 we enter the following a
1 × 6 matrix:
⎡
0 ⎤
= − 1
0 =+ 1 − 2
0 =+ 2
(3.160)
⎢ (0) = 0 ⎥
⎣
(0) = 0
⎦
(0) = 0
(2) To obtain the solution, we select the menu item "SolveODE, Exact". The program
gives the output:
Exact solution is:
⎡
⎢ () = 1 µ
⎤
− 1
0 1 − 0 1 2 + 0 2 − 2
1
1 2 − 1 2 − ⎥ 1
⎢
⎣
− 1
− 2
() = 0 1 − 0 1
2 − 1 2 − 1
() = 1 µ
− 1
0 1 2 − 0 2 − 1
1
1 2 − 1 2 − 1
⎥
⎦
(3.161)
(3) After simplifying, factorizing, and rewriting the output, we obtain
⎡
() = 0
¡ ¢ ⎤
2 − 1 + 1 − 2
− 2 − 1
2 − 1 ⎢
1
¡ ¢
⎣ () = 0
− 2
− − 1 ⎥
1 − ⎦
2
() = 0 − 1
(3.162)
We immediately recognize these results from section 2.
I Numerical solutions: Numerical integration is performed by the different symbolic
algebra programs using one of the methods outlined in the preceding subsection. We
have to specify the DE’s, the initial values, and the values for the rate constants. 31
(1) We adopt the rate constant values 1 =100 and 2 =100.
(2) We define the problem (now using capital letters):
⎡
0 ⎤
= −
0 =+ − 10
0 =+10
⎢ (0) = 1 ⎥
⎣
(0) = 0
⎦
(0) = 0
(3.163)
30 The different software packages differinthewaysinwhichtheproblemshavetobesetup.
31 SWP can handle the problem only if the rate constant values are explicitly entered in the DE’s.
Other software does better.

3.5 Numerical integration 64
, Functions defined:
y 1.0
0.5
0.0
0 2 4
(3) Application of "SolveODE, Numeric" gives the output:
, Functions defined:
This means that the program now “knows” these functions.
(4) We can plot , , and using the Plot2D menu point:
1
c
0.75
0.5
0.25
0
0
1.25
2.5
3.75
5
t
(Weaddtheplotsfor and by dragging both symbols to the plot window)
(5) With 0 =1, 1 =1,and 2 =10, we also plot the exact solutions , , and
= 0
2 − 1
¡
2 − 1 + 1 − 2
− 2 − 1 ¢ (3.164)
1
¡
= 0
− 2
− ¢ − 1
1 − 2
(3.165)
= 0 − 1
(3.166)

3.5 Numerical integration 65
1
c
0.75
0.5
0.25
0
0
1.25
2.5
3.75
5
t
(6) The rest is just some fixing up: We combine the two plots, with an increased
number of points plotted, add axis labels and colors:
1
c(t)
0.75
0.5
0.25
0
0
1.25
Numerical solution of the DE system defined by Eq. 3.163. Full lines: numeric
solutions. Dotted lines: exact solutions.
2.5
t
3.75
5
The exact and the numerical solutions are identical within numeric precision.

3.6 Oscillating reactions* 66
3.6 Oscillating reactions*
The reactant and product concentrations normally approach equilibrium in a smooth
(often exponenential) fashion. Under special circumstances, however, the concentrations
may also show oscillations and/or bistability as they approach equilibrium. Such
behavior can occur only if there is a chemical feedback by autocatalysis.
3.6.1 Autocatalysis
Example for autocatalysis:
A+B → B+B (3.167)
Rate law:
− [A]
= [A] [B] (3.168)
To solve this DE, we substitute for [B ()] using the mass balance relation [A] 0
−[A ()] =
[B ()] − [B] 0
and introduce =[A()] as reaction progress variable:
[B ()] = [A] 0
+[B] 0
− [A ()] (3.169)
=[A] 0
+[B] 0
− (3.170)
This gives the DE
y
Z
−
= ([A] 0 +[B] 0
− ) (3.171)
Z
([A] 0
+[B] 0
) − = − (3.172)
2
Solution:
or
[B ()] =
µ
1 [A]0 [B ()]
ln
= − (3.173)
[A] 0
+[B] 0
[B] 0
[A ()]
[A] 0
+[B] 0
1+([A] 0
[B] 0
)exp[− ([A] 0
+[B] 0
) ]
(3.174)

3.6 Oscillating reactions* 67
I
Figure 3.5: Autocatalysis.
• Ascanbeseenfromthesigmoid(-shaped) concentration-time profile of [B] in
Fig. 3.5, the rate of increase of [B] increases up to the inflection point ∗ .
• This curve is typical for a population increase from a low plateau to a new (higher)
plateau after a favorable change of the environment due to, e.g., the availability
of more food, a technical revolution, a reduction of mortality by a medical
breakthrough, ...
Many interesting games which can be played based on these ideas are described in (Eigen
1975).
3.6.2 Chemical oscillations
The presence of one or more autocatalytic steps in a reaction mechanism can lead to
chemical oscillations.
a) The Lotka mechanism
The origin of chemical oscillations can be understood by considering the reaction mechanism
proposed by Lotka in 1910 (see Lotka 1920):
(3.175)
A+X 1
→ X+X (1)
X+Y 2
→ Y+Y (2)
Y
→
3
Z (3)

3.6 Oscillating reactions* 68
• Both reactions (1) and (2) are autocatalytic.
• The Lotka mechanism illustrates the principle, but it does not correspond to an
existing chemical reaction system.
• A real oscillating reaction is the Belousov-Zhabotinsky reaction; this reaction may
be described using a more complicated mechanism (see below).
In order to model the resulting chemical oscillations, we assume that the reaction takes
place in a flow reactor, which is constantly supplied with new A such that the concentration
of A stays constant ([A] = [A] 0
), while the product Z is constantly removed.
I
Rate equations and steady-state solutions:
[X]
[Y]
=+ 1 [A] 0
[X] − 2 [X] [Y] (3.176)
=+ 2 [X] [Y] − 3 [Y] (3.177)
Steady-state solutions:
[X]
[Y]
=+ 1 [A] 0
[X] − 2 [X] [Y] = 0 (3.178)
=+ 2 [X] [Y] − 3 [Y] = 0 (3.179)
Dividing these equations by [X] and [Y], respectively, we find
2 [Y]
= 1 [A] 0
(3.180)
2 [X]
= 3 (3.181)
Since [A] = [A] 0
, the steady state solutions for [X] and [Y] are independent of time!
I Displacements from steady-state solutions: What happens if the concentrations
are displaced from the steady-state values by small amounts and ?
(1) Ansatz:
[X] = [X]
+ (3.182)
[Y] = [Y]
+ (3.183)

3.6 Oscillating reactions* 69
(2) Rate equations:
=+ 1 [A] 0
([X]
+ ) − 2 ([X]
+ )([Y]
+ ) (3.184)
=+ 1 [A] 0
[X]
+ 1 [A] 0
− 2 [X]
[Y] | {z }
= 1 [A] 0
[X]
− 2 [X]
− 2 [Y]
− 2
| {z }
= 1 [A] 0
= − 2 [X]
− 2 (3.185)
=+ 2 ([X]
+ )([Y]
+ ) − 3 ([Y]
+ ) (3.186)
= 2 [X]
[Y]
+ 2 [X]
+ 2 [Y]
+ 2 − |{z} 3 [Y]
− 3 |{z}
= 2 [X]
= 2 [X]
=+ 2 [Y]
+ 2 (3.187)
(3) Simplified DE’s for small displacements: We may neglect the terms containing
. Thus, we obtain two coupled first-order DE’s
= − 2 [X]
(3.188)
=+ 2 [Y]
(3.189)
which can be converted to a single second-order DE by the ansatz
·
= − (3.190)
·
=+ (3.191)
Second-order DE:
··
= − · = − (3.192)
Formal solution of this second-order DE:
() =[X()] − [X]
= 1 + 2 − (3.193)
(4) Eigenvalues: To determine the eigenvalue belonging to the linear DE’s, we have
to solve the determinant
¯
¯0 − −
+ 0 − ¯ − 2 [X]
¯ 2 [Y]
¯ =0 (3.194)
y
2 + 2 [X]
[Y]
=0 (3.195)
y The solutions for are purely imaginary:
12 = ± ¡ ¢
2 2 12
[X]
[Y]
(3.196)
= ± ( 1 3 [A] 0
) 12 (3.197)

3.6 Oscillating reactions* 70
(5) Solution for ():
() = 1 + 2 − = 0 cos (3.198)
with
=( 1 3 [A] 0
) 12 (3.199)
This means that if the system is displaced from its steady-state, the species concentrations
will start to oscillate and the displacements may even grow in magnitude
until the amplitude reaches the limiting value 0 (see limit cycle diagram in
Fig. 3.6).
I
Figure 3.6: Limit cycle diagram for chemical oscillations according to the Lotka mechanism.
b) The Brusselator 32
Reaction scheme:
A →
1
X (1)
B+X 2
→ R+Y (2)
Y+2X (3.200)
3
→ 3X (3)
X →
4
S (4)
32 The name ‘Brusselator’ stems from the workplace of its inventor Prigogine (Brussels). The model
does not correspond to a real chemical system.

3.6 Oscillating reactions* 71
Here, A and B are reactants, R and S are final products, and X and Y are intermediates.
The autocatalytic reaction is reaction (3).
Rate equations (using “dimensionless time” = 4 ) and steady-state concentrations:
[X]
[X]
Steady-state concentrations:
= ·
[X] = 1 [A] − 2 [B] [X] + 3 [X] 2 [Y] − 4 [X]
4
(3.201)
= ·
[Y] = 2 [B] [X] − 3 [X] 2 [Y]
4
(3.202)
[X]
[Y]
= 1 [A]
4
(3.203)
= 2 4 [B]
1 3 [A]
(3.204)
We now investigate the effect of small displacements from steady-state by [X] and
[Y]. The reaction rates respond to these displacements according to
⎛ ⎞
⎛ ⎞
⎝ [X]
·
⎠ [X] and ⎝ [Y]
·
⎠ [Y] (3.205)
[X]
[Y]
y
• The steady-state is stable if
⎛ ⎞
⎝ [X]
·
⎠
[X]
• The steady-state is unstable if
⎛ ⎞
⎝ [X]
·
⎠
[X]
• Stability criterion:
⎛ ⎞
⎝ [X]
·
⎠
[X]
⎛ ⎞
+ ⎝ [Y]
·
⎠
[Y]
⎛ ⎞
+ ⎝ [Y]
·
⎠
[Y]
⎛ ⎞
+ ⎝ [Y]
·
⎠
[Y]
0 (3.206)
≥ 0 (3.207)
= 2 []
− 2 1 3 [] 2
− 1 (3.208)
4 4
2
c) Belousov-Zhabotinsky reaction
Oxidation of malonic acid to CO 2 by bromate, catalyzed by cerium ions in acidic solution:
2BrO − 3 +3CH 2 (COOH) 2
+2H + −→ 2BrCH(COOH) 2
+3CO 2 +4H 2 O (3.209)
The reaction is probed by addition of the redox indicator ferroin/ferriin: The color varies
between red and blue depending on the concentrations of Ce 4+ and Ce 3+

3.6 Oscillating reactions* 72
I
Figure 3.7: Mechanism of the Belousov-Zhabotinsky reaction.
I
Figure 3.8: Phases I - III of the Belousov-Zhabotinsky reaction.

3.6 Oscillating reactions* 73
I
Figure 3.9: Key steps of the Belousov-Zhabotinsky reaction.
d) The Oregonator 33
TheOregonatormodelcanbeusedtodescribetheBelousov-Zhabotinsky reaction.
I
Schematic model and kinetic equations:
A+Y 1
→ X+R (1)
X+Y 2
→ 2R (2)
A+X 3
→ 2X+2Z (3)
(3.210)
B+Z 5
→ 1 2 Y (5)
2X
→
4
A +R (4)
with A=BrO − 3 B=CH 2 (COOH) 2 + BrCH(COOH) 2 R=HOBr X=HBrO 2
Y=Br − Z=Ce 4+ .
Using dimensionless time ( = 5 [B] ), we obtain
33 The name ‘Oregonator’ stems from the workplace of its inventor Noyes (University of Oregon).

3.6 Oscillating reactions* 74
[X]
[Y]
[Z]
= 1 [A] [Y] − 2 [X] [Y] + 3 [A] [X] − 2 4 [X 2 ]
5 [B]
− 1 [A] [Y] − 2 [X] [Y] + 1
=
2 5 [B] [Z]
5 [B]
= 2 3 [A] [X] − 5 [B] [Z]
5 [B]
(3.211)
(3.212)
(3.213)
I Figure 3.10: Numerical integration of the Oregonator model (rate constants in
lmol −1 s −1 : 1 =0005, 2 =10, 3 =10, 4 =10, 5 = 0004; initial concentrations
in mol l −1 : [A] = 001, [B] = 10, [X] 0
=00, [Y] 0
=00, [Z] 0
=000025.)

3.6 Oscillating reactions* 75
3.7 References
Bader 1983 G. Bader, P. Deuflhard, Numerische Mathematik 41, 373 (1985)
Deuflhard 1983 P. Deuflhard, Numerische Mathematik 41, 399 (1985).
Deuflhard 1985 P. Deuflhard, SIAM Review 27, 505 (1985).
Eigen 1975 M. Eigen, R. Winkler, Das Spiel, Piper, München, 1975.
Gear 1971a C.W.Gear,Commun.Am.Comput.Mach.14, 1976 (1971).
Gear 1971b C. W. Gear, Numerical Initial Value Problems in Ordinary Differential
Equations, Prentice-Hall, Englewood Cliffs, 1971.
Houston 1996 P. L. Houston, Chemical Kinetics and Reaction Dynamics, McGraw-
Hill, Boston, 2001.
Jost 1974 W. Jost, in Physical Chemistry: An Advanced Treatise, Ed.:W.Jost,Vol.
VIa, Academic Press, New York, 1974.
Lotka 1920 A. J. Lotka, Undamped Oscillations Derived from the Law of Mass Action,
J. Am. Chem. Soc. 44, 1595 (1920).
Press 1992 W. H. Press, S. A. Teukolsky, W. T. Vetterling, B. P. Flannery, Numerical
Recipes in Fortran, Cambridge University Press, Cambridge, 1992.
Scott 1994 S. K. Scott, Oscillations, Waves, and Chaos in Chemical Kinetics, Oxford
Science Publications, Oxford, 1994.
Stoer 1980 J. Stoer, R. Bulirsch, Introduction to Numerical Analysis, Springer,New
York, 1980.

3.7 References 76
4. Experimental methods
In this section we consider experimental techniques for determining rate coefficients.
We’ll get to faster reactions as we go down the list.
Special techniques used in quantum state-resolved reaction dynamics, photodissociation,
or transition state spectroscopy (molecular beams, state specific detection, photofragment
translational spectroscopy, photofragment imaging, other pump-probe techniques,
fluorescence up-conversion, fs-CARS, fs-DFWM) will be considered later (in advanced
courses), but we’ll briefly touch photochemistry and “femtochemistry”.
4.1 End product analysis
• firststepforsettingupreactionmechanisms,
• all convenient analytical techniques (HPLC, GC, GC-MS, MS, UV, FTIR, . . . ),
• Problem: No “direct” kinetic information because of lack of time resolution.
I
Figure 4.1: FTIR spectrometer for kinetic studies of photolysis-induced reactions.

4.2 Modulation techniques 77
4.2 Modulation techniques
• especially for photo-induced reactions.
• photolysis light has to be modulated at a frequency that is comparable to that of
the reaction y timescaleof10’sofmsandlonger.
• kinetic information is derived from the “in-phase” and the “in-quadrature” components
of the reactant and product concentrations as function of modulation
frequency.
I Figure 4.2:

4.3 The discharge flow (DF) technique 78
4.3 The discharge flow (DF) technique
DF technique is used for gas phase reactions on ms timescale:
• laminar flow along tubular reactor:
∆ = ∆
with =meanflow speed, ∆ = distance along the flow tube.
(4.1)
• y first-order reaction gives exponential decay of concentration along ∆.
• in reality ⇒ Hagen-Poisseuille flow (parabolic flow profile),
• but fast radial diffusion of reactants (at pressures of 1 to 10 mbar) gives “plug
shaped” radial concentration profiles despite parabolic flow profile.
• more complicated, when “back diffusion” and radial diffusionhavetobetaken
into account (at higher pressures).
• problem: heterogeneous reactions lead to first-order decay of reactive radicals
even in the absence of a reactant.
Combinations with numerous highly sensitive detection techniques, for instance:
• mass spectrometry (MS): universal, but often low sensitivity; problems with fragmentationofmoleculesintheionsource,
• laser induced fluorescence (LIF): excitation to electronically excited state followed
by relaxation by spontaneous emission. very sensitive (e.g., for OH ( 2 Σ ← 2 Π)
detection limit below 10 6 molecules cm 3 ),
• electron spin resonance (ESR): for paramagnetic atoms by exploiting the Zeeman
effect depending on magnetic field
= 0 + (4.2)
∆ =1y ∆ = (4.3)
ESR transitions are in the microwave region (X band: 9GHz).
• laser magnetic resonance (LMR): very sensitive; based on observation of rotational
transitions between Zeeman-shifted rotational levels of paramagnetic radicals
( rot = rotational constant).
00 = rot ( +1)+ 00
00 (4.4)
0 = rot ( +1)+ 0 0 (4.5)
∆ = ±1 ∆ =0 ±1 : (4.6)
y 0 = rot ( 00 +1)( 00 +2)+ 0 0 (4.7)
y ∆ = =2 rot ( 00 +1)+ ( 0 0 − 00
00 ) (4.8)
LMR transitions are in the FIR region (100 m ≤ ≤ 2000 m).

4.3 The discharge flow (DF) technique 79
I
Figure 4.3: Schematic diagram of a DF/MS combination.
I
Figure 4.4: Schematic diagram of a DF/LIF combination.

4.3 The discharge flow (DF) technique 80
I
Figure 4.5: Principles of Electron Spin Resonance (ESR) and Laser Magnetic Resonance
(LMR).
I
Figure 4.6: Laser Magnetic Resonance (LMR).

4.3 The discharge flow (DF) technique 81
I
Figure 4.7: Schematic diagram of a DF/LMR combination
I
Figure 4.8: Flow reactor with laser flash photolysis and time-resolved mass spectrometric
detection of transient reactants and products.

4.4 Flash photolysis (FP) 82
4.4 Flash photolysis (FP)
• “pump-probe” technique:
— generation of transient species by flash photolysis using fast flashlampor
pulsed lasers (“pump”),
— fast detection of transient species by UV absorption spectrometry, laser induced
fluorescence (LIF), cavity ringdown spectroscopy (CRDS) (”probe”).
• for gas and liquid phase reactions on nanosecond timescale (Norrish & Porter).
I
Figure 4.9: Typical “Pump-Probe” set-up for kinetics studies by pulsed laser photolysis
and detection by laser-induced fluorescence.
I
Cavity Ringdown Spectroscopy (CRDS):
• CRDS is an ideal method for absorption spectroscopy with pulsed lasers despite
of their large (±5 %) amplitude fluctuations.
• The absorption measurement is transformed into the time domain:
— time dependence of signal transmitted through a resonantor:
= 0 exp (−) (4.9)

4.4 Flash photolysis (FP) 83
— decay of intensity for pass through cavity (length , mirrorreflectivity ,
absorption path length ):
µ
1 −
= −
−
(4.10)
| {z } | {z }
due to reflection losses at mirrors due to absorption
— with
y
— empty cavity ( =0):
— with absorber ( 6= 0):
= (4.11)
= − [(1 − )+] (4.12)
−1 (1 − )
0 =
(4.13)
−1 =
[(1 − )+]
(4.14)
— The difference of the ringdown times measures the absorption coefficient
and is related to absorber concentration [A] by
• Performance of a CRD spectrometer:
— mirrors:
— measuredemptycavitydecaytime:
— effective absorption pathlength (1 m cell):
• applications to chemical kinetics:
−1 − −1
0 = = [A] (4.15)
=99999% (4.16)
0 ≈ 300 s (4.17)
eff =90km (4.18)
— slow reactions: pump-probe scheme with variable time delay.
first-order reaction with rate constant 1 :
[A] = [A] 0
− 1
(4.19)
y
ln ¡ ¢
−1 − −1
0 = − 1 (4.20)
— fast reactions: kinetics and ringdown occur on the same time scale. y
(extended) Kinetics and Ringdown model ((e)SKaR); see Brown2000 and
Guo2003 for details.

4.4 Flash photolysis (FP) 84
I Figure 4.10:
I Figure 4.11:

4.5 Shock tube studies of high temperature kinetics 85
4.5 Shock tube studies of high temperature kinetics
• Detection of atoms by atomic resonance absorption spectrometry (ARAS),
• Detection of small radicals by frequency modulation absorption spectrometriy
(FM-AS)
I Figure 4.12:
I Figure 4.13:

4.6 Stopped flow studies 86
4.6 Stopped flow studies
• the analogue of the flow tube technique for liquid phase reactions.
• applications mainly kinetics of enzyme reactions, studies of protein folding dynamics
I Figure 4.14:

4.7 NMR spectroscopy 87
4.7 NMR spectroscopy
• line coalescence
• good for interconversion of two conformers (“isomerization”)
I Figure 4.15:

4.8 Relaxation methods 88
4.8 Relaxation methods
• application to fast reactions in the liquid phase (Eigen)
• -jump, -jump, -jump, . . . :
(1) ⇒ displacement from equilibrium
(2) ⇒ measurements of relaxation into new equilibrium state by transient absorption
spectroscopy, conductivity, . . .

4.9 Femtosecond spectroscopy 89
4.9 Femtosecond spectroscopy
I
Figure 4.16: Femtosecond pump-probe spectroscopy for the investigation of ultrafast
reactions (time resolution down to the order of 10 fs).
I
Figure 4.17: Signals in broadband femtosecond transient absorption spectroscopy.

4.9 Femtosecond spectroscopy 90
I Figure 4.18: Broadband femtosecond transient absorption spectroscopy of photochromic
molecular switches in SFB 677.
I
Figure 4.19: Broadband femtosecond transient absorption spectroscopy of Disperse
Red 1.

4.9 Femtosecond spectroscopy 91
I Figure 4.20: Broadband femtosecond transient absorption spectroscopy of Furyl
Fulgides.
I
Figure 4.21: Ultrafast Electronic Deactivation in DNA Building Blocks.

4.9 Femtosecond spectroscopy 92
I
Figure 4.22: Femtosecond fluorescence up-conversion data for DNA strands.
I
Figure 4.23: Ultrafast Electronic Deactivation in DNA Building Blocks: d(pApA) as
Model System.

4.9 Femtosecond spectroscopy 93
4.10 References
Busch 1999 K. W. Busch, M. A. Busch, Cavity-Ringdown Spectroscopy, ACSSymposium
Series, Vol 720, American Chemical Society, Washington, 1999.
Brown 2000 S.S.Brown,A.R.Ravishankara,H.Stark,J.Phys.Chem.A104, 7044
(2000).
Friebolin 1999 H.Friebolin, Ein- und zweidimensionale NMR-Spektroskopie, Wiley-
VCH, Weinheim, 1999.
Guo 2003 Y. Guo, M. Fikri, G. Friedrichs, F. Temps, An Extended Simultaneous Kinetics
and Ringdown Model: Determination of the Rate Constant for the Reaction
SiH 2 +O 2 .Phys.Chem.Chem.Phys.5, 4622 (2003).
OKeefe 1988 A.O’Keefe,D.A.G.Deacon,Rev.Sci.Instrum.59, 2544 (1988).

4.10 References 94
5. Collision theory
So far, we have considered more or less complex chemical reactions following more or
less complicated rate laws, but we have only considered the rate coefficients appearing
in the rate laws as empirical quantities, which had to be determined experimentally.
We shall now ask the question, whether we can rationalize and predict those rate coefficients
using theoretical models. Towards these ends, we have to look at the reactions
at a microscopic, molecular level. In particular, we have to
• consider the relation between the microscopic (individual molecule) properties and
the macroscopic (thermally averaged) rate quantities (rate constants, product
branching ratios)
• develop models for the microscopic (molecular level) progress of individual elementary
reactions, including the different molecular degrees of freedom (quantum
states).
• develop sufficiently simplified models averaging over the molecular degrees of
freedom which still allow us to make practically useful, relevant, and sufficiently
accurate predictions.
We shall do so in the next four chapters.
I Collision theory: To begin with, this chapter considers bimolecular reactions in the
gas phase. These obviously require a collision between two molecules as condition for a
reaction to take place.
• We start with the hard sphere collision model of structureless particles as the
simplest model for bimolecular reactions. Although this model has two major
flaws, it leads to the right result because of a compensation of errors.
• We then consider results of kinetic gas theory, which allow us to properly take
into account the molecular speed distributions.
• We apply the hard sphere collision model to understand transport processes in the
gas phase (diffusion, heat conductivity, viscosity).
• We then proceed with a correct derivation of advanced collision theory models
starting from differential cross sections (which can be measured in crossed molecular
beam experiments). Advanced collision theory is the method of choice for fast
gas phase reactions which can be studied in detail by molecular beam experiments
and handled by quantum theory.

5.1 Hardspherecollisiontheory 95
5.1 Hard sphere collision theory
5.1.1 The gas kinetic collision frequency
I
The gas kinetic collision frequency as upper limit for rate constants of bimolecular
reactions: The gas kinetic collision frequency constitues an upper limit for the
rate constant of bimolecular gas phase reactions. We assume that the molecules A and
B are structureless hard spheres (radii , ) moving around with fixedmeanspeeds
and and look at the number of collisions between A and B per unit time.
I
Figure 5.1: Derivation of the hard sphere gas kinetic collision frequency.
I
Discussion of the molecular quantities affecting the rate constant:
• collision parameter (= projected distance between centers of gravity):
(5.1)
• maximal collision parameter = max value leading to a collision max :
≤ max = + = (5.2)
• collision cross section = 2 (= reactive cross section) 34 :
= 2 = ( + ) 2 (5.3)
34 Deutsch: reaktiver Wirkungsquerschnitt.

5.1 Hardspherecollisiontheory 96
• mean speed of A and B:
=
=
r r
8 8
=
(5.4)
r r
8 8
=
(5.5)
• mean relative speed (2 different molecules, i.e., A 6= B): 35
=
r 8
(5.9)
with the reduced mass
=
+
(5.10)
• mean relative speed (single type of molecules, i.e., A = B):
y
=
+
= 1 2 (5.11)
= √ 2 ×
r
8
(5.12)
I
Hard sphere gas kinetic collision frequency:
• We want to determine the number of collisions of a single molecule A with
molecules B in time . This is given by the product of the (volume that A has
swept in ) × (number density of B):
— volume of cylinder that A has swept in time :
volume = cross section × (5.13)
y
2 × (5.14)
35 The mean relative speed of two molecules with velocities u and u is found by looking at the
square of the difference speed
2 = |u − u | 2 = | | 2 + | | 2 − 2 | || | cos (5.6)
The third term averaged over cos is 0, therefore
Since and
2 = | | 2 + | | 2 = 3
+ 3
= 3 ( + )
= 3
³
2´12 differ only by a constant factor, we also find that
=
s
8
(5.7)
(5.8)

5.1 Hardspherecollisiontheory 97
— number density of B
(5.15)
— Number of collisions of a single A in time : ⇒ number of molecules B in
that volume
= 2 × × (5.16)
• Number of all collisions between A and B:
= 2 × × (5.17)
• Modification for collisions between identical molecules (collisions must not be
counted twice):
= 1 2 × 2 × × (5.18)
=
√
2
2 × 2 × × (5.19)
• Comparison with rate expression:
y
−
= −
= × = (5.20)
• Expression for rate constant for collisions A + B:
max = 2 × = × (5.21)
y
µ 12
max = 2 8
×
(5.22)
• Modification for collisions between identical molecules (A + A):
max = 1 2 2 × = √ 1 µ 12
2 8
×
(5.23)
2
Theseexpressionsgivetheupper limits for the rate constants of bimolecular
reactions. The upper limits are reached only if
• there are no energetic restrictions (no energy barriers),
• all collisions lead to reaction (no steric constraints),
• molecules can be approximated as hard spheres (unrealistic, usually one has
),
• all molecules collide with the same relative speed (in reality there is a distribution
of speeds).

5.1 Hardspherecollisiontheory 98
I
Mean free pathlength:
• The mean free pathlength is the distance that a molecule flies between two successive
collisions:
y
=
distance
# of collisions = distance/time
# of collisions/time =
(5.24)
=
1
2 ×
∝ 1
• Modification for collisions between identical molecules (A + A):
(5.25)
= 1 √
2
1
2 ×
(5.26)
I
Time between two collisions:
1
=
1
2 ×
(5.27)
I
Example 5.1: Collision frequency for O 2 -N 2 @ =298K= 1000 mbar:
Molecular parameters:
2 = 2 2 =3467 Å (5.28)
2 = 2 2 =3798 Å (5.29)
y =36 Å =36 × 10 −10 m (5.30)
= 2 =41Å 2 (5.31)
28 × 32
=
28 + 32 × 10−3 kg mol × 1
s
(5.32)
=
8
=650m s (5.33)
(but (O 2 )=444m s) (5.34)
Rate constant:
µ 12
max = 2 8
×
(5.35)
=26 × 10 −10 cm 3 s (5.36)
=16 × 10 14 cm 3 mol s (5.37)
Mean free pathlength:
=1000mbar (5.38)
=40 × 10 −5 mol cm 3 (5.39)
=25 × 10 25 m 3 (5.40)
=36 × 10 −10 m (5.41)
=1× 10 −7 m (5.42)

5.1 Hardspherecollisiontheory 99
Rule-of-thumb:
Time between collisions:
=03mbary =03mm (5.43)
=1000mbar (5.44)
1 1
=
2 ×
(5.45)
=15 × 10 −10 s (5.46)
≈ 01ns (5.47)
¤
5.1.2 Allowance for a threshold energy for reaction
I The threshold energy E 0 : The majority of reactions exhibit a threshold energy, below
which the reaction does not occur (quantum mechanical tunneling is excluded here).
The molecules must have a minimum energy min to overcome this reaction threshold energy
0 . For our structureless spherical molecules, we consider only translational energy
: 36 0 : no reaction (5.48)
0 : reaction (5.49)
I Boltzmann distribution: Fraction of molecules with translational energy 0 :
µ
( 0 )
=exp − µ
0
=exp −
0
(5.50)
I
Resulting rate constant:
or
µ 12
max = 2 8
×
× − 0
(5.51)
µ 12
max = 2 8
×
× − 0
(5.52)
I Arrhenius activation energy: Applying the definition of the apparent (Arrhenius)
activation energy ,wefind that
= 2 ln
= 0 + 1 (5.53)
2
As we see, is not constant but (slightly) -dependent: The slope of a plot of ln
max
vs. −1 increases as →∞(or −1 → 0). The Arrhenius preexponential factor
would not be constant as well, it is also slightly -dependent.
36 We use and 0 to indicate the energies in molecular units, and and 0 for the respective
values in molar units.

5.1 Hard sphere collision theory 100
5.1.3 Allowance for steric hindrance
I The sterical factor p: In addition, even if 0 , not every collision will lead to
a reaction, because real molecules are not structureless: A reaction requires a specific
orientation of the molecules. This is taken into account by introducing a steric factor
:
µ 12
max = × 2 8
×
× − 0
(5.54)

5.2 Kinetic gas theory 101
5.2 Kinetic gas theory
5.2.1 The Maxwell-Boltzmann speed distributions of gas molecules in 1D and 3D
I
Figure 5.2: Experimental measurement of molecular speed distribution using a rotating
disk velocity selector and an effusive molecular beam (orrifice diameter ¿ ).
I
Rotating disk velocity selector:
• Flight time of molecules with velocity between two rotating disks (distance ):
1 =
(5.55)
• Time delay of second slit with respect to first slit ( =angle, = angular velocity):
2 =
(5.56)
• Transmission only if
y
1 = 2 (5.57)
= ×
(5.58)
In practice, one uses a number of velocity selectors in series. A measurement of the 1D
speed distribution ( ) = ( )
using this method gives

5.2 Kinetic gas theory 102
I The 1D Maxwell-Boltzmann speed distribution (Fig. 5.3):
r
µ
( ) =
2 × exp − 2
2
(5.59)
Note that this distribution is already normalized for
Z ∞
( ) =1 We check this
by substituting
r
r
r
=
2 y =
2 2
y = (5.60)
y
Z ∞
r Z ∞ r
2
( ) =
= 1 Z∞
√
2 −2 −2 = 1 √
√ =1 (5.61)
−∞
−∞
−∞
−∞
I
Figure 5.3: 1D Maxwell-Boltzmann speed distribution.

5.2 Kinetic gas theory 103
I Velocity distribution in 3 dimensions: Since the velocities in direction are
independent, the probability that a molecule has a velocity between and + ,
and + ,and and + is given be the product of ( ) , ( ) ,
and ( ) :
( )
= ( ) ( ) ( )
µ Ã
32
=
× exp − ¡ ¢ !
2 + 2 + 2
(5.62)
2
2
We are, however, not interested in the distribution of velocities pointing into a specific
direction ( ), but in the probability that a molecule has a certain speed || in
any direction, i.e., that the velocity vector ends somewhere on a sphere with radius
q
|| = 2 + 2 + 2 (5.63)
To obtain this distribution, we must integrate over all polar angles and . This
corresponds to replacing the volume element with the volume 4 2 of
a spherical shell with radius and thickness y
I The 3D Maxwell-Boltzmann speed distribution (Fig. 5.4):
() =
µ 32
× 4 2 × exp
µ− 2 (5.64)
2
2
Note that this distribution is also normalized for
Z ∞
() =1 We check this by
substituting
y
Z ∞
0
r
r
µ 12
=
2 y = 2 y = 2
(5.65)
= 2
2 (5.66)
2
() =
µ 32
× 4
2
0
Z ∞
0
0
µ
2 2
2 −2
12
(5.67)
µ 32 Z∞
µ 32 1 1
=4 × 2 −2 =4 × × 1 √ =1 (5.68)
4

5.2 Kinetic gas theory 104
I
Figure 5.4: 3D Maxwell-Boltzmann speed distribution.
I
Notice the differences between the following quantities:
• Value at maximum of ():
• Mean speed:
• Root mean square value:
max =
=
r
2
r
8
³
2´12 =
r
3
(5.69)
(5.70)
(5.71)
I
Table 5.1: Mean molecular speed values for some gases @ =298K
gas [m s]
H 2511
H 2 1776
He 1255
O 2 444
CO 2 379
Hg 178
238 U 163

5.2 Kinetic gas theory 105
I Kinetic energy distribution function: We may substitute for in Eq. 5.64:
= 1 2 2 (5.72)
µ 12 2
y =
(5.73)
y
µ 12 2
= × 1 µ 12 1
2 × ( ) −12 = ( ) −12 (5.74)
2
µ 12 1
y = ( ) −12 (5.75)
2
Insertion into Eq. 5.64 gives (see Fig. 5.5)
( ) =
y
µ 32 µ
µ 12 µ 12
2
× 4 × − 1 1
×
(5.76)
2
2
µ 32 1
( ) =2
( ) 12 − (5.77)
I
Figure 5.5: Maxwell-Boltzmann distribution function for the kinetic energy.
The fraction of molecules with 0 is temperature dependent.

5.2 Kinetic gas theory 106
5.2.2 The rate of wall collisions in 3D
Earlier, we calculated the rate of wall collisions of the molecules of a gas with mass
and number density in a container of volume and the resulting gas pressure using
asimplified model by assuming that 1 r
6 of all molecules have a fixed speed = 8
in + direction (cf. Fig. 5.6). In this simple picture, we neglected the 3D molecular
velocity distribution. We obtained the correct result only due to a cancelation of two
errors (assumption of fixed 1 and 3D distribution). In the following,
6
(1) we briefly review the simplified picture,
(2) we then proceed to derive the correct results by appropriate averaging over the
velocity/speed distribution.
I
Figure 5.6: Simplified model for the calculation of the pressure of an ideal gas.
I
Simplified model for the calculation of the pressure of an ideal gas:
• In the time , all gas molecules (mass ) inthevolume + hit the wall
area .
• Number of wall collisions in time (with number density ):
= × 1 × × (5.78)
6

5.2 Kinetic gas theory 107
• Momentum change due to wall collision, force on wall, and pressure:
=2 × 1 (5.79)
6
• Gas pressure:
y
= = 1
= 1 × 2 × 1 (5.80)
6
= 1 3 2 (5.81)
• Result for 1mol ( =
with =6022 1 × 10 23 mol −1 and kin = 2
2 ):
= 1 3 2 = 2 3 kin = 2 3 × 3 2 = (5.82)
I
Correct derivation of the gas kinetic wall collision frequency:
• Number of molecules with velocities in a volume of with =
p
2 + 2 + 2 (cf. Fig. 5.7) that will hit the area in the time :
( )=| | ( )
( )
( )
(5.83)
• To find the number of all molecules hitting in , we have to integrate over all
positive and over all (positive and negative) and , using the 1D-Maxwell-
Boltzmann distributions ( ) = ( )
,etc.:
µ 32
=
×
2
Z ∞ Z ∞ Z ∞ µ
| | exp − 2
2
0 −∞ −∞
µ µ
exp − 2
exp − 2
2 2
(5.84)
We know that the distributions over and are normalized to 1. Hence, we
only have to solve the integral over , which we do by the substitution
µ 12 µ 12 µ 12
2
=
y =
y =
(5.85)
2
2
y
Z ∞
0
µ
| | exp − 2
=
2
Z ∞
0
= 2
µ 2
Z ∞
0
12
|| −2 µ 2
−2 = 2
1
2 =
12
(5.86)
(5.87)

5.2 Kinetic gas theory 108
• Result: 37
=
µ 12
×
2
µ 12 = ×
(5.88)
2
y
wall = µ 12
= ×
(5.89)
2
This can be rewritten using the mean molecular speed , because
With
we thus obtain the
wall = × 1 4
µ 16
2
=
12
= × 1 4
µ 8
• Final result for the gas kinetic collision frequency:
µ 8
12
(5.90)
12
(5.91)
wall = 1 4 (5.92)
I
Figure 5.7: On the derivation of the gas kinetic wall collision frequency.
µ 12
37 Notice that two of the three
factors in Eq. 5.84 above are compensated by the
2
µ 12
integrations over and ,onlyone
factor remains.
2

5.2 Kinetic gas theory 109
I
Correct derivation of the gas pressure and the ideal gas equation:
• Likewise, we take the differential number of wall collisions for specific speeds(Eq.
5.83) to compute the momentum change ( ) in time
( )=2 | |×| | × ( ) ( ) ( )
(5.93)
• Inserting again the 1D-Maxwell-Boltzmann distributions and integrating over all
positive and all ,wefind the total momentum change in time :
µ 32
=
× 2×
2
Z ∞ Z ∞ Z ∞ µ
2 exp − 2
2
0 −∞ −∞
µ µ
exp − 2
exp − 2
2 2
• Pressure for = µ 12
again using the substitution =
2
y
y
=
=
=
µ 12
× 2
2
µ 1
12
2 2
Z ∞
0
Z ∞
0
(5.94)
µ
2 exp − 2
(5.95)
2
2 −2 =
(5.96)
= (5.97)
Thus, we have fixed our earlier simple derivations by appropriately integrating over the
Maxwell-Boltzmann velocity distribution of the molecules.

5.3 Transport processes in gases 110
5.3 Transport processes in gases
5.3.1 The general transport equation
I Definition of the flux −→ J of a quantity: We consider a generic transport property
Γ (5.98)
which differs in magnitude along the direction, i.e., has the gradient
grad Γ (5.99)
The flux of Γ throughanarea in time is defined as
andcanbewrittenas
−→ Γ = Γ
(5.100)
−→ Γ = − grad Γ (5.101)
Kinetic gas theory allows us to derive a general equation for the proportionality constant
.
I
Figure 5.8: Derivation of the general transport equation.

5.3 Transport processes in gases 111
I The general transport equation: We consider the net flux of Γ through an area
at = 0 that is transported by gas molecules coming from an area above at = 0 +
and from below from an area above at = 0 − (where isthemeanfreepathlength).
Denoting the respective transport quantity per molecule as Γ, these partial fluxes are
given by the number of gas kinetic collisions hitting on ( = 0 ) times the Γ carried
by each gas molecule (number density ):
• Using the gas kinetic wall collision frequency wall = 1 (Eq. 5.92), we obtain
4
Γ + = 1 µ Γ
µΓ
4 × × 0 + (5.102)
and
• Net effect:
Γ − = 1 4 × × µΓ 0 −
µ Γ
(5.103)
Γ = Γ + − Γ − = 1 µ Γ
2 × × (5.104)
• Net flux (with − sign to account for the fact that the flux is directed along the
downhill gradient of Γ):
−→ Γ = Γ
µ Γ
= −1 2
with given by (see the derivation of wall above)
=
µ 8
(5.105)
12
(5.106)
and, for only one type of gas molecules A (i.e., A - A collisons),
= 1 √
2
1
2 ×
(5.107)
• If also varies along (diffusion), Eq. 5.105 may be recast by defining the overall
transport quantity Γ 0 from the transport quantity per molecule Γ as
Γ 0 = Γ (5.108)
into the form
−→ Γ = − 1 2 Ã
!
Γ 0
(5.109)
In the following, ³ we shall apply Eqs. 5.105 or 5.109 to determine the self-diffusion coefficient
Γ =1 Γ 0 =
´, the heat conductivity ¡ Γ = ¢ ,andtheviscosity ¡ ¢
Γ =
of gases.

5.3 Transport processes in gases 112
5.3.2 Diffusion
I
Figure 5.9: The diffusion in a mixture A + B of one sort of molecules A against a
time-independent concentration gradient
is described by Fick’s first law.
• Fick’s first law:
−→ = µ
= −
(5.110)
• In order to derive an expression for the self-diffusion coefficient (diffusion of an
isotope of A in normal A), we make the ansatz
Γ =1 Γ 0 = (5.111)
y
• Result:
−→ = − 1 µ
2
(5.112)
= 1 (5.113)
2
• Fick’s second law (time-dependent concentration gradient):
• Einstein’s diffusion law in 1D:
• Einstein’s diffusion law in 3D:
= 2
2 (5.114)
∆ 2 =2 (5.115)
∆ 2 =6 (5.116)

5.3 Transport processes in gases 113
5.3.3 Heat conduction
I
Figure 5.10: Heat conduction is described by Fourier’s law.
• Fourier’s law (for constant temperature gradient):
−→ =
µ
= −
(5.117)
• Ansatz:
y
• Result:
Γ = = internal energy per molecule (5.118)
−→
= − 1 µ
2
= − 1 µ µ
2
= − 1 µ µ
2
= − 1 µ
2
(5.119)
(5.120)
(5.121)
(5.122)
= 1 2 (5.123)
with = number density and = specific heat per molecule.

5.3 Transport processes in gases 114
5.3.4 Viscosity
I
Figure 5.11: Newton’s law describes the inner friction in laminar flow.
• Newton’s law:
• Ansatz:
µ
= −
(5.124)
Γ = = momentum per molecule in direction (5.125)
y
• Result:
−→ = − 1 µ
2 ( )
= 1 ( )
= 1
=
= 1 (5.126)
2
with = number density and = mass of molecule.

5.4 Advanced collision theory 115
5.4 Advanced collision theory
The hard sphere model is a very primitive model. In reality, we have to account for the
following:
• The reaction probability depends on exact details of the collision, i.e.,
hard sphere model → ( ) (5.127)
• Molecules are not hard spheres with = 2 but “soft”, i.e., the reactive cross
section depends on the relative speed (or kinetic energy)
→ () → ( ) (5.128)
• The speed distribution follows the Maxwell-Boltzmann distribution, i.e.,
→ () → ( ) (5.129)
• Molecules in different quantum states behave differently, i.e.,
→ ( ; ) (5.130)
I Examples: The complexity of the problem is illustrated in Fig. 5.12 by two examples:
D( )+H 2 ( 2 2 2 ) −→ HD( )+H( ) (5.131)
OH( Ω )+H 2 ( 2 2 2 )
→ H 2 O( 2 1 2 3 )+H( ) (5.132)
As can be seen, the reaction rate is expected to depend on (speed, translational
energy), (vibrational quantum numbers ( 1 = symmetric stretch, 2 = bend, 3 =
antisymmetric stretch)), (rotational quantum numbers), Ω (fine structure
state, electronic state). In addition, the products are scattered into the spatial angles
.
I
Figure 5.12: Quantum state-specific elementary reactions.

5.4 Advanced collision theory 116
We shall now derive the fundamental equation of collision theory that allows us to
account for those effects.
5.4.1 Fundamental equation for reactive scatteringincrossedmolecularbeams
• Consider the following reactive scattering process (Fig. 5.13):
A+B→ C (5.133)
• IntensityofbeamofA:
= × (5.134)
Is this reasonable? ⇒ Look at the physical dimensions of × :
[cm s] × £ molecules/ cm 3¤ = £ molecules cm 2 s ¤ =[intensity] (5.135)
• Depletion of intensity of A due to scattering by B:
= − () × × ×
↑
↑
↑
(5.136)
y rate equation:
= −() (5.137)
• Rate coefficient for a given value of :
() =() (5.138)
• Averaging over the speed distribution () leads to the fundamental equation
of all improved collision theories:
( )=h()i (5.139)
i.e.,
( )= ∞ R
0
() ( ) (5.140)
• Frequently used approximation: Since the explicit form of () is usually not
known, we write
( )=hi h()i (5.141)
with
and
hi =
µ 8
12
(5.142)
h()i = -averaged (reactive) cross section (5.143)

5.4 Advanced collision theory 117
I
Figure 5.13: Derivation of the fundamental equation.
5.4.2 Crossed molecular beam experiments
I Differential and integral reactive cross sections: Experiments with crossed molecular
beams allow us to measure differential and integral reactive cross sections
(Herschbach 1987, Lee 1987).
We consider the reaction
( )+ ( ) −→ ( )+ ( ) (5.144)
where denotes the speeds of the molecules and stands for the quantum numbers
of their energy states (electronic, vibrational, rotational, nuclear spin). In an ideal
experiment (which has yet to be performed), all of the reactants are defined by, e.g.,
a specific excitation using laser, and the of the products are determined by some
probe laser. The can be determined in a molecular beam experiment using a rotating
disk velocity selector (see Fig. 5.14).

5.4 Advanced collision theory 118
I
Figure 5.14: Measurement of differential cross sections in crossed molecular beam
experiments.
I
Differential and integral cross sections from crossed molecular beam experiments:
• We define the differential cross section ( ) for scattering
of products C into the solid angle element Ω at such that
( ) Ω
= ( )
× × × ( ) × × ( ) × Ω (5.145)
• Corresponding integral cross sections are obtain by integration over and .
• Depending on the scattering process of interest, we distinguish between
— elastic cross sections ( ): measure of the size of the molecules,
— inelastic cross sections ( ):measureoftheefficiency of energy transfer
processes,
— reactive cross sections ( ): measure of reactivity.

5.4 Advanced collision theory 119
I
Figure 5.15: Measurement of differential cross sections in crossed molecular beam
experiments.

5.4 Advanced collision theory 120
5.4.3 From differential cross sections σ (...) to reaction rate constants k (T )
I
Figure 5.16: The strategy.
I
1 → 2:Integration over angles:
0
( )=
Z 2 Z
0
0
( ) 0
( )sin (5.146)
I 2 → 3:Summation over all product quantum states α 0 :
( )= X 0
0
( ) (5.147)
I 3 → 4:Thermal averaging over initial quantum states α:
( )= X
× ( ) (5.148)
with the Boltzmann factor and the partition function
=
−
P
−
= −
(5.149)
= X
−
(5.150)
y
( )= X
( ) × −
(5.151)

5.4 Advanced collision theory 121
I 5 → 6:Thermal averaging over all ² :
( )= X
× ( )
= X −
× ( )
Z
−
=
× ( )
I
3 → 5 and 4 → 6:Laplace Transformation:
Z ∞
( ) ∝ × ( ) × − (5.152)
0
This expression follows from the fundamental equation (Eq. 5.140) by substituting
for . The exact transformation (giving the correct factors before the integral) will be
carried out below (see Eq. 5.162).
5.4.4 Laplace transforms in chemistry
Formally, the transformation (Eq. 5.152) is a Laplace transformation (see Appendix
E). A Laplace transform generally described the transformation from a microscopic
(“microcanonical”) quantity such as ( ) to a macroscopic (“macrocanonical”, i.e.,
dependent) quantity such as ( ).
I Definition of the Laplace Transformation: The Laplace transform L [ ()] of a
function () is defined as the integral
() =L [ ()] =
Z ∞
0
() − (5.153)
I
Inverse Laplace Transformation:
with an arbitrary real number .
() =L −1 [ ()] = 1
2
Z
+∞
−∞
() (5.154)

5.4 Advanced collision theory 122
I
Notes:
(1) The so far best known Laplace transform to us is the partition function
( )= X
− =
Z ∞
0
() − (5.155)
(2) While the pathway from ( ) to ( ) by forward Laplace transformation is
straightforward, it is not generally straightforward to derive ( ) by inverse
Laplace transformation from ( ), because ( ) would be needed from =0
to = ∞.

5.4 Advanced collision theory 123
5.4.5 Bimolecular rate constants from collision theory
I Transformation from velocity to energy space: Taking our results from kinetic
gas theory, we are now in the position to apply the fundamental equation of collision
theory (Eq. 5.140)
( )=
Z ∞
0
() ( ) (5.156)
using various functional forms for . Towards these ends, we first transform from to
by inserting () from Eq. 5.64 (with the reduced mass instead of ). We thus
obtain
µ
( )=4
2
32
∞
As done above, we substitute for according to
and obtain
( )=
Z
0
3 ()exp
µ− 2 (5.157)
2
= 1 2 2 (5.158)
µ 12 2
y =
(5.159)
y
µ 12 2 1
=
2
µ 1
2
y =
µ 12 µ 1 2
µ 12 µ 12 µ 12 1 1 1
=
(5.160)
2
12 µ 1
12
(5.161)
32
∞
Z
0
( )exp
µ
−
(5.162)
The integration from 0 to ∞ takes into account that a reaction can occur only if
0 .
This expression is interpreted as follows:
• The function − describes the fraction of molecules with energy .
• The function ( ) in the integrand is called the excitation function 38 . This
can, in general, be a rather complicated (and not necessarily smooth) function
(see Fig. 5.17).
• The complete integrand ( ) − is called the reaction function 39 ,
because it describes the reactive fraction of the molecules.
• Therateconstant ( ) corresponds to the area under the reaction function.
38 “Anregungsfunktion”
39 “Reaktionsfunktion”

5.4 Advanced collision theory 124
I
Figure 5.17: Advanced collision theory: The excitation function.
I
Figure 5.18: Advanced collision theory: The reaction function.

5.4 Advanced collision theory 125
I Non-equilibrium energy distribution:* The above expression for ( ) is valid if
the reacting molecules are in internal equilibrium, i.e., the energy states are populated
according to the Boltzmann distribution.
Deviations from the Boltzmann distributions will occur in the case of very fast reactions,
if the reaction is faster than the relaxation between the energy states. In this case, ( )
decreases because the mean energy of the reacting molecule decreases.
Mean energy of the reacting molecules (see Fig. 5.19):
= X
=
Z ∞
0
() (5.163)
(1) For thermal equilibrium:
() =() (5.164)
with () being the Boltzmann distribution.
y
(2) Non-equilibrium:
() () (5.165)
non-eq. eq. (5.166)
I
Figure 5.19: Mean energy of the reacting molecules in thermal equilibrium and for a
non-eqilibrium situation.
I Conclusion: The non-equilibrium distribution leads to a reduced rate coefficient.

5.4 Advanced collision theory 126
a) Advanced collision theory for hard spheres (Fig. 5.20)
I
Figure 5.20: Thehardspheremodel.
(1) Ansatz for the reactive cross section:
( )=
(2) Inserting this ansatz into Eq. 5.162:
( )=
(3) Result for ( ): With
we obtain
y
µ 12 µ 1 2
Z
½
2 0
0 0
(5.167)
32
∞
Z
0
2 exp
µ
−
(5.168)
=
( − 1) (5.169)
2 µ 12 µ 32 1 2
( )=
2
µ
× ( ) 2 × exp − µ
− ∞
¯¯¯¯ − 1 (5.170)
0
( )= 2 µ 8
12 µ
1+ µ
0
exp −
0
(5.171)

5.4 Advanced collision theory 127
(4) Arrhenius activation energy:
y
= 2 ln
ln ∝ 1 µ
2 ln +ln 1+
0
− 0
ln
= 1 µ
1
2 + 1
− µ
0 0
+
1+ 0 2 2
= 0 + 1 2 − 0
1+ 0
(5.172)
(5.173)
(5.174)
(5.175)
b) Line-of-Centers model (Figs. 5.21 - 5.22)
Even for hard spheres, because of angular momentum conservation (see below), not the
entire collision energy, but only the fraction “directed” along the line of centers ()
is effective.
In particular, we have
• for a collision with =0:
= (5.176)
y central collisions are most effective,
• for a collision with = :
=0 (5.177)
y glancing collisions are ineffective because =0,
• general expression (see below):
=
µ
1 − 2
2
(5.178)
y decreases with increasing .

5.4 Advanced collision theory 128
I
Figure 5.21: Collision energy along line-of-centers
Functional expression for ( ):
(1) We assume that a reaction can occur only if ≥ 0 . Thus, we first determine
(see Fig. 5.21):
y
= cos (5.179)
2 = 2 cos 2 (5.180)
= ¡ 2 1 − sin 2 ¢ (5.181)
µ
= 2 1 − 2
(5.182)
2
=
µ
1 − 2
2
(5.183)
(2) We see from this expression that decreases with increasing . Thus, for a
given ,wehaveamaximalvalueof (i.e., a value max ), for which ≥ 0 .
We determine max from the condition that
µ
= 1 − 2
≥
2 0 (5.184)
y
y
2 ≤ 2 µ
1 − 0
= 2 max (5.185)
( )= 2 max = 2 µ
1 − 0
(5.186)

5.4 Advanced collision theory 129
y
⎧
⎨
( )=
⎩
2 µ
1 − 0
0
(5.187)
0 0
( ) is shown as function of in Fig. 5.22. Crossed molecular beam experiments
have indeed shown similar dependencies for a number of reactions.
I
Figure 5.22: Reactive cross section for the line-of-centers model.
(3) Result for ( ) from Eq. 5.162:
y
µ 1
( )=
Z ∞ µ
× 2
0
=
µ 1
×
Z ∞
0
( )=
12 µ 2
32
1 −
0
12 µ 2
32
2 ( − 0 )exp
µ 8
µ
exp −
(5.188)
µ
−
(5.189)
12 µ
2 exp −
0
(5.190)
This result is identical to that of our primitive “hard sphere-fixedmeanrelative
speed” model from Section 5.1!

5.4 Advanced collision theory 130
(4) Arrhenius activation energy:
y
= 2 ln
(5.191)
= 0 + 1 (5.192)
2
c) Allowance for the “steric effect”
In order to account for the “structures” of the molecules which can undergo a reaction
only if the reaction partners have a specificrelativeorientation,weintroducea(generally
-dependent) steric factor ( ):
µ 12 µ
8
( )=( )
2 exp −
0
(5.193)
In ( ), we include all sorts of factors leading to deviations from the predicitions by simple
collision theory. In favorable cases, ( ) can be predicted by theory. Towards these
ends, we define a reaction probability ( ) (also called the opacity function,
andcalculatedbytheory)andintegrateaccordingto
( )=
Z
0 max
( )2 (5.194)
0
In other cases, it may be convenient to define an angle dependent reaction threshold
energy ∗ 0:
∗ 0 = 0 + 0 (1 − cos ) (5.195)
y
∗ 0 = 0 for =0
∗ (5.196)
0 0 for 6= 0
Such models are helpful for explaining steric factors in the range of 1 ≤ ≤ 0001.

5.4 Advanced collision theory 131
I
Figure 5.23: Collision theory: Allowance for steric effects.
I
Table 5.2: Comparison of predictions by collision theory with experimental data.

5.4 Advanced collision theory 132
I Harpooning: The value of ≥ 1 for the steric factor of the K + Br 2 reaction can
be understood by the “harpooning mechanism”: An electron jumps from the K to the
Br 2 at a long intermolecular distance. The subsequent collision and reaction at closer
distance then occurs between a K + and Br − 2 . The collision between these two is strongly
affectedbytheCoulombattraction.
5.4.6 Long-range intermolecular interactions
To this point, we have neglected any interactions between the reacting molecules other
than the the collision itself. Thus, only straight line trajectories were allowed. In reality,
however, there will be attractive or repulsive interactions between the colliding molecules
(see Fig. 5.24). These will inevitably affect the collision dynamics. An extreme example
is the K + Br 2 reaction mentioned above.
We will investigate the potential energy hypersurfaces governing the reactions in Section
6. Here, we will consider only long-range intermolecular interactions which can be easily
implementedincollisiontheory.
I
Figure 5.24: Examples of attractive and repulsive trajectories for the collision between
two molecules.
a) Types of long-range interactions
The long-range interactions between molecules can be described using multipole expansions
of the potentials (charge: 1; dipole: 2; induced dipole: 3; quadrupole: 3 . . . In
general, the interaction potential between two multipoles of ranks and 0 is proportional
to − with = + 0 − 1:
µ
() =−
(5.197)

5.4 Advanced collision theory 133
I
Table 5.3: Types of long-range intermolecular interactions.
system
()
charge-charge (Coulomb) () = 1 2 2
4 0
charge-dipole
( ) = −→ · −→ = − cos
4 0 2
dipole-dipole ( 1 2 )=
charge-quadrupole Θ ∝ −3
dipole-quadrupole
Θ ∝ −4
charge-induced dipole ∝ −4
quadrupole-quadrupole ΘΘ ∝ −5
induced dipole-induced dipole
∝ −6
μ 1 · μ 2 − 3(μ 1 · r)(μ 2 · r)
2
4 0 3
I Table 5.4: Long-range intermolecular interactions ( =15 D, =3Å 3 , =1,
=0;energies in kJ mol; forcomparison: =25kJ mol @ =298K).
³ ´ ³ ´
system =5Å =10Å
ion-ion 278 139
ion-dipole 17.4 4.35
ion-induced dipole 3.34 0.21
dipole-dipole 2.17 0.27
b) Lennard-Jones (6-12) potential for nonpolar molecules
A realistic intermolecular potential for non-polar molecules is the Lennard-Jones (6-12)
potential
∙ ³LJ ´12 ³ LJ
´6¸
LJ () =4 LJ − (5.198)
LJ is usually given in reduced form in units of K as
∗ LJ = LJ
(5.199)

5.4 Advanced collision theory 134
I
Figure 5.25: Schematic plot of the Lennard-Jones potential.
I Estimation of LJ parameters: Values for LJ and LJ are typically derived from
transport properties (e.g., viscosity), thermodynamic data (vdW parameters), or — if
there is no better way — from empirical correlations (see, e.g., Svehla1963) with critical
data, boiling temperatures, . . .
I
Table 5.5: Lennard-Jones for some gases (Mourits1977).
LJ (pm) ( LJ ) (K) LJ (pm) ( LJ ) (K)
He 2560 102 N 2 3738 820
Ne 2822 320 O 2 3480 1026
Ar 3465 1135 CH 4 3790 1421
Kr 3662 1780 CF 4 4486 1673
Xe 4050 2302 CCl 4 5611 4155
H 2 2930 370 CO 2 3943 2009
SF 6 5199 2120

5.4 Advanced collision theory 135
I
Figure 5.26: Lennard-Jones interaction potentials for He, Ne, Ar, Kr, and Xe.
c) Orbital angular momentum and centrifugal barrier*
In the case of a non-central collision, part of the translational energy is converted
to “rotational energy” (orbital angular momentum; see Fig. 5.27). This is seen by
partitioning the velocity vector between the direction ( )andtheradial
direction ( ). The component corresponds to a rotational motion.
I
Figure 5.27: Orbital angular momentum in binary collisions.

5.4 Advanced collision theory 136
• Radial velocity component:
y
• Kinetic energy:
• Angular velocity:
=
cos = =
=
1
2 2 +
| {z }
translation via LC
1
2 2
| {z }
= 1 2 2 2 (rotation)
(5.200)
(5.201)
(5.202)
=
=
=
=
=
2 (5.203)
• Kinetic energy:
= 1 2 ( ) 2 + 1 2 ( ) 2 (5.204)
= 1 2 · 2 + 1 2
2 2 2 (5.205)
= 1 2 · 2 + 1 2 2
2 (5.206)
= 1 2 · 2 + 1 2 2 2
2 2 (5.207)
• Orbital angular momentum: is the orbital angular momentum
= 2
= = (5.208)
y
= 1 2 · 2 +
2
2 2 (5.209)
From quantum mechanics, 2 = ~ 2 ( +1)with as orbital angular momentum
quantum number. y
= 1 2 · 2 + ~2 ( +1)
2 2 (5.210)
= 1 2 · 2 + 2 ( +1) (5.211)
where 2 = ~2
is formally an -dependent rotational constant.
22 • Centrifugal barrier: The centrifugal kinetic energy term ~2 ( +1)
acts as a
2 2
centrifugal barrier
centrifugal = 1 2 2
= ~2 ( +1)
(5.212)
2 2 2

5.4 Advanced collision theory 137
• Withalong-rangeattractivetermoftheform− ¡
¢
, the intermolecular potential
thus has the form
µ 2
() =− + (5.213)
µ 2
= − + ~2 ( +1)
(5.214)
2 2
i.e.,
() ∝− − + −2 (5.215)
— The −2 term is important in practice if 3. In effect, it leads to an
effective energy barrier (“centrifigual” barrier).
— The larger for a given , the larger the orbital angular momentum .
I
Figure 5.28: Potential energy curves with centrifugal barriers for a diatomic system.
d) Langevin “capture” rate constant for ion-molecule reactions*
An nice illustration of the above results is the rate constant for ion-molecule reactions.
The intermolecular interaction is governed by the point charge-induced dipole interaction.
40 In addition, the centrifugal barrier has to be taken into account, so that
() =−
µ 2 ()2
(4 0 ) 2 2 + 4 (5.216)
The resulting expression for the rate constant is
à !
4 () 2 12 µ 1
( )=
(4 0 ) 2 × Γ
(5.217)
2
Typical values of are 100 ×
40 () needs to be checked again!!

5.4 Advanced collision theory 138
6. Potential energy surfaces for chemical reactions
6.1 Diatomic molecules
I
Morse potential (anharmonic oscillator):
() = [1 − exp (− ( − ))] 2 (6.1)
I
Vibrational states of a harmonic oscillator:
µ
= ~ + 1 µ
= + 1
2
2
(6.2)
I
Vibrational states of an anharmonic oscillator:
µ
= ~ + 1 µ
− ~ + 1 2
+ (6.3)
2
2
I Dissociation energies D and D 0 :
= 0 + 1 2 ~ − 1 4 ~ + (6.4)
I
Figure 6.1: Potential energy curve of diatomic molecules.

6.2 Polyatomic molecules 139
6.2 Polyatomic molecules
I
I
The potential energy becomes dependent of the different internal coordinates.
Question: How many internal coordinates (dimensions) do we need?
Answer:
• Total number of degrees of freedom of an -atomic molecule:
• for translational motion of center of gravity:
• for rotational motion around center of gravity
• resulting number of vibrational coordinates:
3 (6.5)
3 (6.6)
2 for linear molecules (6.7)
3 for nonlinear molecules (6.8)
— linear molecules:
— nonlinear molecules:
3 − 5 (6.9)
3 − 6 (6.10)
I
3N − 6 dimensional potential energy hypersurface:
• The internal coordinates are described using the normal coordinates (vibrational
displacements).
• The potential energy function
( 1 2 3−6 ) (6.11)
is a hypersurface in a 3 − 6 (3 − 5) dimensional hyperspace. But we can plot
only in 2-D or 3-D space.

6.2 Polyatomic molecules 140
I
Figure 6.2: Potential energy hypersurface for a simple atom transfer reaction (plot for
fixed angle, e.g., collinear collision and other fixed coordinates depending on dimensionality
of the system).
I
Figure 6.3: Schematic potential energy hypersurface of a triatomic molecule.

6.2 Polyatomic molecules 141
I
Figure 6.4: Potential energy hypersurface of HCO.
I
Figure 6.5: Potential energy hypersurface of HCO.

6.2 Polyatomic molecules 142
I
Models of potential energy surfaces:
(1) Empirical or semiempirical models (e.g., LEPS),
(2) Single points ( ) by ab initio calculations using quantum chemistry. Problem:
Fitting the surface to the computed points.
I
Kinetics and Dynamics:
(1) TST = Transition State Theory (statistical theory),
(2) Classical trajectory calculations (by solving Newton’s equations),
(3) Fully quantum mechanical wave packet calculations (by solving the timedependent
Schrödinger equation).

6.3 Trajectory Calculations 143
6.3 Trajectory Calculations
The “exact” dynamics of the molecules on a potential energy hypersurface can be
followed by classical trajectory calculations.
I Phase Space: 6-dimensional space spanned by the spatial coordinates ( )and
the conjugated momenta ( ):
↔ (6.12)
where
=
=
·
(6.13)
I
I
Hamilton function:
= + = 2
+ () (6.14)
2
Note that depends only on and depends only on .
Equations of motion:
• for a 1 particle system:
(1)
= =
= −
()
y
·
= −
= −
since contains only not .
(2)
y
y
• for an atomic system:
= + =
(6.15)
(6.16)
= 2
2 + (6.17)
= 1 × = 1 µ
×
·
3X
=1
·
=
= ·
(6.18)
(6.19)
2 2 + ( 1 2 3−6 ) (6.20)
·
= −
(6.21)
·
=
(6.22)
Thus we obtain a set of coupled differential equations which can be solved using
the Runge-Kutta or Gear methods.

6.3 Trajectory Calculations 144
• Thermal rate constants ( ) follow by suitable averaging over the initial conditions
( vibrational states, rotational states, angles, ...).
• Major problems: Zero-point energy? Tunneling? Surface crossings?
I
Figure 6.6: Trajectory calculations.
I
Figure 6.7: Trajectory calculations.

6.3 Trajectory Calculations 145
I
Figure 6.8: The reaction H + Cl 2 isareactionwithanearlyTS.Thereactionisknown
to yield HCl with an inverted vibrational state distribution (⇒ HCl-Laser).
I
Figure 6.9: Bimolecular reaction with a late TS.

6.3 Trajectory Calculations 146
7. Transition state theory
7.1 Foundations of transition state theory
The concept of transition state theory (TST) or activated complex theory (ACT) goes
back to H. Eyring and M. Polanyi (1935). TST is a statistical theory; it does not give
information on the exact dynamics of a reaction. It does, however, give (statistical)
information on product state distributions.
There are two forms of TST:
• Macrocanonical TST ⇒ ( ) for a macrocanonical ensemble of molecules with
aconstant (an ensemble of molecules with energy states populated according
to the Boltzmann distribution - each state being equally likely.)
• Microcanonical TST (TST) ⇒ () for a microcanonical ensemble of molecules
with specific energy.
We will consider only the first version.
The detailed application of TST also depends on whether or not the reaction has a
distinctive potential energy barrier (⇒ conventional TST or variational TST).
I
Figure 7.1: Transition State Theory (TST): Fundamental equations and applications.

7.1 Foundations of transition state theory 147
7.1.1 Basic assumptions of TST
(1) The Born-Oppenheimer approximation (separation of electronic and nuclear motion)
is a valid approximation. Thus we can describe the nuclear motion during
the reaction on a single electronic potential energy surface (PES; see Fig. 7.2). 41
(2) The reactant molecules are in internal equilibrium (the quantum states are populated
according to the Boltzmann distribution).
(3) “Transition state” and “dividing surface“:
a) Conventional TST (valid for so-called “type I” PES’s = PES’s with a distinctive
energy barrier along the reaction coordinate (RC)): We can localize
a distinctive “transition state” (TS) for the reaction at the saddle point on
the PES that separates the reactant and product “valleys”. The TS is located
at the maximum of the PES along the minimum energy path (MEP;
see Fig. 7.3).
b) Variational TST (version that can be used for so-called “type II” PES’s =
PES’s without a clear energy barrier along the RC (we can’t localize a simple
TS)): We can localize a 3 − 7-dimensional “dividing surface” (DS) on the
3 − 6-dimensional PES that separates the reactant and product regions
on the PES in such a way that the “reactive flux” through the DS becomes
minimal (Fig. 7.2). The TS is then placed at the point where the MEP
intersects the DS. In this case, the position of the TS depends strongly on
the finer details such as angular momentum (because of centrifugal barriers
which move inwards with increasing or increasing ).
(4) The motion along the minimum energy path across the TS (through the DS) can
be separated from other degrees of freedom and treated as translational motion.
AttheTS,wethereforehave1translationaldegreeoffreedom ‡ (the RC) and
3 − 7 (vibrational) degrees of freedom orthogonal to the RC.
(5) Reactant molecules passing through the DS never come back (no recrossing).
(6) The states of the TS are in quasi-equilibrium with the states of the reactant
molecule even if we have no equilibrium between reactants and products (the
individual molecules don’t know that this equilibrium is not present - this quasiequilibrium
assumption can be dropped in dynamical derivations of TST; see
below).
I
Conditions for TS at the saddle point:
(1) has maximum along ‡ :
=0and 2
0 (7.1)
‡
‡2
y 1 imaginary frequency (criterion for TS in quantum chemical calculations)!
41 TST cannot be used at least in the simple form considered in this lecture when two potential surfaces
are involved that are energetically close to each other (or even become degenerate) and the reaction
involves a “non-adiabatic transition” between these two potentials.

7.1 Foundations of transition state theory 148
(2) has minimum along all 3 − 7 other coordinates:
=0and 2
2
0 (7.2)
y 3 − 7 real other frequencies.
I
Figure 7.2: Transition state theory: The dividing surface with minimal reactive flux.
I
Figure 7.3: Minimum energy path (MEP) and transition state (TS) of a simple atom
transfer reaction.

7.1 Foundations of transition state theory 149
7.1.2 Formal kinetic model
In order to derive the TST expression for ( ), wefirst look at a simple formal kinetic
model. This formal kinetic model has its roots in the original idea that the TS can be
considered as some sort of an “activated complex” with a local energy minimum on
the PES. It is not a good derivation of the TST expression for ( ), but leads to the
correct expression.
I
Activated complex model of TST:
y
A+BC 1 À
2
ABC ‡ 3 → P (7.3)
£
ABC
‡ ¤ = 1 [A] [BC]
2 + 3
(7.4)
I Assumption: 3 ¿ 2 y
£ ¤ ABC
‡ = 1
[A] [BC] =
12 [A] [BC] (7.5)
2
y Equilibrium between the reactants (A+BC)andtheTS(ABC ‡ ):
y
[P]
= 3
£
ABC
‡ ¤
(7.6)
= 3 | {z 12 [A] [BC] (7.7)
}
=
I
ACT expression for rate constant:
= 3 12 (7.8)
Below, we will find that
and
3 = ‡ =
12 ∝ exp
µ− ∆‡
(7.9)
(7.10)
7.1.3 The fundamental equation for k (T )
The above derivation of the TST rate constant was based on an oversimplified picture,
since the TS is not at a local minimum but at a saddle point on the PES. In this section,
we derive the TST fundamental equation more carefully by bringing in some dynamical
aspects. In (1) - (4), we explore the implications of the quasi-equilibrium assumption,
and in (5) - (9) we develop a dynamical semiclassical model: 42
42 Semiclassical because the translation is handled classically while all other degrees of freedom are
described quantum mechanically.

7.1 Foundations of transition state theory 150
(1) We start by assuming equilibrium between the reactants and products:
A+BC 1 À
2
ABC ‡ 3
À
4
P (7.11)
(2) We consider two dividing surfaces at the TS separated by a distance of ≤ .
is the “length” of a “phase space cell” in semiclassical theory which is determined
by Heisenberg’s uncertainty principle:
∆ ‡ × ∆ ‡ = × ∆ ‡ = (7.12)
(3) We now consider the number densities of molecules passing through these surfaces
from left to right ( −→ ‡ )andfromrighttoleft( ←−
‡ ). The total number density ‡
at the TS in equilibrium is their sum, i.e.,
‡ = −→ ‡ + ←−
‡ = ‡ A BC (7.13)
In equilibrium, we must have
Thus,
−→
‡ = ←−
‡ (7.14)
−→
‡ = ←−
‡ = 1 2 ‡ = ‡
2 A BC (7.15)
(4) We now suddenly remove all the products. Then ←−
‡ =0 However, there is no
reason for −→ ‡ to change, because the molecules on the reactant side remain in
internal equilibrium (this is the quasi-equilibrium assumption). 43 y
−→
‡ = ‡
2 = ‡
2 A BC (7.16)
Thus, we can still express −→ ‡ using the equilibrium constant ‡ even if there is
no equilibrium between reactants and products.
(5) We now turn to the reaction rate, which is given by the number density of molecules
−→ ‡ in the time interval that are crossing the DS in the forward direction.
We can write this as
−→ ‡
= −→ −→
‡ ‡ ‡
=
‡ (7.17)
where
• −→ ‡ is the decay rate of the TS in the forward direction (or the frequency
factor for the molecules crossing the DS in the forward direction),
43 The quasi-equilibrium assumption for −→ ‡ is made in canonical TST. In a microcanonical derivation
· ·
−→ ←−
of the TST expression for (), we consider the fluxes ‡ and ‡ through the dividing surface
at the TS. Then, the individual molecules “dont’t know” that the products have been removed and
·
·
←−
−→
‡ =0. Since they don’t know about this, the flux ‡ has to remain unchanged.

7.1 Foundations of transition state theory 151
• −→ ‡ = −→ ‡ is the mean speed of the molecules crossing the DS in the forward
direction.
(6) We now use the equilibrium constant to determine ‡ via
‡ = ‡ ABC
A BC
(7.18)
y ‡ = ‡ ABC = ‡ A BC (7.19)
y
−→ −→
‡
‡
=
‡ A BC = A BC (7.20)
with
−→
‡
=
‡ (7.21)
(7) From statistical thermodynamics (see Appendix G), we have the following expression
for the equilibrium constant ‡ ( ) in terms of the molecular partition
functions:
‡ ( )=
∗
=
µ
exp − ∆
0
(7.22)
• The ’s are the respective molecular partition functions (per unit volume).
• ∗ is written with respect to the zero-point level of the reactants.
• The exp (−∆ 0 ) factor arises subsequently, because we now express
the value of partition function for the TS relative to the zero-point level of
the TS ( ∗ = exp (−∆ 0 )).
(8) Separating the motion along the reaction coordinate ‡ from the other degrees of
freedom, we write
= ‡ ‡
(7.23)
where ‡ is the 1-D (translational) partition function describing the translational
motion of the molecules along ‡ across the TS and ‡
is the partition function
for all 3 − 7 other (rotational-vibrational-electronic) degrees of freedom.
Here,
• the 1-D translational partition function is
‡ =
µ 2
2 12
× (7.24)
• the mean speed along ‡ for crossing the DS in the forward direction is given
by
R
−→ ∞
‡ −2 2 µ
12
0
= R ∞
−∞ −2 2 =
(7.25)
2

7.1 Foundations of transition state theory 152
(9) Collecting and inserting the results into
−→
‡
( )=
( ) (7.26)
we obtain
( )= 1 µ 12 µ 12 µ
2
×
× ‡
exp − ∆
0
(7.27)
2
2
y
( )= ‡ µ
exp − ∆
0
(7.28)
I Result: In order to account () forsomerecrossingoftheTSand() for quantum
mechanical tunneling, we add an additional factor called the transmission coefficient.
For well-behaved reactions, ≈ 1. Thus, we end up with the fundamental equation
of TST in the form:
( )=
‡ µ
exp − ∆
0
Note that ‡
is the partition function of the TS excluding the RC.
Equation 7.29 allows us to calculate ( ) using the following data:
(7.29)
• The threshold energy for the reaction ∆ 0 (or, per mol, ∆ 0 ). ∆ 0 has to be
obtained from ab initio quantum chemistry calculations, and
• structural information on the reactants and the TS (i.e., bond lengths and angles,
vibrational frequencies).
I Figure 7.4: Illustration of ∆ ‡ , ∆ 0 ,and .

7.1 Foundations of transition state theory 153
∆ ‡ : Born-Oppenheimer potential energy barrier (7.30)
∆ 0 : zero-point corrected threshold energy for the reaction (7.31)
: Arrhenius activation energy (7.32)
∆ 0 is given with respect to the zero-point energies of the reactants and the TS,
= 1 2
X
(7.33)
7.1.4 Tolman’s interpretation of the Arrhenius activation energy
I We now understand the difference between E and ∆E 0 :
2 ln ( )
=
Ã
= 2
ln
‡ µ
exp
= ∆ 0 + + 2 ln ‡
−
| {z }
=h reacting moleculesi
− ∆ 0
(7.34)
! (7.35)
µ
2 ln
+ 2 ln
(7.36)
| {z }
=h all molecules i
• ∆ 0 is difference of the zero-point levels of the reactants and the TS,
• the term is the mean translational energy in the RC (from the factor),
• the 2 ln terms are the internal (vibration-rotation) energies of the TS and
the reactants, respectively. y
I Tolman’s interpretation of E (see Fig. 7.4):
= h reacting molecules i − h all molecules i (7.37)

7.2 Applications of transition state theory 154
7.2 Applications of transition state theory
( )=
‡ µ
exp − ∆
0
(7.38)
TST has gained enormous importance in all areas of physical chemistry because it
provides the basis for understanding – and/or calculating –
• quantitative values of preexponential factors,
• ∆ 0 from measured values,
• -dependence of and deviations from Arrhenius behavior,
• gas phase and liquid phase reactions (proton transfer, electron transfer, organic
reactions, reactions of ions in solutions),
• pressure dependence of reaction rate constants (activation volumes),
• kinetic isotope effects,
• structure-reactivity relations (e.g., Hammett relations in organic chemistry),
• features of biological reactions (enzyme reactions),
• heterogeneous reactions (heterogeneous catalysis, electrode kinetics), . . .
7.2.1 The molecular partition functions
In order to evaluate Eq. 7.29 for a given reaction, we need, besides ∆ 0 , the molecular
partition functions for the reactants and the TS. We summarize only the main points
here; further information is given in Appendix G.
I
Definition:
= P −
(7.39)
I
Physical interpretation:
• The molecular partition function is a number which describes how many states
are available to the molecule at a given temperature .
• The ratio
‡
(7.40)
in Eq. 7.29 is thus the ratio of the number of states available to the TS and
‡
the reactant molecules. Since each state is equally likely, is the relative
probability of the TS vs. the probability of the reactants.
Note again that ‡
motion along the RC.
is the partition function of the TS excluding the translational

7.2 Applications of transition state theory 155
I Separation of degrees of freedom: For separable degrees of freedom, we can write
the molecular partition function as
= × × × (7.41)
Nuclear spin ( ) may have to be taken into account in some cases as well.
I
Table 7.1: Molecular partition functions.
type of motion energy partition function magnitude
µ
translation (1-D) 2 2
12 2
10 8 ×
8 2
µ µ 2
translation (3-D) 2
2 32
+ 2
+ 2 2
10 24 ×
8 2 2 2
µ 2
rotation (1-D) a ~ 2
12
2 2
2
10
µ ~ 2
rotation (2-D) b ~ 2
2
( +1)
10 2
2 ~ 2
µ
rotation (3-D) c 12
32 2
or
(
~ 2 ) 12 10 3 10 4
vibration d [1 − exp (− )] −1 1 10
electronic
P exp (− ) 1
a Free internal rotor. For a hindered rotor (torsional vibrational mode), the partition function has to
be determined by an explicit calculation over the hindered rotor quantum states.
b Linear molecule.
c Nonlinear molecule.
d For each harmonic oscillator degree of freedom measured from zero-point level.

7.2 Applications of transition state theory 156
7.2.2 Example 1: Reactions of two atoms A + B → A···B ‡ → AB
This example is somewhat artifical unless there is some mechanism to remove the bond
energy (for example, in associative ionization, A + B → A···B ‡ → AB + + − ).
‡ µ
exp − ∆
0
Ã
! Ã !
= ‡
‡
( )( ) 1
= µ
2 ( + ) 2
2 2
µ
× exp − ∆
0
= µ
2
+
2
µ
× exp − ∆
0
( )=
µ
exp − ∆
0
2
2
32 µ 8 2 ‡
2
32
Ã
!
8 2 ( + ) 2
2 +
y
µ 12 µ
8
( )=
( + ) 2 exp − ∆
0
This result is identical to that obtained by the LC model in collision theory!
(7.42)
(7.43)
(7.44)
(7.45)
(7.46)

7.2 Applications of transition state theory 157
7.2.3 Example 2: The reaction F + H 2 → F···H···H ‡ → FH + H
I
Exercise 7.1: Evaluatetherateconstantforthereaction
F+H 2 → F ···H ···H ‡ → FH + H (7.47)
using the data in Table 7.2 at =200, 300, 500, 1000 and 2000 K, determinethe
Arrhenius parameters, and compare the results with the experimental value of
=2× 10 14 exp (−800 )cm 3 mol s (7.48)
¤
I Table 7.2: Properties of the reactants and transition state of the reaction F + H 2 .
parameter F ···H ···H ‡ a F H 2
2 (F ···H) ( Å) 1602
1 (H ···H) ( Å) 0756
07417 b
1 (cm −1 ) 40076 43952
2 (cm −1 ) 3979
3 (cm −1 ) 3979
4 (cm −1 ) 3108 c
³ (u) 21014 189984 2016
u Å 7433
0277
1 2
d
4 4 1
0 ¡ kJ mol −1¢ 657 000
a The transition state FHH ‡ is assumed to be linear.
b = (H-H).
c One imaginary frequency describes the TS along the RC.
d The electronic ground state of F is 2 32 . We neglect the 2 12 spin-orbit component at ( 2 12 )=
404 cm −1 because it does not correlate with the products H + HF. The electronic state of linear
FHH ‡ is 2 Π. The ground state of H 2 is 2 Σ + .
I Solution 7.1:
( )= ‡ µ
exp − ∆
0
2
Ã
!
= ‡
( )( 2 )
µ
× exp − ∆
0
à !
‡
2
à !
‡
2
(7.49)
à !
‡
2
(7.50)

7.2 Applications of transition state theory 158
The different factors can be evaluated separately:
Ã
!
‡
µ 32 µ
+ 2
2 32
=
(7.51)
( )( 2 )
2 2
à !
‡ µ
8 2 ‡ 2
‡
=
=
2 8 2 2 2 2 2 (7.52)
2
à !
‡
1 − exp (− 2 )
=
(7.53)
2
3Y
´i
à !
‡
2
=1
= 4
4 × 1
h
1 − exp
³
− ‡
(7.54)
Note: The difference between the TST result and experiment can be ascribed (a)
to tunneling and (b) deficiencies of the ab initio PES. Indeed, according to newer ab
initio calculations, the TS is actually slightly bent and the TS parameters differ slightly,
yielding better agreement with experiment. ¥

7.2 Applications of transition state theory 159
7.2.4 Example 3: Deviations from Arrhenius behavior (T dependence of k (T ))*
I
Exercise 7.2: The temperature dependence of bimolecular gas phase reactions of the
type A + B → productscanusuallybeexpressedintheform
Using the TST expression for ( )
( )= × × exp (−∆ 0 ) (7.55)
( )=
‡ µ
exp
− ∆ 0
, (7.56)
determine the values of for the types of reactions in Table 7.3, assuming that the
vibrational degrees of freedom are not excited. ¤
I Table 7.3:
A + B → TS ‡
transl. rot.
atom atom linear TS 1 32 1
+05
32 32 1
atom linear molecule linear TS 1 −32 1
−05
1
atom linear molecule nonlin. TS 1 −32 32
00
1
atom nonlin. molecule nonlin. TS 1 −32 32
−05
32
linear molecule linear molecule linear TS 1 −32 1
−15
1 1
linear molecule linear molecule nonlin. TS 1 −32 32
−10
1 1
linear molecule nonlin. molecule nonlin. TS 1 −32 32
−15
1 32
nonlin. molecule nonlin. molecule nonlin. TS 1 −32 32
−20
32
32
I Solution 7.2: See Table 7.3. ¥
Using the above results, we can recast ( ) and determine the following expression for
the Arrhenius activation energy, with as above:
= ∆ 0 + + 2 ln ‡
− 2
X
Reactants
ln reactants
(7.57)
We see that the value for depends on the vibrational frequencies of the TS and the
reactants. Often, the frequencies of the reactants are very high so that reactant
=1.
However, the TS may have soft vibrations so that ‡
may approach the classical value
=[1− exp (− )] −1 →
for →∞ (7.58)
Thus, each “soft” vibrational mode in the TS increases by 1, each “soft” vibrational
mode in the reactants decreases by 1.

7.3 Thermodynamic interpretation of transition state theory 160
7.3 Thermodynamic interpretation of transition state theory
7.3.1 Fundamental equation of thermodynamic transition state theory
As shown above,
( )=
‡ (7.59)
‡ can often be interpreted using thermodynamic arguments, i.e., ∆ ‡ , ∆ ‡ ,and
∆ ‡ . For liquid phase reactions, this is straightforward. For gas phase reactions, we
have to take into account the difference between ‡ and .
‡
a) Thermodynamic transition state theory for unimolecular gas phase reactions
For unimolecular isomerization and dissociation reactions of the type
or
we have
ABC → ABC ‡ → ACB ‡ (7.60)
ABC → ABC ‡ → A+BC ‡ , (7.61)
£ ¤
ABC
‡
‡ ( )=
[ABC]
= ABC ‡
ABC
= ‡ ( ) (7.62)
‡ ( )= ‡ ( )= ‡ ( ) is dimensionless (∆ ‡ =0) and we do not have to worry
about a difference between ‡ ( )= ‡ ( ) and about units or different standard
states.
I Results: Eq. 7.59 thus gives the following results:
• Rate constant:
y
( )=
‡ ( )= µ
exp − ∆‡
( )= µ
∆
‡
exp exp
µ− ∆‡
(7.63)
(7.64)
• Arrhenius activation energy:
y
= 2 ln
= 2 ∙ µ
ln ∆
‡
exp
exp
¸
µ− ∆‡
(7.65)
= ∆ ‡ + + 2 ∆‡
= ∆ ‡ + + ∆‡
(7.66)
(7.67)
If ∆ ‡ 6= ( ) in the temperature range of interest, we obtain
= ∆ ‡ + (7.68)

7.3 Thermodynamic interpretation of transition state theory 161
• Arrhenius preexponential factor:
∆ ‡ = − (7.69)
y
y
( )= µ µ
∆
‡
exp exp −
= µ
∆
‡
exp
(7.70)
(7.71)
b) Thermodynamic transition state theory for bimolecular gas phase reactions
I Conversion from K
‡ (T ) to K‡
(T ): For bimolecular reactions
A+BC→ ABC ‡ , (7.72)
we have to account for the different numbers of species ∆ ‡ = ‡ − reactants :
We therefore have to convert from ‡ ( ) to ‡ ( ) via
∆ ‡ = −1 . (7.73)
£ ¤
ABC
‡ ABC
‡ ABC ‡ ª
‡ ( )=
[A] [BC] =
A BC
=
A ª BC ª × 1
µ −∆
‡
= ª ‡ ( ) ×
.
ª
(7.74)
Here, ‡ ( ) is conveniently given in units of lmol −1 . 44 However, , ‡ ∆ ‡ , ∆ ‡ and
∆ ‡ arerelatedtothestandardstateat ª =1bar. 45 To keep track of units in the
conversion , it is useful to apply in
lbar by writing
mol K
=83145
J
mol K =83145 m3 Pa
lbar
=0083145
mol K mol K = 0 (7.76)
44 With ‡ ( ) in units of ¡ mol l −1¢ ∆
, we get as close as possible to the standard state ∗ =
1molkg −1 for reactions in aqueous solution.
45 We recall that the thermodynamic equilibrium constant ( )=exp(−∆ ª ) referred to
the standarrd state at ª =1baris dimensionless. Instead of the dimensionless ‡ ( ), onemay
also decide to use the equilibrium constant in pressure units ‡ 0 :
hABC ‡i ABC ‡
‡ ( )=
[A] [BC] =
A BC
= ‡ 0 ( ) × 0 (7.75)

7.3 Thermodynamic interpretation of transition state theory 162
so that, with ∆ ‡ = −1, weobtain
Thus,viaEq.7.59:
‡ ( )=
( )=
ABC ‡ ª
A ª
BC ª
× 1
µ 0
=
ª ‡ ( ) × . (7.77)
ª
0
ª ‡ ( )= µ
0
exp ª
− ∆ª‡
, (7.78)
in units of
µ µ
0
dim ( ( )) = dim × dim = l
(7.79)
ª mol s
and with the standard state for the thermodynamics quantities explicitly indicated by
the ª symbol.
I
Results:
• Rate constant:
( )=
µ
0 ∆
ª‡
exp exp
ª
µ− ∆ª‡
(7.80)
• Arrhenius activation energy:
= 2 ln
(7.81)
If ∆ ª‡ 6= ( ) in the temperature range of interest:
= ∆ ª‡ +2 (7.82)
• Arrhenius preexponential factor:
∆ ª‡ = − 2 (7.83)
y
( )=
= 2
µ
0 ∆
ª‡
exp ª
µ
0 ∆
ª‡
exp ª
µ
exp
exp
−
− 2
µ
−
(7.84)
(7.85)
Comparison with the Arrhenius expression = exp (− ) thus gives
= 2
µ
0 ∆
ª‡
exp ª
(7.86)

7.3 Thermodynamic interpretation of transition state theory 163
• Note: and are given above in units of lmol −1 s −1 (we may denote this by
writing them as and ), while ∆ ª‡ and ∆ ª‡ used above are referred to
the standard state for gases of ª =1bar.
— For ideal gases, ∆ ‡ and ∆ ‡ are related via
∆ ª‡ = ∆ ª‡ + ∆ ‡ ( )=∆ ª‡ − = ∆ +‡ − (7.87)
— Likewise, ∆ ª‡ and ∆ +‡ for are related via
∆ +‡ − ∆ ª‡ =
y
Z
+
ª
Z
+
=
ª
= Z
+
ª
= Z
+
ª
∆ ‡
(7.88)
= ∆ ‡ +
ln
+ 0 = +
ª ∆‡ ln
+ 0 = ª
ª ∆‡ ln (7.89)
+ 0
= ∆ ‡ ln + 0
(7.90)
ª
with ª =1barand + = +
+ =1moll−1 .
∆ ª‡ = ∆ +‡ − ln + 0
ª (7.91)
c) Thermodynamic transition state theory for reactions in the liquid phase
For reactions in condensed phases,
∆ ≈ ∆ (7.92)
and
y
£
ABC
‡ ¤
‡ ( ) =
[A] [BC] = ¡ +¢ ∆ ‡ exp
≈ ¡ +¢
∆ ‡ exp
µ− ∆+‡
exp
• for unimolecular reactions:
( )= µ
exp
• for bimolecular reactions:
( )= µ
+ exp
− ∆+‡
µ− ∆+‡
− ∆+‡
µ− ∆+‡
exp
exp
+‡
∆
exp
µ−
µ− ∆+‡
µ− ∆+‡
(7.93)
(7.94)
(7.95)
(7.96)
Note: The conversion between values for ∆ +‡ and ∆ ª‡ and, likewise, between
values for ∆ +‡ and ∆ ª‡ for a given reaction in the liquid and gas phase requires
some caution. 46
46 See,e.g.,D.M.Golden,J.Chem.Ed.48, 235 (1971).

7.3 Thermodynamic interpretation of transition state theory 164
7.3.2 Applications of thermodynamic transition state theory
a) Comparison of ∆S ‡ -values for different types of transition states
I
Bimolecular reactions:
A+BC→ ABC ‡ (7.97)
y
∆ ‡ 0 (7.98)
or even
∆ ‡ ¿ 0 (7.99)
Furthermore, for chemically otherwise similar molecules
(nonlinear molecule) (linear molecule) À (cyclic molecule) (7.100)
Thus,
¯
¯∆ ‡ (nonlinear molecule)¯¯ ¯¯∆ ‡ (linear molecule)¯¯ ¿ ¯¯∆ ‡ (cyclic molecule)¯¯
(7.101)
For example:
⎛
A ··· B
∆ ‡ ⎜
⎝
¯
y
...
⎞ ‡ ⎛
‡ ⎞ ⎛
⎟ A ··· B
¯
C
⎠¯¯¯¯¯¯¯ ¯∆‡ ⎝ A ··· B ··· C ⎠¯
⎞ ‡ ¯ ¿
∆ ‡ ⎜ ⎟
⎝
. .
¯ C ··· D
⎠¯¯¯¯¯¯¯
(7.102)
cyclic ¿ linear nonlinear (7.103)
I
Unimolecular reactions:
ABC → ABC ‡ (7.104)
• For isomerization reactions with a tight transition state:
∆ ‡ ⎛
⎝
• For bond fission reactions with a loose transition state:
‡
B
⎞
Á Â ⎠ . 0 (7.105)
A – C
∆ ‡ ¡ AB ···C ‡¢ 0 (7.106)
y
loose tight (7.107)

7.3 Thermodynamic interpretation of transition state theory 165
b) Thermochemical estimation of A-factors*
Equations 7.71 and 7.86 allow us to estimate the -factors of chemical reaction rate
constants by the following procedure (Benson1976):
(1) We setup a model for the TS, with a certain structure and vibrational freuqencies.
(2) We take a reference molecule that is similar to the TS and use the ref of the
reference molecule for a zero-order estimation of ∆ ‡ according to
∆ ‡ = ref −
X
( reactants ) (7.108)
reactants
(3) We apply corrections for the differences between ref and ‡ basedonthestatistical
thermodynamics result of
= ln + ln
(7.109)
y
µ
∆ cor = ‡ − ref = ln ‡
‡
+
ref ln (7.110)
ref
Usually, the correction term accounts for differences of the entropies for translation,
rotation, internal rotations, vibrations, symmetry, electron spin, plus sometimes
other terms (e.g., optical isomers).
I Example 7.1: Estimation of the -factor for the reaction O+CH 4 →
(O ···H ···CH 3 ) ‡ → OH + CH 3 .
∆ ‡ = (O ···H ···CH 3 ) ‡ − (O) − (CH 4 ) (7.111)
ref = (CH 3 F) (7.112)
∆ cor thus depends on a comparison of CH 3 F and OHCH ‡ 3 (see Table ). ¤
I Table 7.4: Estimation of the -factor for the reaction O+CH 4 →
(O ···H ···CH 3 ) ‡ → OH + CH 3 .
Contribution ∆ ‡ (Jmol −1 K −1 )
ª (CH 3F) − ª (O) − ª (CH 4)
−1243
Spin: + ln 3 + 91
external rotation: 1 3 ln ( 1 2 3 ) ‡
( 1 2 3 ) ref + 79
translation − 04
CF =1000cm −1 → − 04
OH =2000cm −1 + 00
2 OHC ()
³
=600cm −1 ´
+ 42
∆ ‡ = ª
O ···H ···CH ‡ 3 − ª (O) − ª (CH 4) −1039
Correction ∆ ‡ → ∆ ‡ : + ln ( 0 ) y ∆ ‡ = −198 Jmol −1 K −1 .

7.3 Thermodynamic interpretation of transition state theory 166
Estimated -value:
= 2 µ
∆
‡
exp (7.113)
µ −197Jmol −1
=739 × 625 × 10 12 K −1
× exp
831 J mol −1 K −1 cm 3 mol −1 s −1 (7.114)
=43 × 10 12 cm 3 mol −1 s −1 (7.115)
Experimental -value:
expt =19 × 10 12 cm 3 mol −1 s −1 (7.116)
The difference is probably due to a poor estimate of the TS structure — one can do
much better, I just didn’t try hard enough here. 47
7.3.3 Kinetic isotope effects
I
Figure 7.5: The kinetic isotope effect due to the difference in the zero-point energies
of the reactants and the respective TS structures.
47 For H abstraction reactions from a series of hydrocarbons by 3 CH 2 , the agreement that was achieved
was better than ±30 %.

7.3 Thermodynamic interpretation of transition state theory 167
Kinetic isotope effects occur in particular in the case of H/D atom transfer reactions.
The reason is that the threshold energy ∆ ‡ 0 depends on the vibrational zero-point
energies.
We consider the reactions
where
A+H—R → A—H—R ‡ → products (7.117)
A+D—R → A—D—R ‡ → products (7.118)
∆ ‡ 0 (H—R) = ∆ ‡ + ‡ (A—H—R) − (H—R) (7.119)
∆ 0(D—R) ‡ = ∆ ‡ + ‡ (A—D—R) − (D—R) (7.120)
with
y
=
1
2
X
(7.121)
∆ ‡ 0(DH) = ∆‡ 0(D—R) − ∆ 0(H—R) ‡ (7.122)
=+ ‡ (A—D—R) − ‡ (A—H—R) − ( (D—R) − (H—R))
(7.123)
y
à !
(D—R)
(H—R) =exp − ∆‡ 0(D/H)
(7.124)
I
Example 7.2: R-H/R-D substitution. The vibrational frequencies depend on the force
constants and reduced masses :
= 1 µ 12
(7.125)
2
For H/D substitution:
(HR)
(DR) ≈ 1 (7.126)
2
y
µ 12
(HR) (HR)
(DR) = ≈ √ 2 (7.127)
(DR)
With (HR) = 3000 cm −1 and (DR) = 2100 cm −1 , we find ∆ =
1
2 (3000 − 2100) cm−1 =450cm −1 , by which the DR reactant is stabilized compared
to the HR reactant. For =300K:
µ
(DR)
(HR) =exp − 450
≈ −225 ≈ 01 (7.128)
200
For more realistic estimates, one has to take into account the change of all vibrational
frequencies in the reacting molecules and the TS.
¤

7.3 Thermodynamic interpretation of transition state theory 168
7.3.4 Gibbs free enthalpy correlations
I
Figure 7.6: Linear free enthalpy correlations.
Application of the TST expression
= µ
exp − ∆‡
(7.129)
to a series of related reactions (1) and (2) often gives a relation of the type 48
∆ ‡ 1 − ∆ ‡ 2 = ¡ ¢
∆ ª 1 − ∆ ª 2
(7.130)
y
ln 1
= ln 1
(7.131)
2 2
7.3.5 Pressure dependence of k
Liquid phase reactions often show a pressure dependence of the rate constant. Using
the relation
µ
= (7.132)
we can predict this pressure dependence. Starting with
= µ
exp − ∆‡
(7.133)
we find
µ ln
= − ∆ ‡
(7.134)
48 A well-known example is the Hammett correlation in organic chemistry.

7.3 Thermodynamic interpretation of transition state theory 169
I
Activation volume:
∆ ‡ (7.135)
I Example 7.3: Comparison of the pressure dependencies of the rate constants for 1
and 2 reactions:
(CH 3
) 3
CCl + H 2 O → (CH 3
) 3
C + +Cl − +H 2 O → ∆ ‡ 0
y ↑ y ↓ 1
HO − +CH 3 I
⎡
⎤
|
→ ⎣ HO ···C ···I ⎦
ÁÂ
−
→ ∆ ‡ 0
y ↑ y ↑ 2
¤

7.4 Transition state spectroscopy 170
7.4 Transition state spectroscopy
Until a few years ago, the “transition state” was a rather “elusive and strange” species.
This situation has completely changed (a) because of the rapidly improving quantum
chemical methods for computing TS properties and (b) because of transition state
spectroscopy, in the frequency domain (J. Polanyi, Kinsey), and in the time domain
(Zewail).
I
Figure 7.7: Transition state spectroscopy: The reaction K + NaCl.
I
Figure 7.8: Transition state spectroscopy: Raman emission of dissociating molecules.

7.4 Transition state spectroscopy 171
I Figure 7.9: Transition state spectroscopy: The reaction F + H 2 .
I
Figure 7.10: Transition state spectroscopy: NaI Photodissociation.

7.4 Transition state spectroscopy 172
I
Figure 7.11: Transition state spectroscopy: NaI Photodissociation.
I
Figure 7.12: Transition state spectroscopy: Reaction a in van-der-Waals Complex.

7.4 Transition state spectroscopy 173
I
Figure 7.13: Transition state spectroscopy: Stimulated Emission Pumping.

7.4 Transition state spectroscopy 174
7.5 References
Benson 1976 S. W. Benson, Thermochemical Kinetics, Methods for the Estimation
of Thermochemical Data and Rate Parameters, 2nd Ed., Wiley, New York, 1976.
Evans 1935 M. G. Evans, M. Polanyi, Trans. Faraday Soc. 31, 875 (1935).
Eyring 1935 H. Eyring, J. Chem. Phys. 3, 107 (1935).
Herschbach 1987 D. R. Herschbach, Molekulardynamik einfacher chemischer Reaktionen,Angew.Chem.99,
1251 (1987).
Lee 1987 Y. T. Lee, Molekularstrahluntersuchungen chemischer Elementarprozesse,
Angew. Chem. 99, 967 (1987).
Polanyi 1987 J. C. Polanyi, Konzepte der Reaktionsdynamik, Angew. Chem. 99, 981
(1987).
Wigner 1938 E. Wigner, Trans. Faraday Soc. 34, 29 (1938).

7.5 References 175
8. Unimolecular reactions
Unimolecular reactions are reactions in which only one species experiences chemical
change.
I
Figure 8.1: Examples of unimolecular reactions.
8.1 Experimental observations
(1) There are many reactions governed by a first-order rate law,
[A]
= − ∞ [A] , (8.1)
for example isomerization and dissociation reactions like
Cyclopropane → Propene (8.2)
-Butene → -Butene (8.3)
C 2 H 6 → 2CH 3 (8.4)
Cycloheptatriene → Toluene (8.5)
(2) In these reactions, no other species than the reactants experience any chemical
change.
(3) These reactions have the same rates in the gas phase and in solution,
gas phase ≈ solution phase , (8.6)
i.e., the environment has no effect on the reaction rates!

8.1 Experimental observations 176
(4) There are many similar other reactions, however, for which the rate law is secondorder,
i.e.,
[A]
= − 0 [A] [M] (8.7)
Examples are
Br 2 → Br + Br (8.8)
H 2 O 2 → OH + OH (8.9)
(5) More detailed measurements show that at a given temperature the order of the
reaction may depend on pressure (see Fig. 8.2), for instance for the reaction
one has found that
CH 3 NC → CH 3 CN (8.10)
a) for → 0:
[CH 3 NC]
= − 0 [CH 3 NC] [M] (8.11)
This is called the “low pressure range” for the reaction.
b) for →∞:
[CH 3 NC]
= − ∞ [CH 3 NC] (8.12)
This is called the “high pressure range” for the reaction”.
(6) The transition from the low to the high pressure range depends
a) onthesizeofthemolecule(seeFig.8.3),
b) for a given molecule, it depends on temperature.
(7) The activation energy in the low pressure regime is much smaller than in the high
pressure regime:
= 2 ln
(8.13)
where
0 ∞ (8.14)
and
∞ ≈ 0 (8.15)
(8) The preexponential factors in the low pressure regime are much higher than the
collision frequencies, i.e.,
0 (8.16)
(9) For all these reactions, the molecules must somehow be “energized”. There are
other reactions, in which the molecules are energized chemically (“chemically
activated reactions”) or by photons. At the microscopic, i.e., molecular level,
these reactions should have related mechanisms.

8.1 Experimental observations 177
I
Questions:
(1) What are the differences of the reactions under (1), (4) and (5)?
(2) How can we rationalize their reaction mechanisms?
I
Figure 8.2: Schematic fall-off curve of the unimolecular rate constant.
I
Figure 8.3: Experimentally observed fall-off curves of unimolecular reactions.

8.2 Lindemann mechanism 178
8.2 Lindemann mechanism
First attempts to explain the pressure dependence of unimolecular reaction rate constants
due to radiation and the dissociation dynamics due to centrifugal forces (rotational
excitation) proved to be wrong. In addition, the historic examples for unimolecular reactions,
for instance the N 2 O 5 decomposition, were later proven to have more complex
reaction mechanisms. The first “realistic” model that explained the pressure dependence
of was that proposed by Lindemann (1922).
I
Figure 8.4: Lindemann mechanism.
I
Figure 8.5: Lindemann mechanism: High pressure regime.

8.2 Lindemann mechanism 179
I
Figure 8.6: Lindemann mechanism: Low pressure regime.
I
Figure 8.7: Fall-off curve: The proper way to plot log is vs. log [M]. ab
a
Since 0 ∝ [M], wecanalsoplotlog vs. log 0 . This gives the so-called doubly reduced fall-off
curves.
b In reality, the fall-off regime extends over several orders of magnitude in [M].

8.2 Lindemann mechanism 180
I Half-pressure: We have found the following results:
=
2 1 [M]
−1 [M] + 2
(8.17)
Thus, if −1 [M] = 2 ,wehave
1
∞ = 2
−1
(8.18)
0 = 1 [M] (8.19)
=
2 1 [M]
= 2 1 [M]
−1 [M] + 2 2 −1 [M] = 1 2 1
= 1 2 −1 2 ∞ (8.20)
Half “pressure”:
[M] 12
= 2
−1
(8.21)
I Lifetimes of energized molecules: It is instructive to look at the reciprocal values:
1
−1 [M] = 1 2
(8.22)
1
• is the time between two collisions. This is, in fact, the lifetime of the
−1 [M]
energized molecules with respect to collisional deactivation.
• 1 is the lifetime of the excited molecules with respect to reaction.
2
Thus, at the half pressure, we have
Collisional Deactivation = Unimolecular Decay (8.23)
I Determination of k ∞ :
y
=
2 1 [M]
−1 [M] + 2
(8.24)
1
= −1
+ 1 1
1 2 1 [M]
(8.25)
= 1 + 1 1
∞ 1 [M]
(8.26)
= 1 + 1 ∞ 0
(8.27)
y Plot of −1 vs. [M] −1 should give a straight line with slope 1 −1 and intercept
∞ −1 .

8.2 Lindemann mechanism 181
I
Deficiencies of the Lindemann model:
I Figure 8.8:
• Activation energies:
In particular:
0 ∞ (8.28)
∞ ≈ 0 (8.29)
0 ≈ 0 − ( − 15) (8.30)

8.3 Generalized Lindemann-Hinshelwood mechanism 182
8.3 Generalized Lindemann-Hinshelwood mechanism
8.3.1 Master equation
I
Figure 8.9: Strong collision model.
I
Figure 8.10: Generalized Lindemann-Hinshelwood weak collision model.

8.3 Generalized Lindemann-Hinshelwood mechanism 183
I
Figure 8.11: Generalized Lindemann-Hinshelwood weak collision model.
I
Figure 8.12: Generalized Lindemann-Hinshelwood model: Rate constants in the high
and low pressure (strong and weak collision) limits.

8.3 Generalized Lindemann-Hinshelwood mechanism 184
8.3.2 Equilibrium state populations
As seen from the master equation analysis, the unimolecular reaction rate constant
in the high and in the low pressure regimes depends on the equilibrium (Boltzmann)
state distributions (for discrete states) or () (continuous state distributions). In
a polyatomic molecule, due to the large number of vibrational degrees of freedom, the
number of vibrational states increases very rapidly with increasing vibrational excitation
energy. The quanta of vibrational excitation can be distributed over the oscillators. In
order to calculate the number of combinations, we start with a single oscillator and
expand our scope first to two oscillators and then to =3 − 6 oscillators.
a) Equilibrium state population for a single oscillator:
I
Boltzmann distribution:
with the energy quanta
the degeneracy factor
and the partition function
= −
(8.31)
= × ; =0 1 2 (8.32)
=
(8.33)
1
1 − −
(8.34)
I Limiting case for high T (transition to classical mechanics): ¿ y
classical
=
(8.35)
since
− ≈ 1 − for → 0 (8.36)
Note that this classical description of vibrations is not so bad for unimolecular reactions,
where we are interested in the kinetic behavior at high temperatures.
b) Equilibrium state population for s oscillators:
y
() → () (8.37)
() = () −
(8.38)
with () being the density of vibrational states at the energy and
=
Y
=1
1
1 − −
(8.39)
and
=3 − 6 (resp. =3 − 5) (8.40)

8.3 Generalized Lindemann-Hinshelwood mechanism 185
8.3.3 The density of vibrational states ρ (E)
Thedensityofvibrationalstatesisdefined as
() =
number of states in energy interval
energy interval
(8.41)
i.e.,
() =
()
(8.42)
In a polyatomic molecule, as we shall see in the following, () increases with very
rapidly owing to the large number of overtone and combination states, and reaches truly
astronomical values:
a) s =1:
() = 1
(8.43)
b) s =2:
We consider the ways, in which we can distribute two identical energy quanta between
the two oscillators:
Osc. 1 Osc. 2 Osc. 1 Osc. 2 Osc. 1 Osc. 2 Osc. 1 Osc. 2
0 - - 1
| - - | 2
2 || - | | - || 3
3 ||| - || | | || - ||| 4
4 5
5 6
.
.
.
.
(8.44)
c) s oscillators:
• For simplicity, we consider a molecule with identical harmonic oscillators.
• This molecule shall be excited with quanta, eachofsize, i.e., the molecule
has the excitation energy =
I
Question: How can we distribute these quanta over the oscillators?

8.3 Generalized Lindemann-Hinshelwood mechanism 186
I
Answer: We are asking for the number of distinguishable possibilities to distribute
quanta between oscillators, (). This is identical to the problem of distributing
balls between boxes, for example for =11and =8:
••• • • • •• • •• (8.45)
That number is identical to the number of distinguishable permutations of balls ( • )
and − 1 walls ( | ),
( + − 1)!
() = (8.46)
!( − 1)!
where ( + − 1)! is the total number of permutations, which is corrected by dividing
through !( − 1)! to account for the number of indistinguishable permutations among
the balls (!) andwalls(( − 1)!) among each other (which give indistinguishable results).
I Answer: We make the following approximation for À : 49
( + − 1)!
!( − 1)!
≈
−1
( − 1)!
(8.49)
I
Question: How can we compute ()?
I
Answer:
() =
()
(8.50)
(1) We consider the energy interval
∆ = (8.51)
at the excitation energy
= × (8.52)
In this energy interval, we have the number of states
∆() = () (8.53)
y
() =
()
∆ ()
≈
∆ = 1 ( + − 1)!
!( − 1)!
(8.54)
49 The approximation
( + − 1)!
≈ −1 (8.47)
!
can be derived using Stirling’s formula (ln ! = ln − ) and the power series expansion of
for || ¿1 (see homework assignment).
ln (1 − ) ≈ (8.48)

8.3 Generalized Lindemann-Hinshelwood mechanism 187
(2) Using the approximation
( + − 1)!
!( − 1)!
≈
−1
( − 1)!
(Eq. 8.49) for À gives
() = 1 −1
( − 1)!
(8.55)
(3) Inserting = into (), weobtain
() = 1 () −1
( − 1)!
(8.56)
(4) Result:
() =
−1
( − 1)! () (8.57)
d) Notes:*
(1) The above derivation may seem artificial because in a real molecule not all oscillators
are identical. However, the derivation can be easily generalized.
(2) For instance, we can start from the number of states of the first oscillator 1 ( 1 )
and convolute this with the density of states 2 () of the second oscillator
2 ( 2 )= 2 ( − 1 )= 1
(8.58)
and compute the total (combined) density of states for two oscillators by convolution
according to
() =
Z
etc. up to oscillators. The result is
0
1 ( 1 ) 2 ( − 1 ) 1 (8.59)
() =
−1
( − 1)! Q
=1
(8.60)
(3) We can also use inverse Laplace transforms: Since is the Laplace transform
of () according to
=
Z ∞
0
() − = L [ ()] (8.61)
we can determine () from (which we know) by inverse Laplace transformation.
(4) With corrections for zero-point energy:
() =
( + ) −1
( − 1)! Q
=1
(8.62)

8.3 Generalized Lindemann-Hinshelwood mechanism 188
(5) With (empirical) Whitten-Rabinovitch corrections:
() = ( + () ) −1
( − 1)! Q
=1
(8.63)
where () is a correction factor that is of the order of 1 except at very low
energies.
(6) Exact values of () for harmonic oscillators can be obtained by direct state
counting algorithms.
(7) Corrections for anharmonicity can be applied using different means.
8.3.4 Unimolecular reaction rate constant in the low pressure regime
From the master equation analysis above, we obtained the thermal unimolecular reaction
rate constant in the low pressure regime as
0 = X X
−1() [M]
(8.64)
The reduction of the excited state populations compared to the equilibrium (Boltzmann)
distribution is shown in Fig. 8.13. Two cases can be distinguished:
(1) With the strong collision assumption (h∆ iÀ), we can apply the so-called
equilibrium theories which assume that = for ≤ 0 . Thus, the state
population below 0 remains as in the high pressure regime, whereas it is reduced
above 0 .Then,
y
0 = X
X
−1() [M]
= X
Z ∞
0 = [M]
X
−1() [M] = [M] X
(8.65)
0
() (8.66)
Using the expressions for () () and in the classical limit ( ¿ )
derived above and doing the integration by parts, we obain the Hinshelwood
equation for 0 :
µ −1 0 1
0 = [M]
( − 1)! − 0
(8.67)
This equation is usually used to describe experimental data with as a fit parameter
(usually about half the real ).
Since ∝ 05 and = 2 ln , the low pressure activation energy
becomes
0 = 0 − ( − 15) (8.68)
Thus, may be significantly smaller than 0 . This can be understood as
illustrated in Fig. 8.14.

8.3 Generalized Lindemann-Hinshelwood mechanism 189
(2) With the weak collision assumption (h∆ i ≤ ) made by the so-called nonequilibrium
theories, the bottleneck in the state population falls below the equilibrium
distribution already below 0 (see Fig. 8.14). Using the weak collision
assumption, Troe derived the expression
0 = [M] ( 0 )
− 0
(8.69)
where is the so-called collision efficiency ( ≤ 1) whichcanbeexpressedin
terms of the average energy transferred per collision h∆ i. becomes even
smaller than the Hinshelwood above. A simple estimate of the weak collision
effect based on Fig. 8.14 may lead to
0 ≈ 0 − ( − 05) (8.70)
I
Figure 8.13: State populations.

8.3 Generalized Lindemann-Hinshelwood mechanism 190
I
Figure 8.14: Activation energies for unimolecular reactions in the high pressure, the
low pressure strong collision, and the low pressure weak collision limits.
8.3.5 Unimolecular reaction rate constant in the high pressure regime
From the master equation analysis above, the thermal unimolecular reaction rate constant
in the high pressure regime is (see Fig. 8.12)
with the Boltzmann distribution
asshowninFig.8.13.
∞ =
Z ∞
0
() () (8.71)
() = () −
(8.72)
Averaging over the Boltzmann distribution gives the TST expression
( )=
‡
− 0
(8.73)

8.4 The specific unimolecular reaction rate constants () 191
8.4 The specific unimolecular reaction rate constants k (E)
We may intuitively expect that the specific uinimolecular reaction rate constants ()
depend on the energy content (i.e., the excitation energy ) of the energetically activated
molecules molecules A ∗ . More precisely, we will also have to take into account
the dependence of the specific rate constants on other conserved degrees of freedom,
in particular total angular momentum (⇒ ()), and perhaps other quantities
(symmetry, components of if -mixing is weak, (⇒ (; ; ))).
8.4.1 Rice-Ramsperger-Kassel (RRK) theory
I RRK model: In order to derive an expression for the specific rateconstants (),
werealizethatitisnotenoughthatthemoleculesareenergizedtoabove 0 . Indeed,
for the reaction to take place, the energy also has to be in the right oscillator, namely
in the RC. In particular, for the reaction to take place, of the total excitation energy ,
the amount † has to be concentrated in the RC such that † ≥ 0 . Thus, we can
write the reaction scheme as
∗ †
→ † → (8.74)
where † denotes those excited molecules that have enough energy in the RC to overcome
0 . We assume that once this critical configuration ( † ) is reached, the molecules
will immediately cross the TS and go to the product side.
I Probabilities of A † and A ∗ : The probability of † compared to ∗ can be derived
using statistical arguments by considering the ratio of the density of states. We consider
amoleculewith
• oscillators,
• = quanta which have to be distributed over the oscillators,
• a threshold energy of 0 such that a minimum of = 0 quanta have to be
concentrated in the RC and only − quanta can be distributed freely,
• À and − À .
The probability of † relative to ∗ is given by
¡ †¢
( ∗ ) = ¡ †¢
( ∗ )
(8.75)
where
( ∗ ) ∝ ∗ () = 1 ( + − 1)!
(8.76)
!( − 1)!
and, since in the critical configuration only − quanta can be distributed freely,
¡ †¢ ∝ † () = 1 ( − + − 1)!
( − )! ( − 1)!
(8.77)

8.4 The specific unimolecular reaction rate constants () 192
y
¡ †¢
( ∗ )
( − + − 1)! !( − 1)!
= ×
( − )! ( − 1)! ( + − 1)!
( − + − 1)! !
= ×
( − )! ( + − 1)!
(8.78)
(8.79)
I Specific rate constant: The critical configuration † is reached with the rate coefficient
(≡ frequency factor) † (see the reaction scheme above). Since the energy is now
in place (in the RC), † will cross the TS to products.
The specific rateconstant () is therefore simply
y
() = † ×
() = † × ¡ †¢
( ∗ )
( − + − 1)!
( − )!
×
!
( + − 1)!
(8.80)
(8.81)
I Kassel formula: With the approximations according to Eq. 8.49),
( + − 1)!
!
≈ −1 for À (8.82)
and
we obtain
( − + − 1)!
( − )!
¡ †¢
( ∗ )
≈ ( − ) −1 for − À , (8.83)
=
( − )−1
−1 =
µ −1 −
(8.84)
y
¡ †¢ µ −1 −
( ∗ ) = 0
(8.85)
so that the specific rateconstantbecomes
This expression is known as the Kassel formula for ().
µ −1 −
() = † 0
(8.86)
I
Notes on the Kassel formula:
• Owing to the approximations À and − À , the Kassel formula is valid
only at À 0 .

8.4 The specific unimolecular reaction rate constants () 193
• The Kassel formula describes experimental data qualitatively well. However, instead
of the full value of =3 − 6, one usually has to use an effective number
eff . As suggested by Troe, eff can be estimated from the specific heatofthe
reactant molecules via
=
µ
=
µ
2 ln
using the known expression for ,
Y 1
=
1 − exp (− )
=1
≈ eff (8.87)
(8.88)
• Often, eff is equal to about one half of the number of oscillators,
eff ≈
(3 − 6)
2
(8.89)
I
Figure 8.15: Specific rate constants: RRK formula.
8.4.2 Rice-Ramsperger-Kassel-Marcus (RRKM) theory
RRK theory fails in the following points:
• Itcanbeusedonlywith as an effective parameter ( eff ).
• It leads to the wrong results close to threshold, where → 0 (see the conditions
above).
• It is difficult to derive for oscillators with different frequencies.
• RRK theory is essentially a classical theory. It does not know about zero-point
energy.

8.4 The specific unimolecular reaction rate constants () 194
I RRKM expression: These problems are essentially solved by RRKM theory which
gives the expression
() = ‡ ( − 0 )
()
(8.90)
where
• ‡ ( − 0 ) is the number of open channels (essentially the number of states at
the transition state configuration exluding the RC), counted from the zero-point
level of the TS, at the energy − 0 ,
• () is the density of states of the reacting molecules at the energy .
These quantities are the two critical quantities in microcanonical transition state theory.
I
Figure 8.16: RRKM expression for ().

8.4 The specific unimolecular reaction rate constants () 195
I
Figure 8.17: RRKM expression for () with angular momentum conservation.
I
Figure 8.18: The sum and the density of states.

8.4 The specific unimolecular reaction rate constants () 196
I RRKM expression with angular momentum conservation: Eq. 8.90 can be generalized
to include other conserved quantities, e.g., total angular momentum,
() = ‡ ( − 0 )
()
(8.91)
I Relation between RRKM and RRK theory: To elucidate the relation between Eqs.
8.86 and 8.90, we look at () and ( − 0 ) more closely.
(1) Density of vibrational states ():
() =
number of states in energy interval
energy interval
=
()
(8.92)
a) RRK result for the number of indistinguishable permutations of quanta
over oscillators; = À :
y
() = 1 ( + − 1)!
!( − 1)!
() =
≈ 1 −1
( − 1)! = 1 () −1
( − 1)!
(8.93)
−1
( − 1)! () (8.94)
b) It is relatively easy to show that for oscillators with different frequencies, the
expression for () becomes
() =
−1
( − 1)! Q
=1
(8.95)
c) With corrections for zero-point energy:
() =
( + ) −1
( − 1)! Q
=1
(8.96)
d) With (empirical) Whitten-Rabinovitch correction:
() = ( + () ) −1
( − 1)! Q
=1
(8.97)
where () is a correction factor that is of the order of 1 except at very
low energies.
e) Exact values of () for harmonic oscillators can be obtained by direct
state counting algorithms. Corrections for anharmonicity can be applied
using different means.

8.4 The specific unimolecular reaction rate constants () 197
(2) Number of states () and ‡ ( − 0 ): Since () = () , the
number of states from =0to can be obtained by integration,
a) RRK expression:
y
() =
Z
0
() =
() =
Z
0
Z
0
() (8.98)
−1
( − 1)! () =
() =
Z
1 1
( − 1)! () −1
0
(8.99)
!() (8.100)
b) Allowing for different frequencies, we modify this expression to
() =
! Q
=1
(8.101)
c) As above, a correction for zero-point energy gives
() = ( + )
! Q
=1
(8.102)
d) With the Whitten-Rabinovitch correction, this becomes
() = ( + () )
! Q
=1
(8.103)
e) Correction for → 0: At =0,wemusthaveatleastonestate,i.e.,
‡ () =1+ ( + () )
! Q
=1
− ( (0) )
! Q
=1
(8.104)
(3) Application to unimolecular reactions:
a) At the TS, we have only − 1 oscillators since the RC has to be excluded.
Furthermore, the states range from the zero-point energy 0 of the TS to
, while we count from the zero-point of the reactants. y
‡ ( − 0 )=1+
³
− 0 + ‡ ( − 0 ) ‡ ´−1
−
³
‡ (0) ‡ ´−1
( − 1)! Q −1
=1 ‡ ( − 1)! Q −1
=1 ‡
(8.105)
This expression gives the correct value of ‡ ( − 0 )=1at = 0 .

8.4 The specific unimolecular reaction rate constants () 198
b) Density of states:
() = ( + () ) −1
( − 1)! Q
=1
(8.106)
c) Specific rate constant:
³
() = 1 − 0 + ‡ ( − 0 ) ‡ ´−1
() +
()( − 1)! Q −
−1
=1 ‡
(4) For energies sufficiently above 0 , we can use the expressions
and
to obtain
y
‡ ( − 0 )=
() = ‡ ( − 0 )
()
() =
() =
=
³
− 0 + ‡ ´−1
( − 1)! Q −1
=1 ‡
( + ) −1
( − 1)! Q
=1
³
‡ (0) ‡ ´−1
()( − 1)! Q −1
=1 ‡
(8.107)
(8.108)
(8.109)
³
Q
=1 − 0 + ‡ ´−1
Q −1
=1 ‡
( + ) −1 (8.110)
Q
=1
Q −1
=1 ‡
The resemblence to the Kassel formula is obvious.
Ã
− 0 + ‡
+
! −1
(8.111)
8.4.3 The SACM and VRRKM model
More advanced models allow for the tightening of the TS with increasing (variational
RRKM theory = VRRKM theory) or even for individual adiabatic channel potential
curves (statistical adiabatic channel model = SACM). The latter model has to be used
for reactions with loose transition states.

8.4 The specific unimolecular reaction rate constants () 199
8.4.4 Experimental results
I
Figure 8.19: Experimental methods for the preparation of highly vibrationally excited
molecules with definedexcitationenergy.
I
Figure 8.20: Comparison of theoretically predicted specificrateconstants () for unimolecular
isomerization of cycloheptatrienes with directly measured experimental data.

8.5 Collisional energy transfer* 200
8.5 Collisional energy transfer*
Skipped for lack of time.

8.6 Recombination reactions 201
8.6 Recombination reactions
See homework assignment.

8.6 Recombination reactions 202
8.7 References
Lindemann 1922 F. A. Lindemann, Trans. Faraday Soc. 17, 598 (1922).
Hinshelwood 1927 C. N. Hinshelwood, Proc. Roy. Soc. A113, 230 (1927).
Kassel 1928a J. Phys. Chem. 32, 225 (1928).
Kassel1928b J. Phys. Chem. 32, 1065 (1928).
Kassel 1932 L. S. Kassel, The Kinetics of Homogeneous Gas Reactions, Chem.Catalog,
New York, 1932.
Rice 1927 O. K. Rice, H. C. Ramsperger, J. Am. Chem. Soc. 49, 1617 (1927).
Rice 1928 O. K. Rice, H. C. Ramsperger, J. Am. Chem. Soc. 50, 617 (1928).
Troe1979 J.Troe,J.Phys.Chem.83, 114 (1979).
Troe1994 J. Chem. Soc. Faraday Trans. 90, 2303 (1994).

8.7 References 203
9. Explosions
9.1 Chain reactions and chain explosions
9.1.1 Chain reactions without branching
I
Reaction scheme:
(1) A → X chain initiation
(2) X+B→ P+X chain propagation
(4) X → A chain termination
I
Rate laws:
[X]
= 1 [A] − 4 [X] (9.1)
[P]
= 2 [B] [X] (9.2)
As we see, reaction (2) does not appear in the rate law because → !
I Solutions of Eq. 9.1 for two limiting cases (cf. Fig. 9.1):
(1) Solution for → 0 and [A] ≈ const by substitution, using
= 1 [A] − 4 [X]
y
y
y
y
=0:
y
[X] = − 5 y [X] = −
4
− 1 4
= y
= − 4
= − 4 = 1 [A] − 4 [X] (9.3)
[X] = 1
4
[A] − − 4
(9.4)
[X] = 0 y = 1
[]
4
(9.5)
[X] = 1 [A]
³ ´
1 − − 4
4
(9.6)

9.1 Chain reactions and chain explosions 204
(2) Solution with steady state assumption for [X] at À (i.e, after the induction
time ):
[X]
≈ 0 (9.7)
y
[X]
= 1
[A] (9.8)
4
y Considering the free radical concentration profile [X ()], no desasters are happening.
I
Figure 9.1: Evolution of the free radical concentration in a normal chain reaction.
9.1.2 Chain reactions with branching
I
Reaction scheme:
(1) A → X chain initiation
(3) X+B→ P+ X chain branching
= branching factor
(4) X → A chain termination
I
Most important example for a chain branching reaction:
H+O 2 → OH + O (9.9)
This reaction is responsible for flame propagation in all hydrocarbon flames: Since H is
small, it has the highest diffusion coefficient and diffuses rapidly into the unburnt gases.

9.1 Chain reactions and chain explosions 205
I
Rate laws:
[X]
[P]
= 1 [A] + ( − 1) 3 [X] [B] − 4 [X] (9.10)
= 3 [B] [X] (9.11)
I
Solutions for three limiting cases:
(1) =1or 4 3 ( − 1) [B]:
[X]
≈ 0 (9.12)
y
1 [A]
[X] =
(9.13)
4 − ( − 1) 3 [B]
y
[P]
= 3 [B] [X] = (9.14)
The overall reaction is stable, because the radical concentration is stable. The
product formation rate is final (constant, well behaved).
(2) 4 = 3 ( − 1) [B]: Atshorttimes,[A] [B] ≈ const y
[X]
= 1 [A] = const (9.15)
y
[X] →∞ (9.16)
The overall reaction turns unstable, because the radical concentration goes to
infinity and thus the product formation rate too, leading to chain explosion!
(3) 4 3 ( − 1) [B]: Atshorttimes,[A] [B] ≈ const y
[X]
+ 4 [X] − 3 ( − 1) [X] [B] = 1 [A]
(9.17)
[X]
+[X]( 4 − 3 ( − 1) [B]) = 1 [A]
(9.18)
y
1 [A]
³
´
[X] =
× exp [( − 1) 3 [B] ] − 1 (9.19)
( − 1) 3 [B] − 4
The overall reaction turns unstable, because the radical concentration increases
exponentially, leading to an exponentially growing product formation rate, and
therefore chain explosion!

9.1 Chain reactions and chain explosions 206
I
Figure 9.2: Evolutionofthefreeradicalconcentrationsinachainreactionwithbranching
(three limiting cases).

9.2 Heat explosions 207
9.2 Heat explosions
We consider an exothermic bimolecular reaction in a closed reactor
A+B→ C+D ∆ ª 0 (9.20)
and look at the radial temperature profile in the reactor.
I
Figure 9.3: Temperature profilesinastirredandinanunstirredreactor.
9.2.1 Semenov theory for “stirred reactors” (T =const)
I Heat production in an exothermic reaction (∆ H ª < 0):
• Energy released per volume:
∆ ∆ ª £ ¤ kJ cm
3
(9.21)
• Reaction rate:
• heat release:
= − [][] (9.22)
= ∆ ª
= ∆ ª 0 −
| {z }
| {z }
Arrhenius rate constant in pressure units
(9.23)
(9.24)

9.2 Heat explosions 208
• positive feedback effect:
reaction rate increases
y temperature ↑
y reaction rate ↑
y heat explosion !!
I
Figure 9.4: Heat explosions.
I Heat removal: At constant temperature in the stirred reactor, we use Newton’s law
= ( − 0 ) (9.25)
where
: heat conduction coefficient (depends on material)
: reactor surface area
: temperature
0 : wall temperature
(9.26)
As we see, the heat removal increases (becomes steeper), when we increas the reactor
surface area (at constant reactor volume).
I
Stability limits:
(1) 0 = 1 : Heat removal wins over heat production y negative feedback y stable.
(2) 0 = 2 : Heat removal wins over heat production y negative feedback y stable.
(3) 0 = 3 : Heat production wins over heat removal y positive feedback y heat
explosion, but the system should never reach point [3].
(4) 0 = 4 : tangent y stability limit, the system runs away upon the smallest
perturbation!
(5) 0 = 5 : always unstable!

9.2 Heat explosions 209
I
Conclusions:
• The quantity that determines whether the system is stable or not is the wall
temperature 0 ,because 0 determines the initial point of the heat removal line.
• Smaller surface-to-volume ratio is dangerous, because smaller surface at a given
volume leads to a smaller slope of the heat removal line y instability!
• Security measure: internal cooling system (cold water pipes, heat exchanger).
• Emergency measure: when cooling system fails, the reaction mixture should be
strongly diluted by addition of solvent immediately!
I Quantitative estimation of the stability limits:* At point (4) in Fig. 9.4, we have:
=
(9.27)
y ( − 0 )= ∆ ª 0 − (9.28)
=
(9.29)
y = ∆ ª 0 −
2 (9.30)
Eq
Eq y 2
=( − 0 ) (9.31)
y 2 −
+
y =
2 ± s µ
2
0 =0 (9.32)
2
−
0 (9.33)
Solution is: ⎧
⎪⎨
= 1 ³
+ p ´ ⎫
( 2 2
− 4 0 )
⎪⎬
⎪ ⎩ = 1 ³
− p ´
( 2 2
− 4 0 )
⎪⎭
This can be simplified:
(9.34)
(1) For practically important reactions, we have:
À 0 y 0 ¿
(9.35)
(2) We also can exclude unreasonably hight temperatures y only the −-sign before
∗ s
= µ 2
2 − −
2 0 (9.36)
=
2 − r
1 − 4 0
(9.37)
2
√ counts: y
=
2
à r
1 −
!
1 − 4 0
(9.38)

9.2 Heat explosions 210
(3) We can expand the √ :
= 4 0
; 1; (9.39)
y √ 1 − =1− 1 2 − 1
2 · 4 2 − (9.40)
y ∗ = µ
1 − 1+ 2 0
+ 16 ·
2 0
2
2 2 · 4 · ( ) 2 + (9.41)
y ∗ = 0 + 2 0
(9.42)
y ∗ − 0 = 2 0
(9.43)
(4) Insert into in (I) and power expansion:
· ·
(5) Explosion limit:
2
0
2
· exp
µ
2
0
∙
− ¸
=
0
⎡
= · ∆ ª · 0 · exp ⎣−
´ ⎦ · · (9.44)
0
³1+ 0
∙
= · ∆ ª · 0 · · exp − ¸
· 2 · · (9.45)
0
· ·
· ∆ ª · · · 0 · ·
= (9.46)
⎤
y
∝ ·
(9.47)
Power expansion of the exponent in step (4):
⎡
⎤
∙
exp ⎣−
´ ⎦ =exp −
0
³1+ 1 ¸
1
0
1+
∙
=exp − 1
≈ exp
∙− 1 (1 − )¸
∙
=exp 1 − 1 ¸
∙
= 1 · exp − 1 ¸
∙
= · exp − ¸
0
¡
1 − + 2 ¢¸
= 0
¿ 1(9.48)
(9.49)
(9.50)
(9.51)
(9.52)
(9.53)

9.2 Heat explosions 211
I Figure 9.5:
9.2.2 Frank—Kamenetskii theory for “unstirred reactors”
I
Figure 9.6: Temperature profilesinanunstirredreactor.
Boundary effects are neglected in the Figure. In reality, there is a smooth transition in
the boundary layer.

9.2 Heat explosions 212
9.2.3 Explosion limits
I
Figure 9.7: Explosion limits in the H 2 /O 2 system.
9.2.4 Growth curves and catastrophy theory
I
Exponential growth:
∝ · y
= (9.54)
I
Hyperbolic growth:
∝
1
( − ) y
= 1+ mit 1+1 (9.55)
I
Heat explosion:
∝ quadratic, i.e. we get hyperbolic growth ! (9.56)

9.2 Heat explosions 213
I
Figure 9.8: Growth curves.

9.2 Heat explosions 214
10. Catalysis
The net rate of a chemical reaction may be affected by the presence of a third species
even though that species is neither consumed nor produced by the reaction. If the rate
increases, we speak of catalysis. If the rate decreases, we speak of inhibition.
In general, we have to distinguish
• homogeneous catalysis,
• heterogeoues catalysis.
10.1 Kinetics of enzyme catalyzed reactions (homogeneous catalysis)
The mechanism describing the homogeneous catalytical activity of enzymes has first
been elucidated by Michaelis and Menten (1914).
I
Experimental observations on enzyme catalysis:
(1) For constant substrate concentration [S] 0
, the reaction rate is proportional to
the concentration of the enzyme [E] 0
∝ [E] 0
for [S] 0
= const (10.1)
(2) At an applied enzyme concentration [E] 0
,thereactionrate first increases with
increasing substrate concentrations [S] 0
andthenreachessaturation,asisshown
in Fig. 10.1.
I
Figure 10.1: Enzyme catalysis: Michelis-Menten kinetics.

10.1 Kinetics of enzyme catalyzed reactions (homogeneous catalysis) 215
10.1.1 The Michaelis-Menten mechanism
Basic reaction scheme:
E+S 1
À ES (10.2)
−1
ES 2
→ E+P (10.3)
Typical values of 2 :
2 ≈ 10 2 10 3 s −1 (in some cases up to 10 6 s −1 ) (10.4)
Reaction rate:
[P]
= − [S]
= 2 [ES] (10.5)
[ES]
= 1 [E] [S] − −1 [ES] − 2 [ES] (10.6)
≈ 0 (10.7)
Mass balance:
Steady state ES concentration:
[E] 0
= [E]+[ES] (10.8)
y [E] = [E] 0
− [ES] (10.9)
[S] 0
= [S]+[ES]+[P] (10.10)
y [S] = [S] 0
− [ES] − [P] (10.11)
[ES]
= 1 ([E] 0
− [ES]) ([S] 0
− [ES] − [P]) − ( −1 + 2 )[ES] (10.12)
≈ 0 (10.13)
Approximations:
y
[ES]
[E] 0
¿ [S] 0
(10.14)
[ES] ¿ [S] 0
(10.15)
[P] ≈ 0 for early times (10.16)
= 1 [E] 0
[S] 0
− 1 [S] 0
[ES] − ( −1 + 2 )[ES] (10.17)
≈ 0 (10.18)
Steady state ES concentration:
[ES]
=
1 [E] 0
[S] 0
1 [S] 0
+ −1 + 2
(10.19)

10.1 Kinetics of enzyme catalyzed reactions (homogeneous catalysis) 216
Resulting rate expression:
y
= [P] = 2 [ES] (10.20)
1 2 [E]
=
0
[S] 0
(10.21)
1 [S] 0
+ −1 + 2
2 [E]
=
0
1+ (10.22)
−1 + 2
1 [S] 0
=
with the Michaelis-Menten constant
2 [E] 0
1+ [S] 0
(10.23)
= −1 + 2
1
(10.24)
Limiting cases:
(1) [S] 0
→∞y
[S] 0
¿ 1: y
= [P]
= 2 [E] 0
(10.25)
(2) [S] 0
→ 0 y
[S] 0
À 1: y
= [P]
= 2 [E] 0
[S] 0
(10.26)
Determination of 2 and from a Lineweaver-Burk plot (Fig. 10.2):
1
= 1 +
1
(10.27)
2 [E] 0
2 [E] 0
[S] 0
⇒ Aplotof −1 vs. [S] −1 0
should give a straight line with intercept ( 2 [E] 0
) −1 and
slope 2 [E] 0
. y
slope
intercept = = −1 + 2
(10.28)
1

10.1 Kinetics of enzyme catalyzed reactions (homogeneous catalysis) 217
I
Figure 10.2: Lineweaver-Burk plot.
10.1.2 Inhibition of enzymes:
Extended reaction mechanism:
E+S 1
À ES (10.29)
−1
ES 2
→ E+P (10.30)
E+I 3
À EI (10.31)
−3
The EI complex is catalytically inactive, since I blockstheactivesiteofE.
Inhibition constant:
= [EI]
[E] [I] = 3
−3
(10.32)
I
Reaction rate (Fig. 10.2):
y
[ES]
= ≈ 0 (10.33)
1
= 1 +(1+ [I]) ×
1
(10.34)
2 [E] 0
2 [E] 0
[S] 0
y The slope of the plot of −1 vs. [S] −1 0
goes up (see Fig. 10.2).
Exercise 10.1: Derive Eq. 10.34 with a reasonable assumption for the concentrations
of I. ¤

10.2 Kinetics of heterogeneous reactions (surface reactions) 218
10.2 Kinetics of heterogeneous reactions (surface reactions)
I Examples for heterogeneous reactions (gas-solid interface): Many reactions,
which are very slow in the gas phase, take place as heterogeneous reactions in the
presence of a heterogeneous catalyst:
• catalytic hydrogenation
• NH 3 synthesis (Haber-Bosch)
• catalytic exhaust gas treatment (internal combustion engines)
• catalytic DeNO processes
• CO oxidation
• oxidative coupling reactions of CH 4 (oxide catalysts)
• atmospheric reactions (at gas-solid and gas-liquid interface)
10.2.1 Reaction steps in heterogeneous catalysis
Heterogeneous reactions take place via a series of steps:
• Diffusion in gas (or liquid) phase to catalyst surface.
• Adsorption:
— “physical” adsorption (“physisorption”),
— “chemical” adsorption (“chemisorption”).
• Diffusiononthesurface.
• Chemical reaction on the surface.
• Desorption.
• Diffusion away from surface.
10.2.2 Physisorption and chemisorption
We distinguish between “physical” adsorption (“physisorption”) and “chemical” adsorption
(“chemisorption”) on a surface based upon the adsorption energies (surface binding
energies). ⇒ Fig. 10.3.
• Physisorption:
— van-der-Waals type interaction,
— typical bond energy: ∆ ≤−40 kJ mol −1 ,

10.2 Kinetics of heterogeneous reactions (surface reactions) 219
— example: Ar on a metal or oxide surface.
• Chemisorption:
— formation of a chemical bond between adsorbate and surface.
— typical bond energy: ∆ ≤−100 − 300 kJ mol −1 ,
— example: CO on Pd:
O
||
C
|
− Pd − Pd −
(10.35)
∗ chemical bond by overlap of orbital of the C atom with free bands of
Pd,
∗ back donation from occupied bands into ∗ orbital of the CO substrate.
∗ Experiment shows that (Pd − C=O)≈ 236 cm −1 .
I
Figure 10.3: Potential curves for physisorption and chemisorption.
10.2.3 The Langmuir adsorption isotherm
I
Reaction scheme:
A() +
−
|
S − 1
À
−1
−
A
|
S − (10.36)
− | S− = “active surface site” (10.37)

10.2 Kinetics of heterogeneous reactions (surface reactions) 220
I Fraction of occupied active surface sites Θ:
• Adsorption rate:
• Desorption rate:
• Steady state:
y
y
1 × (1 − Θ) × (10.38)
−1 × Θ (10.39)
1 (1 − Θ) = −1 Θ (10.40)
1 =( −1 + 1 ) Θ (10.41)
I Langmuir isotherm (Fig. 10.4):
Θ =
1
−1 + 1 =
1+
(10.42)
with
= 1
−1
(10.43)
I
Figure 10.4: Langmuir adsorption isotherm.

10.2 Kinetics of heterogeneous reactions (surface reactions) 221
I
Determination of V , V ∞ , and the adsorption enthalpy ∆H : At a series
of temperatures , we need to measure Θ ( )= ( )
∞ ( ) as function of by
pressure/volume measurements in an ultrahigh vacuum apparatus.
• From Eq. 10.42:
• Multiplication of l.h.s. and r.h.s. with
• Aplotof
=
1
Θ =
∞
=1+ 1
1
∞
:
∞
+
∞
(10.44)
= + (10.45)
vs. gives a straight line (see the inset in Fig. 10.4) with slope
=( ∞
) −1 (10.46)
and intercept
=( ∞
) −1 (10.47)
•
• van’t Hoff plot:
=
slope
intercept =
ln
(1 ) = −∆
(10.48)
(10.49)
I Desorption rate: If ∆ 0 0 we can describe the desorption rate by TST (see Figs.
10.3 und 10.6).
−1 = µ
‡
exp − ∆
0
(10.50)
The ’s are partition functions per unit area.

10.2 Kinetics of heterogeneous reactions (surface reactions) 222
10.2.4 Surface Diffusion
I
Figure 10.5: Surface Diffusion.
I
Diffusion is an activated process:
µ
= 0 exp − ∆
∆ = energy barrier of jump of adsorbate from one site to another.
(10.51)
I
Fick’s 2nd law:
( )
= × 2 ( )
2 (10.52)
I
Einstein’s diffusion law:
• 1D:
∆ 2 =2 (10.53)
• 2D:
∆ 2 =4 (10.54)

10.2 Kinetics of heterogeneous reactions (surface reactions) 223
10.2.5 Types of heterogeneous reactions
a) Dissociative Adsorption
I
Scheme:
A 2 () +
−
|
S −
|
S − 2
À
−2
−
A
|
S −
A
|
S − (10.55)
I
Adsorption isotherm:
• Adsorption rate:
• Desorption rate:
y
with
Θ =
2 2 (1 − Θ) 2 (10.56)
−2 Θ 2 (10.57)
( 2
× ) 12
1+( 2 × ) 12 (10.58)
= 2
−2
(10.59)
I
Figure 10.6: Potential energy curves for dissociative adsorption.

10.2 Kinetics of heterogeneous reactions (surface reactions) 224
b) Surface catalyzed unimolecular dissociation
I
Scheme:
A() +
−
|
S − 3
À
−3
−
A
|
S − 4
→ Products (10.60)
I Example: Decomposition of PH 3 on a W metal surface.
I
Decomposition rate and two limiting cases:
[P]
= 4 Θ = 4 × A
1+ A
(10.61)
(1) A ¿ 1: The reaction becomes first-order in A .
[P]
= 4 A (10.62)
(2) A À 1: The reaction becomes zeroth order, i.e., independent of A !
[P]
= 4 (10.63)
c) Bimolecular reactions
I
Scheme:
A() +
−
|
S −
À
−
−
A
|
S − (10.64)
−
A
|
S − + −
B() +
B
−
|
S − 5
→
|
S −
−
À
−
A − B
−
B
|
S − (10.65)
|
S − S − 6
→ Products (10.66)
I Reaction rate: Assuming that A and B compete for the same active sites, we have
(1 − Θ − Θ )= − Θ (10.67)
y
[P]
(1 − Θ − Θ )= − Θ (10.68)
= 5 Θ Θ =
5
(1 + + ) 2 (10.69)

10.2 Kinetics of heterogeneous reactions (surface reactions) 225
I Limiting cases: It is often the case that one species is adsorbed much more strongly
than the other.
• Assume, for instance, that ¿ . Then,
[P]
= 5
(1 + ) 2 (10.70)
(1) À 1:
[P]
= 5
(10.71)
The reaction is inhibited by B, which is strongly adsorbed: At high ,all
active sites are blocked by B.
(2) ¿ 1:
[P]
= 5 (10.72)
The reaction is truly second order.
⇒ We can see from these two cases that, as increases, the rate [P]
increases, reaches a maximum, and then decreases again.
first
d) Example 1: Heterogeneous oxidation of CO
I
Figure 10.7: Heterogeneous oxidation of CO on Pt. (a) Langmuir-Hinshelwood and
Eley-Rideal mechanisms, (b) potential energy diagram (values in kJ mol −1 ).

10.2 Kinetics of heterogeneous reactions (surface reactions) 226
e) Example 2: Haber-Bosch NH 3 synthesis
I Figure 10.8: Potential energy diagram for the catalytic reaction N 2 + 3H 2 → 2NH 3
on a Fe 3 O 4 catalyst according to Ertl and co-workers (1982). Energies in kJ mol −1 .
f) How can we control and accelerate heterogeneous reactions?
The answer depends on which step is rate determining:
• If diffusion to/from the surface is the rate determining step, we can accelerate
the reaction by stirring
• If the adsorption step is rate determining (because of an activation energy), we
can accelerate the reaction by increasing and and, in particular, by increasing
surface area.

10.2 Kinetics of heterogeneous reactions (surface reactions) 227
11. Reactions in solution
“Everyone has problems - chemists have solutions.”
I
Comparison of gas and liquid phase reactions (@ T = 298 K and p =1bar):
• Molecules in the gas phase occupy ≈ 02 % of the total volume, while molecules
in the liquid phase occupy ≈ 50 % of the total volume.
• Molecules in liquid are in permanent contact with each other, and undergo permanent
collisions with each other.
• The crossing of a potential barrier in a reaction is not continuous as in gas phase
but affected by multiple collisions during the crossing.
• As a result of the permanent collisions, we can expect the reactant molecules to
be in thermodynamic equilibrium even above 0 . Thus TST should be applicable.
• However, it is very difficult to evaluate the partition functions in liquids. Hence,
one usually employs the thermodynamic version TST.
• Solvent molecules act as an efficient heat bath.
I Observations on liquid phase reactions: Experience has shown that reactions in
liquids have similar rates as in the gas phase, except if
• the reaction occurs with the solvent,
• there are strong interactions of the reactant molecules with the solvent (e.g., ionic
reactions; ⇒ solvent shells),
• the reactions are diffusion controlled.
11.1 Qualitative model of liquid phase reactions
In contrast to gas phase reactions for which we have advanced theories (⇒ transition
state theory, unimolecular rate theory), the theory for liquid phase reactions is much
less well developed. However, we can understand the required basic reaction steps as
sketched in Fig. 11.1 and a number of limiting cases resulting from these steps.

11.1 Qualitative model of liquid phase reactions 228
I
Figure 11.1: Structure and dynamics of solutions.
I Formal kinetics: We distinguish between the formation of the encounter complex in
the solvent cage (by diffusion) and the actual reaction:
+
À
−
{}
→ (11.1)
y
y
with
[{}]
[ ]
= [][] − − [{}] − [{}] ≈ 0 (11.2)
[{}] =
− +
[][] (11.3)
= [{}] =
− +
[][] = [][] (11.4)
=
− +
(11.5)
I
Rate constant for liquid phase reactions:
=
− +
(11.6)

11.1 Qualitative model of liquid phase reactions 229
I
Two limiting cases:
(1) Diffusion control: À − y
= (11.7)
This case is observed when =0and the diffusion coefficient is small.
We shall find below that
=4 (11.8)
(2) Reaction (or activation) control: ¿ − y
=
−
= {} (11.9)
In this case, the reaction is said to be activation controlled because usually it has
a sizable .
can be described by TST:
=
∆+ −∆+
(11.10)

11.2 Diffusion-controlled reactions 230
11.2 Diffusion-controlled reactions
11.2.1 Experimental observations
The transition to diffusion control can be observed in the gas phase by studying the
kinetics in highly compressed gases (up to =1000bar) or, even better, in supercritical
media. At ≈ 1000 bar, the gas density reaches the density of the liquid phase.
I Smoluchowsky-Kramers limit: At high densities, the rate constant becomes
∝ ∝ 1
(11.11)
I
Figure 11.2: Diffusion control of a unimolecular reaction: At very high pressures, the
fragments recombine before they become separated by diffusion y the reaction becomes
diffusion controlled.
11.2.2 Derivation of k
In order to derive a theoretical expression for , we consider the distribution of the
molecules around a molecule inthecenterofasphereasshowninFig.11.3.

11.2 Diffusion-controlled reactions 231
I
Figure 11.3: Radial distribution of reactants in a diffusion-controlled reaction.
(1) We start by considering the boundary conditions for the concentration [] (cf.
Fig. 11.3):
• At a distance →∞, the concentration of is identical to the mean
concentration in the solution, i.e.,
→∞y []
=[] =∞
=[] (11.12)
• At the distance = for a diffusion controlled reaction, the concentration
of vanishes because of its fast reaction with , i.e.,
→ 0 y [] =0
=0 (11.13)
(2) The result is a concentration gradient []
around , which gives rise to a net
flux ←− by diffusion as described by Fick’s first law:
= × × []
(11.14)
We define this flux, measured in units of molecules/s, in the direction from = ∞
(high concentration) to = (low concentration); thus there is no − sign in
the usually written form of Fick’s law.
is the mean diffusion coefficient, measured in m 2 s,
= + (11.15)

11.2 Diffusion-controlled reactions 232
(3) With =4 2 as the surface of the sphere with radius around , weobtain
= × 4 2 × []
(11.16)
(4) To determine the concentration profile []
, we integrate over []
:
Z []∞ Z ∞
[]
=
[]
4 2
(11.17)
y
∞
[]| ∞
= −
4
¯¯¯¯
(11.18)
y
[] − []
= − 0 (11.19)
4
(5) is found from the boundary conditions that []
=0at = :
=4 × [] (11.20)
(6) The concentration gradient of [] is obtained by inserting the above result for
into Eq. 11.19:
[] − []
= []
(11.21)
y
³
[]
=[] 1 − ´
(11.22)
This expression is plotted in Fig. 11.3.
(7) To determine , we write the rate of the diffusion controlled reaction as
[ ]
= [] (11.23)
Inserting the expression for from Eq. 11.20, we obtain
[ ]
=4 × [][] (11.24)
The diffusion-limited rate constant is thus (in molecular units)
=4 (11.25)
(8) Nernst-Einstein relation between diffusion constant and viscosity for spherical particles
with radius :
=
(11.26)
6
y
∝
(11.27)
(9) Keepinmindthatdiffusion is an activated process:
= 0 −
Typical value for H 2 O: ≈ 15 kJ mol −1 .
(11.28)

11.2 Diffusion-controlled reactions 233
I
Typical values:
y Typical magnitude of :
≈ 2 × 10 −5 cm 2 s −1 (11.29)
≈ 5 × 10 −8 cm (11.30)
≈ 8 × 10 9 lmol −1 s −1 (11.31)
I
Table 11.1: Diffusion constants of ions in aqueous solution at =298K.
ion (cm 2 s −1 ) ion (cm 2 s −1 )
H + 91 × 10 −5 OH − 52 × 10 −5
Li + 10 × 10 −5 Cl − 20 × 10 −5
Na + 13 × 10 −5 Br − 21 × 10 −5
I
Examples of diffusion controlled reaction in solution:
• radical recombination reactions, e.g.
+ → 2 : =13 × 10 10 lmol −1 s −1 (11.32)
• electronic deactivation (quenching) in solution
• H + transfer reactions
+ + − → 2 : =14 × 10 11 lmol −1 s −1 (11.33)
This reaction is ten times faster than a typical diffusion controlled reaction, because
H + transfer is a concerted process (Grotthus mechanism, see Fig. 11.4)!
I
Figure 11.4: Structures involved in fast H + transfer processes in liquid water.

11.3 Activation controlled reactions 234
11.3 Activation controlled reactions
The rate constant for an activation controlled reaction is
= {} (11.34)
I Estimation of K {} : We can estimate the equilibrium constant {} using simple
arguments based on the coordination number. Accordingly, the concentration of the
encounter complex is
[{}] =[] × (probability of finding next to ) (11.35)
=[] × []
[]
(11.36)
where [] is the solvent concentration and is the coordination number. y
{} = [{}]
[][] =
[]
(11.37)
For H 2 Oat =298K, [] =555moll −1 . Thus, taking for example =8,wefind
that
{} =014 l mol −1 (11.38)
I Application of thermodynamic TST: TSTcanbeappliedintwoways:
(1) Considering the overall reaction, we can write
=
exp ¡ −∆ ‡ ¢ (11.39)
where ∆ ‡ is the overall Gibbs free enthalphy for the reaction.
(2) We can also separate the effects from and {} by writing
´
{} =exp
³−∆ ª {}
(11.40)
and
=
exp ¡ −∆ + ¢ (11.41)
∆ ª {}
is the Gibbs free enthalpy change for the formation of the encounter
complex, and ∆ + is the Gibbs free enthalphy of activation from the encounter
complex. Thus,
= ³
exp −
³∆ ∆+´ ´
ª
{} + (11.42)
I Pressure dependence of the rate constant: ⇒ See section 7.3.5.

11.4 Electron transfer reactions (Marcus theory) 235
11.4 Electron transfer reactions (Marcus theory)
Electron transfer is what happens in oxidation-reduction reactions. Hence, electron
transfer reactions are extremely important. They are strongly affected by the solvent
because the change of the charge distribution results in a large reorganization energy.
I Figure 11.5:
I Marcus theory: For a simplified derivation of ( ) we refer to Fig. 11.5.
(1) The energy of the donor and acceptor molecules will generally depend on all 3−6
internal coordinates. We consider the dependence on an effective coordinate ,
which describes the minimum energy path from reactant to product (Fig. 11.5).
(2) The exergonicity of the ET reaction ∆ ª is negative for an exergonic reaction
(convention; cf. Fig. 11.5).
(3) For the donor (D) and the acceptor (A), the Gibbs energy profiles () and
() shall be of parabolic form, which we may write as
() = 2 (11.43)
and
() =( − ) 2 + ∆ ª . (11.44)
The origin =0has been placed at the minimum of the donor; the acceptor has
its minimum at = . 50
50 We assume that ∝ 2 , taking any proportionality constant into account by proper rescaling of the
abscissa.

11.4 Electron transfer reactions (Marcus theory) 236
(4) The ET rate constant is described by TST as
=
exp ¡ −∆ + ¢ (11.45)
The Gibbs free energy of activation
∆ + = 2 (11.46)
is the difference from the donor minimum to the TS at the donor/acceptor curve
crossing at point .
(5) In order to relate ∆ + to the potential curves, we define the “reorganization
energy” as the energy that the acceptor would have at the equilibrium
geometry of the donor. This is the energy of the acceptor parabola () at
=0.
If we count from the minimum of the acceptor parabola (), wehave
=(− ) 2 = 2 (11.47)
(6) We now determine the energy of the intersection point of the two parabolas
from the condition
( )= ( ) . (11.48)
Inserting into Eqs. 11.43 and 11.44 yields, and using Eq. 11.47, yields
2 =( − ) 2 + ∆ ª (11.49)
= 2 − 2 + 2 + ∆ ª (11.50)
= 2 − 2 + + ∆ ª (11.51)
y
y
y
y
2 = + ∆ ª (11.52)
∆ + = 2 = ( + ∆ ª ) 2
( )=
= + ∆ ª
2
(11.53)
= ( + ∆ ª ) 2
(11.54)
4
2 4
Ã
!
exp − ( + ∆ ª ) 2
(11.55)
4
(7) Limiting cases: We can distinguish between
• −∆ ª : Regular region – the reaction becomes faster, the more
exergonic the reaction becomes.
• −∆ ª = : turning point
• −∆ ª : Inverted region – the reaction becomes slower, the more
exergonic the reaction becomes.

11.4 Electron transfer reactions (Marcus theory) 237
I Conclusion: Equation 11.55 predicts the ET rate constant to increase with increasing
exoergicity (−∆ ª ) of the reaction. However, once −∆ ª , the rate constant
is predicted to decrease again (see Figs. 11.6D and 11.7). The prediction of this
unexpected inverted Marcus regime before any experimental data were available is
the hallmark of Marcus theory.
I Figure 11.6:
I Figure 11.7:

11.5 Reactions of ions in solutions 238
11.5 Reactions of ions in solutions
Reactions of ionic species in liquids are one exception to the rule that reactions in the
liquid phase usually have similar rates as in the gas phase (the other exception being
electron transfer reactions). The reason is that the charged species are very sensitive
to their environment, in particular solvation. Changes of the charge distribution are
accompanied by correspondingly large solvation shell rearrangements.
11.5.1 Effect of the ionic strength of the solution
The main effect arises from the stabilization of shielding every ion in solution by the oppositely
charged ”ionic atmosphere” or “ion cloud” (⇒ Debye-Hückel theory, Appendix
??).
I
Equilibrium constant for
hAB ‡ i
:
‡ = ‡
= ‡
£
‡ ¤
[][]
(11.56)
I
Activity coefficient from Debye-Hückel theory:
log = − 2 12
(11.57)
with (for water at =298K, according to Debye-Hückel theory)
=0509 l 12 mol −12 (11.58)
I
Ionic strength:
= 1 X
2 (11.59)
2
11.5.2 Kinetic salt effect
I
TST rate constant:
[ ]
= £ ‡ ‡¤ = ‡
‡ [][] (11.60)
‡
Defining 0 as the rate constant for the ideal solution, where all =1, this becomes
= 0
‡
(11.61)

11.5 Reactions of ions in solutions 239
I
Dependence of the rate constant on the ionic strength:
³
´
log =log 0 − 2 + 2 − ‡ 2
12
(11.62)
with ‡ = + .
Inserting ‡ = + ,weobtain
log =log 0 − ¡ 2 + 2 − ( + ) 2¢ 12
(11.63)
y
log 0
=+2 12
(11.64)
I
Conclusions:
• If and havethesamesign,then will increase with increasing , because
the repulsion is reduced by the shielding effect of the ion cloud.
• If and have opposite sign, then will decrease with increasing , because
the attraction is reduced by the shielding effect of the ion cloud.
I
Figure 11.8: Kinetic salt effect for reactions of ions in solution.

11.5 Reactions of ions in solutions 240
11.6 References
Eigen 1954a M. Eigen, Disc. Faraday Soc. 17, 194 (1954).
Eigen 1954b M.EigenZ.Physik.Chem.NF1, 176 (1954).

11.6 References 241
12. Photochemical reactions
I
Primary photochemical processes:
• Absorption of a photon (depends on transition dipole moment ¯¯ ¯¯2),
• Intramolecular processes:
— fluorescence (≡ spontaneous emission of a photon, FL),
— stimulated emission (SE), (⇒ Einstein coefficients)
— chemical reaction in the excited state (R),
— intramolecular vibrational (energy) redistribution (IVR),
— internal conversion (IC),
— intersystem crossing (ISC),
— phosphorescence (PHOS).
• Intermolecular (i.e., collisional) processes:
— electronic deactivation (quenching, Q),
∗ may affect fluorescence as well as phosphorescence,
— vibrational energy transfer (VET),
— rotational energy transfer (RET),
— collision-induced intersystem crossing (CIISC).
12.1 Basic radiative processes in a two-level system; Einstein coefficients
I
Figure 12.1: Absorption, stimulated emission, and spontaneous emission.
2
= − 1
= 12 () 1 − 21 () 2 − 21 2 (12.1)

12.2 Fluorescence quenching (Stern-Volmer equation) 242
12.2 Fluorescence quenching (Stern-Volmer equation)
I
Kinetic model:
A + → A ∗ 1
A ∗ → A +
A ∗ +Q → A + Q
(12.2)
If any other radiationless processes can be neglected, we find
[A ∗ ]
= 1 [A]
+ [Q]
(12.3)
I Fluorescence intensity I 0 in the absence of the quencher Q:
0 ∝ [A ∗ ]
= 1 [A] (12.4)
I Fluorescence intensity I in the presence of the quencher Q:
∝ [A ∗ ]
=
1 [A]
+ [Q]
(12.5)
I
Stern-Volmer equation:
y
0
= 1 [A]
1 [A]
+ [Q]
= + [Q]
=1+ [Q] (12.6)
0
=1+ 0 [Q] (12.7)
0 =( ) −1 is the radiative lifetime.
I Conclusion: Aplotof 0 vs. [Q] should give a straight line with intercept 1 and
slope 0 [Q].

12.3 The radiative lifetime 0 and the fluorescence quantum yield Φ 243
12.3 The radiative lifetime τ 0 and the fluorescence quantum yield Φ
I Determination of the radiative lifetime τ 0 : Assuming that there are no competing
nonradiative processes, we can determine the radiative lifetime 0 by
(1) time-resolved measurements of the (exponential) fluorescence decay after pulsed
excitation,
(2) high-resolution measurement of the natural (Lorentzian) linewidth according to
∆ ∆ = ~ (12.8)
y
0 =
1
2 ∆ 0
(12.9)
(3) evaluation from the molar absorption coefficient (Lambert-Beer) using the relationship
between the Einstein coefficients 21 and 21 , 51
21 = 83
3 21 (12.10)
(4) calculate the value of ¯¯ ¯¯2 (⇒ 12 21 21 )fromfirst principles.
I Fluorescence quantum yield: Apart from quenching, the fluorescence is reduced by
other radiationless processes (IC, ISC, R, . . . ). The experimentally measured fluorescence
lifetime is therefore smaller (often much smaller) than the radiative lifetime 0 .
The ratio is the fluorescence quantum yield:
Φ =
+ + + + [Q] + = 0
(12.11)
51 The actual procedure, which involves integration over the absorption band, corrects for the influence
of the refraction index of the solvent, etc., is called the Strickler-Berg analysis.

12.4 Radiationless processes in photoexcited molecules 244
12.4 Radiationless processes in photoexcited molecules
I
Figure 12.2: Jablonski Diagram.
Elementary Photochemical Processes: Jablonski Diagram
Energy
RXN
(10 -14 .. 10 -6 s)
S 2
T 2
IC
S 2 -S 1 Abs.
T 1
T 2 -T 1 Abs.
IVR
IC
VET* S 1
ISC
(10 -9 ..10 -3 s)
S 0 VET* (10 -11 s)
IVR (10 -13 ..10 -9 s)
Absorption
(10 -18 s)
Fluorescence
(10 -9 ..10 -8 s)
Phosphorescence
(10 -3 ..10 -6 s)
*also known as VR
Physical Chemistry III: Chemical Kinetics • © F. Temps, IPC Kiel 253
12.4.1 The Born-Oppenheimer approximation and its breakdown
a) The full Hamiltonian
We consider a non-moving, non-rotating molecule in the laboratory framework. The
molecule is described by the Schrödinger equation (SE)
ˆ(r R) = (r R) (12.12)
or ³
ˆ − ´
(r R) =0 (12.13)
with being the energy eigenvalue associated with the wavefunction (r R) as function
of the electronic (e) coordinates r and the nuclear (N) coordinates R.
The Hamilton operator ˆ appearingintheSEcanbewrittenas
ˆ = ˆ + ˆ = ˆ (r)+ ˆ (R)+ (r R) (12.14)
where
ˆ = − ~2
2
X
=1
∇ 2 (12.15)
ˆ = − ~2 X
∇ 2 (12.16)
2
=1

12.4 Radiationless processes in photoexcited molecules 245
and
with
(r R) = (R)+ (r R)+ (r) (12.17)
=+ 2
4 0
X
X
0 0 =1
= − 2
4 0
X
=1
X
=1
=+ 2 X X
4 0
0
0
0 0 0
=1
(12.18)
(12.19)
1
(12.20)
Electronic spin, spin-orbit interaction, and nuclear spins are neglected here. Further,
we have left aside the so-called electronic mass and electronic nuclear cross polarization
terms, which appear in the case of a moving molecule as result of the separation of the
center-of-mass.
b) Adiabatic separation of nuclear and electronic motion
We now want to separate the slow motion of the nuclei from the the fast motion of
the electrons. This separation can be made because the fast electrons can be assumed
to very rapidly follow the slow nuclei. The kinetic energy of the nuclei is assumed to be
small compared to the electronic energy.
Thus, for every nuclear configuration R, there is an electronic wavefunction el
(r R)
belonging to the electronic state |i specified by quantum number . This el
(r R)
depends on the nuclear coordinates R, but only very little on the nuclear velocities.
We say that the electrons adiabatically follow the periodic vibrational motion of the
nuclei. This approximation is therefore called adiabatic approximation.
I Ansatz for Ĥ:
with
ˆ 0 = ˆ + (r R) =− ~2
2
ˆ = ˆ 0 + ˆ 0 (12.21)
X
=1
∇ 2 + (r R) (12.22)
ˆ 0 = ˆ = − ~2 X
∇ 2 (12.23)
2
=1
I The zero-order Hamiltonian Ĥ 0 :
• The zero-order Hamiltonian ˆ 0 depends only on the electron kinetic energy operator
and the potential energy and describes the rigid molecule (all R are
fixed!) via the unperturbed SE
ˆ 0 el (r; R) = (0) (R) el (r; R) (12.24)

12.4 Radiationless processes in photoexcited molecules 246
• The zero-order electronic wave functions el
(r; R) describe the electrons and
el
their distribution
¯¯ (r; R)¯¯2 for the case of fixed nuclei in the electronic state
|i withtheelectronicenergy el
.
• The el
(r; R) depend on R only as a parameter, not as a variable (we do not
differentiate or integrate with respect to the in the zero-order SE).
• The el
(r; R) form a complete orthonormal system.
• The eigenvalues (0) (R) of the electronic SE are what we have come to know as
the molecular potential energy functions (or hypersurfaces) (R) of the molecule
with all nuclei at rest.
I The perturbation Ĥ 0 :
• The perturbation ˆ 0 is the nuclear kinetic energy operator ˆ .
c) Solution of the complete SE
• To solve the complete SE
³
ˆ 0 + ˆ 0 − ´
(r R) =0 (12.25)
we use the ansatz
(r R) = X (R) · el (r; R) (12.26)
with the expansion coefficients = (R) depending only on R but not on r.
• Inserting the ansatz 12.26 into the complete SE,
³
ˆ 0 + ˆ 0 − ´ X
(R) el
(r; R) =0 (12.27)
and doing some algebraic manipulation, we arrive at a
• system of coupled equations for the electronic wavefunctions el (r; R) for the
electronic state el
(r; R) and the nuclear wavefunctions (R):
and
ˆ 0 el
(r; R) =
(0)
(R) el
(r; R) (12.28)
ˆ 0 (R)+ X (R) (R) = ¡ − (0)
(R) ¢ (R) (12.29)
with the coupling coefficients

12.4 Radiationless processes in photoexcited molecules 247
(R) =
Z
∗ (r; R) ˆ 0 (r; R) r
r
−~ 2 Z
r
∗ (r; R) X
1
(r; R) r
(12.30)
• We interprete these Eqs. as follows:
(1)
(0) (R) can be interpreted as the zero-order potential functions of the th
zero-order electronic state that is given by the solution of the electronic SE
(Eq. 12.24) for fixed nuclei R.
(2) Without the (R), Eq. 12.29 describes the kinetic energy of the nuclei
in the potential (0) (R). 52 Eq. 12.29 is thus simply the vibrational SE for
the electronic potential
(0) (R).
(3) The coefficients (R) defined by Eq. 12.30 are coupling terms that connect
(i.e., mix) the electronic states and . They describe how the
different electronic states and are coupled by the nuclear motion (the
two terms on the RHS resulting from the nuclear kinetic energy operator
ˆ ;seeEq.12.30).
d) Born-Oppenheimer approximation
The Born-Oppenheimer (BO) approximation completely neglects the coupling coefficients
. The complete SE is therefore reduced to two uncoupled equations, which
we denote as electronic SE
ˆ 0 el
(r) =
(0)
(R) el
(r) (12.31)
and nuclear (vibrational) SE
h i
ˆ +
(0) (R) = (R) (12.32)
which describe, respectively, the electronic wavefunction el
for fixed nuclear coordinates
R in the electronic state el
and the set of nuclear wavefunctions (R) for the energy
state of the nuclei in the electronic state el
.
e) Breakdown of the Born-Oppenheimer approximnation
When two (or more) electronic states el
(R), el
(R) become nearly isoenergetic or,
at a crossing, even degenerate, the coupling matrix element (R) may no longer be
neglected, because through the (R), thenuclear motion leads to a mixing between
the different electronic BO-states.
52 We used to denote these potential energy functions (≡ potential energy hypersurfaces) as (R).

12.4 Radiationless processes in photoexcited molecules 248
I
This breakdown of the BO approximnation leads to non-adiabatic coupling of
electronic states:
• For a diatomic molecule, the coupling leads to an avoided crossing of the two
electronic potential curves.
• For polyatomic molecules, the coupling can lead to a conical intersection (CI)
between the two electronic potential energy hypersurfaces.
As will be outlined in the following, the existence of conical intersections between different
excited electronic potential energy hypersurfaces and between an excited and
the ground electronic potential energy hypersurface has dramatic consequences for the
dynamics of photochemical reactions.
The recognition since about the years 2000 - 2005 that conical intersections in polyatomic
molecules are the rule rather than the exception has indeed revolutionized our
understanding of photochemical reactions.
12.4.2 Avoided crossings of two potential curves of a diatomic molecule
I
I
Illustration of an avoided crossing using two diabatic model potential functions
+ coupling term:
1 () = 1 [1 − exp [− 1 ( − 1 )]] 2 (12.33)
2 () = 2 + exp [− 2 ( − 2 )] (12.34)
12 () = 1 [1 + tanh [− 2 ( − 12 )]] (12.35)
12 () is just a convenient model function with the physically correct behavior
12 () → 0 for →∞.
Figure 12.3: Diabatic model potential curves for a diatomic molecule.
Avoided Crossing of Two Potentials of a Diatomic Molecule
Diabatic model potential curves: V1
( R) 0
H
0 V2
( R)
1exp ( )
2
V R D
R R
1 1 1 1
exp ( )
V R D R R
2 2 2 2
Physical Chemistry III: Chemical Kinetics • © F. Temps, IPC Kiel 254

12.4 Radiationless processes in photoexcited molecules 249
I
Figure 12.4: Avoided crossing of two potential curves of a diatomic molecule.
Avoided Crossing of Two Potentials of a Diatomic Molecule
Adiabatic potential curves: V1
( R) W( R)
H
W( R) V2( R)
V
V1
( R) W( R)
R
Eigenvalues
WR ( ) V2( R)
Physical Chemistry III: Chemical Kinetics • © F. Temps, IPC Kiel 255
I
Figure 12.5: Avoided crossing of two potential curves of a diatomic molecule. At the
avoided crossing the characters of the adiabatic potential curves switch.
Avoided Crossing of Two Potentials of a Diatomic Molecule
V E W V V V V
1 2 1 2 1 2
0 E
W
with: ;
W V2
E
2 2
2
V ( R) ( R) ( R) W( R)
"Avoided Crossing" with
Splitting of 2 |W( R x ) |
Physical Chemistry III: Chemical Kinetics • © F. Temps, IPC Kiel 256

12.4 Radiationless processes in photoexcited molecules 250
I
Figure 12.6: Conical intersection of potential energy surfaces of a polyatomic molecule.
Conical Intersections of Potential Surfaces in Polyatomic Molecules
V E W V V V V
1 2 1 2 1 2
0 E
W
with: ;
W V2
E
2 2
V
S 2
Conical Intersections
Diatomic Molecules: s = 1
2
V ( R) ( R) ( R) W( R)
S 1
S 0
Q 1
Q 2
Polyatomic Molecules: s = 3N-6
V ( Q , Q ) ( Q , Q )
1 2 1 2
( Q , Q ) W( Q , Q )
1 2 1 2
2
Physical Chemistry III: Chemical Kinetics • © F. Temps, IPC Kiel 257
12.4.3 Quantum mechanical treatment of a coupled 2×2 system
We can easily verify the “avoided crossing” of two coupled potential energy curves as
follows:
I
Schrödinger equation:
( − ) |i =0 (12.36)
I
Hamilton-Operator:
= (0) + (1) (12.37)
I
Ansatz:
|i = 1 | 1 i + 2 | 2 i (12.38)
with orthonormal basis vectors that are eigenvectors of (0) , i.e.,
½
­ ® 1
| = =
0
for =
for 6=
(12.39)
and the zero-order energies
(0) | 1 i = (0)
1 | 1 i (12.40)
(0) | 2 i = (0)
2 | 2 i (12.41)

12.4 Radiationless processes in photoexcited molecules 251
I Solution of the Schrödinger equation: Insertion of the ansatz (Eq. 12.38) into the
SE and multiplication from the left by either h 1 | or by h 2 | gives the two equations:
1 h 1 | ( − ) | 1 i + 2 h 1 | ( − ) | 2 i =0 (12.42)
1 h 2 | ( − ) | 1 i + 2 h 2 | ( − ) | 2 i =0 (12.43)
This is a system of coupled linear equations (“secular equations”):
with the “matrix elements” (integrals over all r resp. R):
Specifically, we have
1 ( 11 − )+ 2 12 =0 (12.44)
1 21 + 2 ( 22 − ) =0 (12.45)
= h | | i (12.46)
= h | (0) + (1) | i (12.47)
= h | (0) | i + h | (1) | i (12.48)
11 = h 1 | (0) | 1 i = (0)
1 (12.49)
22 = h 2 | (0) | 2 i = (0)
2 (12.50)
and
12 = h 1 | (1) | 2 i = h 2 | (1) | 1 i = 21 (12.51)
Therefore, we obtain the secular equations as follows:
I
Secular equations:
³ ´
1 (0)
1 − + 2 12 =0 (12.52)
³ ´
1 21 + 2 (0)
2 − =0 (12.53)
I Secular determinant: A non-trivial solution for the secular equations requires that
(0)
1 − 12
¯ 21 (0)
=0 (12.54)
2 − ¯
I
Eigenvalues:
0=
³ ´³
(0)
1 −
= (0)
1 (0)
2 −
´
(0)
2 − − | 12 | 2 (12.55)
³ ´
(0)
1 + (0)
2 + 2 − | 12 | 2 (12.56)
with
because (1) is Hermitian.
| 12 | 2 = 12 21 (12.57)

12.4 Radiationless processes in photoexcited molecules 252
With some rewriting 53 ,weobtain
± = 1 ³ ´
(0)
1 + (0)
2 ± 1 r ³
(0)
1 + (0)
´2
(0)
2 − 4 1 (0)
2 +4| 12 | 2 (12.58)
2
2
y
± = 1 2
³ ´
(0)
1 + (0)
2 ± 1 r ³
(0)
1 − (0)
´2
2 +4|12 | 2 (12.59)
2
I Shorthand form: Using the abbreviations
± (R) = ± (12.60)
1 (R) = (0)
1 (12.61)
2 (R) = (0)
2 (12.62)
2 (R) =| 12 | 2 (12.63)
Σ (R) = 1 2 ( 1 (R)+ 2 (R)) (12.64)
∆ (R) = 1 2 ( 1 (R) − 2 (R)) (12.65)
we can rewrite the solution for the eigenvalues + and − in a compact form as
± (R) =Σ (R) ±
q
(∆ (R)) 2 +( (R)) 2 (12.66)
12.4.4 Radiationless processes in polyatomic molecules: Conical intersections
I “Traditional” view on the cis-trans isomerization of C 2 H 4 :
• Reaction coordinate can be approximated by torsion angle (0 ≤ ≤ 180 ◦ ).
• Reaction proceeds via a perpendicular transition state ( =90 ◦ ), in which the C
atoms are connected by a single bond.
• Correlation diagram shows that the orbital of the trans-isomer transforms into
the ∗ orbital of the cis-isomer and vice versa.
• Crossing of the two diabatic states corresponds to an avoided crossing of the
adiabatic states.
53
³
(0)
1 + (0)
´2 ³
(0)
2 − 4 1 (0) 2 =
(0)
1
´2
+
³
³
=
(0)
2
(0)
1
´2
+2
(0)
´2
+
³
1 (0) 2 − 4 (0)
1 (0) 2
(0)
2
´2
− 2
(0)
1 (0) 2 =
³
(0)
1 − (0)
´2
2

12.4 Radiationless processes in photoexcited molecules 253
• The reaction is thus thermally forbidden because of the high energy barrier at the
avoided crossing between the trans and cis isomers on the left and right.
• The reaction is photochemically allowed because of the barrierless adiabatic correlation
between the ∗ states on the left and right.
I
Figure 12.7: Cis-trans isomerization of C 2 H 4 : Transformation of the and ∗ orbitals.
Photochemical cis-trans isomerization of C 2 H 4
Traditional picture: Diabatic potentials
diabatic crossing
Topics in Chemical Kinetics and Dynamics • © F. Temps, IPC Kiel 35
I
Figure 12.8: Cis-trans isomerization of C 2 H 4 : HMO picture with an avoided crossing.
Photochemical cis-trans isomerization of C 2 H 4
Traditional picture: Avoided crossing
avoided crossing
Topics in Chemical Kinetics and Dynamics • © F. Temps, IPC Kiel 36

12.4 Radiationless processes in photoexcited molecules 254
I
Figure 12.9: Internal coordinates relevant for the cis-trans isomerization of C 2 H 4 .The
out-of-plane bending angle plays a role as coupling mode leading to a CI; the angles
and play a role in the isomerization to H 3 CCH (also by a CI).
Photochemical cis-trans isomerization of C 2 H 4
Important internal coordinates:
Torsion
Out-of-plane bend
Barbatti et al., 2004
Physical Chemistry III: Chemical Kinetics • © F. Temps, IPC Kiel 258
I Figure 12.10: Diabatic model PES’s for C 2 H 4 .
Photochemical cis-trans isomerization of C 2 H 4
Diabatic PES:
(s1)-dimensional
crossing seem
Physical Chemistry III: Chemical Kinetics • © F. Temps, IPC Kiel 259

12.4 Radiationless processes in photoexcited molecules 255
I Figure 12.11: Adiabatic model PES’s for C 2 H 4 with the conical intersection at =90 ◦ .
Photochemical cis-trans isomerization of C 2 H 4
Adiabatic PES:
conical intersection
(s-2) dimensional
Physical Chemistry III: Chemical Kinetics • © F. Temps, IPC Kiel 260
I
Figure 12.12: Wave packet motion in a complex photo-induced reaction.
Dynamics of Photo-Induced Reactions: Photochemical Funnels
Conical Intersection =
"Photochemical Funnel"
Energy
Q b
S 0
S 1
S 0
Reaction Coordinate
Physical Chemistry III: Chemical Kinetics • © F. Temps, IPC Kiel 262

12.4 Radiationless processes in photoexcited molecules 256
13. Atmospheric chemistry*
(skipped for lack of time in the semester)

13 Atmospheric chemistry* 257
14. Combustion chemistry*
(skipped for lack of time in the semester)

14 Combustion chemistry* 258
15. Astrochemistry*
(skipped for lack of time in the semester)

15 Astrochemistry* 259
16. Energy transfer processes*
(See lecture in M.Sc. course.)

16 Energy transfer processes* 260
17. Reaction dynamics*
(See lecture in M.Sc. course.)

17 Reaction dynamics* 261
18. Electrochemical kinetics*
(skipped for lack of time in the semester)

Appendix B 262
Appendix A: Useful physicochemical constants
I Table A.1: List of useful physical constants (Cohen 1986).
physical constant
symbol value
Loschmidt (or Avogadro) number =60221367 × 10 23 mol −1
gas constant =8314511 J mol −1 K −1
Boltzmann constant =1380658 × 10 −23 JK −1
electron charge =160217733 × 10 −19 C
Faraday constant = 96485309 C mol −1
speed of light =299792458 × 10 8 ms −1
Planck’s constant =66260755 × 10 −34 Js
~ ~ =10547266 × 10 −34 Js
electron mass =91093897 × 10 −31 kg
proton mass =16726231 × 10 −27 kg
neutron mass =16749286 × 10 −27 kg
atomic mass unit =16605402 × 10 −27 kg
Rydberg constant e e =10973731534 × 10 7 m −1
Bohr radius 0 0 =529177249 × 10 −11 m
electric field constant 0 0 =1( 0 2 )
0 =8854187817 × 10 −12 Fm −1
magnetic field constant 0 0 =4 × 10 −7 NA −2
0 =125664 × 10 −6 NA −2
I
References:
Cohen 1986 E. R. Cohen and B. N. Taylor, The 1986 Adjustment of the Fundamental
Constants, CODATA Bull. 63, 1 (1986). An updated list is contained in every
August issue of Physics Today.

Appendix B 263
Appendix B: The Marquardt-Levenberg non-linear least-squares
fitting algorithm
With the omnipresence of the computer, the Marquardt-Levenberg algorithm (Bevington
1992, Press 1992) has become an extremely important data analysis tool. It is generally
themethodofchoiceincaseswhereitisnotpossibletomakesimplelinearplotsto
determine rate constants.
I Problem: Experimentally, we measure the values of some quantity at the points ,
, . Supposewetake measurements.Nowsupposewewouldliketodescribeour
data points by some model. Towards these ends, we take a physically reasonable
model function
= ( ;( 1 ))
(B.1)
which we would like to fit to our data in such a way that () describes the data
points in “the best possible way”. The model function depends directly on the
variables , and it also depends parametrically on nonlinear parameters .In
kinetics, for example, is usually some concentration, e.g. , which depends on time
i.e., = (). The parameters may be the rate constants (or decay times )
and the initial concentration 0 . Our job is to adjust and optimize these parameters
by somehow “fitting” them. The method of choice is, of course, “least-squares fitting”:
We start by defining the so-called figure-of-merit or goodness-of-fit function 54
2 =
"
#
X [ − ( )] 2
=1
2
(B.2)
Here, the and the independent variables , are the measured quantities, the
are the experimental uncertainties of the ,and ( ) is the calculated value
of the model fit function at the point { } (the parameters 1 are
omitted here in ). What we have to do is to adjust the parameters in the model
function such that 2 reaches a minimum, i.e. we have to search the -dimensional
parameter space for the (global) minimum of 2 .
I Solution: The Marquardt-Levenberg algorithm is a method to determine this minimum
of 2 as function of the fit parameters in the model fit function.
The minimum of 2 as function of the fit parameters a is defined by the conditions
"
#
2
= X [ − ( )] 2
=0 (B.3)
2
=1
X
∙ ¸
[ − ( )] ( )
= −2
(B.4)
2
=1
Thus, taking the partial derivatives of 2 with respect to each of the parameters
,weobtainasetof coupled equations in the unknown parameters . These
equations are, in general, nonlinear in the .
54 Deutsch: Fehlerquadratsumme.

Appendix B 264
The Marquardt-Levenberg algorithm gives a recipy for finding the minimum of 2 using
a combination of the method of steepest descent (following the gradient of the hypersurface
defined by 2 as a function of the parameters 1 and, near the minimum
of 2 , a parabolic Taylor series expansion of the fitting function around the minimum).
At the end of a fit, we shall usually report the (± 2 standard deviations), the
standard deviation of the ’s, and the so-called reduced- 2 ,whichis 2 divided by the
number of degrees of freedom, = − , i.e.,
"
#
2 = 1 X [ − ( )] 2
(B.5)
−
2
=1
I Example 1: Exponential decay. Suppose we measure the time dependent concentration
() of a molecule in some quantum state, which is populated at some time 0
by a laser pulse and then decays monoexponentially. As a model fit function, we use
the expression
() = 1 exp [− 2 ( − 3 )] + 4 + 5
(B.6)
The parameters we are interested in are 2 (the decay rate coefficient) and 1 (the
initial amplitude or initial concentration). The parameter 3 just describes the trigger
delay time (= 0 ). We have also added the parameters 4 and 5 to describe a constant
background term (due to some offset of our experimental signal) and a linear drift in
this background term (perhaps due to some experimental instability, such as a gradual
temperature increase in the lab). Our figure of merit is then
" #
X
2 [ − ()] 2
=
(B.7)
=1
The (in general nonlinear) conditions for the minimum of 2 , from which we determine
the parameters ,are
" #
2
= X [ − ()] 2
X
∙ ¸
[ − ()] ()
= −2
=0
1 1
2
=1
2
=1 1
(B.8)
2
2
= (B.9)
(B.10)
Figure B.1 below shows a simple two-parameter fit (parameters 1 and 1) tosucha
signal.
2

Appendix B 265
I
Figure B.1: Exponential decay curve of a particular vibration-rotation state of the
CH 3 O radical resulting from the unimolecular dissociation reaction of the radical according
to CH 3 O → H+H 2 CO (Dertinger 1995). The small box is the output box
from a fit usingtheORIGINprogram.
I Example 2: Multiexponential decay. In femtosecond spectroscopy, we often observe
ultrafast multiexponential decays of laser-excited molecules. The laser prepares
an excited wavepacket, which usually does not decay single-exponentially. Further, we
need to take into account the final duration of the pump laser pulse (by deconvolution,
or by forward convolution).
• Model function to be fitted to measured data:
() = X exp (− )
(B.11)
with adjustable parameters and .
• Instrument response function (IRF): Often represented by a Gaussian centered at
time 0 Ã !
1
() = √ exp − ( − 0) 2
(B.12)
IRF 2 2 2 IRF
with width parameter IRF related to the full width at half maximum (FWHM)
of the IRF by
FWHM = √ 8ln2≈ 2355 IRF
(B.13)

Appendix C 266
• Convolution of the molecular intensity () and () gives the signal function
() =
Z
+∞
−∞
( 0 ) ( − 0 ) 0 =
where ⊗ denotes the convolution.
Z
+∞
−∞
( − 0 ) ( 0 ) 0 = () ⊗ ()
(B.14)
• Resulting model function to be fitted to the data:
() = 1 X
∙ 1 2 IRF
exp − ( − ¸ ∙ µ ¸
0)
( − 0 ) − 2 IRF
1+erf √ +
2
2 2
2IRF
(B.15)
where erf () is the error function and is a simple constant background term (can
be replaced by background + drift + ).
I
Figure B.2: Excited-state relaxation dynamics of the adenine dinucleotide after UV
photoexcitation.
I
References:
Bevington 1992 P. R. Bevington, D. K. Robinson, Data Reduction and Error Analysis
for the Physical Sciences, McGraw-Hill, Boston, 1992.
Dertinger 1995 S. Dertinger, A. Geers, J. Kappert, F. Temps, J. W. Wiebrecht,
Rotation-Vibration State Resolved Unimolecular Dynamics of Highly Vibrationally
Excited CH 3 O( 2 ): III. State Specific Dissociation Rates from Spectroscopic Line
Profiles and Time Resolved Measurements, Faraday Discuss. Roy. Soc. 102, 31
(1995).
Press 1992 W. H. Press, S. A. Teukolsky, W. T. Vetterling, B. P. Flannery, Numerical
Recipes in Fortran, Cambridge University Press, Cambridge, 1992. Versions are
also available for C and Pascal.

Appendix C 267
Appendix C: Solution of inhomogeneous differential equations
C.1 General method
An inhomogeneous DE is a DE of the type
0 + () × = ()
(C.1)
In order to solve Eq. C.1, we first find a solution of the homogeneous DE
0 + () × =0
(C.2)
and then determine a particular solution of the inhomogeneous DE:
• Solution of the homogeneous DE by separation of variables:
0 + () × =0
0 = − () ×
(C.3)
(C.4)
y
ln
= − () (C.5)
= × − ()
(C.6)
• Determination of a particular solution of the inhomogeneoues DE by variation of
constant:
→ ()
(C.7)
0 () =
(C.8)
Insertion of () and 0 () into Eq. C.1 and integration gives
=
(C.9)
The general solution of Eq. C.1 is then given by
= +
(C.10)

Appendix C 268
C.2 Application to two consecutive first-order reactions
Consider the reaction system
A 1
−→ B
B 2
−→ C
(C.11)
(C.12)
which is described by the following rate equations
[A]
[B]
[C]
= − 1 [A] (C.13)
=+ 1 [A] − 2 [B]
(C.14)
=+ 2 [B]
(C.15)
The solution for Eq. C.13 is
[A] = [A] 0
− 1
(C.16)
Thus we have to find the solution of the inhomogeneous DE for [B] (Eq. C.14)
[B]
+ 2 [B] = 1 [A] 0
− 1
(C.17)
• Solution of the homogeneous DE by separation of variables:
y
[B]
= − 2 [B] (C.18)
[B]
= × − 2
(C.19)
• Determination of a particular solution by variation of constant:
y
[B]
= ()
[B]
= () × − 2
= ()
× − 2 − () × 2 − 2
(C.20)
(C.21)
(C.22)
Insertion of [B]
and [B]
into Eq. C.17 gives
()
which simplifies to
× − 2 − () × 2 − 2 + () × 2 − 2 = 1 [A] 0
− 1
()
= 1 [A] 0
( 2− 1 )
This DE is easily integrated to obtain ():
(C.23)
(C.24)

Appendix C 269
— If 1 6= 2 we obtain
and thus
() = 1 [A] 0
2 − 1
( 2− 1 )
(C.25)
[B]
= () − 2
= 1 [A] 0
2 − 1
( 2− 1 ) − 2
= 1 [A] 0
2 − 1
− 1
(C.26)
(C.27)
(C.28)
— If 1 = 2 we obtain
()
= 1 [A] 0
( 2− 1 )
(C.29)
= 1 [A] 0
(C.30)
y
y
() = 1 [A] 0
(C.31)
[B]
= () − 2
= 1 [A] 0
− 2
(C.32)
(C.33)
• General solution:
y
Initial value condition at =0:
[B] = [B]
+[B]
(C.34)
⎧
⎪⎨ × − 2 + 1 [A] 0
− 1
if 1 6= 2
[B] =
2 − 1
(C.35)
⎪⎩
× − 2 + 1 [A] 0
− 2
if 1 = 2
[B ( =0)]=0 (C.36)
y
⎧
⎪⎨ − 1 [A] 0
if 1 6= 2
=
2 − 1
(C.37)
⎪⎩
0 if 1 = 2
Final solutions for [B ()]:
⎧
1 [A] ⎪⎨ 0
¡
− 1 − ¢ − 2
if 1 6= 2
[B ()] =
2 − 1
(C.38)
⎪⎩
1 [A] 0
− 2
if 1 = 2

Appendix C 270
C.3 Application to parallel and consecutive first-order reactions
Consider the reaction system
A 1
−→ P
A 1
−→ B 2
−→ P
(C.39)
(C.40)
which is described by the following rate equations
[A]
[B]
[P]
= − ( 1 + 1 )[A]=− 1 [A] (C.41)
=+ 1 [A] − 2 [B]
(C.42)
=+ 1 [A] + 2 [B]
(C.43)
with 1 =( 1 + 1 ). 55
The solution for Eq. C.41 is
[A] = [A] 0
−( 1+ 1 ) =[A] 0
− 1
(C.44)
Thus we have to find the solution of the inhomogeneous DE for [B] (Eq. C.42)
[B]
+ 2 [B] = + 1 [A]
0
− 1
using the above standard procedure:
(C.45)
• Solution of the homogeneous DE by separation of variables:
y
[B]
= − 2 [B] (C.46)
[B]
= × − 2
(C.47)
• Determination of a particular solution by variation of constant:
y
[B]
= ()
[B]
= () × − 2
= ()
× − 2 − () × 2 − 2
(C.48)
(C.49)
(C.50)
55 An example is the radiationless deactivation of a ∗ electronically excited molecule A directly to
the ground state P or via an intermediatate optically dark ∗ state B, which decays to the ground
statemoreslowly.

Appendix C 271
Insertion of [B]
and [B]
into Eq. C.45 gives
()
× − 2 − () 2 − 2 + 2 () − 2 = 1 [A] 0
− 1
(C.51)
The terms with ± () 2 − 2 cancel, so that
()
× − 2 = 1 [A]
0
− 1
which we rewrite in order to solve () as
()
= 1 [A] 0
− 1 × + 2
or
()
= 1 [A]
0
( 2− 1 )
This DE is easily integrated to obtain ():
(C.52)
(C.53)
(C.54)
— If 2 6= 1 =( 1 + 1 ) we obtain
() = 1 [A] 0
2 − 1
( 2− 1 )
(C.55)
and thus
[B]
= () − 2
= 1 [A] 0
2 − 1
( 2− 1 ) − 2
= 1 [A] 0
2 − 1
− 1
(C.56)
(C.57)
(C.58)
— The case 2 = 1 does not interest us here, as the ∗ state B should be a
longer-lived one.
• General solution for 2 6= 1 :
y
Initial value condition at =0:
[B] = [B]
+[B]
[B] = × − 2 + 1 [A] 0
2 − 1
− 1
(C.59)
(C.60)
[B ( =0)]=0 (C.61)
y
= − 1 [A] 0
2 − 1
(C.62)

Appendix D 272
Final solutions for [B ()]:
i.e.
[B ()] = − 1 [A] 0
2 − 1
× − 2 + 1 [A] 0
2 − 1
− 1
= 1 [A] 0
2 − 1
− 1 − 1 [A] 0
2 − 1
× − 2
[B ()] = 1 [A] 0
1 − 2
× ¡ − 2 − − 1 ¢
(C.63)
(C.64)
(C.65)
• In the time-resolved fluorescence measurements, we observe
() ∝ [A ()] + [B ()]
(C.66)
with some unknown factor accounting for the (lower) transition probability of
state B, i.e., the fluorescence-time profile () is
() =[A] 0
− 1 + 1 [A] 0
1 − 2
× ¡ − 2 − − 1 ¢
=[A] 0
− 1 + 1 [A] 0
1 − 2
× − 2 − 1 [A] 0
1 − 2
× − 1
= µ
1 [A] 0
× − 2 + [A]
1 − 0
−
1 [A] 0
× − 1
2 1 − 2
= 2 × − 2 + 1 × − 1
(C.67)
(C.68)
(C.69)
(C.70)
This is the same result as would be obtained by a fit toasumoftwoexponentials
with the rate constants 1 = 1 + 1 and 2 , i.e., time constants
1 =( 1 + 1 ) −1 and 2 = 2 .

Appendix D 273
Appendix D: Matrix methods
This appendix has been taken from the “Introduction to Molecular Spectroscopy” script,
not all subsections apply to chemical kinetics.
D.1 Definition
I Matrices: A matrix is a rectangular, × dimensional array of numbers :
A =
⎛
⎜
⎝
⎞
11 12 13 1
21 22 23 2
⎟
. . . . ⎠
1 2 3
(D.1)
The are called the matrix elements.
We will need to consider only × dimensional square matrices, which we can think
of as a collection of column vectors a :
⎛
⎞
11 12 1
A = ⎜ 21 22 2
⎟
⎝ . . . ⎠
(D.2)
1 2
I Notation: Matrices will be denoted by bold symbols:
A
(D.3)
I Matrix manipulation by computer programs: Matrix manipulations are carried
out efficiently using computer programs (see, e.g., Press1992) or with symbolic algebra
programs. 56
56 The most commmon symbolic math programs are MathCad, MuPad, Maple, andMathematica.
MathCad and Mathematica are available in the PC lab.

Appendix D 274
D.2 Special matrices and matrix operations
I Identity matrix: I = 1
I = 1 =
⎛
⎜
⎝
10 0
01 0
. . .
00 1
⎞
⎟
⎠
(D.4)
We frequently use the Kronecker :
=
(D.5)
where
½
1 for =
=
0 for 6=
(D.6)
I
Diagonal matrices:
⎛
⎜
⎝
⎞
11 0
0 22 0
⎟
. . . . ⎠
0 0
(D.7)
I
Block diagonal matrices:
⎛
⎞
11 12 0
⎜ 21 22 0
⎟
⎝ . . . . ⎠
0 0
(D.8)
I Complex conjugate (c.c.) of a matrix: A ∗
To obtain the complex conjugate A ∗ of the matrix A, change to −:
( ∗ )
=( ) ∗ (D.9)
I Transpose of a matrix:
To obtain the transpose A of the matrix A, interchange rows and colums:
¡
¢ =( ) (D.10)

Appendix D 275
I Hermitian conjugate (“transpose conjugate”) of a matrix: A †
To obtain the Hermitian conjugate (or “transpose conjugate” or “adjoint”) A † of the
matrix A, take the complex conjugate and transpose:
† = ¡ ∗¢ = ¡ ¢ ∗
(D.11)
where ¡
† ¢ =( ) ∗ (D.12)
I Hermitian matrices: AmatrixA is Hermitian, if
A † =(A ∗ ) = A
(D.13)
i.e., ¡
† ¢ =( ) ∗ = (D.14)
I Inverse of a matrix: A −1
AmatrixA −1 is the inverse of the original matrix A, if
AA −1 = A −1 A = I
(D.15)
I Orthogonal matrices: AmatrixA is orthogonal, iftheinverseofA equals its transpose:
A −1 = A
(D.16)
I Unitary matrices: AmatrixU is unitary, iftheinverseofU equals its transpose
conjugate:
U −1 = U † =(U ∗ )
(D.17)
y
U † U = I
(D.18)
I
Trace of a matrix:
tr (A) =
X
=1
(D.19)
I
Matrix addition:
C = A + B
(D.20)
with
= +
(D.21)

Appendix D 276
I
Multiplication by a scalar:
C = A
(D.22)
with
=
(D.23)
I
Matrix multiplication:
C = A · B = AB
(D.24)
with
=
X
(D.25)
=1
Example:
µ µ µ µ
12 56 1 × 5+2× 71× 6+2× 8 19 22
=
34 78 3 × 5+4× 73× 6+4× 8 43 50
(D.26)
Notes:
(1) Matrix multiplication is defined only if the number of columns of the first matrix
is identical to the number of rows of the second matrix.
(2)
AB6= BA
(D.27)
I Singular matrices: AmatrixA is called singular, if its determinant vanishes:
det (A) =|A| =0
(D.28)

Appendix D 277
D.3 Determinants
I Determinant of a matrix: A determinant of a matrix is, in general, a number which
is evaluated according to the rule
det (A)=|A| =
X
=1
(D.29)
The are called the co-factors which are obtained by omitting the ’th row and ’th
column (marked below in red) and multiplying by (−1) + :
11 1 1
. . .
=(−1)
¯¯¯¯¯¯¯¯¯¯¯
+ 1
. . .
1
¯¯¯¯¯¯¯¯¯¯¯
(D.30)
I
Evaluation of simple determinants:
¯
¯ = −
(D.31)
11 12 13
21 22 23 = (D.32)
¯ 31 32 33
¯¯¯¯¯¯
= 11 22 33 + 12 23 31 + 13 21 32
− ( 13 22 31 + 11 23 32 + 12 21 33 )
where we “extended” the original determinant according to the scheme
11 12 13 11 12
21 22 23 21 22
(D.33)
¯ 31 32 33
¯¯¯¯¯¯ 31 32

Appendix D 278
D.4 Coordinate transformations
I Multiplication of a vector by a matrix: x 0 = Ax
The multiplication of an -dimensional column vector x by an × matrix A gives a
new vector x 0 with components corresponding to those of x in a transformed coordinate
system.
I Example: Multiplication of a 2-dim vector
x =
µ
1
2
(D.34)
by a matrix
gives
µ
cos − sin
A =
sin cos
(D.35)
µ µ
cos − sin 1
Ax =
(D.36)
sin cos 2
µ
1 cos −
=
2 sin
(D.37)
1 sin + 2 cos
= x 0 (D.38)
I
Figure D.1: Rotation of a coordinate system.

Appendix D 279
I
The result is a rotation:
• Intheoriginalcoordinatesystem,thenewvectorx 0 corresponds to the result of
an anti-clockwise rotation of x around .
• Alternatively, we can say that x 0 isgiveninanewcoordinatesystemthatis
obtained by a clockwise rotation of the old coordinate system around .
I Rotation matrices and rotation operators: A is called a rotation matrix. Weshall
also call A a rotation operator.
I
Exercise D.1: Show that Eqs. D.36 - D.38 does describe an anti-clockwise rotation
of the vector x to the new vector x 0 and determine its coordinates using trigonometric
arguments. ¤

Appendix D 280
D.5 Systems of linear equations
I Solutions of linear equations (I): Cramer’s rule. Matrices and determinants can
be generally used for solving systems of linear equations of the type
11 1 + 12 2 + 13 3 + 1 = 1
21 1 + 22 2 + 23 3 + 2 = 2
=
(D.39)
1 1 + 2 2 + 3 3 +
=
These equations can be written in matrix form:
⎛
⎞ ⎛ ⎞ ⎛ ⎞
11 12 1 1 1
⎜ 21 22 2
⎟ ⎜ 2
⎟
⎝ . . . ⎠ ⎝ . ⎠ =
⎜ 2
⎟
⎝ . ⎠
1 2
or
Ax= c
(D.40)
(D.41)
Simple algebraic manipulations (see any math textbook) gives the expressions
|A| = |A |
(D.42)
where |A| is the determinant of the coefficient matrix A and |A | is the respective
determinant of |A| in which the th column is replaced by the 1 2 3 ,e.g.
|A 2 | =
⎛
⎞
11 1 1
⎜ 21 2 2
⎟
⎝ . . . ⎠
1
(D.43)
Solutions for the : From Eq. D.42 we obtain the non-trivial solutions for the
according to
= |A |
(D.44)
|A|
under the condition that the A matrix is not singular, i.e., the determinant of coefficients
does not vanish
|A| 6= 0
(D.45)
(The trivial and uninteresting solutions are 1 2 3 =0).
I Solutions of linear equations (II): In the following, we shall only consider the special
linear equations of the type
11 1 + 12 2 + 13 3 + 1 =0
21 1 + 22 2 + 23 3 + 2 =0
= .
1 1 + 2 2 + 3 3 + =0
(D.46)

Appendix D 281
or
Ax=0
(D.47)
The trivial and uninteresting solutions of this equation are 1 2 3 =0.
The interesting solutions require that the matrix of coefficients A is singular, i.e., the
determinant of coefficients has to vanish
|A| =0
(D.48)
This type of equations is encountered in eigenvalue problems.

Appendix D 282
D.6 Eigenvalue equations
Consider a set of linear equations as considered in the previous section which can be
writtenintheform
Ax= x
(D.49)
where
(1) A is an × -dimensional matrix,
(2) is a scalar constant, called an eigenvalue of A ( is any one of the set of
eigenvalues of A), and
(3) x is an -dimensional column vector (which can in general be complex), called
the eigenvector of A belonging to the particular eigenvalue.
Eq. D.49 is called an eigenvalue equation. It has the following property: The multiplication
of x by (and hence that of x by the matrix A) changes the length of x (by
the factor ), but not the direction.
I Eigenvalues and eigenvectors of the molecular Hamiltonian: The determination
of eigenvalues and eigenvectors of the molecular Hamiltonian H, i.e., the “energy
matrix”, or the matrix of the Hamilton operator b , is the most important problem in
spectroscopy (in general, it is the key problem!!).
I Secular equation and eigenvalues λ : Eq. D.49 can be rewritten as
(A − I) x = 0 (D.50)
In order not to obtain the trivial solutions 1 = 2 = 3 = =0, det (A − I) has
to vanish, i.e., det (A − I) =0. This is the so-called characteristic equation or
secular equation:
|A − I| =
¯
The secular equation has roots
which are called the eigenvalues.
11 − 12 1
21 22 − 2
. . .
=0 (D.51)
1 2 − ¯
{ 1 2 } (D.52)

Appendix D 283
I Eigenvectors x : For each eigenvalue , we can write an eigenvalue equation
In matrix notation, this becomes
Ax 1 = 1 x 1 (D.53)
Ax 2 = 2 x 2 (D.54)
. (D.55)
Ax = x (D.56)
with the diagonal eigenvalue matrix
Λ =
AX= X Λ
⎛
⎜
⎝
⎞
1 0 0
0 2 0
⎟
. . . ⎠
0 0
(D.57)
(D.58)
The different eigenvectors x are usually normalized to unity, i.e.,
|x| = ¡ x † x ¢ 12
=1
(D.59)
I Matrix diagonalization: The matrix of the eigenvectors X can be determined by
multiplying the equation
AX= X Λ
(D.60)
from the left with X −1 ,whichgives
X −1 AX= X −1 X Λ
(D.61)
i.e.,
y
X −1 AX= Λ
(D.62)
The determination of the eigenvector matrix is equivalent to finding a matrix X that
transforms the matrix A into diagonal form (Λ).
I
Example:
A =
µ
2 3
310
(D.63)
Eigenvalue equations:
Ax = x
(D.64)
or in matrix form:
AX= X Λ
(D.65)

Appendix E 284
Secular equation:
det (A − I) =
¯ 2 − 3
3 10− ¯
=(2− )(10− ) − 9=0
(D.66)
(D.67)
y
0=20− 2 − 10 + 2 − 9= 2 − 12 +11 (D.68)
12 =+ 12 2 ± s µ12
2
2
− 11 = 6 ± √ 36 − 11 (D.69)
=6± 5 (D.70)
y
Eigenvalues:
1 =1 (D.71)
2 =11 (D.72)
Eigenvectors: Inserting the eigenvalues one by one into the eigenvalue equation yields:
(1) µ µ µ µ
2 3 11 11 11
=
310 1 =11
21 21 21
y
µ
−3
1
(2) µ µ µ µ
2 3 12 12 12
=
310 2 =1
22 22 22
y µ
1
3
(D.73)
(D.74)
(D.75)
(D.76)
Unnormalized eigenvalue matrix:
µ
−31
1 3
(D.77)
Normalized eigenvalue matrix:
X =
⎛
⎜
⎝
−3 1
√ √
10 10
1
√
10
3
⎞
⎟
⎠
√
10
(D.78)

Appendix E 285
I Final note: In practice, matrix manipulations such as matrix inversion, the evaluation
of determinants, the determination of eigenvalues and eigenvectors, etc., arecarriedout
numerically by using computers.
I
References
Press 1992 W. H. Press, S. A. Teukolsky, W. T. Vetterling, B. P. Flannery, Numerical
Recipes in Fortran, Cambridge University Press, Cambridge, 1992. Versions are
also available for C and Pascal.

Appendix E 286
Appendix E: Laplace transforms
The concept of Laplace and inverse Laplace transforms is extremely useful in chemical
kinetics for two reasons:
(1) They connect microscopic molecular properties and statistically averaged quantities:
a) () ↔ ( ): Collision cross section vs. thermal rate constant for bimolecular
reactions (section ??),
b) () ↔ ( ): Specific rateconstantvs. thermal rate constant for unimolecular
reactions (section 8),
c) () ↔ ( ): Density of states vs. partition function in statistical rate
theories 8).
(2) They provide a convenient method for solving of differential equations (section
3.4.2).
I
Definition E.1: The Laplace transform L [ ()] of a function () is defined as the
integral
Z ∞
() =L [ ()] = () −
(E.1)
where
0
• is a real variable,
• () is a real function of the variable with the property () =0for 0,
• is a complex variable,
• () =L [ ()] is a function of the variable .
I
Definition E.2: The inverse Laplace transform L −1 [ ()] of the function () is
defined as the integral
() =L −1 [ ()] = 1 Z
2
+∞
−∞
()
(E.2)
where
• is an arbitrary real constant.

Appendix F 287
I
Notes:
• Since L −1 [ ()] recovers the original function (), the pair of functions ()
and () is said to form a Laplace pair.
• ThepropertiesoftheLaplaceandinverseLaplacetransformscanbederivedfrom
those of Fourier transforms (Zachmann 1972).
• Laplace transforms can be computed using symbolic algebra computer programs
(Maple, Mathematica, etc.). In favorable cases, it is also possible to obtain inverse
Laplace transforms by computer programs. However, it is often more convenient
to take the Laplace and inverse Laplace transforms from corresponding Tables. 57
I
Table E.1: Table of Laplace transforms.
() () =L [ ()] () () =L [ ()]
()
()
() = R ∞
0
() −
() − ( =0) 1
−
1
1
1
( − ) 2
1
2
sin
2 + 2
2 2
1
3
cos
2 + 2
−1
Γ ()
1
1 ¡
− ¢ 1
( − )
( − )( − )
Γ () is the Gamma function. For integer , Γ () =( − 1)! (see Appendix F).
I
References:
Houston 1996 P. L. Houston, Chemical Kinetics and Reaction Dynamics, McGraw-
Hill, Boston, 2001.
Steinfeld 1989 J. I. Steinfeld, J. S. Francisco, W. L. Haase, Chemical Kinetics and
Dynamics, Prentice Hall, Englewood Cliffs, 1989.
Zachmann 1972 H. G. Zachmann, Mathematik für Chemiker, Verlag Chemie, Weinheim,
1972.
57 Extensive Tables of Laplace transforms are given by (Steinfeld 1989).

Appendix F 288
Appendix F: The Gamma function Γ (x)
• If is a positive integer, the Gamma function is
Γ () =( − 1)!
(F.1)
• The Gamma function smoothly interpolates between all points ( ) given by
=( − 1)! for positive non-integer (cf. Fig. F.1).
• The Gamma function is defined for all complex numbers with a positive real
part as
Z ∞
Γ () = −1 −
(F.2)
0
This integral converges for complex numbers with a positive real part. It has poles
for complex numbers with a negative real part (cf. Fig. F.2).
I
Figure F.1: The Gamma function.

Appendix G 289
I
Figure F.2: For negative arguments the Gamma function has poles (at all negative
integers).

Appendix G 290
Appendix G: Ergebnisse der statistischen Thermodynamik
In diesem Kapitel sollen wichtige Ergebnisse der statistischen Thermodynamik zusammengefasst
werden. Im Vordergrund steht die Berechnung der thermodynamischen Zustandsfunktionen
und die Berechnung von chemischen Gleichgewichten aus molekularen
Eigenschaften. Wir benutzen die Ergebnisse der Quantenmechanik:
• Ein Molekül kann nur bestimmte Energiezustände einnehmen.
• Diese Energiezustände sind im Prinzip nicht kontinuierlich sondern diskret verteilt.
• Jeder Energiezustand hat ein bestimmtes statistisches Gewicht . Das statistische
Gewicht gibt an, wie oft ein Zustand bei einer bestimmten Energie vorkommt.
Beispiele:
=1 eindimensionaler harmonischer Oszillator
=2 isotroper zweidimensionaler harmonischer Oszillator
(z.B. Biegeschwingung von CO 2 )
=2 +1Rotation eines linearen Moleküls
(G.1)
• Die Ergebnisse der statischen Thermodynamik unterscheiden sich, je nachdem
ob das betrachtete System aus unterscheidbaren oder aus ununterscheidbaren
Teilchen besteht:
Beispiele für unterscheidbare Teilchen: Atome im Kristallgitter
Beispiele für ununterscheidbare Teilchen: 1.) Gasmoleküle
2.) Elektronen (Fermi-Statistik)
(G.2)
• Die anzuwendende Statistik hängt vom Spin der Teilchen ab:
Spin gerade: Bose-Einstein-Statistik
Spin ungerade: Fermi-Dirac-Statistik
(G.3)
Auf die Unterschiede zwischen der Bose-Einstein- und der Fermi-Dirac-Statistik
wird an anderer Stelle eingegangen (⇒ Vorlesung “Statistische Thermodynamik”).
• Im Grenzfall À gilt die Boltzmann-Statistik.
• Im Folgenden werden behandelt: Elektronische Anregung, Schwingungsanregung,
Rotationsanregung, Translationsanregung.
Die Translation erfordert abhängig vom Problem manchmal eine etwas andere
Behandlung (klassische statistische Thermodynamik bzw. Semiklassik). Quantenmechanik
und klassische Mechanik stimmen im Ergebnis für À überein.

Appendix G 291
G.1 Boltzmann-Verteilung
Die Besetzungswahrscheinlichkeit eines Zustands wird durch die Boltzmann-Verteilung
angegeben. Diese ergibt sich als die wahrscheinlichste Verteilung aus dem Boltzmann-
Ansatz für die Entropie
= ln
(G.4)
wird als die thermodynamische Wahrscheinlichkeit bezeichnet.
Die Boltzmann-Verteilung kann hergeleitet werden, wenn man entsprechend dem 2.
Hauptsatz den Maximalwert für sucht (Methode der Lagrange’schen Multiplikatoren).
Hier setzen wir die Boltzmann-Verteilung als gegeben voraus (z.B. als empirisches, experimentelles
Ergebnis):
I Boltzmann-Verteilung: Für diskrete Energiezustände ist die Besetzungswahrscheinlichkeit
eines Zustands proportional zu − :
=
∝ −
(G.5)
Bei Berücksichtigung einer möglichen Entartung von Zuständen (Entartungsfaktor bzw.
statistisches Gewicht ) und mit einer Proportionalitätskonstante 0 zur Normierung
gilt
=
= 0 × × −
(G.6)
Dabei ist die Zahl der Teilchen im Energiezustand (mit dem statistischen Gewicht
) und die Gesamtzahl der Teilchen.
Die Funktion gibt die Wahrscheinlichkeit der Besetzung des Zustands an. Die
Normierungskonstante 0 ergibt sich aus der Normierungsbedingung, dass die Gesamtwahrscheinlichkeit
zu 1 herauskommt:
P
= X = 0 × X − =1
(G.7)
y 0 1
= P . (G.8)
−
y
I
Normierte Boltzmann-Verteilungsfunktion:
=
=
−
P −
(G.9)
I Molekulare Zustandssumme: Die Summe P − im Nenner der o.a. Gleichung
wird als molekulare Zustandssumme bezeichnet:
= P −
(G.10)
Die Zustandssumme ist die zentrale Grösse der statistischen Thermodynamik. ist
von der Temperatur abhängig.

Appendix G 292
I
Anschauliche Bedeutung der molekularen Zustandssumme:
= Zahl der bei der Temperatur im Mittel besetzten Energiezustände eines Molekül.
I
Boltzmann-Verteilungsfunktion für kontinuierlich verteilte Energiezustände:
Für kontinuierlich verteilte Energiezustände ersetzt man durch () und erhält
analog eine kontinuierliche Besetzungswahrscheinlichkeit:
() = ()
= () × −
(G.11)
I Zustandsdichte: () = Zustandsdichte (=Anzahl der Zustände pro Energie-
Intervall):
() = ()
(G.12)

Appendix G 293
G.2 Mittelwerte
I Mittlere Energie ²: In der Thermodynamik interessieren Mittelwerte, z.B. die mittlere
Energie . Für Größen, die durch eine Verteilungsfunktion beschrieben werden,
erhält man den Mittelwert nach
=
P
P
=
P
P
=
P
= X P
= P
mit =
und P =1.
Übergang zu einer kontinuierlichen Verteilungsfunktion liefert
Z ∞
()
(G.13)
=
0
Z ∞
0
()
(G.14)
Diese Ausdrücke vereinfachen sich, wenn () auf 1 normiert ist.
I
Allgemeine Berechnung der Mittelwerts y einer Größe y(²):
X
=
X
(G.15)
bzw.
Z ∞
()()
=
0
Z ∞
()
(G.16)
0
I
Example G.1: Berechnung der mittleren Energie für ein 3-Niveausystem:
0 =0 1 = 1 2 =2 1
0 =3 2 =2 2 =1
= X =6
(G.17)
(G.18)

Appendix G 294
Ergebnis:
=
1 X
= X
= X
(G.19)
= 1 6 (3 0 +2 1 +1 2 ) (G.20)
= 1 6 (3 0 +2 1 +2 1 ) (G.21)
= 1 6 (0 + 4 1) (G.22)
= 4 6 1 (G.23)
= 2 3 1
(G.24)
¤
I Mittlere Energie eines Moleküls: Bei Gültigkeit der Boltzmann-Verteilung kann die
mittlere Energie eines Moleküle aus der molekularen Zustandssumme berechnet werden:
=
=
1 X
= X
P
−
P
−
= X
(G.25)
(G.26)
y
=
P
−
= 2
= 2
= 2
= 2
X
X
2 −
|× 2
2
(G.27)
(G.28)
¡
−
|(durch differenzieren zeigen) (G.29)
X
−
(G.30)
= 2 ln
(G.31)
(G.32)

Appendix G 295
G.3 Anwendung auf ein Zweiniveau-System
I
Chemisch relevante Beispiele für Zweiniveau-Systeme:
• Paramagnetische Atome u. Moleküle
• NMR-Spektroskopie: Atomkerne mit Kernspin. Typischer Frequenzbereich:
100 MHz.
• Molekül mit cis-trans-Isomeren oder 2 Konformeren
I
Figure G.1: Zwei-Niveau-System: Energiezustände
The Two Level System: Energy Levels
g
g
Zustandssumme:
Q
i
0 / kT 1
/ kT
1e
1e
1e
i
/ kT
i
g e
1
/ kT
Besetzungsverhältnis:
0/
kT
0 g0e
N
N
N
N
Q
1
/ kT
1 g1e
N
N
Q
1
/ kT
1 g1e
0
/ kT
0 ge 0
e
1
/ kT
Physical Chemistry III: Chemical Kinetics • © F. Temps, IPC Kiel 173
I
Besetzungsverhältnis:
mit
1
= 1 − 1
(G.33)
= 0 − 0 + 1 − 1
=1+ − 1
(G.34)
(G.35)
(1) Grenzfall für → 0:
1
→∞y − 1 → −∞ =0
y Bei → 0 ist nur der tiefste Energiezustand besetzt.
(G.36)

Appendix G 296
(2) Grenzfall für →∞:
1
→ 0 y − 1 → −0 =1
(G.37)
y Bei →∞ergibt sich eine Gleichbesetzung der Energiezustände. Dies ist die
maximale Besetzung im oberen Zustand!
• Eine Besetzungsinversion ( 1 0 ) würde formal eine negative Temperatur
erfordern (→ Laser).
I
Figure G.2: Zwei-Niveau-System: Boltzmann-Besetzung der Zustände (rot: unterer
Zustand, blau: oberer Zustand).
The Two Level System: Boltzmann Distribution
N i
N
1
= 1000 J/mol
0.75
0.5
0.25
0
0
250
500
750
1000
T /(K)
Physical Chemistry III: Chemical Kinetics • © F. Temps, IPC Kiel 174
I
Mittlere thermische Energie:
(1) 1 =125kJ mol (entsprechend =1000cm −1 ; z.B. C-C-Schwingung im IR
oder Energieabstand zwischen cis-trans-Isomeren):
½
1
= − 1 8 × 10 =
−3 @ = 300K
(G.38)
0 023 @ =1000K
(2) 1 =004 kJ mol (entsprechend Energieabstand zwischen Kernspinzuständen bei
100 MHz NMR):
1
0
= − 1 =0999984 @ =300K
(G.39)
Reihenentwicklung:
−1 =1− 1
+ ≈ 1 − 1
(G.40)
y sehr kleiner Besetzungsunterschied y Wunsch der NMR-Spektroskopiker zu
immer größeren Frequenzen (Rekord heute 950 MHz).

Appendix G 297
I
Figure G.3: Zwei-Niveau-System: Mittlere thermische Energie.
The Two Level System: Energy
E
1000
= 1000 J/mol
750
500
250
0
0
250
500
750
1000
T /(K)
Physical Chemistry III: Chemical Kinetics • © F. Temps, IPC Kiel 175
I
Spezifische Wärme:
=
(G.41)
I
Figure G.4: Zwei-Niveau-System: Spezifische Wärme.
The Two Level System: Specific Heat
c v
5
= 1000 J/mol
4
3
2
1
0
0
250
500
750
1000
T /(K)
Physical Chemistry III: Chemical Kinetics • © F. Temps, IPC Kiel 176

Appendix G 298
G.4 Mikrokanonische und makrokanonische Ensembles
Die statistische Thermodynamik macht Aussagen über statistische Ensembles
(Ansammlungen von Systemen, Molekülen). Wir betrachten verschiedene Ensembles
von miteinander wechselwirkenden Molekülen.
I Der Begriff des Ensembles: Wir betrachten ein abgeschlossenes System bei konstantem
Volumen, konstanter Zusammensetzung und konstanter Temperatur (bzw. bestimmter
Energie (s.u.)). Ein Ensemble erhalten wir, wenn wir uns N Replikationen
dieser Systeme denken, die wir unter Einhaltung bestimmter Regeln erhalten (Regel =
Canon). Alle Systeme in dem Ensemble sollen im thermischen Gleichgewicht miteinander
(sie dürfen zwischen einander Energie austauschen).
Es werden verschiedene Sorten von Ensembles unterschieden:
• Mikrokanonisches Ensemble: Ensemble von Systemen (z.B. Atomen,
Molekülen) mit konstanter Teilchenzahl, konstantem Volumen, konstanter Energie.
• Makrokanonisches Ensemble (kurz kanonisches Ensemble genannt): Ensemble
von Systemen (z.B. Atomen, Molekülen) mit konstanter Teilchenzahl, konstantem
Volumen, konstanter Temperatur.
• Großkanonisches Ensemble: Ensemble von offenen Systemen (z.B. Teilchen,
Molekülen) mit konstantem Volumen, konstanter Temperatur, die zwischeneinander
Teilchen austauschen können.
I Systemzustandssumme (= kanonische Zustandssumme): Molekulare Zustandssumme:
= P alle Molekülzustände −
(G.42)
Systemzustandssumme (= kanonische Zustandssumme):
= P alle Systemzustände −
(G.43)
Innere Energie des Systems:
= 2 ln
(G.44)
Bei mehreren Teilchensorten (A, B, C, . . . ): Zustandsummen sind multiplikativ,
da die Energien additiv sind, y
= × × ×
(G.45)

Appendix G 299
• Systemzustandssumme für unterscheidbare Teilchen der Sorte (z.B.
Kristall):
Y
= =
(G.46)
y für 1mol:
= 2 ln
2
ln
=
(G.47)
• Systemzustandssumme für nicht unterscheidbare Teilchen der Sorte (z.B.
Gas):
y für 1mol ( = ):
=
!
(G.48)
= 2 ln
= 2
ln
!
(G.49)

Appendix G 300
G.5 Statistische Interpretation der thermodynamischen Zustandsgrößen
I
Innere Energie:
= 2 ln
(G.50)
I
Helmholtz-Energie:
= −
(G.51)
y
y
µ
=
µ
2
= −
2
−
−
µ
−
= − ln
= −
2
= − ln
(G.54)
= − ln
(G.52)
(G.53)
(G.55)
I
Entropie:
= −
µ
= −
= (G.56)
I
Druck:
= −
µ
µ ln
=
= (G.57)

Appendix G 301
G.6 Chemisches Gleichgewicht
y
bzw.
+ À
=exp
µ− ∆
ª
∆ ª = − ln
(G.58)
(G.59)
(G.60)
I Gleichgewichtskonstante: Wir berechnen zunächst
= − ln ,
(G.61)
=
indem wir
!
ln − anwenden:
etc.. . . einsetzen und die Stirling’sche Näherung ln ! =
= − [ln +ln +ln ]
"
#
= − ln
! +ln
! +ln
!
= − [( ln − ln !)
+( ln − ln !)
+( ln − ln !)]
= − [( ln − ln + )
+( ln − ln + )
+( ln − ln + )]
(G.62)
(G.63)
(G.64)
(G.65)
I Gleichgewichtsbedingung: Im chemischen Gleichgewicht muss gelten
µ
=0 (G.66)
y
0= [( ln − ln + )
+( ln − ln + )
+( ln − ln + )]
(G.67)
µ
1
= ln − − ln +1
µ
1
+ ln − − ln +1
µ
1
− ln − − ln +1
(G.68)
=(ln − ln )+(ln − ln ) − (ln − ln ) (G.69)
=ln
− ln
(G.70)

Appendix G 302
y
=
(G.71)
Übergang auf Teilchendichten und volumenbezogenen Zustandssummen:
y
( )
( )( ) = ( )
( )( )
=
( )
( )( ) = ∗
∗ ∗
(G.72)
(G.73)
Bezug auf die getrennten molekularen Nullpunktsenergien:
=
0
0 0
× −∆ 0
(G.74)
mit 0 = die auf das Volumen bezogene Molekülzustandssumme:
0 =
× × × (G.75)
I Endergebnis: Die Indices ∗ und 0 wurden hier zur Klarheit benutzt. Der Einfachheit
halber werden diese üblicherweise weggelassen.
y
=
× −∆ 0
(G.76)

Appendix G 303
G.7 Zustandssumme für elektronische Zustände
Die elektronische Zustandssumme muss unter Einschluss aller elektronischen Entartungen
explizit berechnet werden:
= X
−
(G.77)
Speziell sind zu berücksichtigen:
• Spinentartung (Spinmultiplizität ):
=2 +1
(G.78)
• elektronischer Bahndrehimpuls ( (Atome) bzw. Λ (lin. Moleküle) bzw. Symmetrierasse
oder für nichtlin. Moleküle):
• elektronische Feinstrukturkomponenten (Index im Atom-Termsymbol):
=2 +1
(G.79)
Nur niedrig liegende Elektronenzustände liefern einen Beitrag. Die Zustandsumme ist
explizit auszurechnen.

Appendix G 304
G.8 Zustandssumme für die Schwingungsbewegung
I
Harmonischer Oszillator:
= =0 1 2 (G.80)
I
Schwingungszustandssumme:
∞X
= − =
y
=0
∞X
=0
−·
mit =
(G.81)
= © 1+ − + −2 + −3 + ª mit = − 1 (G.82)
=1+ + 2 + 3 +
(G.83)
= 1 für 1
1 −
(G.84)
=
∙ µ
1
1 − = 1 − exp −
¸−1
−
(G.85)
I
s Oszillatoren:
=
Y
=1
∙ µ
1 − exp − ¸−1
(G.86)
I
typische Werte bei Zimmertemperatur:
≈ 1 10
(G.87)
I Grenzwert µ von Q für T →∞bzw. für ν → 0: Für →∞bzw. für → 0
kann exp −
entwickelt werden:
µ
exp −
≈ 1 −
+
(G.88)
y
1
1
= µ
1 − exp − = µ
1 − 1 − (G.89)
+
y
≈
(G.90)
Für klassische Oszillatoren:
≈
Y
=1
(G.91)

Appendix G 305
I
Abweichungen vom harmonischen Oszillator:
• Für anharmonische Oszillatoren ist die Zustandssumme am besten durch explizite
numerische Berechnung nach P ∞
=0 − zu ermitteln.
• Torsionsschwingungen können bei tiefen Temperaturen als harmonische Oszillatoren
beschrieben werden, bei hohen Temperaturen als freie 1D-Rotatoren. Im
Zwischenbereich muss die Zustandssumme ebenfalls explizit berechnet werden
(wobei die Energiezustände für den gehinderten 1D-Rotator zugrundegelegt werden
müssen).

Appendix G 306
G.9 Zustandssumme für die Rotationsbewegung
I
lineare Moleküle:
= ~2
( +1) =0 1 2 (G.92)
2
~ 2
2 =
(G.93)
I
Rotationszustandssumme:
(2) = 1
= 1
= 1
∞X
−
(G.94)
=0
∞X
(2 +1) −~2 (+1)2
=0
Z ∞
(G.95)
(2 +1) −~2 (+1)2 (G.96)
=0
y
(2) = 2
~ 2 =
(G.97)
I
Symmetriezahl:
(G.98)
I Rotationszustandssumme für nichtlineare Moleküle: ist nur für symmetrische
Kreisel analytisch darstellbar, nicht für asymmetrische Kreisel. Für nicht zu tiefe Temperaturen
ist
(3) = 12
µ 2
~ 2 32
( ) 12 = 12
()32 () −12 (G.99)
I
typische Werte bei Zimmertemperatur:
≈ 100 1000
(G.100)

Appendix G 307
G.10 Zustandssumme für die Translationsbewegung
I
Teilchen im 1D-Kasten:
³
= 2
´2
8 × =1 2 3 (G.101)
I
1D-Translationszustandssumme:
y
=
=
∞X
− =
Z ∞
1
−2 2 8 2
(G.102)
µ 2
2 12
× ¡ in cm −1¢ (G.103)
= µ 2
2 12
(G.104)
I
Teilchen im 3D-Kasten:
" µ 2 µ 2 µ # 2
= 2
8 ×
+ +
(G.105)
I
3DTranslationszustandssumme:
y
µ 32 2
=
×
2
=
¡
in cm
−3 ¢ (G.106)
µ 2
2 32
(G.107)
I
typische Werte:
≈ 10 24 ×
(G.108)
y
≈ 10 24 cm −3
(G.109)