Introduction to microkinetic modeling

Citation for published version (APA):

Filot, I. A. W. (2018). Introduction to microkinetic modeling. Eindhoven: Technische Universiteit Eindhoven.

Document status and date:

Published: 24/12/2018

Document Version:
Publisher’s PDF, also known as Version of Record (includes final page, issue and volume numbers)

Please check the document version of this publication:

• A submitted manuscript is the version of the article upon submission and before peer-review. There can be
important differences between the submitted version and the official published version of record. People
interested in the research are advised to contact the author for the final version of the publication, or visit the
DOI to the publisher's website.
• The final author version and the galley proof are versions of the publication after peer review.
• The final published version features the final layout of the paper including the volume, issue and page
numbers.
Link to publication

General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners
and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.

• Users may download and print one copy of any publication from the public portal for the purpose of private study or research.
• You may not further distribute the material or use it for any profit-making activity or commercial gain
• You may freely distribute the URL identifying the publication in the public portal.

If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license above, please
follow below link for the End User Agreement:
www.tue.nl/taverne

Take down policy

If you believe that this document breaches copyright please contact us at:
[email protected]
providing details and we will investigate your claim.

Download date: 18. . 2019

 I NTRODUCTION TO
M ICROKINETIC M ODELING

by
Dr. ir. I.A.W. Filot
This book was carefully produced. Nevertheless, author and publisher do not warrant the
information contained therein to be free of errors. Readers are advised to keep in mind
that statements, data, illustrations, procedural details or other items may inadvertently be
inaccurate.

Filot, I.A.W.
Introduction to Microkinetic Modeling
Eindhoven University of Technology

A catalogue record is available from the Eindhoven University of Technology Library

ISBN: 978-90-386-4520-9

Subject headings: microkinetic modeling, kinetics, langmuir-hinshelwood, physical chemistry,

statistical thermodynamics.

Copyright © 2017-2019 by Ivo Filot

All rights reserved
Version: 1.5.0

Git commit id: 4144ce8

Commit date: January 22, 2019
Compile date: January 22, 2019
 i

Preface

This book started out as two separate documents. One was a set of exercises for the
Advanced Thermodynamics and Catalysis course and the other was a method and theory
section at that time envisioned for my PhD thesis. Only a very small part of the material in
this book eventually made it into the thesis, as the whole would be much too elaborate!
Since I started out in 2010 as a teaching assistant for the Advanced Thermodynamics and
Catalysis course, I have been collecting useful questions to prepare the students for the final
examination. And as is well known, if you give students questions, they demand answers in
the form of detailed solutions. I believe in learning by example, so I was more than happy
to write these solutions down on paper and I have spent many evenings behind my laptop
typing the solutions in LATEX.
At the end of 2015, I was appointed as an assistant professor (tenure track). I was asked
to take over the Advanced Thermodynamics and Catalysis course from Prof. Emiel Hensen,
which I happily accepted. I changed the learning goals a bit to adopt my current research into
the educational track. As a result, I had to come up with new lecture materials and I decided
to bundle everything I made so far: the questions and solutions, the (envisioned) theory and
method section for my thesis and the notes I made when I followed the kinetics course when I
was a student. The result was a monster! My writing style had obviously changed significantly
over the years and there was no clear story. Thus I spent the following two years rewriting the
text.
In 2016, a draft version only containing a set of useful exercises and solutions were given
to the students to practice. In 2017, the first version of the book was ready and about one
hundred copies were printed and given to the second year Bachelor students as lesson material.
I found out that students are really marvelous proofreaders. They will point out every little
mistake, whether it is a spelling mistake or a miscalculation. In these two years, I received
about 500 corrections. The current version is yet another iteration further down the road and
my intention is to continuously update and improve the material in this book.
Within this book, I will elaborate on the connection between the smallest quantum scale
and the larger macroscopic scale, providing a theoretical framework on which many kinetic
studies in heterogeneous catalysis are based. I will focus on the underlying quantum chemical
mechanism leading to the observed overall rate laws in chemistry and process engineering.
This book is organized in several chapters. I will start by introducing general kinetics
and the equations governing the rate of change in chemical composition. Next, a statistical
approach is formulated to deduce macroscopic observables from the constituting quantum
chemical properties by means of averaging over many states. Energy, temperature and
pressure are defined as our important parameters and we show how these concepts are
connected with observables such as chemical composition in equilibrium as well as the
chemical kinetics leading to these equilibria. Finally, we explain how the above-mentioned
concepts can be used within the framework of microkinetic modeling in a predictive and
illustrative manner to gain insights in how real-life catalytic processes work.
All chapters have exercises to help you practice with the lesson material. Solutions are
provided in the appendix as well as more challenging exercises at the same difficulty as what
ii

can be expected on the final exam.

Since I am not a native English speaker, there is always the chance that I have used some
odd grammatical structures or made (despite the careful proofreading of many students and
myself ) spelling errors. Feedback is thus always very welcome and I invite you to deliver it by
e-mail to [email protected]. Any useful comment will result in your name being mentioned
in the acknowledgment section (of course, with your permission).
I sincerely hope this book is useful to you and provides you with a more in-depth under-
standing of the beautiful field of (micro-)kinetics!

Ivo Filot Eindhoven, 30th March 2018

Preface to version 1.5

With version 1.5, we have modified the book quite a bit. First of all, we got a lot of
constructive feedback on the lay-out of the book. It was hard to see where a question or
exercise would start or end. Evidently, the solutions to the exercises suffered from the same
problem. We have changed the lay-out in such a way that the exercises and questions are
clearly highlighted in the book, which hopefully makes it easier to find what you are looking
for. While we were at it, we also added tabs on the sides of each page so you can efficiently
find the chapter you are looking for.
Note that I have been saying we, instead of I. In the past three months, I have been
receiving a lot of help from Tom van den Berg, who has been working on Chapter 5 of the
book. In Chapter 5, a hands-on tutorial to use MKMCXX will be given. This was previously the
purpose of Chapter 4, yet it felt short in the sense that it did not provide sufficient explanation
in how to set-up such simulations from scratch. As such, Chapter 4 has been adapted to
explain the algorithm behind microkinetic simulations using a series of simple Python scripts
and Chapter 5 now deals to performing simulations using MKMCXX. Tom has also made
a series of exercises to help the student practice using MKMCXX. We hope that these are
beneficial to your understanding of the topic.
Finally, a lot of spelling, grammar, and other kinds of errors have been resolved on the
basis of the feedback of a lot of people. As is traditional, the list of students can be found on
the acknowledgment page. Enjoy the revision of the book and please keep sending in any
questions and comments you might have!

Ivo Filot Geldrop, 24th December 2018

 iii

Acknowledgements

This book was not possible without the help of some very talented people. First of all, I
need to mention Prof. Emiel Hensen and Prof. Hans Niemantsverdriet. These two excellent
teachers gave the Catalysis and Kinetics course (Reactiekinetiek en Katalyse) when I was a
young Bachelor student. I fondly remember the enthusiasm of Prof. Hensen and the very
pedagogically correct learning style of Prof. Niemantsverdriet. During my PhD studies, I
helped out as a teaching assistant for the kinetics course. In this period, I wrote many of the
solutions to the questions which are embedded in this reader. These solutions were used and
corrected by Remco Lancee, Alessandro Coumans and Lennart van Haandel. Often, Lennart
offered alternative solution strategies to the questions, which I found very helpful.
When I became the teacher for the kinetics course, I was lucky to have very dedicated
teaching assistants on my side as well. I wish to thank Bart Zijlstra, Robin Broos, Evert van
Noort and Jasper Bouwman for helping out with the supervised self-study lessons. Michel
van Etten not only helped out, but was also willing to host a couple of lectures when I was
unavailable, which is highly appreciated. I also should mention that Tobias Kimpel was always
available to step in when we needed an extra hand.
Last but not least, I wish to mention all of the students and colleagues who helped me in
any way with improving this reader, whether it was by mentioning errors, offering suggestions
or pointing out difficult parts:

• Class of 2014/2015: Marle Vleugels, Lisa Struik, Martin van Son.

• Class of 2015/2016: Jasper Bouwman, Camiel Steffanie, Remco Fijneman.

• Class of 2016/2017: Steven Kuijpers, Teun Nevels.

• Class of 2017/2018: Luuk Kempen, Stefan Zondag.

• Class of 2018/2019: Vera Crommentuijn, Jeroen Vennix, Cynthia Struijk, Stacey Fun,
Lisa Cattenstart, Frank van Hertrooij, Nico van Rijswijk, Danny van den Eertwegh, Job
Nijhuis.

• Colleagues: Kevin van Eeten, Jan Wiesfeld.

Do you miss your name in the list, but did you provide me with some feedback? Please let me
know!
 Table of Contents

1 Kinetics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.1 Introduction 1
1.2 The rates of reactions 2
1.2.1 Time-evaluation of a single elementary reaction step . . . . . . . . . . . . . . . . . . . . 4
1.2.2 Sequential elementary reaction steps . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

1.3 Chain reactions 9

1.4 Catalytic reactions 11
1.4.1 Langmuir adsorption isotherms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

1.5 Reaction mechanisms in catalysis 16

1.5.1 Potential energy diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
1.5.2 Rate-determining step and overall reaction rate . . . . . . . . . . . . . . . . . . . . . . 19
1.5.3 Reaction orders . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
1.5.4 Apparent activation energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

1.6 Questions and Exercises 28

1.6.1 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

2 Transition State Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

2.1 Introduction 39
2.2 Statistical method 40
2.3 Maxwell-Boltzmann distribution law 41
2.4 Partition functions of subsystems 46
2.5 Thermodynamic variables 46
2.6 Partition functions for molecules 49
2.6.1 Translations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
2.6.2 Vibrations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
2.6.3 Rotations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54

2.7 Collision theory 55

2.7.1 Maxwell-Boltzmann velocity distribution . . . . . . . . . . . . . . . . . . . . . . . . . . 55
2.7.2 Collision effectiveness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
2.7.3 Unimolecular reactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58

2.8 Transition state theory 59

2.8.1 Comparison between Transition State Theory and Collision theory . . . . . . . . . . . . . 62
2.8.2 Fundamental objection against collision theory . . . . . . . . . . . . . . . . . . . . . . 63
 ii TABLE OF CONTENTS

2.9 The potential energy surface 63

2.10 Dimensionality of equilibrium constants and reaction rate constants 65
2.11 Questions and Exercises 67
2.11.1 Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
2.11.2 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
2.12 Challenges 71

3 Elementary Reaction Steps . . . . . . . . . . . . . . . . . . . . . . . 75

3.1 Introduction 75
3.2 Adsorption 76
3.3 Desorption 78
3.4 Surface reactions 78
3.5 Hindered rotors 79
3.6 Final remarks 80
3.7 Questions and Exercises 81
3.7.1 Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
3.7.2 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81

4 Microkinetic Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
4.1 Introduction 83
4.2 Microkinetic modeling of a simple catalytic reaction 84
4.2.1 Unimolecular catalytic reaction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
4.2.2 Bimolecular catalytic reaction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
4.3 Reaction orders 94
4.4 Apparent activation energy 96
4.5 Degree of rate control 99
4.6 Degree of selectivity control 101
4.7 Comparison with Langmuir-Hinshelwood kinetics 101
4.8 Questions and Exercises 106
4.8.1 Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
4.8.2 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106

5 Performing microkinetics with MKMCXX . . . . . . . 107

5.1 Introduction 107

5.2 Installation 108
5.3 The basic input 108
5.3.1 Compounds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
5.3.2 Reactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
5.3.3 Settings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
5.3.4 Runs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
 TABLE OF CONTENTS iii

5.4 Kinetic analysis tools 114

5.4.1 Reaction order . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
5.4.2 Apparent activation energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
5.4.3 Degree of rate control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
5.5 Data interpretation 116
5.5.1 Operating temperature and production . . . . . . . . . . . . . . . . . . . . . . . . . 118
5.5.2 Reaction orders . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
5.5.3 Apparent activation energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
5.5.4 Degree of rate control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
5.6 Convergence 121
5.7 Numerical versus analytical 122
5.8 Exercises 123

A Fundamental Constants . . . . . . . . . . . . . . . . . . . . . . . . . 127

B Mathematics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129

B.1 Stirling Approximation 129

B.2 Gaussian integrals 130
B.3 Taylor expansion for the exponential function 130
B.4 Geometric series 131
B.5 Lagrange’s method of undetermined multipliers 131
B.6 l’Hôspital’s Rule 133
B.7 Matrix diagonalization 133
B.8 Sum of degree of rate control coefficients 134

C Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135

C.1 Solutions of Chapter 1 135

C.1.1 Solutions to exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
C.2 Solutions of Chapter 2 166
C.2.1 Solutions to questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
C.2.2 Solutions to exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
C.2.3 Solution to challenges . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
C.2.4 Challenge: Maxwell-Boltzmann distribution . . . . . . . . . . . . . . . . . . . . . . . 183
C.3 Solutions of Chapter 3 186
C.3.1 Solutions to questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186
C.3.2 Solutions to exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186
C.4 Solutions of Chapter 4 189
C.4.1 Solutions to questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
C.4.2 Solutions to exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190
C.5 Solutions of Chapter 5 192
C.5.1 Solutions to Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192
 iv TABLE OF CONTENTS

D Exam Practice Questions . . . . . . . . . . . . . . . . . . . . . . . . 195

D.1 Explanation 195

D.2 Kinetics 195
D.3 Statistical Thermodynamics 197
D.4 Collision theory 198
D.5 Transition state theory 199

E Exam Question Solutions . . . . . . . . . . . . . . . . . . . . . . . . 201

E.1 Introduction 201

E.2 Kinetics 201
E.3 Statistical Thermodynamics 204
E.4 Collision theory 207
E.5 Transition state theory 208

F MKMCXX . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211

F.1 Settings 211

F.2 Troubleshooting 211
C HAPTER 1


CHAPTER
KINETICS

Contents
1.1 Introduction 1
1.2 The rates of reactions 2
1.2.1 Time-evaluation of a single elementary reaction step . . . . . . . . . . . . . . . . . . . . 4
1.2.2 Sequential elementary reaction steps . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.3 Chain reactions 9
1.4 Catalytic reactions 11
1.4.1 Langmuir adsorption isotherms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.5 Reaction mechanisms in catalysis 16
1.5.1 Potential energy diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
1.5.2 Rate-determining step and overall reaction rate . . . . . . . . . . . . . . . . . . . . . . 19
1.5.3 Reaction orders . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
1.5.4 Apparent activation energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
1.6 Questions and Exercises 28
1.6.1 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

1.1 Introduction
Kinetics is part of the science of motion. It deals with the rate of chemical reactions and
treats chemistry from a dynamic viewpoint. The counterpart of kinetics is thermodynamics,
which deals with the static point of view on chemistry. Thermodynamics is interested in the
initial and final states of a system. Within kinetics, the mechanism whereby the system is
converted from one state to another and the time wherein this process occurs is described.
Thermodynamics deals with equilibria, whereas in kinetics equilibrium is a special situation
wherein the forward and backward rates are equal to each other. As such, kinetics can be seen
as a more fundamental science than thermodynamics. Unfortunately, the complexities that
occur within the kinetic treatment of chemistry are such that it is difficult to always apply
it with large accuracy and sometimes we find that for a more accurate treatment of certain
properties, thermodynamics gives the more accurate answer.
 C HAPTER 1
2 Chapter 1. Kinetics


Underlying both kinetics and thermodynamics are more fundamental theories such as
statistical mechanics and quantum chemistry. The quantum description of matter is only valid
at the smallest length and time scales and the description of large ensembles of particles over
extended periods of time require a statistical description. Hence, kinetics and thermodynamics
are the result of the merge between statistical mechanics and quantum chemistry.
In this chapter, we will start with a detailed description on kinetics with a strong focus on
the description of dynamic processes within heterogeneous catalysis. This chapter serves as an
overview and sets the stage for the contents of Chapter 2, where we zoom in on a fundamental
aspect within kinetics which is the elementary reaction step. In Chapter 1, the elementary
reaction step is considered as the fundamental building block within a kinetic mechanism,
and in Chapter 2, we elaborate on the quantum chemical and statistical foundations of the
elementary reaction step.
At the end of this Chapter, you should have a conceptual understanding of what chemoki-
netic networks are and how we can describe these in terms of elementary reaction steps. You
are able to construct overall rate expressions for typical chemical processes and derive such
expressions on the basis of a reaction mechanism. Finally, you will be able to express the
dependence of these rate expressions on important boundary conditions such as temperature
and pressure.

1.2 The rates of reactions

Consider a chemical reaction between molecules A and B that give products C and D according
to the following reaction equation

k+
−−
νa A + νb B )−*
− νc C + νd D. (1.1)
− k

Herein, νx is the stoichiometric coefficient of component X and k+/− are the forward and
backward rate constants. The rate of a chemical reaction can hence be defined as either the
rate of disappearance of the reactants or the rate of formation of the products in the following
fashion:

1 d[A] 1 d[B] 1 d[C] 1 d[D]

r=− =− = = (1.2)
νa dt νb dt νc d t νd dt

where [X] is the concentration of component X and r is the reaction rate. Expression 1.2
should only be used for so-called elementary reaction steps. A detailed discussion on what
exactly entails an elementary reaction step is given further on in this book1 , but for the time
being it is sufficient to know that an elementary reaction step is a chemical reaction in which
one or more chemical species react directly to form products in a single reaction step and
with a single transition state.
An alternative way to define such an elementary reaction step is given by

 
X 
Rj : νi Xi . (1.3)
 
i

1
A detailed description of an elementary reaction step is provided in section 2.9.
C HAPTER 1
1.2 The rates of reactions 3


When νi is negative, compound Xi is a reactant. On the other hand, when νi is positive,

compound Xi is a product of the reaction. Using the same notation, equation 1.1 is given by

 
X 
(1.4)

R1 : νi Xi = −νa A − νb B + νc C + νd D .
 
i

The general rate expression for an elementary reaction step is then defined as

1 d[Xi ]
rj = , for i = 1, 2, . . . , n (1.5)
νi dt

Note that the above considers the rate of a single chemical reaction. Often, we simply wish
to know the rate of chemical change of a particular compound. From the general definition,
the rate of change of a single compound in a single elementary reaction step is then given by

 
d[X]
= νx k+ [Xi ]νi − k− [Xi ]νi  .
Y Y
rx = (1.6)
dt
νi <0 νi >0

For example, the rate of change of compound A in equation 1.1 would be

d[A]  
rA = − = νA k+ [A]νA [B]νB − k− [C]νC [D]νD . (1.7)
dt

And in a similar fashion, for product C we would obtain the following rate expression
d[C]  
rC = = νC k+ [A]νA [B]νB − k− [C]νC [D]νD . (1.8)
dt

Clearly, the part between the parentheses of equations 1.7 and 1.8 stays the same for each
compound in the reaction. Hence, we can define the reaction rate of a compound within an
elementary reaction step as follows

rx = νx rj . (1.9)

This equation should not come as a surprise as it is basically a rewritten version of equation
1.5. Furthermore, from the above equations, it should be apparent that rj can be written as

 

r j = k + [Xi ]νi − k− [Xi ]νi  .

Y Y
(1.10)
νi <0 νi >0

Thus, we have formally defined how to construct the reaction rate expression of a single
elementary reaction step and of a compound inside such a step given the stoichiometric coef-
ficients and concentration of the compounds and the reaction rate constant of the elementary
reaction step. The thing that remains is a motivation why we use an expression such as 1.10
to calculate the rate of chemical change.
A chemical reaction between compounds A and B can only occur when A and B meet each
other in a chemical mixture. If we consider a tiny volume inside the mixture, the probability
that compound A is inside that volume is directly proportional to the number density of A
inside that volume. This number density is typically expressed as a concentration. In line
with this reasoning, the probability of A and B being in the same volume is proportional to
 C HAPTER 1
4 Chapter 1. Kinetics


the product of the concentration of A and B. Hence, we see in formula 1.10 the product of the
concentrations of all compounds on either side of the elementary reaction step.
This leaves us with a motivation for the reaction rate constant k. Whether or not a reaction
event occurs does not only depend on whether A and B find each other in a chemical mixture.
It is for instance also affected by the particular orientation of A and B with respect to each
other or the (kinetic) energy of the compounds. Hence, the term k reflects the chance that a
meeting of A and B results in a chemical reaction event.
For now, we simply use k as some given constant, but later in this book, we will pro-
vide a detailed discussion how the value of k can be calculated on the basis of statistical
thermodynamics and quantum chemistry.2

1.2.1 Time-evaluation of a single elementary reaction step

Equation 1.6 is an ordinary differential equation which can be integrated over time to obtain
an expression for the concentration of [X] as a function of time. For simple ordinary dif-
ferential equations, analytical solutions can be constructed. Here, we show an example
on the procedure. For more complex chemical systems, we have to resort to numerical
approximations using a procedure known as microkinetic modeling. This is the topic of
chapter 4 of this book.
For now, assume we have the following simple unimolecular elementary reaction step,
given by

k+
A−
)−
−*
− B. (1.11)
k−

Let us further assume that k+ >> k− . In other words, the reaction from A to B is
considered to be irreversible3 In that scenario, we can rewrite the above equation to

A → B. (1.12)

This gives us the following differential equations4 for compounds A and B

d[A]
rA = = −k[A] (1.13)
dt
d[B]
rB = = k[A] (1.14)
dt

To solve this system of ordinary differential equations, we need to have two initial values
for [A] and [B] at time t = 0. For simplicity, let us assume that [A]t=0 = 1 and [B]t=0 = 0.

2
See section 2.8 on 59 and further.
3
This is termed the irreversibility assumption and is denoted by a single arrow within the reaction equation. In some
situations, the reversibility of a reaction is taken into account but described by separate forward and backward reactions.
In the text we will make clear which of the above options is used.
4
Note that we have replaced k+ by k for simplificity.
C HAPTER 1
1.2 The rates of reactions 5


Thus we obtain the following expression for A

d[A]
= −k[A] (1.15)
dt
d[A]
= −k dt (1.16)
[A]
Z [A]t Z t
d[A]
= −k dt (1.17)
[A]0 [A] 0
ln[A]t − ln[A]0 = −kt (1.18)
!
[A]t
ln = −kt (1.19)
[A]0
[A]t = [A]0 exp (−kt) (1.20)
[A]t = exp (−kt) (1.21)

and similarly, the following expression for B

d[B]
= k[A] (1.22)
dt
d[B]
= k exp (−kt) (1.23)
dt
d[B] = k exp (−kt) dt (1.24)
Z [B]t Z t
d[B] = k exp (−kt) dt (1.25)
[B]0 0
exp (−kt) exp (0)
 
[B]t − [B]0 = k − + (1.26)
k k
[B]t = 1 − exp (−kt) . (1.27)

1.2.2 Sequential elementary reaction steps

The product of one elementary reaction step can be the reactant of another elementary reaction
step. From this, we can easily envision series of connected elementary reaction steps. For
example, consider the overall transformation from some particular reactant R to product P.

R→P (1.28)

that proceeds via the following elementary reaction steps

RA (1.29)
AB (1.30)
BC (1.31)
C  P. (1.32)

Note that all elementary reaction steps considered here are unimolecular. If we assume
that these reactions mainly operate in the forward direction (i.e. kf >> kb ), then we can
analytically derive the expression for the following overall reaction rate.
For each compound in our system, we can construct a differential equation from the rate
expressions. That is,
 C HAPTER 1
6 Chapter 1. Kinetics


d[R]
= −k0 [R] (1.33)
dt
d[A]
= k0 [R] − k0 [A] (1.34)
dt
d[B]
= k0 [A] − k0 [B] (1.35)
dt
d[C]
= k0 [B] − k0 [C] (1.36)
dt
d[P]
= k 0 [C ] (1.37)
dt
(1.38)

In a similar fashion as the previous section, we obtain a set of ordinary differential

equations, which is also called a system of ordinary differential equations. This system is
somewhat more complex than the previous system, but it can still be integrated analytically.
For simplicity, let us assume that the initial values are as follows: [R](t = 0) = 1 and all other
concentrations at t = 0 are zero.
Let us introduce some additional terminology. The set of all possible concentrations ci is
termed the phase space of the system. Here, this space is defined by

ci ∈ [0, 1] (1.39)

and

X
ci = 1 (1.40)
i

Integration in time of the differential equations is then called propagation in phase space.
The trajectory is the path in phase space corresponding to this time-integration.
The derivation of the analytical expression goes beyond the scope of this book, but the
interested reader may consult any engineering mathematics book.5 The methodology that we
have used to obtain the solution was by using the so-called Laplace transformation.[1] The
result is:

 
[R] = exp −k0 t (1.41)
 
[A] = k0 t · exp −k0 t (1.42)
1  
[B] = (k0 t)2 · exp −k0 t (1.43)
2
1  
[C] = (k0 t)3 · exp −k0 t (1.44)
6
  1 1

[P] = 1 − exp −k0 t 1 + k0 t + (k0 t)2 + (k0 t)3 (1.45)
2 6

The concentration as a function of time is depicted in Figure 1.1. In this graph, k0 = 1

s−1 . From this Figure, it can be seen that the concentration of B increases at a slower rate
than the concentration of A and that the concentration of C increases at a slower rate than the
concentration of B. This is of course logical. Each subsequent species in the kinetic network
has to overcome an additional barrier that slows the rate of formation. Despite these barriers,
C HAPTER 1
1.2 The rates of reactions 7


1.0 R
A
B
0.8 C
P
Concentration in mol/l

0.6

0.4

0.2

0.0
0 2 4 6 8 10
Time t in s

Figure 1.1: Concentration as a function of time. At t = 0, Rt=0 = 1. The system converges to the final result
where P = 1 at t → ∞.

the final product P is still formed, because the reactions were considered irreversible. The
driving force in this system is thus the strong negative enthalpy of the reaction.
Conclusively, we have seen how to construct a system of ordinary differential equations
from a set of sequential elementary reaction steps. By setting appropriate initial conditions,
we were able to integrate this system over time.
The above example also illustrates another important concept. In equation 1.6, we formu-
lated the rate of change for a single molecule in a single elementary reaction step. Here, we
have shown the situation wherein compound X takes part in multiple elementary reaction
steps. In such scenarios, the rate of change of compounds X is simply the sum of all the rates
of change of all elementary reaction steps wherein compound X takes part:

d[X] X
= νx,i ri , (1.46)
dt
i

where νX,i is the stoichiometric coefficient of compounds X in elementary reaction step

i and ri is the reaction rate of elementary reaction step i. Note that if compound X is not
involved in reaction step i, νX,i is equal to zero and hence ri does not contribute to the rate
of change of compound X.
Combining the above equation with equation 1.10 leads to the general expression for the
rate of change of compound X in a chemokinetic network:

  
d[X] X
+
Y
ν −
Y
ν
= νx,j k [Xi ] i − k [Xi ] i  . (1.47)

dt
j νi <0 νi >0
5
A good source is Advanced Engineering Mathematics from Erwin Kreyszig.
 C HAPTER 1
8 Chapter 1. Kinetics


The steady state approximation
The procedure shown in the previous section is a powerful approach to understand the time-
dependent behavior of a chemical network. The major drawback of this approach is that
when the system of ordinary differential equations cannot be solved analytically, we have to
resort to numerical approximations. Although the system used in the previous session could
be solved analytically, this tends to be more of an exception than a rule. Here, we wish to
introduce a powerful assumption, known as the steady state assumption, that helps us to
obtain analytical expressions for systems of ordinary differential equations. In turn, these
analytical expressions help us to evaluate the complex kinetic behavior.
Let us consider the following overall reaction6 , which relates to the decomposition of
ozone.

2O3 → 3O2 (1.48)

The above overall reaction is a chemokinetic network which can be represented by two
elementary reaction steps as shown below.

k1
O3 −−→ O2 + O (1.49)
k2
O + O3 −−→ 2O2 (1.50)

Let us for simplicity assume that both these reactions are irreversible, i.e. they only
proceed in the forward direction. The three differential equations that describe the rate of
change of all compounds in the network are given by

d[O3 ]
= − k1 [O3 ] − k2 [O][O3 ] (1.51)
dt
d[O]
= k1 [O3 ] − k2 [O][O3 ] (1.52)
dt
d[O2 ]
= k1 [O3 ] + 2k2 [O][O3 ] (1.53)
dt
We can solve the above system of ordinary differential equations numerically by providing
initial value conditions, but an alternative method to gain insight in the network is to construct
an overall rate expression. To obtain a simple analytical expression for the rate of production
of oxygen from ozone, we apply an assumption. The assumption we are going to apply is
the steady state approximation. This approximation assumes that the rate of change of one
particular compound is equal to zero. The motivation for this assumption is that the particular
compound for which the approximation is applied is much more reactive than the other
compounds in the network in such a way that its change in concentration over time is zero. It
makes sense to apply this assumption to O as an oxygen radical is known to be very reactive.
This leads to the following expression

d[O]
= k1 [O3 ] − k2 [O][O3 ] = 0 (1.54)
dt
Solving the above equation for [O] gives

k
[O] = 1 (1.55)
k2
6
An overall reaction is a general kinetic expression that conveys the reactivity of one or more elementary reaction
steps and reflects the stoichiometry of the overall process.
C HAPTER 1
1.3 Chain reactions 9


Inserting this result back into equation 1.53 results in

d[O2 ]
= 3k1 [O3 ]. (1.56)
dt

The above expression emphasizes another important point. We mentioned that equation
1.48 was an overall reaction. Had we treated it as an elementary reaction step, then we could
readily derive the following differential equation for the rate of change in O2 :

d[O2 ]
= 3k[O3 ]2 . (1.57)
dt

Clearly, equations 1.56 and 1.57 are inherently different and lead to significantly different
kinetics. This simple example illustrates a very important concept in kinetics. An overall
reaction only expresses the stoichiometry (i.e. the mole balance) between the reactants and
the products of a chemical process, but it cannot tell us a priori anything about the rate of
change at which the process takes place (i.e. the kinetics). In contrast, the set of elementary
reaction steps that constitute the process does not only allow us to investigate the kinetics of a
reaction, but it is also a detailed representation of the reaction mechanism.

1.3 Chain reactions

A chain reaction is a sequence of elementary reaction steps wherein a reactive product, for
instance a radical, causes additional elementary reaction steps to take place. An illustrative
example is the reaction between hydrogen and chlorine to form hydrochloric acid

H2 + Cl2 → 2HCl (1.58)

which proceeds by the following set of elementary reaction steps:

k1
Cl2 −−→ 2Cl · (1.59)
k2
Cl · + H2 −−→ HCl + H · (1.60)
k3
H · + Cl2 −−→ HCl + Cl · (1.61)
k4
2Cl · −−→ Cl2 (1.62)

Note that for this particular set, the reactions only occur in the forward direction. This
is either because the backward reaction is negligible compared to the forward one (which is
the case for the second and third reaction), or because we explicitly model the forward and
backward reaction as separate reactions (which is the case for the first and fourth reaction).
In chain reactions, we can differentiate between initiation, propagation and termination steps.
The first step, the formation of two chlorine radicals, is the initiation step. The second and
third elementary reaction steps are the propagation steps. The fourth step is the reverse of
the first step and is the termination step.
From the above four elementary reaction steps, we are able to derive an analytical expres-
sion for the rate of formation of hydrochloric acid by application of the previously proposed
steady state approximation. The reasoning stays the same: on the time scales where the
reaction takes place, we consider the concentration of the highly reactive compounds (i.e. the
radicals) to be constant in time, hence we assume
 C HAPTER 1
10 Chapter 1. Kinetics


d[Cl · ]
=0 (1.63)
dt
d[H · ]
= 0. (1.64)
dt

Given the above four elementary reaction steps, we obtain the following five differential
equations (i.e. one for each compound in the system of elementary reaction steps)

d[Cl2 ]
= − k1 [Cl2 ] − k3 [H · ][Cl2 ] + k4 [Cl · ]2 (1.65)
dt
d[H2 ]
= − k2 [Cl · ][H2 ] (1.66)
dt
d[HCl]
= k2 [Cl · ][H2 ] + k3 [H · ][Cl2 ] (1.67)
dt
d[Cl · ]
= 2k1 [Cl2 ] − k2 [Cl · ][H2 ] + k3 [H · ][Cl2 ] − 2k4 [Cl · ]2 = 0 (1.68)
dt
d[H · ]
= k2 [Cl · ][H2 ] − k3 [H · ][Cl2 ] = 0. (1.69)
dt

To find a rate expression for the change in concentration of HCl, we need to obtain an
equation for the concentration of the radicals (which is constant over time) and plug these
into the rate expression for HCl. The trick to solve this problem, is to realize that given the
steady-state approximation, we are allowed to add and/or subtract equations 1.68 and 1.69
from any of the above expressions since equations 1.68 and 1.69 equate to zero.
Summing 1.68 and 1.69 provides us with an expression for [Cl · ]:

s
k1
[Cl · ] = [Cl2 ] (1.70)
k4

Plugging equation 1.70 back into equation 1.68 provides us (after some algebra) the
following expression for H ·

q
k1
k2 k [H 2 ]
[H · ] = p4 (1.71)
k3 [Cl2 ]

Finally, inserting the result of equations 1.70 and 1.71 into equation 1.67 gives us the
following expression for the rate of change in HCl:

s
d[HCl] k1 1
= 2k2 [H ][Cl2 ] 2 (1.72)
dt k4 2

If 1.58 would have been an elementary reaction step, the rate expression would be

d[HCl]
= 2k[H2 ][Cl2 ]. (1.73)
dt

Hence, we encounter again a situation where the rate expression derived from the set of
elementary reaction steps is significantly different from the simple rate expression assuming
that the overall reaction is an elementary reaction step.
C HAPTER 1
1.4 Catalytic reactions 11


1.4 Catalytic reactions

A catalyst is a compound which is added to the reaction mixture that accelerates the reaction
without itself being consumed in the process. In turn, a reaction that uses a catalyst is termed
a catalytic reaction. The key concept of catalysis is thus that the reactants associate with the
catalytic material, undergo a chemical transformation and finally dissociate from the catalyst.
There are different kinds of catalysts, but in this reader, we will focus on heterogeneous
catalysts and the corresponding gas-solid kinetics.
A heterogeneous catalyst can be envisioned as a relatively large extended surface. This
surface is composed of active sites, which are local positions on which a molecule can adsorb,
react and desorb. The total number of active sites is considered constant and equal to N .
Furthermore, we assume for the time being that all sites are equivalent and each site can only
be occupied by a single species or be vacant. If an adsorbate X is adsorbed on a catalytic site,
this is denoted by X*. The fractional coverage of sites covered by X is in turn denoted by θX .
In summary, our model assumes the following about active sites:

• The total number of sites is constant.

• All sites are equivalent.

• A single site can only adsorb a single molecule or atom.

From here on, we can define elementary reaction steps which deal with a catalytic surface.
Let us consider the adsorption of reactant R on a catalytic site denoted by *

k1+
R+∗)
−−
−− R∗
* (1.74)
k1−

and the reaction in which R* can be converted to product P, which immediately desorbs
from the catalytic surface

k2+
R∗ −−→ P + ∗. (1.75)

In order to derive kinetic equations for this system, we have to consider the dimensionality
of the reaction rates. For some compounds, the rate of change will be expressed in the three-
dimensional space of the gases. For the species which are associated with the catalyst, their
rate of change pertains to the two-dimensional space of the catalyst surface. Thus, we have to
define the following macroscopic equation for our rates

d[R]
= N k1+ 1 − θR [R] − N k1− θR , (1.76)


−V
dt
where V is the volume of the gases, N is the total number of catalytic sites in the volume
V , θR is the fractional coverage of sites covered by R and 1 − θR the number of empty sites.

The latter can also be denoted by θ∗ .

For the rate of change for adsorbed R, we obtain the following expression

dθR  
= k1+ 1 − θR [R] − k1− + k2+ θR (1.77)


dt
and for the rate of change of product P

d[P ]
V = N k2+ θR (1.78)
dt
 C HAPTER 1
12 Chapter 1. Kinetics


To obtain a kinetic expression for the rate of production of P, we can again make use of
the steady state approximation. Here, we apply the approximation for the fractional coverage
of R on the surface θR :

dθR  
= k1+ 1 − θR [R] − k1− + k2+ θR = 0 (1.79)


dt
From this, we can directly establish a relation for θR as a function of the concentrations
of R:

k1+ [R]
k1− +k2+
θR = . (1.80)
k1+ [R]
1+
k1− +k2+
Finally, plugging this result back into equation 1.78 gives

k1+ [R]
d[P ] k1− +k2+
V = N k2+ . (1.81)
dt k1+ [R]
1+
k1− +k2+
In conclusion, we have seen how to model a reaction that uses a catalytic surface. The
novelty here was that we had to introduce additional terms into the kinetic equations to deal
with the dimensionality of the space wherein the species reside (i.e. three-dimensional for
the gas-phase species and two-dimensional for the adsorbates).

1.4.1 Langmuir adsorption isotherms

The adsorption of compounds on a catalytic surface is a pivotal step in any catalytic cycle, as
without adsorption, no use is made of the catalytic material. The relationship between the
surface coverage of a particular species and its corresponding gas-phase pressure at constant
temperature is known as a Langmuir adsorption isotherm. This isotherm is named after Irving
Langmuir, who studied the deterioration of tungsten filaments in incandescent light bulbs.
For this purpose, he constructed a detailed theoretical framework which he later used to build
a kinetic description of catalytic reactions. For his accomplishment in the field of catalysis, he
was awarded with the Nobel Prize in Chemistry in 1932.
Using the theoretical framework as devised by Irving Langmuir, we are here going to
explain different types of adsorption. The different types of adsorption are named after the
nature of adsorption, which are
• Direct or associative adsorption: A gas-phase species adsorbs directly on the surface
and retains its internal chemical bonding. A key example is the adsorption of N2 or CO.

• Dissociative adsorption: A gas-phase species adsorbs on the surface and simultaneously

dissociates. The adsorption of H2 on many transition metals directly leads to the
dissociation of the H2 molecule by which the adsorbed state are two separate hydrogen
atoms bonded to the surface.

• Competitive adsorption: Basically a form of adsorption wherein multiple species com-

pete for the same type of active site. A typical example is the competitive adsorption of
CO and H2 .7
In the next subsections, we will derive the Langmuir isotherm for these three types of
adsorption.
7
There is also a form of competition wherein adsorbates on adjacent active sites tend to repel (or attract) each other.
Such kind of lateral interactions are not meant here.
C HAPTER 1
1.4 Catalytic reactions 13


Direct adsorption
In the direct adsorption mechanism, a gas-phase species A adsorbs on the surface wherein its
molecular form stays intact. The corresponding elementary reaction step is

k+
A+∗−
)−
−−A∗.
* (1.82)
k−

This gives the following differential equation8 for the rate of change in the surface coverage
of A:

dθA
= pA k+ θ∗ − k− θA . (1.83)
dt

If we assume that the catalytic surface is composed of only one type of site, than we can
formulate the following mass balance for the fractional coverages

θA + θ∗ = 1. (1.84)

Applying equation 1.84 to equation 1.83 results in

dθA
= p A k + 1 − θA − k − θA . (1.85)


dt

Applying the steady state approximation to the above expression gives us an expression
for the fractional coverage as a function of the gas-phase pressure

k+ p
k− A KpA
θA = = (1.86)
k+ p
1+ k 1 + KpA
− A

This functional form is termed a Langmuir adsorption isotherm. Note that in the above
expression, we have used the equilibrium constant K , which is the ratio of the forward reaction
rate constant and the backward reaction rate constant as given by

k+
K= . (1.87)
k−

In Figure 1.2, several Langmuir adsorption isotherms are plotted using equation 1.86
with different values for the equilibrium constant K . From this Figure, we can see that
with increasing value for the equilibrium constant, the surface coverage is higher at a given
pressure.

8
Instead of the concentration of A in the gas phase as shown in the previous example, we here use the pressure as
for gas-solid interactions, the latter is used more often. Both approaches are of course equally valid, as the collision
chance scales linearly with both concentrations as well as pressure.
 C HAPTER 1
14 Chapter 1. Kinetics


1

0.8
Coverage θ

0.6

0.4

0.2

0
0 2 4 6 8 10
Pressure p in bar

Figure 1.2: Langmuir adsorption isotherms. The top line corresponds to an equilibrium constant of K=10 bar−1 ,
the middle one has an equilibrium constant of 1 bar−1 and the bottom line has an equilibrium constant of 0.1
bar−1 .

Dissociative adsorption
The adsorption of certain di- or polyatomic molecules results in the immediate dissociation
of these molecules upon adsorption on a catalytic surface. A typical example is the adsorption
of H2 , which for a broad range of transition metals gives dissociative adsorption.
The elementary reaction step for the dissociative adsorption of species A2 on a catalytic
surface is given by

k+
A2 + 2∗ )
−−
−− 2A∗
* (1.88)
k−

and the corresponding differential equation is

dθA
2
= p A k + 1 − θA − k − θA
2. (1.89)
dt

Note that we have the same molar balance as shown in the previous section by which we
can replace θ∗ with 1 − θA .
Applying the steady state approximation to this differential equation and solving for θA
gives the following Langmuir adsorption isotherm

q
KpA2
θA = q . (1.90)
1 + KpA2

In Figure 1.3, a comparison is shown between associative and dissociative adsorption.

Competitive adsorption
An interesting case occurs when two species compete for the same active sites. Consider the
situation where compounds A and B both adsorb associatively on the same catalytic surface
as given by the following two elementary reaction steps
C HAPTER 1
1.4 Catalytic reactions 15


0.8

Coverage θ 0.6

0.4

0.2

0
0 2 4 6 8 10
Pressure p in bar

Figure 1.3: Langmuir adsorption isotherms of associative versus dissociative adsorption. The dotted line cor-
responds to associative adsorption whereas the solid line represents dissociative adsorption. The equilibrium
constant K is the same for both isotherms.

A k+
A+∗−
)−
−*− A∗ (1.91)
−
kA
+
kB
−−
B+∗)−*
−B∗. (1.92)
−
kB

Applying the steady state approximation to both θA and θB gives the following expressions

θA = KA pA θ∗ (1.93)
θB = K B p B θ∗ (1.94)

To solve for θ∗ , we introduce a site balance which for this situation is given by

θA + θB + θ∗ = 1. (1.95)

Plugging equations 1.93 and 1.94 into equation 1.95 gives

KA pA θ∗ + KB pB θ∗ + θ∗ = 1 (1.96)

Collecting all terms in θ∗ gives

(1.97)


KA pA + KB pB + 1 θ∗ = 1

from which we can readily derive

1
θ∗ = . (1.98)
1 + KA pA + KB pB
 C HAPTER 1
16 Chapter 1. Kinetics


Coverage B (dimensionless)
10 1.05

0.90
8
0.75
Pressure of B (pB) in bar

6 0.60

4 0.45

0.30
2
0.15

0 0.00
0 2 4 6 8 10
Pressure of A (pA) in bar

Figure 1.4: Langmuir adsorption isotherms of competitive adsorption of A and B.

Plugging this expression back into 1.93 and 1.94 gives

K A pA
θA = (1.99)
1 + KA pA + KB pB
KB pB
θB = . (1.100)
1 + KA pA + KB pB

The above two equations are the two Langmuir adsorption isotherms for competitive
adsorption.
In Figure 1.4, a contour plot is given for competitive adsorption. The equilibrium con-
stants KA and KB are both set to 1 bar−1 . The competition between the two adsorbates
is clearly seen from the result that the coverage of B decreases with increasing pressure
of A. Furthermore, because the equilibrium constants for both adsorption isotherms are
equal to each other, the surface coverage of both components at elevated pressures equals
θA = θB = 12 .

1.5 Reaction mechanisms in catalysis

Quite often, one is interested in the net production of a particular compound in a chemical
reaction. In process engineering, it is common to consider part of the chemical process as a
black box and model the system (or chemical reactor) purely on the basis of the long-living
and stable compounds within the system. In other words, one neglects the intermediates or
radicals as these are very short-lived and would be difficult to measure experimentally.
Given these limitations, macroscopic reaction rates are employed to express the production
rates in terms of the concentration of the reactants. These production rates (or for that matter
reaction rates) are often so-called power laws and are based on the overall reaction. An
example of such an overall reaction is the oxidation of carbon monoxide over a car-exhaust
clean-up catalyst as given by the following overall reaction equation:
C HAPTER 1
1.5 Reaction mechanisms in catalysis 17


1
CO + O2 −−→ CO2 . (1.101)
2
A simple approach for expressing overall reaction rates is to set up a rate based on a power
law where the exponents in these power laws are set equal to the stoichiometric coefficients
in the reaction equation. For example,

1
r = k[CO][O2 ] 2 . (1.102)

However, many experiments showed that such an approach was too naive. It is rarely the
case that the stoichiometric coefficient of the reactants are equal to the exponents in these
power laws. The underlying reason for this is that such power laws do not properly describe
the kinetics of the reaction. In order to have a proper description, we need to describe the
kinetics of the elementary reaction steps constituting the overall reaction.
For example, the catalytic oxidation of CO can be decomposed in the following elementary
reaction steps:9

CO + ∗  CO∗ (1.103)
O2 + 2∗  2O∗ (1.104)
CO ∗ +O∗  CO2 ∗ (1.105)
CO2 + ∗  CO2 ∗ (1.106)

Describing the kinetics of these four elementary reaction steps can give an overall rate
expression for this reaction. Constructing a set of elementary reaction steps constituting
the overall reaction is typically based on a mixture of chemical intuition, logics and carefully
conducted experiments. For heterogeneous catalytic reactions, a good approach to construct a
set of elementary reaction steps is to decompose the reactants on the catalytic surface to their
constituting elements and subsequently (re-)assemble these elements into their final product.
To exemplify, let us consider ammonia synthesis. Herein, nitrogen and hydrogen gas is
converted to ammonia over a catalytic surface. The first step is decomposing the nitrogen
and hydrogen molecules to nitrogen and hydrogen atoms. In other words, the nitrogen and
hydrogen molecules are dissociated over the catalytic surface. The corresponding elementary
reaction steps are:

N2 + 2∗  2 N∗ (1.107)
H2 + 2∗  2 H∗ (1.108)

You might argue that nitrogen does not adsorb dissociatively10 and that the nitrogen
dissociation actually proceeds in two elementary reaction steps:

N2 + ∗  N2 ∗ (1.109)
N2 ∗ +∗  2 N∗ (1.110)

In principle, both paths are valid. In order to resolve whether this happens in one or two
steps is by using the definition of an elementary reaction step. An elementary reaction step
9
Note that in equation 1.105 a vacant site is formed after recombination of adsorbed CO and O.
10
In fact, I gave this as an example for associative adsorption previously. Moreover, it is known that the N2 bond is
very strong, hence making such a pathway unlikely.
 C HAPTER 1
18 Chapter 1. Kinetics


has a single transition state. As it turns out, for some metals, nitrogen adsorption immediately
results in dissociation via a single transition state, while for other metals, nitrogen first
molecularly adsorbs and then dissociates.
After adsorption and dissociation, ammonia is made by subsequent hydrogenation of the
nitrogen atom on the surface. Each of these hydrogenation steps are elementary reaction
steps as the formation of a single N-H bond occurs via a single transition state.

N ∗ +H∗  NH ∗ + ∗ (1.111)
NH ∗ +H∗  NH2 ∗ + ∗ (1.112)
NH2 ∗ +H∗  NH3 ∗ + ∗ (1.113)

Finally, after the ammonia has been formed on the surface, it needs to desorb. This
occurs in a single elementary reaction step. For some reactions, the final hydrogenation
step immediately leads to desorption (such as is the case for methane). In that case, both
hydrogenation and desorption occur in the same elementary reaction step, because it happens
via a single transition state.

NH3 ∗  NH3 + ∗ (1.114)

We started this chapter by mentioning that the construction of power laws is a poor
approach. You might argue against this, as in principle, you can still construct a single power
law expression by fitting the exponents in the power law to a number of experiments. Such an
approach however would not reveal much of the underlying process. Consider now what kind
of advantages one would gain by loosing the black-box assumption and developing a complete
(micro)kinetic model. One would have a better description of the short-lived intermediates
or radicals. The effect of temperature and pressure could be studied in much greater detail.
One could identify the elementary reaction step that limits the overall reaction and look for
new catalytic materials that lower the reaction barrier of this elementary reaction step.

1.5.1 Potential energy diagram

If the set of elementary reaction steps that describe the mechanism is relatively simple,
it is possible to construct a potential energy diagram (sometimes also termed a reaction
energy diagram) from this set. The potential energy diagram describes the change in energy
between the different thermodynamic states in the reaction and also shows the barrier of each
elementary reaction step.
The potential energy diagram of a single elementary reaction step is given in Figure 1.5.
In this Figure, the initial, transition and final state of the reaction are shown. The difference
in energy between the initial and transition state corresponds to the activation energy. This is
the barrier in terms of energy that has to be crossed in order for this reaction to proceed. The
difference between the initial and final state corresponds to the reaction energy and is the
amount of heat released by the elementary reaction step.
If you would approach this reaction from the right hand side, i.e. from the final state
towards the initial state, then you would still need to cross the barrier imposed by the transition
state. This barrier for the backward reaction is simply named the activation energy in the
backward direction and is defined as the difference between the transition state and the final
state. Conclusively, all elementary reaction steps can proceed in both the forward and the
backward direction and regardless of the direction, you will encounter a barrier. The principle
behind this is termed microscopic reversibility.
The potential energy diagram for CO oxidation is shown in Figure 1.6. In this diagram,
on the left hand side, CO and 12 O2 in the gas phase are given. First, CO and O adsorb on
C HAPTER 1
1.5 Reaction mechanisms in catalysis 19


Transition state

∆Eact

Initial state
∆HR

Final state

Figure 1.5: Potential energy diagram of a single elementary reaction step

CO + 12 O2
∆HCO
∆HCO2
CO2
∆Eact CO2 *
CO* 1 ∆H
2 O2

CO* + O*

Figure 1.6: Conceptual potential energy diagram for the CO oxidation reaction. ∆HX corresponds to the
adsorption energy of compound X . ∆Eact is the activation energy for the recombination reaction between
adsorbed CO and O.

the surface. Recall that O2 adsorbs dissociatively and only a single oxygen is required for the
CO oxidation, hence we only need to adsorb a single O atom. CO* can react with O* to form
CO2 * on the surface, which in turn can leave the surface. Note that adsorption reactions are
exothermic as new chemical bonds with the catalytic surface are formed. Similarly, dissociation
steps are endothermic as chemical bonds with the catalytic surface are broken.
The potential energy diagram is a powerful concept to study the energetics of a reaction
mechanism. Despite this, it only conveys the relative energy levels of the thermodynamic
states and has therefore limited information about the kinetics of the reaction. To study the
kinetics, one has to construct an expression for the overall reaction rate in terms of the set of
elementary reaction steps, which will be the topic of the next section.

1.5.2 Rate-determining step and overall reaction rate

Given that we have a full understanding of the set of all elementary reaction steps of a catalytic
reaction, how can we then construct an analytic expression for the overall reaction rate? The
key assumption to use is to consider that one of the elementary reaction steps, typically a step
over the catalytic surface, is the rate-determining step. In other words, the rate of the overall
reaction equals the rate of the slowest elementary reaction step.
Let us consider again the example of CO oxidation towards CO2 , which occurs via the
following four elementary reaction steps
 C HAPTER 1
20 Chapter 1. Kinetics


CO + ∗  CO∗ (1.115)
O2 + 2∗  2O∗ (1.116)
CO ∗ +O∗  CO2 ∗ + ∗ (1.117)
CO2 + ∗  CO2 ∗ (1.118)

We wish to develop an analytical expression for the rate of production for CO2 for this
system. From the set of elementary reaction steps as defined above we are of course able
to construct a set of ordinary differential equations and solve this set numerically given
appropriate boundary conditions. However, if we wish to pursue an analytical expression, we
are going to make a series of assumptions:

• We assume that the surface oxidation step (CO ∗ +O∗  CO2 ∗ +∗) is the rate-
determining step. Thus, the overall reaction rate equals the rate of this elementary
reaction step.

• We assume that all steps other than the rate-determining step are in quasi-equilibrium.
This means that on the time scales by which the overall reaction occurs, all steps other
than the rate-determining step have already reached a pseudo- or quasi-equilibrium.
In other words, their rate of change is zero.

• We assume that CO and CO2 adsorb associatively whereas O2 adsorbs dissociatively.

We have previously seen how to construct Langmuir adsorption isotherms for such a
situation. Note that we had to assume a quasi-equilibrium in order to establish these
Langmuir adsorption isotherms. In this situation, we have competitive adsorption
of three components, but we will shortly see that it is not more complicated than
competitive adsorption for two components.

• Finally, we employ a mean-field approximation. In this assumption, we neglect the

local topology of the catalyst surface and assume that every compound on the catalytic
surface can interact with every other compound on the catalytic surface. This concept is
further illustrated in Figure 1.7. In this Figure, we note that all adsorbates (denoted by
squares and circles) are randomly placed on the catalytic surface. In reality, an adsorbate
sitting at the left bottom corner of the catalytic surface will never react with an adsorbate
at the right top of the catalytic surface, however, within the mean-field approximation,
we assume that everything is continuously randomly distributed and such interactions
do occur. In other words, we neglect the local topology and thus we can describe the
rate of reaction using simply the surface fractions of the adsorbed species.

Using the above assumptions, i.e. the quasi-equilibrium approximation, the rate-determining
step approximation and the mean-field approximation, we obtain the following three equations
for the surface coverage of CO, O2 and CO2 :

KCO pCO
θCO = q (1.119)
1 + KCO pCO + KO2 pO2 + KCO2 pCO2
q
K O2 p O2
θO = q (1.120)
1 + KCO pCO + KO2 pO2 + KCO2 pCO2
KCO2 pCO2
θCO2 = q (1.121)
1 + KCO pCO + KO2 pO2 + KCO2 pCO2
C HAPTER 1
1.5 Reaction mechanisms in catalysis 21


Figure 1.7: Schematic depiction of the mean-field approximation. In the mean-field approximation, there is no
interaction between the adsorbed species and all species are distributed randomly over the surface.

The overall rate is equal to the rate of the rate-determining step, hence

rCO2 = k3+ θCO θO − k3− θCO2 θ∗ (1.122)

k3+ KCO pCO KO2 pO2 − k3− KCO2 pCO2
q

=  2 (1.123)
q
1 + KCO pCO + KO2 pO2 + KCO2 pCO2

This equation can be further simplified when we assume that the rate determining step
is irreversible. In such circumstances, the forward rate is much larger than the backward rate
and hence the equation simplifies to

q
k3+ KCO pCO K O2 p O2
rCO2 =  2 . (1.124)
q
1 + KCO pCO + KO2 pO2 + KCO2 pCO2

With the irreversible step approximation set, the reverse reaction is not occurring, but any
product formed can still readsorb on the surface. Thus, we see a term in the denominator
corresponding to adsorbed CO2 . On top of the irreversible step approximation we are allowed
to make an even more stringent assumption termed the zero-conversion approximation. As the
name implies, within this assumption no products are being formed and hence all components
that are formed after the rate-determining step can be ignored and will thus not occur on the
catalytic surface. The reaction rate then further simplifies to

q
k3+ KCO pCO KO2 pO2
rCO2 =  2 . (1.125)
q
1 + KCO pCO + KO2 pO2

Note that despite that the zero-conversion approximation and the irreversible step ap-
proximation are different approximations, under the zero-conversion limit the same kind of
effects are present as under the irreversible step approximation. Whereas the zero-conversion
limit does not state that the rate-determining step cannot proceed in the reverse direction,
due to the fact that no product has been formed (i.e. we are working at zero conversion),
 C HAPTER 1
22 Chapter 1. Kinetics


there is no product present for the reverse reaction to occur. Admittedly, the difference is
subtle, yet the reader is advised to use the approximations with caution and not to treat them
as synonymous.
With these additional approximations installed, we can even further simplify this equation
by making another assumption. Often, catalytic reactions have a surface compound which
adsorbs much stronger than the other adsorbates. Thus, we expect to find that the surface is
mainly covered with that compound and some free sites. For example, let us assume that CO
binds much stronger that the other compounds. In that case,

q
KCO pCO  KO2 pO2 , KCO2 pCO2 (1.126)

and the above equation then simplifies to

q
k3+ KCO pCO KO2 pO2
rCO2 = . (1.127)
(1 + KCO pCO )2

In this situation, we say that CO is the MARI, which stands for the Most Abundant Reaction
Intermediate. The final assumption we can make is to consider that we are working at very low
temperature by which the surface coverage of CO is not merely the most abundant intermediate
among the reaction intermediates, the coverage of CO is also greater than the amount of
available empty sites. In that case, the reaction rate simplifies to

q
k3+ KO2 pO2
rCO2 = . (1.128)
KCO pCO

At this point, we have found the equation under the condition that the whole surface is
covered with CO and thus that CO is poisoning the catalytic reaction. It should be noted
that this assumption is rarely valid, except at very low temperature, i.e. well below typical
operating conditions.

1.5.3 Reaction orders

The overall reaction rate depends on the pressures of the reactants and this dependency is
reflected by the reaction order. The reaction order is basically a number which defines the
scaling behavior of the reaction, i.e. if for instance the reaction order is 2, the rate of the
reaction will quadruple if the pressure is doubled.
The reaction order can be calculated using the following formula

∂ ln r
nX = pX . (1.129)
∂pX

To exemplify this, let us calculate the reaction order in CO for the CO oxidation reaction.
We assume that the quasi-equilibrium assumptions holds and furthermore we consider CO
to be the MARI.
C HAPTER 1
1.5 Reaction mechanisms in catalysis 23


∂ ln r
nCO =pCO (1.130)
∂pCO
q
k3+ KCO pCO KO2 pO2
∂ ln
(1+KCO pCO )2
=pCO (1.131)
∂pCO
 
∂ 1 1
=pCO ln k3+ + ln KCO + ln pCO + ln KO2 + ln pO2 − 2 ln (1 + KCO pCO )
∂pCO 2 2
(1.132)
∂
ln pCO − 2 ln (1 + KCO pCO ) (1.133)


=pCO
∂pCO
∂ ln (1 + KCO pCO )
=1 − 2pCO (1.134)
∂pCO
pCO ∂ (1 + KCO pCO )
=1 − 2 (1.135)
(1 + KCO pCO ) ∂pCO
KCO pCO
=1 − 2 (1.136)
(1 + KCO pCO )
=1 − 2θCO (1.137)
(1.138)
The above result can be interpreted as follows. At very low surface coverage of CO, the
reaction order in CO is equal to 1. This means that when we double the CO pressure, we
expect that the reaction rate will double as well. Alternatively, at very high surface coverage
of CO, the reaction order will be -1. In this case, doubling the CO pressure will result in a
reduction of the reaction rate by a factor 2. The latter condition is the poisoning regime. Here,
the CO surface coverage is so high that there are no available free sites for O2 to adsorb and
hence the reaction rate decreases with increasing CO pressure.
If CO is not the MARI, we have the following equation for the reaction rate.

q
k3+ KCO pCO K O2 p O2
rCO2 =  2 (1.139)
q
1 + KCO pCO + KO2 pO2 + KCO2 pCO2

In a similar fashion as shown above, we can calculate the reaction orders to be

nCO =1 − 2θCO (1.140)

1
nO2 = − θO (1.141)
2
nCO2 = − θCO2 (1.142)
The reaction rate as a function of coverage is shown in Figure 1.8 and the corresponding
reaction order is given in Figure 1.9. From these Figures, we can see that the reaction order in
O2 is constant as a function of temperature, whereas the reaction order in CO changes from -1
to +1. In Figure 1.10, the surface coverages for CO*, O* and * are given. Using this Figure, we
can rationalize the results obtained for the reaction orders. At low temperature, the surface
is mainly covered with CO, hence the reaction order in CO is negative. With increasing
temperature, the surface coverage of CO decreases and the amount of available sites increases.
As a consequence, the reaction order in CO increases from -1 to +1. Because the surface
coverage of O is very low in the temperature range under consideration, its corresponding
reaction order is constant and amounts to nO = 12 for the whole range.
 C HAPTER 1
24 Chapter 1. Kinetics


CO2 production
8

6
1
Rate in s

0
600 700 800 900 1000 1100
Temperature T in K

Figure 1.8: Rate of CO2 formation as a function of temperature

1.5.4 Apparent activation energy

To express the dependence of the overall rate on the temperature, typically the concept of the
apparent activation energy is employed. The higher the energy of the apparent activation
energy, the higher the energy barrier that the overall reaction has to cross. The expression for
the apparent activation energy is as follows

app ∂ ln r
∆Eact = RT 2 , (1.143)
∂T
where R is the gas constant and T is the temperature. Applying the above equation to the
(simplified) reaction rate for CO2 formation as given in equation 1.127

 q 
k3+ KCO pCO KO2 pO2
∂ ln  
(1+KCO pCO )2
app
∆Eact =RT 2 (1.144)
∂T
 
∂ 1 1
=RT 2 ln k3+ + ln KCO + ln pCO + ln KO2 + ln pO2 − 2 ln (1 + KCO pCO )
∂T 2 2
(1.145)
 
∂ 1
=RT 2 ln k3+ + ln KCO + ln KO2 − 2 ln (1 + KCO pCO ) (1.146)
∂T 2

Let us take one step back from the above equation and study the results obtained so
far before we continue deriving the final result. In the last step, we note that we have four
terms that we have to differentiate. One term corresponds to the derivative towards T of the
natural logarithm of the reaction rate constant, two similar terms but then for the equilibrium
constant and a final term that corresponds to the part of the denominator of the overall
reaction equation. To solve the last term, we need to apply the chain rule. Let us first solve
C HAPTER 1
1.5 Reaction mechanisms in catalysis 25


1.00 CO
O2
0.75
Reaction order (dimensionless)

0.50
0.25
0.00
0.25
0.50
0.75
1.00
600 700 800 900 1000 1100
Temperature T in K

Figure 1.9: Reaction order in CO and O2 as a function of temperature

the first three terms for which we need to know how k and K depend on the temperature.
From thermodynamics (we will discuss this in more detail in Chapter 2), these terms depend
on temperature in the following fashion:

 
−∆Eact
k = ν exp , (1.147)
RT

where ν is the pre-exponential factor and ∆Eact is the reaction barrier for the correspond-
ing elementary reaction step. Plugging the above in our expression and solving for the first
term gives

 !  
∂ −∆Eact ∂ −∆Eact
ln ν exp = ln (ν) + (1.148)
∂T RT ∂T RT
∆Eact
= . (1.149)
RT 2

Similarly, the temperature dependence of the equilibrium constant is reflected by the

following equation

       
−∆GR −∆HR + T ∆SR −∆HR ∆SR
K = exp = exp = exp exp , (1.150)
RT RT RT R

where ∆GR is the Gibbs Free energy, ∆HR is the reaction enthalpy and ∆SR is the
reaction entropy. Thus, we can now solve for the second and third term in equation 1.146
 C HAPTER 1
26 Chapter 1. Kinetics


1.0 CO*
O*
* (free sites)
Surface coverage (dimensionless)

0.8

0.6

0.4

0.2

0.0
600 700 800 900 1000 1100
Temperature T in K

Figure 1.10: Surface coverage in CO and O2 as a function of temperature

   !
∂ −∆HR ∆SR
ln exp exp (1.151)
∂T RT R
 
 !  !
∂  −∆HR ∆SR
= ln exp + ln exp  (1.152)
∂T RT R
 
∂ −∆HR ∆SR
= + (1.153)
∂T RT R
∆HR
= (1.154)
RT 2

Finally, we have to solve for the fourth term in equation 1.146 which can be done by using
the chain rule and applying the result of equation 1.154.

∂ ∂
2 ln (1 + KCO pCO ) =2 ln (1 + KCO pCO ) (1.155)



∂T ∂T
∂ ln α ∂α
=2 with α = 1 + KCO pCO (1.156)
∂α ∂T
1 ∂α
=2 with α = 1 + KCO pCO (1.157)
α ∂T
pCO ∂KCO
=2 (1.158)
1 + KCO pCO ∂T
   
pCO ∂ exp −∆HRT
R
exp ∆S
R
R

=2 (1.159)
1 + KCO pCO ∂T
pCO KCO ∆HCO
=2 (1.160)
1 + KCO pCO RT 2
C HAPTER 1
1.5 Reaction mechanisms in catalysis 27


pCO KCO
Note that the term 1+K corresponds to the Langmuir adsorption isotherm for CO
CO pCO
under the conditions that CO is the MARI, hence, we can further reduce the above equation to

∂ ∆HCO
2 ln (1 + KCO pCO ) = 2 θCO (1.161)


∂T RT 2
Combining all four terms provides us with the following equation for the apparent
activation energy

 
app ∂ 1
∆Eact =RT 2 ln k3+ + ln KCO + ln KO2 − 2 ln (1 + KCO pCO ) (1.162)
∂T 2
1
=∆Eact + ∆HCO + ∆HO2 − 2∆HCO θCO (1.163)
2

1
=∆Eact + ∆HCO 1 − 2θCO + ∆HO2 (1.164)
2
We can interpret this equation on the basis of Figure 1.6. The overall activation energy
of the kinetic network depends on the barrier of the rate-determining step. The barrier is
decreased by the half of the adsorption heat of O2 as oxygen adsorption is exothermic (i.e. ∆H
for adsorption is always negative) and releases heat. The barrier is further decreased by the
adsorption heat of CO, however, with increasing amount of CO on the surface, the reduction
of the barrier by CO adsorption decreases to the point that at coverages larger than 0.5, the
reaction barrier is increased by the heat of adsorption of CO.
 C HAPTER 1
28 Chapter 1. Kinetics


1.6 Questions and Exercises
The answers to the questions and exercises are given at the end of this book in Appendix C
(page 135).

1.6.1 Exercises
The exercises are marked by a number of gears to indicate their difficulty levels.

 Exercise 1.1 

The decomposition of acetaldehyde

CH3 CHO → CH4 + CO (1.165)

proceeds in the gas phase via methyl radicals according to

1. CH3 CHO → CH3 * + CHO*

2. CH3 * + CH3 CHO → CH4 + CH3 CO*
3. CH3 CO* → CH3 * + CO
4. 2CH3 * → C2 H6 (1.166)

Assume that all reactions are irreversible. Derive the rate expressions for the formation
of CH4 and C2 H6 by using the steady-state approximation.

 Exercise 1.2 

N2 O5 is an unstable compound formed in the atmosphere upon interaction of NO2 with

oxygen. Its decomposition

2N2 O5 → 4NO2 + O2 (1.167)

proceeds according to the following rate equation

d[O2 ]
r= = k[N2 O5 ] (1.168)
dt
Show that the following set of elementary reaction steps leads to this rate equation and
provide an explicit expression for the overall rate constant k. Identify all reaction intermediates
and apply the steady state approximation on these intermediates.

1. N2 O5  NO2 + NO3
2. NO2 + NO3 → NO + NO2 + O2
3. NO + NO3 → 2NO2 (1.169)

 Think deeper...
C HAPTER 1
1.6 Questions and Exercises 29


• Explain why the reaction order with respect to N2 O5 is unity and not 2 as suggested by
the overall rate equation.

• Derive a rate equation for the formation of NO2 and relate it to the rate equation for
O2 formation.

 Exercise 1.3 

Derive the rate expression for N2 O for the reaction

2NO + H2 → N2 O + H2 O (1.170)

using the steady-state expression. The mechanism constitutes the following elementary
reaction steps

1. 2NO  N2 O2
2. N2 O2 + H2 → N2 O + H2 O

 Exercise 1.4  

The exposure of sunlight to a mixture of methane and bromine gas results in a violent
reaction releasing bromomethane and hydrogen bromide. This reaction proceeds according
to the following mechanism:

1. Br2 → 2Br*
2. CH4 + Br* → CH3 * + HBr
3. CH3 * + Br2 → CH3 Br + Br*
4. 2Br* → Br2

• Identify the type of elementary reaction step in the above chain reaction. Distinguish
between initiation, propagation and termination reactions.

• Derive a rate expression for the formation of CH3 Br. Clearly explain the assumptions
you have used in the derivation.

 Exercise 1.5 

Derive the Langmuir adsorption isotherms for the following situations

(a) Molecular adsorption of CO.

(b) Dissociative adsorption of CO.

 C HAPTER 1
30 Chapter 1. Kinetics


(c) Competitive adsorption of molecularly adsorbed CO and dissociatively adsorbed H2 .
Assume there occurs no reaction between CO and H2 .

(d) Consider the mechanism of methanol formation through consecutive hydrogenation

of CO. The rate-determining step is the first hydrogenation of CO with one H atom.
This step is irreversible. All other steps are fast, except for the desorption of methanol
which is considered to be in equilibrium with the gas phase. CO and H2 adsorption
are also quasi-equilibrated.

• Write down the mechanism.

• Reason why you only need to take the surface coverages of CO, H and CH3 COH
into account.
• Derive an expression for the reaction rate.
• Give the limits (the domain) of the reaction orders for H2 , CO and methanol.

 Exercise 1.6   

Recent work has shown that the rate of catalytic synthesis of methanol from CO2 and H2
scales first-order with the partial pressure of CO2 and 32 order with the partial pressure of H2 .
The overall reaction equation is

CO2 + 3H2 → CH3 OH + H2 O (1.171)

The mechanism is thought to proceed in the following manner

1. H2 + 2∗  2H*
2. CO2 + ∗  CO2 *
3. CO2 * + H*  HCOO* + ∗
4. HCOO* + H*  H2 COO* + ∗
5. H2 COO* + H*  H3 CO* + O*
6. H3 CO* + H*  H3 COH* + *
7. H3 COH + *  H3 COH*

a) Complete the mechanism by adding three elementary reaction steps which remove O* as
H2 O involving equilibrium between water in the gas phase and the adsorbed state.

b) Determine which step is most likely the rate determining step considering the reaction
orders as obtained from the experimental results.

c) Derive the corresponding rate equation for methanol formation, assuming that the rate-
determining step is irreversible (proceeds only in the forward direction) and all other steps
are in quasi-equilibrium. Furthermore, assume that the surface is nearly empty. Show that
nH2 = 32 .

 Think deeper...

At which molar fraction of H2 is the reaction rate at its optimum?

C HAPTER 1
1.6 Questions and Exercises 31


 Exercise 1.7  

During the course lectures, CO oxidation was studied as a typical example of surface
catalysis. It relates to the clean-up of exhaust gases from the combustion of transport fuels. The
rate equation has been derived assuming that the surface reaction is the rate determining step.
Furthermore assume that the rate-determining step is irreversible. Consider the following
mechanism:

1. CO + ∗  CO*
2. O2 + 2* → 2O*
3. CO* + O* → CO2 + 2* (1.172)

a) Which assumptions do you propose to compute the surface coverage of O atoms? Recall
that we are considering in this question car-exhaust clean-up in the three-way automotive
catalyst. Should the surface oxygen coverage be high or low in this process and why?

b) Derive the overall rate equation for the above mechanism.

 Exercise 1.8  

Consider the steam reforming of methane, which is a large-scale industrial process for
the production of synthesis gas:

CH4 + H2 O → CO + 3H2 (1.173)

For a Ni catalyst supported on alumina, we find that the reaction orders are the following

• nCH4 > 0

• nH2 O < 0

• nH2 > 0

a) What is synthesis gas and name at least two chemical processes that use synthesis gas as
its feedstock.

b) In case we wish to produce H2 , for instance in ammonia production, what other important
chemical reaction can be utilized to increase the amount of H2 that can be produced from
methane?

c) Why is steam reforming carried out under moderate pressure and high temperature?

d) Propose a mechanism involving recombination of adsorbed C* and adsorbed O* to form

adsorbed CO* as the rate-determining step. Derive a reaction rate equation assuming that O*
is the MARI. Assume that the adsorption of CH4 is dissociative, whereas the adsorption of
H2 O is molecular.

e) Are the experimental orders for Ni consistent with this model?

 C HAPTER 1
32 Chapter 1. Kinetics


f ) Derive another rate equation assuming that methane dissociation is rate-determining and
assume again that O* is the MARI. What are the reaction orders in this case?


 Exercise 1.9  

The catalytic hydrogenolysis of ethane into methane is composed of the following ele-
mentary reaction steps

1. C2 H6 + 2∗  C2 H5 * + H*
2. C2 H5 * + H* → 2CH3 *
3. CH3 * + H*  CH4 + 2∗
4. H2 + 2∗  2H*

Assume that step (2) is rate determining and that all other elementary reaction steps are
in quasi-equilibrium.
a) If a small amount of D2 is added to the reactants, C2 H6−n Dn is observed in the gas phase.
Explain this phenomenon.

b) Deduce the rate equation for step (2).

c) Derive the expressions for the surface coverages of ethyl, methyl and hydrogen.

d) Give the full rate equation including the terms relevant when the reaction approaches
equilibrium. In other words, assume that the rate-determining step is reversible.

e) Give the simplified rate equation in the case that H* is the MARI and very low conversion
of ethane.

 Think deeper...

Derive an expression for the apparent activation energy for the latter case.


 Exercise 1.10  

Consider the reaction between NO and CO in the presence of O2 on a Rh catalyst. NO

and CO adsorb molecularly. The surface NO dissociation step is much slower than all other
elementary reaction steps. The rate of CO* + O* is comparable to the rate of dissociative O2
adsorption. The other steps are quasi-equilibrated.

1. NO + ∗  NO*
2. CO + ∗  CO*
3. O2 + 2∗ → 2O*
4. N2 + 2∗  2N*
5. NO* + ∗ → N* + O*
6. CO* + O* → CO2 + 2∗
C HAPTER 1
1.6 Questions and Exercises 33


Derive the rate equation for N2 and CO2 formation. Use the steady-state approximation
on O* to derive this expression.

 Think deeper...

In which technological application are these reactions important?

 Exercise 1.11  

Ethylene oxide is an important feedstock in the production of ethylene glycol. Ethylene

oxide is made by partial oxidation of ethylene (C2 H4 ) over a Ag catalyst. In this process,
ethylene adsorbs molecularly whereas oxygen adsorbs dissociatively. The transition state for
the formation of ethylene oxide has such a high barrier that this elementary reaction step is
considered to be rate-determining. Assume that ethylene oxide immediately and irreversibly
desorbs from the catalytic surface after formation. Further assume that all other elementary
reaction steps are in quasi-equilibrium.

a) Provide an expression for the surface coverage of ethylene as a function of the partial
pressure of ethylene and oxygen.

b) Derive an expression for the rate of formation of ethylene oxide as a function of the partial
pressures of ethylene, ethylene oxide and oxygen.

c) At low temperature, it is found that oxygen is strongly adsorbed. Derive a simplified

expression for this situation. What are the reaction orders in ethylene and oxygen?

d) Describe the surface composition for the situation described in item (c).

e) Derive an expression for the apparent activation energy for the situation described in item
(c).

 Exercise 1.12   

Consider the catalytic oxidation of sulfur dioxide with molecular oxygen to sulfur trioxide.
The overall reaction equation is

1
SO2 + O2 → SO3 (1.174)
2
The catalytic reaction proceeds over a catalytic surface containing only one type of surface
sites. SO2 adsorbs molecularly, whereas O2 adsorbs dissociatively. SO3 is formed on the
catalytic surface by bond formation between adsorbed SO2 and O. Adsorbed SO3 is in quasi-
equilibrium with SO3 in the gas phase.

a) Give the elementary reaction steps of this proceses.

b) Assume that the surface reaction between SO2 and O is the rate-determining step and is
irreversible. Derive an expression for the rate of formation of SO3 .

c) Give the limits of the reaction orders in SO2 , SO3 and O2 .

 C HAPTER 1
34 Chapter 1. Kinetics


Assume from here on that O2 is strongly adsorbed whereas SO2 and SO3 are only weakly
adsorbed.

d) Provide an expression for the rate of formation of SO3 and derive the reaction orders in
terms of the surface composition (fractional occupancies).

e) Derive an expression for the apparent activation energy in terms of the surface composition
(fractional occupancies).

Using a different catalyst, the following rate expression was found

 
! 1/2 1/2
KSO2 pSO2 KO pO
r=k 2 2
(1.175)
 
1/2 1/2
 
1 + KSO2 pSO2 + KSO3 pSO3 1 + K O pO
2 2

f ) Provide a reasoning why the above rate expression differs from the previously described
catalyst. What is the essential difference?

 Exercise 1.13   

Platinum is an efficient catalyst for oxidative dehydrogenation of ethanol towards acetalde-

hyde. The oxidizing agent is molecular oxygen. Mechanistic studies have shown that ethanol
adsorbs molecularly, whereas oxygen adsorbs dissociatively. The rate-limiting step is the
dehydrogenation of adsorbed ethanol to acetaldehyde coinciding with the release of a water
molecule.

C2 H5 OH* + O* → C2 H4 O + H2 O + 2* (1.176)

a) Derive an expression for the surface coverage of ethanol and oxygen in terms of their
corresponding partial pressures.

b) Derive an expression for the rate of formation of acetaldehyde as a function of the partial
pressures of ethanol, oxygen, acetaldehyde, and water.

Assume that at very low temperature, ethanol adsorbs strongly.

c) Provide an expression for the rate of formation considering the above assumption. What
are the reaction orders in ethanol and in oxygen?

d) Derive an expression for the apparent activation energy and explain this result in terms of
the surface composition and the reaction profile.

Assume the temperature is increased.

e) Describe the surface composition in this scenario.


C HAPTER 1
1.6 Questions and Exercises 35


 Exercise 1.14   

The synthesis of water from H2 and O2 over a Pt surface proceeds via the following
elementary reaction steps:

1. H2 + 2∗  2H*
2. O2 + 2∗  2O*
3. O* + H*  OH* + ∗
4. OH* + H*  H2 O* + ∗
5. H2 O + ∗  H2 O*

a) Assume that step (3) is the rate-determining step and provide a rate expression for this
step containing both the forward as well as the backward rate.

b) Show that the rate expression can also be expressed as the equation shown below. Provide
an equality for the constant Kequilibrium in terms of the equilibrium constants of the individual
elementary reaction steps.

!
q pH2 O
r = k3+ K1 K2 pH2 pO2 1− √ θ*2 (1.177)
Kequilibrium pH2 pO2

c) Assume that oxygen is the MARI. Provide an expression for θ* using this assumption.

d) Provide an expression for the reaction orders in hydrogen, oxygen and water under the
assumption that O is the MARI and that the reaction is far from equilibrium.

e) Assume that the surface is nearly empty. Calculate the optimal gas phase composition (the
ratio between hydrogen and oxygen) to find the best rate. Again assume that the reaction is
operated far from equilibrium.

 Think deeper...

Derive for case (e) an expression for the apparent activation energy and explain your
results in terms of the surface processes and surface composition.

 Exercise 1.15   

Consider the reaction between NO and CO over a Pt surface. NO and CO adsorb molec-
ularly. The surface NO dissociation step is considered to be much slower than all other
elementary reaction steps in the system. As such, assume that NO dissociation is the rate-
limiting step and that all other steps are in quasi-equilibrium. Note that step (5) is reversible
and as such the reverse reaction needs to be taken into account as well.
 C HAPTER 1
36 Chapter 1. Kinetics


1. NO + ∗  NO*
2. CO + ∗  CO*
3. N2 + ∗  N2 *
4. N2 * + ∗  2N*
5. NO* + ∗  N* + O* (1.178)
6. CO* + O*  CO2 * + ∗
7. CO2 + ∗  CO2 *

a) Derive the rate equation for N2 formation. Take the reversibility of the reaction into
account and use an overall equilibrium constant Keq in the term corresponding to the reverse
reaction.

Assume that O* is the MARI

b) Derive a simplified rate equation for this case and give expressions for the reaction orders
in NO, CO, N2 and CO2 . Explain these dependencies in terms of the surface coverages.

The apparent activation energy is determined under the condition that the surface is
nearly empty.

c) Derive an expression for the apparent activation energy in terms of the activation energy
of the rate-determining step and the enthalpy changes of the other steps.

 Think deeper...

What is the expression for the apparent activation energy for case (b)?

 Exercise 1.16   

The synthesis of hydrogen peroxide from hydrogen and oxygen over a metal surface
proceeds via the following elementary reaction steps:

1. H2 + 2∗  2H*
2. O2 + ∗  O2 *
3. O2 * + H*  OOH* + ∗
4. OOH* + H*  H2 O2 * + ∗
5. H2 O2 + ∗  H2 O2 *

a) Derive an expression for the rate of H2 O2 formation assuming that step (3) is the slowest
step. Take the reverse reaction explicitly into account (i.e. do not assume zero-conversion or
an irreversible step approximation).

b) Assume now that the surface is nearly completely occupied with O2 and that the reaction
is conducted at very low conversion; simplify the above expression and determine the reaction
orders with respect to H2 , O2 and H2 O2 .
C HAPTER 1
1.6 Questions and Exercises 37


c) Derive an expression for the apparent activation energy under these conditions.

d) Derive also an expression for the apparent activation energy in the high-temperature limit
when the surface is nearly empty.

 Exercise 1.17    

Methanol synthesis can proceed in the direct pathway by fourfold hydrogenation of CO

to methanol. For this process, a catalyst is used which contains two types of active sites
indicated by θ and τ . The active sites have a specific surface topology by which carbonaceous
compounds, i.e. CHx O, can only adsorb on site θ, but H can adsorb on both these sites. An
asterisk (*) is used to indicate adsorbed compounds on site θ, whereas a pound sign (#) is
used to indicate adsorbed compounds on site τ .
Assume the following:
• Methanol is formed by threefold hydrogenation of C to CH3 O and finally hydrogenating
the O moiety to form methanol.

• The rate-determining step in this reaction is the hydrogenation of CHO to form CH2 O.

• The rate-determining step is irreversible and the system is assumed to operate in the
zero conversion limit.

• Hydrogen adsorbs dissociatively at both the θ as well as the τ site. These sites are
oriented in such a fashion that a single hydrogen molecule cannot adsorb on both these
sites simultaneously.

• There is no migration of H between the θ and τ sites.

• All elementary reaction steps on the surface, i.e. between CHx O* and H#, proceed
between the two different active sites.

• Although H* will not directly react with any CHx O* species, the adsorption of H* does
result in an inhibiting term which needs to be modeled adequately.

a) Construct the set of elementary reaction steps that define this chemo-kinetic network. Use
an asterisk (*) to indicate θ sites and a pound sign (#) to denote τ sites.

b) Derive the Langmuir adsorption isotherm for dissociative adsorption of hydrogen on the τ
sites.

c) Derive an expression for the overall reaction rate as a function of the partial pressures of
the reactants, the reaction rate constant of the rate-determining step and the equilibrium
constants of the relevant elementary reaction steps. Identify the inhibiting term corresponding
to adsorption of H on a θ site.

d) Derive the reaction order in H2 and in CO.

e) Derive the apparent activation energy as a function of the relevant partial surface coverages.


 C HAPTER 1
C HAPTER 2


CHAPTER
TRANSITION STATE THEORY

Contents
2.1 Introduction 39
2.2 Statistical method 40
2.3 Maxwell-Boltzmann distribution law 41
2.4 Partition functions of subsystems 46
2.5 Thermodynamic variables 46
2.6 Partition functions for molecules 49
2.6.1 Translations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
2.6.2 Vibrations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
2.6.3 Rotations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
2.7 Collision theory 55
2.7.1 Maxwell-Boltzmann velocity distribution . . . . . . . . . . . . . . . . . . . . . . . . . . 55
2.7.2 Collision effectiveness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
2.7.3 Unimolecular reactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
2.8 Transition state theory 59
2.8.1 Comparison between Transition State Theory and Collision theory . . . . . . . . . . . . . 62
2.8.2 Fundamental objection against collision theory . . . . . . . . . . . . . . . . . . . . . . 63
2.9 The potential energy surface 63
2.10 Dimensionality of equilibrium constants and reaction rate constants 65
2.11 Questions and Exercises 67
2.11.1 Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
2.11.2 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
2.12 Challenges 71

2.1 Introduction
Fundamental to the construction of microkinetic models lies transition state theory, which
describes the reaction rates of elementary reaction steps. To get a sound understanding of this
 40 Chapter 2. Transition State Theory

theory, first a short introduction into statistical thermodynamics is given. Next, the concept
of partition functions is explained and the relevant partition functions for molecules (i.e.

C HAPTER 2
translational, rotational and vibrational) are derived.
Transition state theory was derived by Henry Eyring in 1935, building upon the earlier


works of Max Trautz and William Lewis based on the kinetic theory of gases, known as
collision theory. Since collision theory serves as a pedagogical introduction to the topic of
transition state theory, we will also briefly discuss it in this chapter.
At the end of this Chapter, you are able to identify the degrees of freedom of molecules
in order to calculate their partition functions. From there on, you are able to construct rate
expressions from first-principles employing transition state theory. From here on, you can
use this methodology to obtain quantitative rate constants necessary for the construction of
microkinetic models.

2.2 Statistical method

On the basis of quantum chemistry, it is well known that by means of the Schrödinger equation
the microscopic laws of a particle or a system with a number of particles (N ≈ 100) can be
defined. The question now arises what the observable properties are of a system containing
an extremely large number of such particles. That is, we want to discuss the relationship
between these microscopic dynamics and the observed properties of a large system.
The task of solving the Schrödinger equation for a many-body system (say N = 1023 par-
ticles) is daunting to say the least and far from feasible for modern day computers. However,
although one might think that the complexity and perhaps the obscurity would increase by
increasing the number of particles, the opposite is true. It is already known from thermo-
dynamics that a large system can be modeled by only a handful of parameters and we can
consider macroscopic systems to be quite orderly.
We shall assume that the distinctive regularities are a consequence of the statistical laws
governing the behavior of systems consisting of very many particles. Therefore we will avoid
direct evaluation of the precise N -particle dynamics and assume that probability statistics
provide the correct description of what we see during a macroscopic measurement.
We assume that we could observe a many-body system in a particular microscopic state.
Such a characterization would require a huge amount of variables. So we will assume that
that our system is quantal and obeys the Schrödinger equation:

∂
i~ |ψi = H |ψi (2.1)
∂t

To specify the state |ψi the system is in, we would require a number of variables of the
N th order, where N is the number of particles in the system. To exemplify, if we would
consider the stationary states

H |ψν i = Eν |ψν i (2.2)

then the index ν would represent the collection of quantum numbers specifying the
state |ψν i. The space spanned by all the state vectors |ψi is termed the Hilbert space. The
equivalent term for classical systems is a Phase space.
Time evolution in the above-mentioned Hilbert space is governed by time-integration of the
time-dependent Schrödinger equation. The path through the Hilbert space that encompasses
the set of microstates is termed the trajectory. If we fix the total energy E , the total number of
particles N and the total volume V of the system, then the trajectory will move on a ”surface”
of state space, although the dimensionality of this space is still enormously high.
 2.3 Maxwell-Boltzmann distribution law 41

The basic concept within statistical mechanics is the following: If we wait long enough, the
system will eventually flow through all the microscopic states consistent with the constraints
C HAPTER 2

we have imposed to control the system. If this is the case, then any observable property G
asserted from measurements is given by


N
1 X
Gobs = Ga (2.3)
N
a=1

where Ga is the ath measurement whose time duration is so short that it can be considered
that the system is only in one microscopic state. If this is the case, then we can write Gobs as

X 1 
Gobs = nν Gν (2.4)
ν N

where nν is the number of times the state ν is observed in N observations and Gν is the
expectation value for G when the system is in state ν . After a considerable amount of time,
when all possible states have been visited, we can define the probability or fraction of time
spent in ν and rewrite equation 2.3 as

X
Gobs = Pν G ν , (2.5)
ν

where Pν is the probability of finding the system in state ν .

From this we can formulate the following. If the system is of considerable size (i.e.
N ≈ 1023 ) and we perform measurements long enough, then the time average (eq. 2.3)
equals the ensemble average (eq. 2.5). Dynamical systems that obey this equivalence are said
to be ergodic. Although it is difficult to establish whether a system is ergodic, it is thought
that this property holds for many-body systems encountered in Nature.
This leads to the following postulate:

Postulate 1. The (long) time average of a mechanical variable M in the thermodynamic system of
interest is equal to the ensemble average M , in the limit as N → ∞, provided that the system of the
ensemble replicates the thermodynamic state and environment of the actual system of interest.[2, 3, 4]

2.3 Maxwell-Boltzmann distribution law

As was pointed out in the previous section, during a measurement of a microscopic system, a
time average would give the same result as an ensemble average. To quantify this idea, we need
to define our ensemble as well as the probability or distribution of the various microscopic
states leading to a second postulate:

Postulate 2. In an ensemble (N → ∞) representative of an isolated thermodynamic system, the

particles in the ensemble are distributed uniformly, that is, with equal probability or frequency, over
the possible quantum states consistent with the specified values of N, V and E .[2, 3, 4]

With the above assumptions, the number of possible states or probability ΩD for a given
E and V is

Y (gi )ni
ΩD = N ! , (2.6)
ni !
i
42 Chapter 2. Transition State Theory

where N is the total number of particles, the product loops over all possible states i, gi
is the degeneracy of state i and ni is the number of molecules (or particles) put in state i

C HAPTER 2
(occupancy number). Moreover, the total energy of such a system would be the sum of the
energies of the individual molecules:


X
E= ni i (2.7)
i

where i loops over each of the molecules, i is the energy per molecule and ni is the
number of molecules i.
Before we continue deriving the Maxwell-Boltzmann distribution, let us first explore the
concept of thermodynamic probabilities. Consider a system consisting of three micro states
(A, B, C) of the same energy (i.e. threefold degenerate). What are the probabilities of having
three particles (1, 2 and 3) distributed equally over all three states over having all particles in a
single micro state? The total number of probabilities can be easily calculated using equation
2.6 which yields

33
ΩD = 3! = 27. (2.8)
3!
To determine how many configurations there are for having all three particles in different
micro states, consider the following rules for Maxwell-Boltzmann statistics.

• You can put zero or more particles in any given micro state.

• All particles can be distinguished from each other

There are six different configurations wherein each micro state has a single particle

1. A1 B2 C3

2. A1 B3 C2

3. A2 B1 C3

4. A2 B3 C1

5. A3 B1 C2

6. A3 B2 C1

On the other hand, there are only three different configurations wherein a single micro
state has all three particles

1. A1,2,3 BC

2. AB1,2,3 C

3. ABC1,2,3

This simple example shows that we expect that a configuration wherein the particles are
spread out over the different micro states is more likely (i.e. it has more probabilities) than
a configuration wherein all three particles are in a single micro state. Let us now use this
concept to derive a general expression for very large systems.
If all states associated with a certain E and V value are equally probable, then the most
probable state is the one where ΩD has the largest value. Thus, it is required to find the
 2.3 Maxwell-Boltzmann distribution law 43

values for n1 , n2 , · · · , ni which make ΩD (from eq. 2.6) a maximum. Therefore, we try to
obtain the values for ni which gives
C HAPTER 2

∂ΩD
(2.9)


= 0,
∂ni

for all ni under the restriction that the values for ni adhere to equations 2.10 and 2.11.

X
n1 + n2 + ... ≡ ni = N (2.10)
i

X
n1 1 + n2 2 + ... ≡ ni i = E (2.11)
i

where N are the number of particles and E the energy of the system as assumed by
properties of the system mentioned at the beginning of this section.
By changing the values of n1 , n2 , · · · ni with a small amount ∂ni , the value ΩD is changed
by the small amount ∂ΩD :

! ! !
∂ΩD ∂ΩD ∂ΩD
∂ΩD = ∂n1 + ∂n2 + · · · + ∂ni (2.12)
∂n1 ∂n2 ∂ni

According to equation 2.10, the above gives

∂n1 + ∂n2 + · · · + ∂ni + · · · = 0 (2.13)

and similarly equation 2.11 yields

1 ∂n1 + 2 ∂n2 + · · · + 1 ∂ni + · · · = 0. (2.14)

Maximizing ΩD subject to these restrictive conditions can be done by means of Lagrange’s

method1 of undetermined multipliers. We therefore introduce two new variables β and γ by
which we obtain

! !
∂ΩD ∂ΩD
∂ΩD = − γ − β1 ∂n1 + − γ − β2 ∂n2 (2.15)
∂n1 ∂n2
!
∂ΩD
+ ··· + − γ − βi ∂ni . (2.16)
∂ni

When ΩD has a maximum value, i.e.

∂ΩD = 0, (2.17)

the values of ni must therefore be such that the right-hand side of equation 2.15 vanishes.
Suppose that we choose arbitrary values for ∂n3 , ∂n4 , · · · , ∂ni , · · · , then ∂n1 , ∂n2 should be
chosen in such a way that equations 2.13 and 2.14 hold. Since equation 2.15 holds for any γ
and β , let us choose values for γ and β such that:
1
See Appendix B.5 on 131.
44 Chapter 2. Transition State Theory

C HAPTER 2
∂ΩD
− γ − β1 = 0 (2.18)
∂n1


∂ΩD
− γ − β2 = 0 (2.19)
∂n2

The only way to ensure that the right-hand side of equation 2.15 is zero for any small
values of ∂n3 , ∂n4 , · · · , ∂ni , · · · is to equate all values between the brackets to zero, which
generalizes to

∂ΩD
− γ − βi = 0 (i = 1, 2, 3, · · · ). (2.20)
∂ni

By not minimizing ΩD but ln ΩD instead, we obtain

∂ ln(ΩD )
− γ − βi = 0 (2.21)
∂ni

By applying Stirling’s approximation (see appendix B.1) to equation 2.6 we obtain the
following equation

X X
ln ΩD = N ln N − N + (ni ln gi ) − (ni ln ni − ni ) (2.22)
i i

and taking the first derivative of the above yields

∂ ln ΩD
= ln gi − ln ni . (2.23)
∂ni

By combining expression 2.21 and 2.23 this yields

ln gi − ln ni − γ − βi = 0 (2.24)

and with some rearranging, this gives

ln ni = ln gi − γ − βi . (2.25)

Finally taking the exponent on both sides, we obtain

ni = gi e−γ e−βi . (2.26)

Note that the above equation is the Maxwell-Boltzmann distribution law, which gives the
values for each ni for the most probable distribution of molecules among the energy levels.
We will be using this result in the next section to define the entropy of a system and the
concept of partition functions.
On the basis of equation 2.13, the sum of the most probable distribution of molecules
should be equal to unity, thus giving:
 2.3 Maxwell-Boltzmann distribution law 45
C HAPTER 2

X
ni = N, (2.27)
i
where ni is the number of particles in state i for the most probable distribution ΩD .


Furthermore,

N = g1 e−γ e−β1 + g2 e−γ e−β2 + · · · + gi e−γ e−βi + · · · (2.28)

 
= e−γ g1 e−β1 + g2 e−β2 + · · · + gi e−βi + · · ·

= e−γ gi e−βi
X

i
Hence,

N N
e−γ = P = , (2.29)
−βi f
i gi e
where

gi e−βi .
X
f = (2.30)
i
The sum f is coined by Planck as the Zustandssumme, but is also called the molecular
partition function or sum-over-states. The term partition function becomes obvious when
one wants to evaluate the most probable number of molecules in two different levels, for
example:

ni g e−βi
= i (2.31)
nj gj e−βj

In other words, for a given assembly of particles, the partition function shows how the
molecules are allocated among the different energy levels or equivalently, how the total energy
is partitioned among the molecules.
The only remaining matter is to define β . From thermodynamic considerations it is
obvious that there is some relation between the value β and the temperature T . In order to
give a clear relation between T and β , the absolute thermodynamic temperature [4, 5] needs to
be defined according to:

dQ = T dS (2.32)
wherein dS is the increase in entropy of a system when it adsorbs reversibly an amount
of heat dQ at the temperature T . By defining the above, from experimental observations, the
relation between β and T is given by

1
β= (2.33)
kb T
where kb is the Boltzmann constant. Thus, the partition function becomes

!
X −i
f = gi exp . (2.34)
kb T
i
 46 Chapter 2. Transition State Theory

2.4 Partition functions of subsystems

C HAPTER 2
Sometimes it makes sense to calculate the overall partition function of a system by considering
the individual partition functions of its subsystems. If a system can be divided into N
subsystems with negligible interaction, then the overall partition function is given by


Y
Z= fi . (2.35)
i

For example, the total partition function for an ensemble of N distinguishable particles
which all have the same partition function, the overall partition functions can be calculated
from the product of the partition function of its subsystem. If the partition function of each
of these subsystems is the same, the overall partition function is given by

Z = fN . (2.36)

If the subsystems are identical particles2 , which are impossible to distinguish from each
other, the partition function is calculated by

Y f Ni
Z= i , (2.37)
Ni !
i

which reduces to

fN
Z= , (2.38)
N!
when the individual partition functions for the particles are the same.

2.5 Thermodynamic variables

The usefulness of the partition function becomes clear when we wish to calculate any thermo-
dynamic properties. Such properties are the ensemble averages as given by

X
hV i = Vi p i , (2.39)
i

where Vi is the thermodynamic value V for micro state i and pi is the probability of micro
state i. For the total energy E for a system, this becomes

X
hEi = i pi (2.40)
i
 
−i
i i exp
P
kb T
=   (2.41)
−i
i exp kb T
P
 
−i
i i exp
P
kb T
= (2.42)
f
2 ∂ ln f
= kb T . (2.43)
∂T
2
We mean with identical here in the quantum mechanical sense.
 2.5 Thermodynamic variables 47

Other useful relationships between the partition function and thermodynamic variables
are for the entropy
C HAPTER 2

S = kb ln ΩD (2.44)


∂
kb T ln f , (2.45)


=
∂T
the pressure p

∂ ln f
 
p = kb T , (2.46)
∂V

and the chemical potential µ

∂ ln f
 
µ = −kb T . (2.47)
∂N

The equilibrium constant K for a reaction A + B  C + D can be defined as

!
f f −∆GR
K = C D = exp . (2.48)
fA fB kb T

Here we introduce a simple example to illustrate the above. Consider two systems (A and
B) which each have two states separated by an energy  and where the ground state energy of
A lies  lower in energy than system B. The partition function for system A is then given by

!
X i
fA = gi exp − (2.49)
kb T
i

= 1 + exp(− ) (2.50)
kb T

and for system B by

!

− i
X
fB = gi exp (2.51)
kb T
i
 2
= exp(− ) + exp(− ). (2.52)
kb T kb T

In Figure 2.1, the partition function of A and B as well as the occupancy (n) of energy
states A0 and A1 are shown as function of temperature. At very low temperature (please
note the logarithmic plot), only the ground state of A is occupied as this is the lowest state.
With increasing temperature, the first excited state of A becomes increasingly occupied. At
very high temperature, system A is half the time in the ground state and half the time in the
first excited state. From the ergodicity principle, this is the same as saying that if we would
construct a large ensemble of particles of A, half of these particles would be in the ground
state and half in the first excited state at any point in time. These occupancies are reflected by
the partition function of A. As at low temperature only the ground state of A is occupied, the
partition functions equals unity as this is the only state that A can reside in. With increasing
temperature, the first excited state of A becomes accessible and hence the partition function
increases towards a value of 2 with increasing temperature.
48 Chapter 2. Transition State Theory

partition function f or occupancy n (dimensionless)

C HAPTER 2
2
fA
fB


nA,0
1.5 nA,1

0.5

0
10−2 10−1 100 101 102
log(kb T /)

Figure 2.1: Partition function f for system A and B and occupancies of the energy states of A as function of
temperature.

For system B, a similar reasoning is valid. Since the ground state of B lies higher than the
ground state of A, at very low temperature it is unoccupied. Hence, the partition function of
B is around zero at very low temperature as there are no states the system can occupy and
hence there are no configurations possible. With increasing temperature, both the ground
state as well as the first excited state increase in occupancy and the total partition function for
B increases towards 2.
We can define an equilibrium between A and B with a corresponding equilibrium constant

f exp(− k T ) + exp(− k2T )

K= B = b b
. (2.53)
fA 1 + exp(− k T )
b

The value for the equilibrium constant K as function of temperature is depicted in Figure
2.2. Herein, we can see that at very low temperature the value for K is almost 0 as at this
temperature only states of A are occupied. At the very high temperature limit, all states of A
and B are equally occupied. This is reflected by K , which has a value of unity.
We could for example also calculate the average energy, which for system A would be

∂ ln f
 
EA = kb T 2 (2.54)
∂T
exp(− k T ) · 
= b
(2.55)
1 + exp(− k T )
b

and for system B

∂ ln f
 
EB = kb T 2 (2.56)
∂T
exp(− k T ) ·  + exp(− k2T ) · 2
= b b
. (2.57)
exp(− k T ) + exp(− k2T )
b b
 2.6 Partition functions for molecules 49

equilibrium constant K (dimensionless)

K
C HAPTER 2

0.8


0.6

0.4

0.2

0
10−2 10−1 100 101 102
log(kb T /)

Figure 2.2: Equilibrium between A and B as function of temperature.

Careful examination of the above two formula shows that the average energy is simply
the occupation fraction of a particular state multiplied by the energy of that state. This is
of course exactly what you would expect for the average energy. At the lowest temperature,
the average energy of a molecule corresponds to the energy of the ground state and with
increasing temperature, the average energy increases as well as more states at higher energy
are occupied.
From the above, it should be clear that once the partition function of a system has been
defined, we are able to calculate many meaningful thermodynamic properties. In the next
section, we will look into more detail what constitutes the partition function for molecules
and which corresponding energy terms are relevant.

2.6 Partition functions for molecules

In section 2.3 we derived the formula (eq. 2.34 for the partition function for a Maxwell-
Boltzmann distribution. To evaluate the partition function for a molecule, we need to investi-
gate the possible energy levels inside a molecule. The total molecular energy  is composed
of a series of energy terms

 = trans + vib + rot + el + nuc , (2.58)

where trans , vib , rot , el , and nuc are the translational, vibrational, rotational, electronic
and nuclear energies, respectively. All these energies are evaluated with respect to the lowest
state allowed, thus we can refer to the energy  as the zero-point energy or residual energy.
This definition has particular consequences for the evaluation of the vibrational energies,
which we will see shortly.
Furthermore, each of these energies can be coupled to a partition function, thus we obtain
translational, vibrational, rotational... partition functions, by means of the following equation

 
X −k
i
fk = exp  , (2.59)
kb T
i
 50 Chapter 2. Transition State Theory

where k can be any of the above mentioned types (translational, vibrational...).

Remember that a partition function represents the number of configurations possible

C HAPTER 2
at a particular energy. From this definition, it should be apparent that the total molecular
partition function is simply the product of the individual partition functions3 as given by


Y
ftotal = fi . (2.60)
i

Simply speaking; if a molecular has x configurations given a translation energy trans and
y configurations for a vibrational energy vib , the total number of configurations are z = x · y
as the translational and vibrational configurations of the molecule are independent of each
other.
The number of configurations for each type of partition function critically depend on the
temperature as is evident from equation 2.31. As such, the number of configurations for each
energy term differs significantly. The energy of the excited nuclear states is extremely large as
compared to the ground state and for almost all terrestrial purposes (relatively low T ), only
the ground state of the nuclear energy is occupied. Therefore, within this work, we assume
that the influence of the nuclear partition function is negligible and that the term cancels out.
For the electronic excited states we will provide a similar reasoning. On average, the
electronic states are separated by about 2eV (equivalent to a wave number of 15000 cm−1 ).
Therefore, for low T , only the ground state is occupied and the electronic partition function
is equal to the ground state electronic energy.
That leaves us to derive the partition functions for translation, rotation and vibration,
which will be covered in the upcoming subsections.4

2.6.1 Translations
The translational kinetic energy of a molecule confined to a rectangular box of dimensions a
x b x c depends on three quantum numbers, p, r, s as given by the following equation:

!
h2 p2 r2 s2
p,r,s = 2
+ 2 + 2 (2.61)
8m a b c

where h is Planck’s constant and m is the mass of the particle and the labels p, r, s are
particular translational quantum states such that

e−x (p)/kb T e−y (r)/kb T e−z (s)/kb T

X X X
ftrans = fx fy fz = (2.62)
p r s

Combining the two equations 2.61 and 2.62 yields

2 2 2
e−Ap e−Br e−Cs
X X X
ftrans = (2.63)
p=1 r=1 s=1

where A, B, C are
3
Similar to how we evaluate the overall partition function given a set of partition functions for its subsystems as
shown in section 2.4.
4
To get a feeling of the values and meaning of the translational, rotational and vibrational partition functions, have a
look at exercise 2.4 on page 68.
 2.6 Partition functions for molecules 51
C HAPTER 2

h2
A= (2.64)
8ma2 kb T
h2


B= (2.65)
8mb2 kb T
h2
C= (2.66)
8mc2 kb T

Now assuming that the quantum states are a continuum and we take into account all the
quantum states (up to infinity), than we can approximate the sum over states by an integral as

Z ∞ Z ∞ Z ∞
2 2 2
ftrans = e−Ap dp e−Br dr e−Cs ds. (2.67)
0 0 0

By applying the standard integral

Z ∞  1
2 1 π 2
I= e−cz dz = (2.68)
0 2 c

the expression for the translational partition function becomes

3
3
2πmkb T 2 abc 2πmkb T 2 V
ftrans = = . (2.69)
h3 h3
Note that the above formula covers the full three-dimensional partition functions, however,
in further chapters it is convenient to use the translational partition function for a single
dimension. In that case, the equation reduces to:

p
L 2πmkb T
ftrans = (2.70)
h

2.6.2 Vibrations
To obtain a value for the vibrational partition function, we will first elaborate how the vibrational
partition function is obtained for a diatomic molecule and continue to extent this result to
polyatomic molecules. Let the equilibrium internuclear distance be r0 . Then according to
Hooke’s law, perturbation in this internuclear distance r gives result to an increase of the
potential energy of the molecule by:

1
V = k (r − r0 )2 (2.71)
2
where k is the force constant.
The variation of r with time is in classical mechanics the simple harmonic motion, for
which the frequency ω is given by

s
1 k
ω= (2.72)
2π µ
52 Chapter 2. Transition State Theory

where µ is the reduced mass of the molecule

C HAPTER 2
mA mB
µ= (2.73)
mA + mB


and mA , mB are the masses of the two nuclei.
The permitted vibrational energy levels are found according to quantum mechanics to be

 
1
vib = ν+ hω (2.74)
2

where ν is the vibrational quantum number which can have any non-negative integer
value.
As explained in the beginning of this section, the vibrational energy is evaluated with
respect to the lowest accessible vibrational state, thus losing the term 12 in the value for the
vibrational energy vib . Therefore, we can write the vibrational energies as

vib = νhω (2.75)

hence the partition function for a diatomic molecule is

∞
e−νhω/kb T
X
fvib = (2.76)
ν=0

To evaluate this summation, the right hand side of the above equation is written as

e−νhω/kb T = xν (2.77)

and equation 2.76 can be written as

∞
xν = 1 + x + x2 + x3 + x4 + · · ·
X
fvib = (2.78)
ν=0

This type of an infinite series is known as the geometric series (see section B.4 in the
Appendix), which converges to

1
fvib = (2.79)
1 − e−hω/kb T

The quantity hω/kb has the dimensions of temperature and is often denoted as the
characteristic temperature for vibration and written as Θ, so that we can rewrite the previous
equation as

1
fvib = (2.80)
1 − e−Θ/T

The question that now arises is how to define equation 2.80 for a polyatomic molecule. For
any given (non-linear) polyatomic molecule with a fixed centre of mass, 3N − 6 independent
coordinates can be given for the relative position of each atom with respect to one another,
where N is the number of atoms within the molecule.
 2.6 Partition functions for molecules 53

The positions of the nuclei are defined by a set of coordinates qi , by means of a change of
basis, these variables qi can be transformed (or rearranged) in such a way that the potential
C HAPTER 2

energy of a molecular configuration, relative to that of the equilibrium configuration, can be

expressed as:


3N
X −6
V = V1 + V2 + V3 + · · · + V3N −6 = Vi (2.81)
i=1

where

1  2
Vl = kl ql − ql0 (2.82)
2

where kl is the force constant and ql0 the equilibrium value of ql . 5

Coordinates chosen in such a way that equations 2.81 and 2.82 are valid are termed normal
coordinates. The associated frequencies of these normal coordinates are similar to those of
the diatomic molecule as given by equation 2.75 and are termed the normal modes.
Considering that we are dealing with 3N − 6 normal modes for a given polyatomic
molecule, we can rewrite equation 2.80 for the full vibrational partition function as:

3N −6
Y 1
fvib = (2.83)
i 1 − e−Θi /T

However, it is more convenient to use the vibrational partition function per normal mode,
which is given by:

1
fvib = (2.84)
1 − e−hω/kb T

To conclude this subsection, a small note is in order here. In the beginning of this section
we mentioned that we wrote, by definition, the partition functions with respect to the lowest
accessible state allowed. This means, that we write the vibrational partition functions with
respect to the zero point energy. The true vibrational partition function however, has this zero
point energy term in the numerator and looks like:

e−hω/2kb T
fvib = (2.85)
1 − e−hω/kb T

 Remember

Whether or not the vibrational partition functions are defined with respect to the zero
point energy or to the true electronic ground state, has an important consequence
for the definition of the activation energy in transition state theory.

5
For a linear polyatomic molecule, the number of independent coordinates is given by 3N − 5, leading to only
3N − 5 vibrational degrees of freedom. This is explained in more detail in section 2.9 on 65.
 54 Chapter 2. Transition State Theory

2.6.3 Rotations
Finally, we have to define the rotational partition function. The rotational energy for a diatomic

C HAPTER 2
molecule is given by


J(J + 1)h2
rot(J) = (2.86)
8π 2 I

where J is any non-negative integer and I is the moment of inertia about an axis per-
pendicular to the molecular axis passing through the centre of mass of the molecule. As
the energy level rot(J) is degenerate with degeneracy grot = 2J + 1, the rotational partition
function is

∞
2 2
(2J + 1)e−J(J+1)h /8π Ikb T
X
frot = (2.87)
J=0

Similar as for the translational partition function, we replace the sum over states by an
integral, thus obtaining

Z ∞
2 2
frot = (2J + 1)e−J(J+1)h /8π Ikb T dJ (2.88)
0

yielding

8π 2 Ikb T
frot = (2.89)
h2

The above result is for a diatomic molecule. Furthermore note that this partition function
contains both degrees of freedom (i.e. in two dimensions).
Just like we have seen for the vibrational frequencies, a part of the above formula has
the dimension of temperature and we can define the characteristic temperature for rotation
represented by Θrot as

h2
Θrot = (2.90)
8π 2 Ikb

which turns equation 2.89 into

T
frot = (2.91)
Θrot

Note that the above derivation accounts for diatomic molecules. For polyatomic molecules,
without further derivation, the rotational partition function is given by

 3/2 r
kb T π
frot = , (2.92)
hc IA IB IC

where IA , IB , IC , are the moments of inertia around the three orthogonal rotation axes
A, B and C .[6]
 2.7 Collision theory 55

σAB
C HAPTER 2

~
r

Figure 2.3: Schematic depiction of the cylinder space swept out by molecule A. All molecules whose center are


within the volume as defined by the cylinder will undergo a collision.

2.7 Collision theory

Before we discuss transition state theory and derive the corresponding rate equation within
this theory, we will first treat collision theory which is a somewhat simpler theory enabling us
to calculate reaction rates, yet conveys a lot of important principles behind transition state
theory.
Consider a mixture composed of molecules which are able to react with each other. Let us
furthermore assume that these molecules can be represented as rigid spheres with diameters
σA and σB for two types of molecules A and B with nA and nB the corresponding number
of molecules per cubic meter. We can then define a bimolecular collision as the situation
where there is contact of the surfaces of the two spheres.
We wish to calculate the number of collisions per second for one molecule of A. For a
mixture of A and B, we can consider that one molecule of A will be moving in an arbitrary
direction ~r with a mean velocity relative to a molecule of type B. A collision between A and B
will occur in the situation wherein the center of molecule B is at a position within a distance
σAB = σA + σB /2 of the line of flight of the center of molecule A during the passage of

A. This is schematically depicted in Figure 2.3.

The total number of collisions of molecule A with those of type B per second can then be
estimated from the volume swept out by a sphere of radius σAB multiplied by the number of
molecules of type B per cubic meter, nB . The required volume is

2 r
V = πσAB (2.93)

and the collision number

2 rn
ncol = πσAB (2.94)
B

2.7.1 Maxwell-Boltzmann velocity distribution

The appropriate mean relative velocity ~r can be calculated from the distribution of the velocities
of molecules A and B. For this, we need to know the distribution of velocities of A and B, which
can be obtained as special cases of the Boltzmann distribution of energy as given in equation
2.31. According to the latter, the probability that a molecule is in a level of energy i and
statistical weight gi is proportional to gi exp i /kB T . The probability may be represented

by dn/n0 , where dn is the number of molecules with energy i and n0 is the total number of
molecules in the mixture.
Velocity has three components, i.e. in the ẋ, ẏ and ż direction. The translational energy
in each of the directions is given by

1
i = mẋ2 , (2.95)
2

where m is the mass of a molecule and ẋ is the first derivative towards time of the position
in the x direction. The statistical weight is the volume in phase space in units of hf for f
56 Chapter 2. Transition State Theory

degrees of freedom, which is given by

C HAPTER 2
m dẋ dx
gi = . (2.96)
h


Therefore, the fractional number of molecules with x component of velocity between ẋ
and ẋ + dẋ and x coordinate between x and x + dx is given by

!
dn m dẋ dx −mẋ2
=A exp , (2.97)
dn0 h 2kB T

where A is a proportionality constant to ensure that the total fraction of molecules is unity
(i.e. a molecule has to have a specific velocity). A can thus be determined by integrating
with respect to ẋ and x over their complete ranges from −∞ to ∞ and 0 to a, respectively,
where a is the length of the corresponding edge of the container in phase space, which is here
assumed to be rectangular. Thus, we can find A by solving the following integral

!
Aam ∞ −mẋ2
r
dn
Z Z
Aam 2πkB T
=1= exp = (2.98)
dn0 h −∞ 2kB T h m

and thus

h
−1/2
A= 2πmkB T (2.99)
a
by which the final result is

!1/2 !
dn m −mẋ2
= exp dẋ. (2.100)
dn0 2πkB T 2kB T

In three directions, the result is simply

!3/2 !
dn m −mv 2
= exp dẋ dẏ dż, (2.101)
dn0 2πkB T 2kB T

where

q
v = |~
r| = ẋ2 + ẏ 2 + ż 2 . (2.102)

Equation 2.101 can be rewritten for the case that we are not interested in which particular
direction the molecules are moving, but only in their absolute velocity. The ”volume” element
dẋ dẏ dż then becomes v 2 sin ϑ dv dϑ dϕ. Integrating over the angles gives

Z π Z 2π
sin ϑ dϑ dϕ = 4π (2.103)
0 0

and applying this result to equation 2.101 gives

!3/2 !
dn m −mv 2
= 4π v 2 exp dc. (2.104)
dn0 2πkB T 2kB T
 2.7 Collision theory 57

Equation 2.104 is the common form of the Maxwell-Boltzmann velocity distribution law.
The average velocity v is obtained by multiplying equation 2.104 by v and integrating from 0
C HAPTER 2

to ∞ which gives


 1/2
8kB T
v= . (2.105)
πm

If we want to know the mean relative velocity between two distributions, we have to
multiply the two distribution equations 2.101 for each of the two distributions and change the
velocity coordinates to relative velocity coordinates by the following substitution

2
2
2
r2 = ẋA − ẋB + ẏA − ẏB + żA − żB . (2.106)

Integrating this expression over all directions and velocities and multiplying by r gives
the following mean relative velocity

 1/2
8kB T
r= , (2.107)
πµ

where m in equation 2.105 is replaced by the reduced mass µ which is given by

mA mB
µ= . (2.108)
mA + mB

Note that in the special case that mA = mB = m,

m
µ= (2.109)
2

and

√
r= 2v (2.110)

Finally, we can plug equation 2.107 into equation 2.94 to obtain

 1/2
8kB T 2 n .
ncol = σAB B (2.111)
µ

From this, we can easily calculate the number of collisions per unit volume per molecule
of type A with molecule B

 1/2
8πkB T 2 n n .
ZAB = σAB A B (2.112)
µ

In the case that molecule A and B are like molecules, the above equation simplifies to

 1/2
πkB T
Z=2 σ 2 n2 . (2.113)
m
 58 Chapter 2. Transition State Theory

2.7.2 Collision effectiveness

Not all collisions will result in a reaction and there are a number of important factors that

C HAPTER 2
determine the effectiveness of a collision. The most important one we will consider here is
that the molecules that collide should have a minimum energy to overcome the underlying


reaction barrier. This minimum energy corresponds to a minimum velocity and hence we can
calculate using the Maxwell-Boltzmann velocity distribution an effectiveness factor q between
0 and 1 such that

r = qZ, (2.114)

where r is the rate of the elementary reaction step and Z is the collision density. If we
consider an energy ∆Eact corresponding to the minimum energy of activation by which a
reaction will occur, we can calculate the value q as follows

3/2  
R∞  2
µ 3 exp −µv
4π v 2kB T dv
!
vmin 2πkB T −∆Eact
q= = exp (2.115)
R∞ 
µ
3/2 
−µv 2
 kB T
0 4π 2πk T v 2 exp 2k T dv
B B

by which the final expression becomes

 1/2 !
8πkB T 2 n n exp −∆Eact
rcol = σAB A B (2.116)
µ kB T

for unlike molecules

and

 1/2 !
πkB T −∆Eact
rcol = 2 σ 2 n2 exp (2.117)
m kB T

for like molecules.

2.7.3 Unimolecular reactions

There is a special case which deserves a bit of additional attention. Let us consider a uni-
molecular reaction. You might argue that such a reaction does not require molecules to
meet rendering the concept of collisions as not very useful. For example, in an isomerization
reaction, the molecule itself transforms to another configuration which in principle does not
require a collision.
According to the hypothesis of Lindemann and Christiansen, all molecules acquire and
lose energy by collisions with surrounding molecules. If we assume this hypothesis to be
true, then we can construct the following set of elementary reaction steps for the reaction of
A to P activated by a collision of another molecule M:

1 k+
A+M −
)−
−*− A ∗ +M (2.118)
k1−

k+
A∗ −−2→ P (2.119)

Note that in the above equation, the asterisk designates an activated complex rather than
an adsorbed species. Application of the steady state approximation to the reaction intermediate
A* yields
 2.8 Transition state theory 59
C HAPTER 2

d[A∗]
= k1+ [A][M ] − k2+ [A∗] − k1− [A∗][M ] = 0 (2.120)
dt
k+ [A][M ]


[A∗] = +1 (2.121)
k2 + k1− [M ]

and hence the rate of reaction becomes

d[P ] k+ k+ [A][M ]
= 1+ 2 − . (2.122)
dt k2 + k1 [M ]

For a mixture only containing A, [M] is [A]. In other words, the reaction of A to P is
activated by a collision with another molecule of A and hence

d[P ] k1+ k2+ [A]2

= . (2.123)
dt k2+ + k1− [A]

For sufficiently high pressures of A, this reaction will be first order. However, in the low
pressure regime the above equation can be approximated by

d[P ] k1+ k2+ [A]2

= ≈ k1+ [A]2 . (2.124)
dt k2+ + k1− [A]

The above dependency of the rate on the pressure can be verified by experimental means
and it was found that the equation holds for several isomerization and decomposition reactions.
Interestingly, this implies that unimolecular reactions in fact do not exist at all, because
collisions with surrounding molecules are needed to bring the reacting molecule to a suf-
ficiently high energetic state that it is capable of crossing the reaction barrier of the elementary
reaction step.

2.8 Transition state theory

Within this section, we will derive the formula for the chemical rate using transition state
theory. In principle, many similar derivations exist.[4, 7, 8] Here, we will employ a derivation
wherein we assume that the reaction coordinate can be modeled as a translation.6
To derive a general formula for the reaction rate constant, let us introduce a simple model
system. In Figure 2.4, the potential energy surface of the three-body complex H-H-H for the
reaction H2 + H → H + H2 is shown.[9]
Within this Figure, the two dark regions (denoted by A and B) designate the stable states,
being the complex HB -HC with the atomic hydrogen HA far away and the situation where
HA − HB are bonded and HC as atomic hydrogen. One can construct a path between these
two states (regions) as denoted by the dotted white line. This path that connects the two
stable states wherein all the points on the path are higher in energy than the two stable
states represents the reaction coordinate. At the center of this line lies a meta-stable state
which is termed the transition state (denoted in the Figure as TS). The transition state is
characterized as being a maximum in energy along the trajectory of the reaction coordinate
and a minimum in all other directions (i.e. in any direction perpendicular to the reaction
6
There is a somewhat simpler derivation, but we leave this as an exercise to you. See exercise 2.12 for an alternative
derivation of the Eyring equation on page 70
60 Chapter 2. Transition State Theory

C HAPTER 2
2.4


2.2

2.0
rBC in a.u.

1.8

1.6

1.4

1.4 1.6 1.8 2.0 2.2 2.4

rAB in a.u.
Figure 2.4: Potential energy surface of the one-dimensional tri-atomic hydrogen system. rAB and rBC are the
distances between HA and HB and between HB and HC , respectively. The white dotted line indicates the
reaction coordinate of the elementary reaction step wherein a hydrogen atom is transferred between the two other
hydrogen atoms.

coordinate). In mathematical terms, such a point on the potential energy surface, which has
different curvatures in different directions, is known as a saddle point.

An illustrative video showing the curvature of the PES of H3 can be found

on Youtube using this link:
 https://www.youtube.com/watch?v=5y0DQhu1-CY

We want to derive a rate constant which is defined as the speed at which species residing
in state A go to state B . Using a statistical assumption, it is obvious that such a rate is
proportional to the number of molecules residing in state A.

rate = k[A] (2.125)

where k is termed the reaction rate constant and has the appropriate dimensions given the
elementary reaction step.7 In the derivation, we are going to use the following assumptions:
7
This definition for the dimensionality of the rate constant is actually a source of confusion. We will elaborate on the
topic in section 2.10 on page 65.
 2.8 Transition state theory 61

1. The transition state and initial state are in thermal equilibrium in such a way that we
can define an equilibrium constant K to describe the ratio of the species between the
C HAPTER 2

two states.

2. The transition state has a particular ”width” in the direction of the reaction coordinate


and species have a particular velocity at which they can cross this width.

3. Species that have passed the transition state from the initial state will immediately
move towards the final state.

Given these three assumptions, an elementary reaction step can be envisioned as the
following reaction sequence

RIS  RTS →irrev P (2.126)

for which we can construct the following expression for the reaction rate constant k:

k = νK, (2.127)

where ν is a crossing frequency (in s−1 ) at which species at the transition state migrate
to the final state.8
One could object to the above assumption as although the above is true for a system in
equilibrium, we are looking for reaction rates, which are also valid out of equilibrium. How
can we be sure that the equation 2.127 also holds when we study the transient behavior? From
experiment though, it is found that rate constants do not change when the system approaches
equilibrium[7], therefore, we believe that equation 2.127 holds, irrespective of whether our
system is in equilibrium or not.
Going back to our model system, for the elementary reaction step H2 + H → H + H2 ,
the equilibrium constant K can be defined as

f0
!
f −∆0
K = TS = TS
0 exp . (2.128)
fIS fIS kB T

Within this equation, ∆0 is the zero-point energy corrected energetic difference between
the stable state A and the transition state. fIS and fTS are the total molecular partition
functions of the complex (see equation 2.60) in the initial and transition state, respectively.
The prime indicates that we have extracted the electronic partition function including the zero-
point energy correction from the total
 molecular
 partition function and used the Boltzmann
equation to construct the term exp − ∆k T
0
.
b
For the derivation of ν we wish to evaluate some characteristic time τ for the system to
pass through the transition state (and then set ν as the reciprocal value of the average τ ).
Given a width δ of the transition state and an average velocity v of the species, the time τ is
given by

δ
τ = . (2.129)
v
8
It is unfortunate that the variable ν is also used for the stoichiometric coefficient, but we hope that from the context
it is always clear what ν represents.
 62 Chapter 2. Transition State Theory

Thus, the reaction rate constant equation so far yields

C HAPTER 2
0
!
v fTS −∆0
k= 0 exp . (2.130)


δ fIS kB T

Next, we extract the partition function corresponding to the reaction coordinate from the
total molecular partition function of the transition state. We have assumed that the partition
function corresponding to the direction of the reaction coordinate is a translational partition
function. By combination of equation 2.70 and the previous equation we obtain

0
p !
vδ 2πmkb T fTS −∆0
k= 0 exp . (2.131)
δ h fIS kB T

The average velocity v in one dimension can be obtained in a similar fashion as shown
for the Maxwell-Boltzmann velocity distribution in section 2.7.1:

 
R∞ −mx2
ẋdx
0 exp 2kb T
v= R (2.132)
∞
 
−mx2 dx
0 exp 2kb T
r
2kb T
= . (2.133)
πm

Plugging this equation into equation 2.131, and noting the introduction of the ”2” in the
denominator of the first term as only half of the crossings are from the IS to the FS, yields

q
2kb T p 00
!
πm δ 2πmkb T fTS −∆0
k= 0 exp . (2.134)
2δ h fIS kB T

Note that a second prime is introduced for the total motional partition function for the
transition state to indicate that we have extracted one translational partition function. The
above equation can be rewritten to

‡ !
k T fTS −∆0
k= B exp . (2.135)
h fIS kB T

Herein, we have dropped the primes and introduced a ‡ to indicate that one partition
function is extracted from the total motional partition function for the transition state.
Equation 2.135 is known as the Eyring equation or transition state theory equation.

2.8.1 Comparison between Transition State Theory and Collision theory

The fundamental difference between transition state theory and collision theory is that the
latter considers molecules to be rigid spheres and hence only the translational motion of these
rigid spheres is considered. Hence, the degrees of freedom due to rotation and vibration are
not considered in collision theory. Nevertheless, for a diatomic molecule with an internuclear
distance of σAB without any vibrational degrees of freedom, transition state theory should
 2.9 The potential energy surface 63

give the same result as collision theory. If mA and mB are the masses of the two reactant
molecules, the moment of inertia I is
C HAPTER 2

mA mB
I = σAB (2.136)


mA + mB

and thus k can be written using expression 2.135 as

‡ !
k T fTS −∆0
k= B exp . (2.137)
h fIS kB T
(3) (2) !
k T ftrans frot ∆Eact
= B exp (2.138)
h f (3) f (3) kB T
trans trans
2 2 A B m m
(2π(mA +mB )kB T )3/2 8π σAB mA +mB kB T !
k T 3 h2 ∆Eact
= B h exp (2.139)
h (2πmA kB T )3/2 (2πmB kB T )3/2 kB T
h3 h3
!1/2 !
mA mB 2 exp ∆Eact .
= 8πkB T σAB (2.140)
mA + mB kB T

The above formula is equal to equation 2.116 when multiplied by the number density of
A and B nA · nB , showing that both theories provide the same rate expression.

2.8.2 Fundamental objection against collision theory

Despite the fact that collision theory and transition state theory give the same results when the
underlying assumption of collision theory is applied to the choice of the partition functions
within transition state theory, collision theory is in conflict with thermodynamics because it
neglects vibrational and rotational degrees of freedom. This becomes immediately apparent
if we derive the equilibrium constant from collision theory

k+
K= (2.141)
k−
1/2
−∆Eact,f
  
8πkB T 2 exp
µAB σAB kB T
=  1/2 (2.142)
−∆Eact,b
 
8πkB T 2
µCD σCD exp kB T
2 r
!
σAB µCD −∆H
= 2
exp . (2.143)
σCD µAB kB T

The above formula correctly describes the relation between K and the enthalpy of the
reaction, yet such an expression is only valid at 0K or in the case when there is no entropy
change. In principle, this discrepancy could be resolved by introduction of a steric factor P
in collision theory, however such an introduction would result in a loss of the fundamental
understanding.

2.9 The potential energy surface

Although covered very briefly in the previous section (see Figure 2.4), we wish to extend our
discussion on the concept of the potential energy surface of a molecule and its degrees of
freedom. Given the geometry of the molecule, an energy corresponding to that that specific
configuration can be defined or calculated. To uniquely define such a geometry, n = 3N − 6
64 Chapter 2. Transition State Theory

coordinates for non-linear molecules and n = 3N − 5 for linear molecules need to be defined.
These numbers correspond to the degrees of freedom that the molecule has. You might expect

C HAPTER 2
that each molecule has 3N degrees of freedom, as each atom can be positioned independently
along each of the three Cartesian axes. The reduction comes from the fact that by a change of


basis, three such coordinates pertain to the central position of the molecule (its translation with
respect to an origin) and another three relate to the rotation of the molecule. A linear molecule
only has two unique rotational axes, as it is impossible to define a rotational axis along the
direction of the linearity of the molecule, hence for linear molecules the configurational
degrees of freedom equal 3N − 5. Thus, the energy of a (here, non-linear) molecule is a
function of its coordinates as given by

 
E = f x0 , x1 , · · · , x3N −6 = f (~
x) . (2.144)

And the above function defines the potential energy surface in multidimensional space.
The function f that calculates the energy is thus a map from high-dimensional space to a
scalar value (i.e. the energy) as give by

f : R3N −6 → R. (2.145)

Functions that depend on a single variable are easily plotted in a graph. Two-dimensional
functions can be plotted using a contour plot or a heat map. For three-dimensional functions
you could create an isosurface plot, but for any higher-dimensional function you need to
reduce the dimensionality in order to make a plot. Hence, potential energy surfaces can rarely
be depicted in a graph. The reason we are able to construct a contour plot as seen in Figure
2.4 is because the one-dimensional H3 system only has 2 degrees of freedom.
A molecule is considered to be stable when the Cartesian coordinates of its atoms are
in such a geometry that the molecule resides in a local minimum on this potential energy
surface. Such states are often coined initial and final states with regard to elementary reaction
steps. A transition state is defined as being a maximum in one direction, i.e. the direction
of the reaction coordinate, while being a minimum in all other directions on this potential
energy surface. Here, we will address how we can analyze in which particular state (e.g. initial,
transition or final state) a molecule is and what the relation of this state is with regard to its
partition functions.
Mathematically, initial and final states as well as transition states are extrema on the
potential energy surface. Extrema are points where the gradient equals zero:

 
∇E = ∇f x0 , x1 , · · · , x3N −6 = ~0 (2.146)

To identify the type of the extreme, either a local minimum or a saddle point, we need to
probe the second derivatives by constructing a Hessian matrix for which each element in the
matrix is given by

 
∂2E ∂ 2 f x0 , x1 , · · · , x3N −6
Hi,j = = . (2.147)
∂xi ∂xj ∂xi ∂xj

Diagonalization of this Hessian matrix (see page 133) yields a series of eigenvalues and
eigenvectors which are the force constants and the directions for the normal vibrations,
respectively. If all of these force constants are positive, this means that all vibrations are real
(see eq. 2.72) and the molecule is considered to be in a local minimum. To look at this from
 2.10 Dimensionality of equilibrium constants and reaction rate constants 65

another perspective, in principle the force constants tell us to what extend the energy of a
system increases as a result of a perturbation in one of the directions of the normal vibrations.
C HAPTER 2

If all these constants are positive, this means that by perturbing the molecule in any direction,
its energy will increase. This is another way of saying that the molecule is in a local minimum.


In line with the above reasoning, a transition state is characterized as being a maximum in
energy in the direction of the reaction coordinate, therefore, a transition state should have
one and only one negative force constant. This negative force constant then results in an
imaginary frequency.
This now brings us to the relation of the potential energy surface of a molecule with
respect to its partition functions. The partition functions of a molecule represent the motional
degrees of freedom of that molecule. We can imagine that these motional degrees of freedom
tell us something about the ways the molecule can be represented, or in other words, give us
a number of the different configurations a molecule can have. Therefore, it is obvious to see
that motional degrees of freedom add to the entropy of the species. Every molecule has 3N
partition functions. If the molecule is in the gas phase and is linear, it has 3N − 5 vibrational
degrees of freedom. When the molecule is not linear (and in the gas phase), it has 3N − 6
vibrational degrees of freedom. Linear molecules have one vibrational degree of freedom
more than non-linear molecules in exchange of having one rotational degree of freedom less.
Both linear as well as non-linear molecules have three translational degrees of freedom.
In contrast, molecules which are in the transition state have exchanged one of their degrees
of freedom into an imaginary frequency. This imaginary frequency is represented in the
Eyring formula (eq. 2.135) by the term kbhT . To derive a formula for the macroscopic rate for
any elementary reaction step, one needs to identify the different degrees of freedom for initial,
transition and final state in order to establish the required partition functions as well as the
difference in energies between the states on the potential energy surface (i.e. the reaction and
activation electronic energies). For the most basic types of reactions (adsorption, desorption
and surface reaction), this procedure is discussed in the next chapter.

2.10 Dimensionality of equilibrium constants and reaction rate constants

From section 2.5 it should be clear that the equilibrium constant K is dimensionless. This
is in contrast to many textbooks and articles wherein a unit is reported for the equilibrium
constant, where it seems that the unit actually depends on the particular reaction. For example,
one could define an equilibrium constant K for the gas phase reaction A + B → C as

pC
K= , (2.148)
pA pB

which should clearly have the dimensionality of inverse pressure (e.g. bar−1 ). The dif-
ference lies in whether K is a so-called standard equilibrium constant or not. The key word is
standard. A standard equilibrium constant, often denoted with a superscript ”0”, refers to an
equilibrium constant given a particular standard (e.g. at T = 293.15K and p = 1 atm). The
standard equilibrium constant is defined as

!
−∆G0r Y ν
K 0 = exp = ai i , (2.149)
kb T
i

where ai refers to the activity of species i. For ideal gases, the activity is given by

p
ai = γi 0i , (2.150)
p
66 Chapter 2. Transition State Theory

where p0 is the standard pressure and γi is the activity coefficient. As these activities are
(made) dimensionless, so is the equilibrium constant.

C HAPTER 2
Similar to the equilibrium constant, we can also define reaction rate constants in the
forward and backward direction for the gas phase reaction A + B → C . The forward and


backward rates are defined by

r + = k + pA pB (2.151)

and

r − = k − pC . (2.152)

To construct any meaningful mass balance, r+ and r− should have the same units.
This can only be achieved if the unit of k+ and k− differs. But if that is the case, than the
equilibrium constant K is no longer dimensionless, since

k+
K= . (2.153)
k−
Again, the solution lies in the definitions. For reaction rates, we typically use pressures
and concentrations, but in principle we should use activities and the forward and backward
rates should be

r+ = k+ aA aB (2.154)

and

r− = k− aC . (2.155)

However, as activities are assumed to scale linearly with concentrations and pressures and
since the same standard is applied to all species, we often loose the strict definition for the
sake of convenience. Thus, on the basis of the Eyring equation (equation 2.135), the reaction
rate constants should have the unity of s−1 .
 2.11 Questions and Exercises 67

2.11 Questions and Exercises

C HAPTER 2

The answers to the questions and exercises are given at the end of this book in Appendix C
(page 135).


2.11.1 Questions
a) What type(s) of partition functions are relevant to chemistry? How do these types differ?
To what kind of configurational freedoms do they relate?

b) What are the units of a partition function?

2.11.2 Exercises
The exercises are marked by a number of gears to indicate their difficulty levels.

 Exercise 2.1  

Consider the equilibrium for the dissociation of A in B and C according to

AB+C (2.156)

A has four energy levels. One ground state, two degenerate excited states ∆E above its
ground state and a second excited state 2∆E above the ground state. B has two energy levels
of which the lowest energy level is ∆E above the ground state of A and another energy level
that lies 2∆E above the ground state of A. The ground state of C lies at the same level as the
ground state of B . Furthermore, there are two degenerate excited states that lie 2∆E above
the ground state of C .

a) Provide the equilibrium condition for this system.

b) Express the equilibrium constant K in terms of the partition functions of A, B , and C .

c) What are the limiting values of K at zero and at infinite temperature?

 Exercise 2.2  

A molecule A has two energy states separated by ∆E .

a) Derive an expression for the partition function of A and calculate the value of the partition
function at 0 K and infinitely high temperature.

b) Suppose that A is in equilibrium with its isomer B which has the following energy levels
with respect to the ground state of isomer A: ∆E/2, 3∆E/4 and ∆E . Derive an expression for
the equilibrium constant K and compute the limiting values of K at low and high temperature.

 Exercise 2.3   

Two isomers A and B are in equilibrium and possess the following spectroscopically
determined energy levels
68 Chapter 2. Transition State Theory

C HAPTER 2
A
i = i∆E (2.157)
 
i
B = + 1 ∆E (2.158)


i 2

with i = 0, 1, 2, 3, ....
Provide an expression for the equilibrium constant for the isomerization between A
and B as a function of temperature and provide values for this constant at T = 0.1∆E/kb ,
T = 2∆E/kb , and T = 10∆E/kb and at infinitely high temperature. You will get an infinite
geometric series for which an exact solution is known. If you forgot about geometric series,
have a look at Appendix B.4 on page 131.

 Exercise 2.4   

a) Calculate the translational partition function of a N2 molecule at 298 K in a container

sized such that it contains one mole of ideal gas.

b) How does qtrans change with temperature, pressure and volume?

c) The wave number for the stretching frequency of N2 is 2330 cm−1 . Calculate the vi-
brational partition function corresponding to this vibrational mode and with respect to the
vibrational ground state.

d) Calculate the rotational partition function of N2 at 298 K. The moment of inertia is

1.407 · 10−46 kg · m2 and the symmetry number of N2 is 2.

e) Calculate the total partition function of N2 .

 Exercise 2.5   

a) Calculate the average velocity of an N2 molecule at room temperature (T = 298K). Make

use of the Maxwell-Boltzmann velocity distribution as given in equation 2.104 on 56. Feel
free to directly use equation B.6 on page 130 to solve the integral.

b) At which temperature is the average velocity of N2 equal to that of He at T = 298K?

c) Calculate the average translational energy of 1 mol of N2 at 100, 298 and 1000 K. Feel free
to directly use equation B.5 on page 130 to solve the integral.

 Exercise 2.6  

Given a mixture of N2 and H2 in a 1:3 ratio at 1 bar and T = 298K.

a) Calculate the number of molecules N2 and H2 in 1 m3 of this mixture.

 2.11 Questions and Exercises 69

b) The collision diameter of H2 is 0.271 nm and the collision diameter of N2 is 0.373 nm.
How many collisions are there per second between the H2 molecules?
C HAPTER 2

c) How many collisions are there between the N2 molecules?



d) How many collisions are there between N2 and H2 molecules?

e) What is the total number of collisions in 1 m3 of this mixture?

 Exercise 2.7 

Consider the gas phase isomerization of cyclopropane to propylene. Experimentally, it is

found that the rate expression is given by the equation below

!
−274000 J/mol
r = 2 · 1015 exp pcyclopropane (2.159)
RT

a) Give values for the Arrhenius pre-exponential factor and the activation energy in kJ/mol.

b) Draw an energy diagram for this reaction from transition-state theory.

c) Why does the rate of this reaction decrease when the pressure is lowered?

 Exercise 2.8  

a) Provide the general equation for the reaction of a molecule R via the transition state R#
to a product P according to transition state theory. Indicate in detail in which direction the
reaction steps are allowed to proceed and provide a rate expression in terms of the relevant
reaction constant and equilibrium constant. Draw an energy diagram which clearly shows
the energy levels of R, R# and P , as well as the barrier energy ∆E .

b) Give the general expression for the reaction rate according to transition state theory in
terms of the partition functions (no explicit expressions are required here) and ∆E .

c) What is the essential difference between transition state theory and collision theory?

 Exercise 2.9  

Consider the direct dissociative adsorption of methane on a Ru surface to give adsorbed

CH3 and H.

a) Give the main assumptions of transition state theory and draw the corresponding reaction
energy diagram for this adsorption reaction.

b) Give the general expression for the rate constant based on transition state theory.
70 Chapter 2. Transition State Theory

c) Qualitatively discuss the different factors in this equation, paying specific attention to the
relevant partition functions and the activation energy.

C HAPTER 2
d) The entropy of methane in the gas phase is about 130 J/mol/K. Estimate the pre-exponential
factor for the dissociative adsorption of methane at 800 K.


e) Assuming that the activation energy is 50 kJ/mol, estimate the rate constant for dissociative
adsorption at 800 K.

 Exercise 2.10   

a) Derive an expression for the rate of O2 dissociation into two oxygen fragments in the gas
phase according to transition state theory. Provide explicit expression for all relevant partition
functions. Clearly convey what assumptions you use and relate the partition functions to the
motional degrees of freedom of the system in the initial and transition state.

b) Rewrite your expression in the form of the Arrhenius equation and give the activation
energy and the pre-exponential factor according to the Arrhenius equation.

 Exercise 2.11   

a) Assume Hooke’s law to be valid and use a harmonic approximation of the potential energy
surface. Calculate the force constant in the direction of the reaction coordinate and the
2
force constant (k = ∂∂xV2 ) in the direction perpendicular to the reaction coordinate. Use a
numerical approximation to calculate these force constants. Obtain the required values for
your approximation from the contour lines in Figure 2.4. Do not worry if you cannot calculate
it exactly, we are interested in rough values. In fact, if you are not able to even calculate a
rough value, give at least the sign of the force constant (i.e. positive or negative).
q
b) Calculate the frequencies by ω = mk . Use the rest mass of a H atom as the value for the

mass m. What property does the frequency in the direction of the reaction coordinate have?

c) Based on your frequency analysis for a two-dimensional potential energy surface, propose
a definition for a transition state on a multi-dimensional potential energy surface. Hint: What
are the properties of a stable point on a potential energy surface to be a transition state?

 Exercise 2.12   

a) How does the vibrational frequency of a chemical bond relate to the strength of that
chemical bond?

b) Would the frequency corresponding to a vibration in the direction of the reaction coordinate
in the transition state be strong or weak? Relate your answer to the strength of a (to be formed
/ broken) bond in the transition state.
 2.12 Challenges 71

c) Assume that k = νK , where k is the rate constant according to transition state theory, ν is
the frequency factor relating to crossing the transition state and K is the equilibrium constant
C HAPTER 2

between the initial state and the transition state. We are going to derive the frequency factor
ν using a different assumption.


Assume that ν corresponds to the vibrational frequency in the direction of the reaction
coordinate (note that the frequency factor ν and the vibrational frequency ω have the same
dimensions!). Write down the expression for K in terms of the partition functions and
separate the partition functions corresponding to motions that are perpendicular to the
reaction coordinate from the single partition function corresponding to a motion in the
direction of the reaction coordinate. Assume that the motion in the direction of the reaction
coordinate is a loose vibration. Show that the answer is:

Q !
1 f Ea
K=   · Q i i exp − (2.160)
hω
1 − exp − k T j fj kb T
b

where fi are the partition functions in the transition state and fj are the partition functions
in the initial state.

d) Assume that the frequency in the direction of the reaction coordinate is weak and that
hω
kb T  1. Use a Taylor series (see Appendix section B.3) that is terminated after the linear
term in x to rewrite the expression for the partition function of the loose vibration. Show that
the answer is:

kb T
(2.161)
hω

e) Plug the results you have obtained for K in the expression for k and assume that ω = ν .
Show that you get the correct expression for the rate constant in transition state theory. Explain
the difference between this derivation of the rate constant in transition state theory and the
derivation earlier in the chapter. What is the similarity between a loose vibration and a
translation?

2.12 Challenges
Challenges are relatively difficult topics aimed towards a more fundamental understanding
of the theory. We suggest to tackle these challenges within a group. It is advisable to make
use of a laptop as typically the use of Spreadsheet software such as Excel or a programming
environment such as Python is recommended.

 Challenge 2.1 Maxwell-Boltzmann distribution

In this challenge, we are going to explore the Maxwell-Boltzmann distribution in greater
detail by considering a relatively simple example. Consider an ensemble of 5 particles among
which you can distribute 6 units of energy. There are a number of ways that 6 units of energy
can be distributed over these 5 particles, which we are going to explore in this challenge. One
way to distribute 6 units of energy among 5 particles is by placing all 6 energy units into a
72 Chapter 2. Transition State Theory

single particle and leaving the other four particles in the ground state. This can be done for
each of the 5 particles, hence there are 5 ways of doing this.

C HAPTER 2
To make this exercise a bit more tangible, we will differentiate between macrostates and
microstates. We define a macrostate by only stating the number of particles in each energy


level. Placing all 6 energy units into a single particle and leaving the other four in the ground
state is in this terminology termed as a macrostate. Exactly which particle is excited and
which are left in the ground state is in turn termed as a microstate. One way to look at this
differentiation is by considering whether or not we treat the particles as distinguishable.9 If
the particle are indistinguishable, then we could only differentiate between the macrostates,
but once we consider particles to be distinguishable, we can differentiate between many more
states. Those additional states are termed microstates and each microstate can be categorized
as a representation of a particular macrostate.
The description above is schematically represented in Figure 2.5. In this Figure, three
macrostates and their corresponding number of microstates are given.

5 20 20
6 6 6

5 5 5

4 4 4

3 3 3

2 2 2

1 1 1

0 0 0

Figure 2.5: Schematic depiction of the first three macrostates for distributing 6 units of energy over 5 particles.
Above each subfigure is the number of microstates for that particular macrostate given.

Within a Maxwell-Boltzmann distribution, all particles are distinguishable from each

other. Hence, for the first macrostate as shown in Figure 2.5, there are 5 microstates. For the
second and third macrostate, there are 20 microstates.

a) Show that the number of microstates for each macrostate is given by

N!
Ωi = Q , (2.162)
i ni !
9
Recall that within a Maxwell-Boltzmann distribution, particles are considered to be distinguishable.
 2.12 Challenges 73

where N is the total number of particles and ni is the number of particles in energy level
i. You only need to show that this formula is correct for the first three macrostates. You
C HAPTER 2

are not expected to derive this formula (although of course extra credits if you are able to do so).


b) Find the other macrostates (there are 10 in total) and show that the total number of
microstates of all macrostates amounts to 210. You can identify the remaining macrostates
by using the same schemes as shown in Figure 2.5, but another way of classifying them is
by making ”sumrows”. A sumrow is constructed using a set of numbers which are added
together to create the sum. In our case, the sum is always equals 6 and you can only use
in total 5 digits. To exemplify this, the macrostates as shown in Figure 2.5 are the following
sumrows:

(a) 6+0+0+0+0
(b) 5+1+0+0+0
(c) 4+2+0+0+0

Choose the method of your preference and show all 10 macrostates and calculate their corre-
sponding number of microstates.

c) Show that the total number of microstates for distributing q levels of energy over N particles
is given by

!
q+N −1 (q + N − 1)!
Ω(N, q) = = . (2.163)
q q! (N − 1)!

Again, you only have to show that equation 2.163 is valid for this example.

d) Calculate the average number of particles in each energy level using the following formula

X Ωi
nj = ni,j , (2.164)
Ω
i

where Ωi is the number of microstates in macrostate i, Ω is the total number of microstates

and ni,j is the number of particles in energy level j in macrostate i. Plot this average number
as a function of energy level i and determine the parameters A and c by fitting the following
distribution function

ni = A · exp (−c · i ) . (2.165)

Compare your distribution function with the Maxwell-Boltzmann distribution function. How
well does the distribution function you constructed deviate from the exact Maxwell-Boltzmann
distribution function? For example, you can calculate the R2 of your fit, but a visual inspection
is sufficient. Why does your distribution deviate from the exact result?

e) Use the equation you have obtained to calculate the temperature of this ensemble given
that the energy levels are separated by 10 kJ/mol.


 C HAPTER 2
 3

CHAPTER
C HAPTER 3


ELEMENTARY REACTION STEPS

Contents
3.1 Introduction 75
3.2 Adsorption 76
3.3 Desorption 78
3.4 Surface reactions 78
3.5 Hindered rotors 79
3.6 Final remarks 80
3.7 Questions and Exercises 81
3.7.1 Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
3.7.2 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81

3.1 Introduction
In the previous chapter, we covered on the theory behind elementary reaction rates and the
rate laws describing these. Moreover, we derived on the basis of statistical thermodynamics
the general expression for an elementary reaction step:

‡ !
k T fTS −∆0
k= B exp . (3.1)
h fIS kB T

To calculate the reaction rate constant, we need to plug in the partition functions cor-
responding to the degrees of freedom of the initial and transition state. For convenience,
we repeat the formulas for the rotational, vibrational and translational partition functions in
Table 3.1.
In this chapter we will extend on this theory and apply it to specific examples such as
adsorption, desorption and surface reactions.
 76 Chapter 3. Elementary Reaction Steps

Table 3.1: Formulas for the translational, vibrational and rotational partition functions. Note that the vibrational
partition function is given with respect to the zero-point energy.

Partition function Formula

p
L 2πmkb T
Translational (1D)
h

C HAPTER 3
1
Vibrational
1 − e−Θi /T
8π 2 Ikb T


Rotational (diatomic)
h2
 3/2 r
kb T π
Rotational (polyatomic)
hc IA IB IC

3.2 Adsorption
When a molecule absorbs from the gas phase to a (catalytic) surface, it undergoes a appreciable
change in entropy, which is represented in formula 3.1 by the partition functions. In the gas
phase, any non-linear molecule possesses three translational degrees of freedom as well as
three rotational degrees of freedom. All other degrees of freedom correspond to internal
vibrations. However, when the molecule absorbs, it loses several degrees of motional freedom.
In general, upon (chemical) adsorption, the molecule no longer has any translational or
rotational degrees of freedom and only vibrational degrees of freedom.
In the transition state, between the gaseous and absorbed state, the molecule has some
number of motional degrees of freedom left, and with respect to transition state theory, the
change in the degrees of freedom between initial and transition state can be correlated to the
reaction rate. To exemplify this with respect to equation 3.1, we will denote the number of
degrees of freedom of any specific type (either rotational, translational or vibrational) by a
number written between round braces in superscript:

‡(2) ‡(3) ‡(3N −6)

k T ftrans frot fvib
k= b (3.2)
h f (3) f (3) f (3N −6)
trans rot vib

The above formula represents non-activated adsorption, i.e. its activation energy 0 = 0.
Note that therefore the exponential term of equation 3.1 has dropped out as the exponent of
zero gives unity. Furthermore, within this equation, it can be seen that the transition state
(the nominator) has two degrees of translational freedom, three rotational degrees of freedom
and otherwise 3N − 5 vibrational degrees of freedom. As discussed in the previous section,
one of the partition functions corresponding to an imaginary frequency in the direction of the
reaction coordinate has been dropped out, resulting in a total number of partition functions
of 3N − 1 for the transition state.
In contrast, the initial state (the denominator) represent the molecule in the gas phase,
which contains three rotational and three translational degrees of freedom as denoted by
the number between round braces. Furthermore, all other 3N − 6 degrees of freedom are
vibrational. In other words, the molecule loses one translational degree of freedom from
migrating from the gas phase to the surface.
From the above assumptions, we can derive a general rate of adsorption, which is known
in Literature as Hertz-Knudsen adsorption. If we substitute the terms for the partition
functions by their respective formulas, as given in Table 3.1 and furthermore assume that (1)
the vibrational partition functions can be neglected as they approximate unity under typical
 3.2 Adsorption 77

reaction conditions and (2) there is no appreciable change in the rotational partition functions
upon adsorption, we obtain the following formula:

k T L2 2πmkb T h3


kads = b (3.3)
h L3 2πmk T
32 h2
b
C HAPTER 3

Rearrangements and cancellation of terms yields:



k T
kads = p b (3.4)
L 2πmkb T

Finally, assuming the ideal gas law

k T
P = b (3.5)
V

and noting that

V
A= (3.6)
L

PA
kads = p (3.7)
2πmkb T

In the above equation, P denotes the partial pressure of the molecule in the gas phase
and A denotes the area of the surface site on which the molecule absorbs.
Recall that the equation is valid only when the underlying assumptions are valid:

1. The transition state is a loose transition state in which the molecule still has several
translational degrees and rotational degrees of freedom left

2. The molecule loses only one translational degree of freedom between initial and transi-
tion state

3. The ideal gas law is assumed to be valid

4. The rotational degrees of freedom do not change appreciable upon adsorption

5. The vibrational degrees of freedom can be neglected as they approximate unity upon
this temperature

Although quite some assumptions are used (on top of the assumptions already made in
the derivation of transition state theory!), nevertheless, from lots of experimental results it was
found that the above equation holds. However, it can be the case that the kind of adsorption
you are studying might very well go through an activated state or has a greater loss in degrees
of freedom between initial and final. If that is the case, than you will have to re-derive equation
3.1 using the right partition functions and energies corresponding to your particular problem.
 78 Chapter 3. Elementary Reaction Steps

3.3 Desorption
Desorption is the opposite of adsorption: a bound compound leaves the catalytic surface and
goes to the gas phase. In the previous section we considered that in the transition state, the
compound had three rotational degrees of freedom, two translational degrees of freedom and
3N − 6 vibrational degrees of freedom. Furthermore, we assume that in the initial state (the
fully absorbed state), the compound only contains vibrational degrees of freedom. This gives
us the following expression for the reaction rate:

C HAPTER 3
‡(2) ‡(3) ‡(3N −6) !


k T ftrans frot fvib −Edes
k= b exp (3.8)
h (3N ) kb T
f vib

We can work this out by plugging in the formulas for the partition functions and assuming
that the vibrational partition functions approximate unity.

!
k T A 2πmkb T 8π 2 Ikb T


−Edes
k= b exp (3.9)
h h2 σh2 kb T

2
After reduction of the above equation by noting that we can express the term 8πh2 Ik as a
b
temperature, known as the characteristic temperature for rotation. After plugging this value
in and combining similar terms, this yields:

!
k T 3 A 2πmkb


−Edes
k = b3 exp (3.10)
h σθrot kb T

Note that in the above derivation, we have introduced the symmetry number σ , which
represents the number of indistinguishable orientations of the particular molecule as a result
of rotation. [10] For diatomic molecules, it is easily seen that the symmetry value is either
two (when the atoms are identical) or one (when the atoms are different). However, if we are
dealing with a rotational partition function of a polyatomic molecule, the symmetry factor
can be larger than two.
The symmetry number for a particular compound can be easily derived from group theory.
Once the point group of the complex is known, the symmetry number equals the sum of
the different rotational symmetry operators.[11] To exemplify this: Ammonia belongs to the
C3v point group, which has one E and two C3 symmetry operations. As such, the symmetry
number for ammonia is 3. Likewise, benzene belongs to the D6h symmetry group, which
has one E , two C6 , two C3 and seven C2 rotational axes. This yields a symmetry number of
12 for benzene.

3.4 Surface reactions

Complexes absorbed on a catalytic surface typically only have vibrational degrees of freedom,
as such, for reactions on a surface the pre-exponential factor is mainly governed by the kbhT
term in eq. 3.1. This term is roughly in the order of 1013 .
Surface reactions are activated, i.e. they proceed through an transition state. The dif-
ference between initial and final state is known as the activation energy or reaction barrier.
With reference to the partition functions as given in Table 3.1 on page 76, the reaction barrier
is equal to the difference in the electronic energy corrected for the zero point energy.
 3.5 Hindered rotors 79

∆Ea = ∆elec + ∆zpe (3.11)

Although for many reactions the zero point energy corrections tend to be in the order
of 1-5 kJ/mol in comparison to electronic energies of roughly 30-100 kJ/mol and therefore
negligible, they become especially important when dealing with (de-)hydrogenation reactions
C HAPTER 3

and can no longer be neglected.

If we can neglect the contribution of the vibrational partition function, the equation for a
surface reaction becomes


!
k T −Ea
k = b exp (3.12)
h kb T

3.5 Hindered rotors

In the previous chapter we stated that normally complexes absorbed on a surface only have
vibrational degrees of freedom. This is not always true. A well-known circumstance when we
can no longer model the motional degrees of freedom for strongly absorbed complexes as
vibrations is when they contain one or more rigid rotors. One can imagine a rigid rotor as a
non-free rotation, such as is the case for a methyl (CH3 ) or ammonia (NH3 ) absorbed on the
surface. Another example is when the molecule contains free methyl groups, which have no
interaction with other complexes or with the catalytic surface.
Although it is not possible to give a (general) closed-form expression of the partition
function, it is relatively easy to calculate the partition function either using a classical or a
quantum mechanical approximation.[12] The latter is beyond the scope of this book, but we
wish to give a couple of examples using the classical approximation.
Assume we have a NH3 group absorbed on a catalytic surface in such a way that the
rotation is parallel to the surface, then we can imagine that this rotation is hindered. The
complex has a higher energy when the hydrogen atoms lie above the bonds between the metal
atoms as when the hydrogen atoms lie in between the bonds. Furthermore, the energetic
barrier for rotation is very low.
We will describe such a system using a Lagrangian:

L=T −V (3.13)

note that we assume that V ≈ 0 and that we can express T as the kinetic energy for
rotation

1 2 1X
T = Iω = mi Ri2 ϕ̇2 (3.14)
2 2
i

where I is the rotational moment of inertia and ϕ the angular velocity.

The partition function for this Lagrangian system is given by

Z ∞ !
1 2π −E(γ)
Z
f = dϕ dγ exp (3.15)
h 0 −∞ kb T
 80 Chapter 3. Elementary Reaction Steps

where ϕ is the angle (and therefore represents position) and γ is the momentum. Note
that E depends on the momentum, but not on the position of the atoms in the system.
We can derive an expression for γ by noting that

∂L
γ= = Iϕ (3.16)
∂ ϕ̇

C HAPTER 3
Plugging the expression for γ into the expression for the kinetic energy gives


γ2
E= (3.17)
2I
Using the expression for E(γ), we can calculate the partition functions by means of
equation 3.15, obtaining:

2π p
fhind. rot = 2πIkb T (3.18)
h

In the beginning of this subsection, we considered the ammonia molecule as a hindered

rotor. The rotational moment of inertia of this molecule is equal to

INH3 = 3mH R2 (3.19)

where mH is the mass of a hydrogen atom and R the length of a N-H bond. Plugging
this expression into equation 3.18 gives

3
(2π) 2 R p
fhind. rot = 3kb mH T (3.20)
h

3.6 Final remarks

The previous sections showed explicit formulas for adsorption, desorption and surface reac-
tions. In principle, you can use these formulas to study describe the majority of the systems
out there.
However, we are sometimes interested in that particularly peculiar system, wherein none
of the above expressions are valid. In such a case, one has to start re-deriving the expression
for the reaction rate by first inventorying the applicable partition function, plugging these
into the general rate expression and finally deriving the reaction rate.
In the end, this chapter has shown you how to set up reaction rates. In the upcoming
chapter we will show that the sum of all these elementary reaction steps is the basis of a
microkinetic model which can be used to study the behavior of a chemical system.
 3.7 Questions and Exercises 81

3.7 Questions and Exercises

The answers to the questions and exercises are given at the end of this book in Appendix C
(page 135).

3.7.1 Questions
1. What is the difference between a rotational degree of freedom and a hindered rotor?
What are the similarities?
C HAPTER 3

2. The zero point energy originates from the occupancy of the ground state of a vibration.
When is this zero point energy correction more significant, for atoms with a low mass


or with a high mass? Explain your reasoning.

3.7.2 Exercises
The exercises are marked by a number of gears to indicate their difficulty levels.

 Exercise 3.1 

a) Calculate the rate constant for CO dissociation in the forward and the backward direction
using appropriate assumptions regarding the nature of the partition functions. Take a forward
barrier of ∆Ef = 80 kJ/mol and a backward barrier of ∆Eb = 120 kJ/mol. Use a temperature
of 500K.

b) Calculate the rate constant for CO desorption. Use equation 3.10. For the surface area of
the adsorption site, assume that CO is adsorbed on a threefold site of an FCC Rh lattice. The
Rh-Rh distance is 2.71 Å. Use a rotational temperature (θ) of 2.8K, a desorption energy of 120
kJ/mol and a temperature of 500K.

c) Calculate the rate constant for CO adsorption. Assume that the partial pressure of CO
is 20 bar. Use the same surface area as you used for the calculation of the desorption rate
constant.

d) Why is the rate constant for desorption higher than for adsorption?

 Exercise 3.2 

a) Calculate the forward rate constant of CH2 hydrogenation to CH3 . Assume that CH3 has
one hindered rotor in the transition state and that all other configurational degrees of freedom
are vibrational. Use an activation energy of 50 kJ/mol. Use a C-H bond length of 1.2 Å and a
temperature of 500K.

b) Calculate the apparent activation energy of this elementary reaction step using the following
expression: Ea = RT 2 ∂∂T ln k . By what amount is the apparent activation energy higher than

the electronic activation energy?


 C HAPTER 3
 4

CHAPTER
MICROKINETIC MODELING
C HAPTER 4


Contents
4.1 Introduction 83
4.2 Microkinetic modeling of a simple catalytic reaction 84
4.2.1 Unimolecular catalytic reaction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
4.2.2 Bimolecular catalytic reaction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
4.3 Reaction orders 94
4.4 Apparent activation energy 96
4.5 Degree of rate control 99
4.6 Degree of selectivity control 101
4.7 Comparison with Langmuir-Hinshelwood kinetics 101
4.8 Questions and Exercises 106
4.8.1 Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
4.8.2 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106

4.1 Introduction
In chapter 2, we explained how a system containing different chemical compounds can be
described by only a very small number of macroscopic parameters, such as temperature,
pressure and concentration. At the end of that chapter, we introduced transition state theory
by which we are able to derive the Eyring equation which expresses the rate of interchange
between the chemical compounds once the activation energy and the partition functions are
known. In chapter 3, we provided a couple of examples of the rate of change for particular
types of elementary reaction steps.
Within this chapter, we will show how to connect multiple elementary reaction steps
together to set up a microkinetic model from scratch. We show how to construct the fun-
damental set of ordinary differential equations describing the system and how this set can
be time-integrated using the appropriate boundary conditions and initial values. Finally, we
show what observables one can derive from a microkinetic model and to which experimental
data these observables can be related.
 84 Chapter 4. Microkinetic Modeling

4.2 Microkinetic modeling of a simple catalytic reaction

Within this section, we will show how to set up a microkinetic model by providing a simple
example of a catalytic reaction. For the moment, we will stick to unimolecular elementary
reaction steps to keep things simple. More elaborate chemical systems will be shown further
on in this chapter. The best way to understand how microkinetic simulations work is by
constructing a microkinetic simulation by yourself. For this purpose, we have constructed a
small set of Python scripts by which we will slowly introduce the methodology to you. These
Python scripts are by no means the most efficient way to construct or perform a microkinetic
simulation, but it serves as a handy tool to learn the principles behind such simulations.1 The
reader is encouraged to download the Python scripts and use these while reading this chapter.
Look out for the QR codes with the download links in this chapter. We recommend running
the scripts in a Python IDE such as Spyder.2

C HAPTER 4
The steps we need to take to construct a microkinetic model are as follows:
1. Construct the set of elementary reaction steps.


2. Derive rate expressions for each of the elementary reaction steps.

3. Convert the set of rate expressions to a set of ordinary differential equations.

4. Define boundary conditions for the system (e.g. partial pressures), the initial values
(e.g. initial surface concentrations) and any model parameters (e.g. temperature).

5. Solve the system of ordinary differential equations.

6. Interpret the results (using our chemical intuition).

4.2.1 Unimolecular catalytic reaction

A unimolecular reaction on a catalytic surface is given by the kinetic network below. In this
network, A adsorbs on the catalytic surface, is then converted to B and finally B desorbs from
the surface.

A −−→ A ∗ −−→ B ∗ −−→ B. (4.1)

The overall reaction for this chemokinetic network is

A −−→ B, (4.2)

which is composed of the following three elementary reaction steps

A + ∗  A∗ (4.3)
B + ∗  B∗ (4.4)
A∗  B ∗ . (4.5)

The above system has two adsorption-desorption steps and one surface step. We will
express the rates for adsorption-desorption steps by using Hertz-Knudsen kinetics (see equa-
tions 3.7 and 3.10 on pages 77 and 78, respectively). Furthermore, we will use the assumptions
as discussed in section 3.4 on page 78 for the surface reactions. The parameters for the
adsorption and desorption steps are given in Tables 4.1 and 4.2, respectively.
1
In the next Chapter, the program MKMCXX will be introduced. This program is highly optimized to perform
microkinetic simulations.
2
More information about Spyder is found in this link: https://www.spyder-ide.org/
 4.2 Microkinetic modeling of a simple catalytic reaction 85

Table 4.1: Kinetic parameters for the adsorption and desorption steps of A and B in the unimolecular reaction.

Compound A [m2 ] m (a.u.) σ [-] θrot [K] Edes [kJ/mol]

A 1 · 10−20 1 1 1 120
B 1 · 10−20 1 1 1 220

Table 4.2: Kinetic parameters for the conversion of A to B over the catalytic surface in the unimolecular reaction.

f
Reaction νf [s−1 ] νb [s−1 ] Ea [kJ/mol] Eab [kJ/mol]
A* −−→ B* 1013 1013 50 150

Note from these two Tables that adsorbed B is significantly more stable than adsorbed A
C HAPTER 4

(120 versus 220 kJ/mol, respectively). Furthermore, the reaction on the surface from A to B
has a relatively low activation barrier (50 kJ/mol), while the barrier for the reverse reaction is


quite high (150 kJ/mol). On the basis of the fundamental equations and the data in Tables 4.1
and 4.2, we can construct the following three Python functions to calculate the reaction rate
constants for adsorption, desorption and the surface reaction as shown in Listing 4.1.

The Python code to time-integrate the unimolecular reaction at a specific

temperature can be found using the following link:  https://www.
mkmcxx.nl/downloads/input/solve_mkm_ode_unimol.py

The function calc_k_arr as shown in Listing 4.1 calculates the Arrhenius-type rate
constant as given by equation 3.12 on page 79. The functions calc_kads and calc_kdes
calculate the reaction rates for adsorption and desorption, respectively. These functions are
based on equations 3.7 and 3.10 on pages 77 and 78, respectively. The input parameters
and the units of the variables are explained in the comment section of the corresponding
functions.
In order to set up a microkinetic model using the above-mentioned kinetic parameters,
we have to specify initial values, boundary conditions and model parameters. These are as
follows

• Initial values: The initial values are the initial surface concentrations at time t = 0. For
our system, we will consider the initial surface concentrations of compound A and B at
t = 0 to be θA = θB = 0. The fraction of empty surface sites will be set to θ∗ = 1.

• Boundary conditions: The partial pressure of A is set to pA = 1 atm and the partial
pressure of B is set to pB = 0 atm. In other words, we operate the reaction at zero
conversion. This choice deserves some additional explanation. Within our microkinetic
simulation, we wish to study the behavior of our catalyst. We wish to compare our
theoretical model with a so-called initial rate experiment, where the production rate is
measured close around X = 0.3 In this situation, the reactants have only barely settled
on the catalytic surface and only the adsorbed species are in a steady state. To ensure
zero conversion, we keep the partial pressures in the gas phase fixed. Nevertheless, if
one is interested in the solution at any non-zero conversion, it is easy to modify these
settings by simply changing the gas phase pressures.

• Model parameters: As all rate expressions explicitly depend on the temperature, we

also have to set the temperature as a model parameter. Typically, this is the only
3
This is often termed as that one operates the reaction at differential conditions.
86 Chapter 4. Microkinetic Modeling

Listing 4.1: Python functions to calculate the reaction rate constants.

def calc_k_arr(T, nu, Eact):

C HAPTER 4
1
2 """
3 Calculate reaction rate constant for a surface reaction
4


5 T - Temperature in K
6 nu - Pre-exponential factor in s^-1
7 Eact - Activation energy in J/mol
8 """
9 R = 8.3144598 # gas constant
10 return nu * np.exp(-Eact / (R * T))
11
12 def calc_kads(T, P, A, m):
13 """
14 Reaction rate constant for adsorption
15
16 T - Temperature in K
17 P - Pressure in Pa
18 A - Surface area in m^2
19 m - Mass of reactant in kg
20 """
21 kb = 1.38064852E-23 # boltzmann constant
22 return P*A / np.sqrt(2 * np.pi * m * kb * T)
23
24 def calc_kdes(T, A, m, sigma, theta_rot, Edes):
25 """
26 Reaction rate constant for desorption
27
28 T - Temperature in K
29 A - Surface area in m^2
30 m - Mass of reactant in kg
31 sigma - Symmetry number
32 theta_rot - Rotational temperature in K
33 Edes - Desorption energy in J/mol
34 """
35 kb = 1.38064852e-23 # boltzmann constant
36 h = 6.62607004e-34 # planck constant
37 R = 8.3144598 # gas constant
38 return kb * T**3 / h**3 * A * (2 * np.pi * m * kb) / \
39 (sigma * theta_rot) * np.exp(-Edes / (R*T))
 4.2 Microkinetic modeling of a simple catalytic reaction 87

mandatory model parameter, although one could imagine that for more complex types
of simulations other model parameters are required. For example, if one models an
electrochemical reaction, one could consider the pH of the solution or the potential as
model parameters.

Using the initial values, boundary conditions and model parameters, the system of
ordinary differential equations is completely defined and can be time-integrated. In contrast
with the types of equations we saw in Chapter 1, this set of equations can no longer be solved
analytically4 and we have to use a numerical method to integrate this system over time. This
can be done using an ordinary differential equation solver (in short: ODE solver). In this
chapter, we will be using the SciPy library that readily provides an ODE solver.5
The set of ordinary differential equations for this system to solve is given in equations
4.6 - 4.8. Please note that in these equations kb refers to the reaction rate constant in the
C HAPTER 4

backward direction and not to the Boltzmann constant.



∂θA
= kads,A · pA · θ∗ − kdes,A · θA − kf · θA + kb · θB (4.6)
∂t
∂θB
= kads,B · pB · θ∗ − kdes,B · θB + kf · θA − kb · θB (4.7)
∂t
∂θ∗
= −kads,A · pA · θ∗ + kdes,A · θA − kads,B · pB · θ∗ + kdes,B · θB (4.8)
∂t

Using this set of ordinary differential equations and the proper boundary conditions, we
can readily construct our Python code to time-integrate our chemokinetic system. The code
is provided in Listing 4.2. By plotting the return variables of solve_odes(T) we can obtain
a graph. This is done in the main() function of the script as can be seen in Listing 4.3.

Listing 4.3: Python functions to perform the time-integration of the set of ordinary differential equation.

1 def main():
2 plt.figure()
3 x,y = solve_odes(1200)
4 labels = ['A','B','*']
5 for i in range(0, len(labels)):
6 plt.semilogx(x, y[:,i], label=labels[i])
7 plt.legend()
8 plt.show()

The ODE solver is constructed using a generic interface as explained in detail on its
webpage.6 First, we need to construct a function that takes as input the two parameters
t (time) and y (surface concentrations). Optionally, a list of parameters can be provided.
Here, we provide as additional parameters the boundary conditions, which are the partial
pressures and the model parameters, which is the temperature. In dydt(T, y, params),
the first derivative of the surface concentrations is calculated and returned as a vector. The
time-integration is performed in the function solve_odes(T). Herein, an integration object
is constructed and the function dydt(T, y, params) is passed as input. We use the vode
keyword, which stands for Real-valued Variable-coefficient Ordinary Differential Equation. We
set the absolute and relative tolerances (atol and rtol) to 10−8 . The maximum number
of steps between two output time steps of the integrator object is set to 1000, although
the default value for this variable would probably suffice. An important thing to remember
about chemokinetic systems is that their underlying set of ordinary differential equations
4
Recall that for constructing an analytical expression, we require a series of assumptions.
5
https://docs.scipy.org/doc/scipy/reference/generated/scipy.integrate.ode.html
6
https://docs.scipy.org/doc/scipy/reference/generated/scipy.integrate.ode.html
88 Chapter 4. Microkinetic Modeling

Listing 4.2: Python functions to perform the time-integration of the set of ordinary differential equation.

1 def solve_odes(T):
2 # initial conditions
3 y0 = [0,0,1]
4 t0 = 0
5 t1 = 1e-6 # total integration time
6 T = 1200 # temperature in K
7 pa = 1e5 # pressure of A in Pa

C HAPTER 4
8 pb = 0 # pressure of B in Pa
9
10 # construct ODE solver
11 r = ode(dydt).set_integrator('vode', method='bdf',


12 atol=1e-8, rtol=1e-8, nsteps=1000, with\_jacobian=True)
13 r.set_initial_value(y0, t0).set_f_params([T, pa, pb])
14
15 # integrate on a logaritmic scale
16 xx = np.linspace(-12.0, np.log10(t1), int((np.log10(t1) + 12.0) * 10))
17 yy = []
18 tt = []
19 for x in xx:
20 tnew = 10.0**x
21 tt.append(tnew)
22 yy.append(r.integrate(tnew))
23
24 return tt, np.matrix(yy)
25
26 def dydt(t, y, params):
27 """
28 Set of ordinary differential equations
29 """
30 T = params[0]
31 pa = params[1]
32 pb = params[2]
33
34 dydt = np.zeros(3)
35
36 ma = 1.66054e-27
37 mb = 1.66054e-27
38
39 k_ads_a = calc_kads(T, pa, 1e-20, ma)
40 k_des_a = calc_kdes(T, 1e-20, ma, 1, 1, 120e3)
41
42 k_ads_b = calc_kads(T, pb, 1e-20, mb)
43 k_des_b = calc_kdes(T, 1e-20, mb, 1, 1, 220e3)
44
45 kf = calc_k_arr(T, 1e13, 50e3)
46 kb = calc_k_arr(T, 1e13, 150e3)
47
48 dydt[0] = k_ads_a * y[2] - k_des_a * y[0] - kf * y[0] + kb * y[1]
49 dydt[1] = k_ads_b * y[2] - k_des_b * y[1] + kf * y[0] - kb * y[1]
50 dydt[2] = -k_ads_a * y[2] + k_des_a * y[0] - k_ads_b * y[2] + k_des_b * y[1]
51
52 return dydt
 4.2 Microkinetic modeling of a simple catalytic reaction 89

is a so-called stiff system. Stiff systems require special integration methods, such as bdf
which stands for Backward Differentation Formulas. The bdf is one of the best methods to
solve such systems. The other important thing to take care of is to either explicitly specify
to calculate the Jacobian using finite differences (with_jacobian=true) or to construct a
Jacobian function yourself. Here, we have opted for using the finite difference method, as it is
much easier to use. An explicit function to calculate the Jacobian is however computationally
much faster and is the recommended procedure.7
We have chosen to perform our integration on a logarithmic scale. The main advantage of
this procedure is that we get a fixed number of data points for every time scale, beginning at
the smallest time scales (i.e. corresponding to the fastest reactions) and ending at the longest
time-scales (i.e. the slowest reactions). The results would be exactly the same if a linear time
scale is chosen, it is just a preference we like to employ.
Before we start performing the simulations and interpreting its output, we would like to
C HAPTER 4

repeat some nomenclature here. All possible sets of concentrations applicable to our system is
termed the phase space. To exemplify this statement with respect to our system, the complete


phase space is defined by:

θA , θB , θ∗ ∈ [0, 1] (4.9)

under the constraint (remember that there is conservation of mass)

θA + θB + θ∗ = 1 (4.10)

Furthermore, if our system has no peculiarities such as strong non-linear behavior or

bifurcations (if you have never heard of these terms before, just ignore them), then our system
will most likely converge to a steady state solution, which is defined as:

∂θi
=0 (4.11)
∂t
for all i, where i is each compound in the system.
Let us apply this newly learned nomenclature to our system which models a unimolecular
reaction. In Figures 4.1a-4.1d, the transient behavior of our system (i.e., the state of our system
as a function of time) has been studied at four different temperatures. In Figures 4.1a-4.1c, a
similar final result is obtained, which is the catalytic surface fully covered with B. Note that in
Figure 4.1a, the surface concentrations no longer changes at t ≥ 10. Therefore, we can say
that at t ≥ 10 the system is at its steady-state solution.
From the other Figures (4.1b-4.1d), we can see that the steady state solution is reached
at different time t given a different temperature. At higher temperatures, the steady state
solution is obtained in a shorter amount of time. This can easily be explained due to the
fact that all elementary reaction rates increase exponentially with increasing temperature.
Therefore, our system converges to the steady-state solution faster at elevated temperatures.
One might wonder then as to why the system in 4.1d (i.e., at a significantly high tem-
perature, does not converge to a similar-steady solution as the other Figures. In 4.1d, it is
found that the steady state solution has a nearly empty catalytic surface. This has to do with
the temperature dependence of the pre-exponential factors for adsorption and desorption.
From equations 3.7 and 3.10 √
on pages 77 and 78, respectively, it can be seen that the rate
for adsorption drops with T , while the rate for desorption increases by T 3 . As the latter
becomes dominant at high temperature, at highly elevated temperatures the desorption term
dominates resulting in a nearly empty surface.
7
Consult the web page of SciPy for more information how this is done.
 90 Chapter 4. Microkinetic Modeling

1.0 A* 1.0 A*
B* B*
* (free sites) * (free sites)
Surface coverage (dimensionless)

Surface coverage (dimensionless)

0.8 0.8

0.6 0.6

0.4 0.4

0.2 0.2

0.0 0.0
10 11 10 9 10 7 10 5 10 3 10 1 101 10 11 10 9 10 7 10 5 10 3 10 1 101
Time t in s Time t in s

(a) T = 200K. (b) T = 350K.

C HAPTER 4
1.0 A* 1.0 A*
B* B*
* (free sites) * (free sites)
Surface coverage (dimensionless)

Surface coverage (dimensionless)

0.8 0.8


0.6 0.6

0.4 0.4

0.2 0.2

0.0 0.0
10 11 10 9 10 7 10 5 10 3 10 1 101 10 11 10 9 10 7 10 5 10 3 10 1 101
Time t in s Time t in s

(c) T = 800K. (d) T = 1500K.

Figure 4.1: Surface coverage as a function of time for the unimolecular catalytic surface reaction.

In Figure 4.2, the consumption of A and the production of B are shown. Due to the
strong difference in adsorption and desorption rates at different temperatures, a phenomenon
similar to the is seen. The reasoning goes as follows: at very low temperature, all compounds
adsorb easily on the surface, but due to the low temperature, there is not enough energy to
overcome the reaction barrier and no products are formed. At too high temperature, the
desorption rate is several orders of magnitude higher than the adsorption rate and almost
no reagents are found on the surface. The kinetic energy of the gaseous species is too high,
thus the molecules just ricochet off the surface. Due to the existence of these two competing
effects, i.e. rate of dissociation versus surface coverage, the production of compound B with
respect to temperature shows an optimum (at T ≈ 1175K ).
Finally, the uptake of compound A equals the production of compound B, as to be expected
on the basis of the stoichiometry of the reaction and the inherent mass conservation of the
method.
To summarize, we have shown that the overall (catalytic) unimolecular reaction is in fact
a composition of three elementary reaction steps. By means of transition state theory, we
are able to derive rate equations for these elementary reaction steps. The implementation
of these rate equations reveal interesting transient phenomena and enable us to study the
kinetic system as function of temperature. Despite that the unimolecular catalytic reaction is
very illustrative, it lacks the details of a more realistic overall reaction. As such, we are going
to describe a catalytic bimolecular reaction.

4.2.2 Bimolecular catalytic reaction

A typical bimolecular reaction in catalysis is the oxidation of CO to CO2 . This particular

system is relevant to car-exhaust gas clean-up. The overall reaction is
 4.2 Microkinetic modeling of a simple catalytic reaction 91

A
B
4000000

2000000
1
Rate in s
0

2000000

4000000

800 900 1000 1100 1200 1300 1400 1500

C HAPTER 4

Temperate T in K

Figure 4.2: Production rates of compounds A and B as a function of temperature.



2CO + O2 → 2CO2 (4.12)

and constitutes the following elementary reaction steps

CO + ∗ CO∗ (4.13)
O2 + 2∗ 2O∗ (4.14)
CO ∗ +O∗ CO2 ∗ +∗ (4.15)
CO2 + ∗ CO2 ∗ (4.16)

These elementary reaction steps lead to the following set of ordinary differential equations

∂θCO
= k1,ads θ∗ − k1,des θCO − k3,f θCO θO + k3,b θCO2 θ∗ (4.17)
∂t
∂θO
= 2k2,ads θ∗2 − 2k2,des θO
2 −k
3,f θCO θO + k3,b θCO2 θ∗ (4.18)
∂t
∂θCO2
= k4,ads θ∗ − k4,des θCO2 + k3,f θCO θO − k3,b θCO2 θ∗ (4.19)
∂t
∂θ∗
= −k1,ads θ∗ + k1,des θCO − 2k2,ads θ∗2 + 2k2,des θO
2
∂t
+ k3,f θCO θO − k3,b θCO2 θ∗ − k4,ads θ∗ + k4,des θCO2 (4.20)

This reaction is bimolecular in the sense that adsorbed CO needs to recombine with
adsorbed O in order to form CO2 . The numerical values for the relevant rate expressions are
given in tables 4.3 and 4.4.

The Python code to simulate CO oxidation as a function of temperature can

be found using this link:  https://www.mkmcxx.nl/downloads/
input/solve_mkm_ode_bimol.py

Combining the data in Tables 4.3 and 4.4 and the set of ordinary differential equations
leads to the Python code as shown in Listing 4.4. Note that we have introduced two auxiliary
92 Chapter 4. Microkinetic Modeling

Listing 4.4: Python functions to construct the first derivative towards time of the surface concentrations. Note
that we have introduced a set of auxiliary variables (i.e. r1f - r4b) to capture similar terms for the reaction rates.

1 def dydt(t, y, params):

C HAPTER 4
2 """
3 Set of ordinary differential equations
4 """
5 T = params[0]


6 pa = params[1]
7 pb = params[2]
8 pc = params[3]
9
10 dydt = np.zeros(4)
11
12 ma = 28 * 1.66054e-27
13 mb = 32 * 1.66054e-27
14 mc = 80 * 1.66054e-27
15
16 # calculate all reaction rate constants
17 k_ads_1 = calc_kads(T, pa, 1e-20, ma)
18 k_des_1 = calc_kdes(T, 1e-20, ma, 1, 2.8, 80e3)
19 k_ads_2 = calc_kads(T, pb, 1e-20, mb)
20 k_des_2 = calc_kdes(T, 1e-20, mb, 2, 2.08, 40e3)
21 kf = calc_k_arr(T, 1e13, 120e3)
22 kb = calc_k_arr(T, 1e13, 80e3)
23 k_ads_4 = calc_kads(T, pc, 1e-20, mc)
24 k_des_4 = calc_kdes(T, 1e-20, mc, 1, 0.561, 10e3)
25
26 # collect similar terms in new variables
27 r1f = k_ads_1 * y[3]
28 r1b = k_des_1 * y[0]
29 r2f = k_ads_2 * y[3]
30 r2b = k_des_2 * y[1]**2
31 r3f = kf * y[0] * y[1]
32 r3b = kb * y[2] * y[3]
33 r4f = k_ads_4 * y[3]
34 r4b = k_des_4 * y[2]
35
36 dydt[0] = r1f - r1b - r3f + r3b
37 dydt[1] = 2.0 * r2f - 2.0 * r2b - r3f + r3b
38 dydt[2] = r3f - r3b + r4f - r4b
39 dydt[3] = -r1f + r1b - 2.0 * r2f + 2.0 * r2b + r3f - r3b - r4f + r4b
40
41 return dydt
 4.2 Microkinetic modeling of a simple catalytic reaction 93

Table 4.3: Parameters for the adsorption and desorption steps in CO oxidation. Note that the values used here
are not necessarily representative for the actual process.

Compound A [m2 ] m (a.u.) σ [-] θrot [K] Eads [kJ/mol]

CO 1 · 10−20 28 1 2.8 80
O2 1 · 10−20 32 2 2.08 40
CO2 1 · 10−20 80 1 0.561 10

Table 4.4: Parameters for the surface reaction of CO oxidation. Note that the values used here are not necessarily
representative for the actual process.

f
Reaction νf [s−1 ] νb [s−1 ] Ea [kJ/mol] Eab [kJ/mol]
CO* + O* −−→ CO2 * + * 1013 1013 120 180
C HAPTER 4

variables for each elementary reaction step to efficiently collect terms and calculate the first


derivatives with respect to time.

For the boundary conditions, we have used the stoichiometric ratio of 1:2 for the partial
pressure of O2 and CO and set the total pressure to 20 atm.

0.010 CO
O2
CO2

0.005
1
Rate in s

0.000

0.005

0.010
300 400 500 600 700 800 900 1000
Temperate T in K

Figure 4.3: Production rate of CO2 and consumption rates of CO and O2 as a function of temperature.

The result for this simulation is given in Figure 4.3. From this Figure, we can see that
the uptake of CO is the same as the production of CO2 , whereas the uptake of O2 is half the
uptake of CO. This shows that this reaction proceeds according to the stoichiometry, as was
to be expected. Again, we note that there is an optimum in the production as a function of
temperature. Similarly to the unimolecular reaction, this relates to the Sabatier’s principle. If
the temperature is too low, no coupling between CO and O will occur, if the temperature is
too high, CO and O2 will no longer adsorb on the surface in order to react.
In Figure 4.4, the surface coverage as a function of temperature is shown. Here we can
see that at low temperature, the surface is mainly covered by CO. This is evident from the
relatively strong adsorption energy of CO (80 kJ/mol). With increasing temperature, the
surface coverage of CO decreases, whereas the number of free sites increases. This confirms
our earlier reasoning regarding the optimal temperature for the reaction. If the temperature
is too low, no reaction can occur between CO and O. In fact, the temperature is so low that
CO poisons the surface (due to its high adsorption energy) by which the surface coverage of
O is too low for an appreciable rate of coupling between CO and O to occur. With increasing
temperature, the CO coverage decreases and consequently, the rate increases. At elevated
 94 Chapter 4. Microkinetic Modeling

1.0 CO*
O*
CO2*

Surface coverage (dimensionless)

0.8 *

0.6

0.4

0.2

0.0
300 400 500 600 700 800 900 1000

C HAPTER 4
Temperate T in K

Figure 4.4: Surface coverage as a function of temperature.


temperatures, the CO coverage decreases even more and the surface becomes empty. At that
point, the overall rate decreases again.

4.3 Reaction orders

In the previous section we mentioned that at low temperature, CO poisons the surface. Here,
we will show another technique that confirms this fact. If CO indeed poisons the surface at
low temperature, then we expect that the partial pressure of CO negatively influences the
rate. A systematic way of investigating the influence of the reactants and the products on
the reaction is by looking at the reaction order in these components. The reaction order is
defined as follows (see also section 1.5.3 on page 22):

∂ ln r+
ni = pi , (4.21)
∂pi

where ni is the reaction order in component i, r+ is the rate in the forward direction (i.e.
the rate corresponding to an initial rate experiment) and pi the partial pressure of component
i. The reaction orders are calculated using a linear fitting procedure. A short code snippet
how this is done is shown in Listing 4.5. At the end of this chapter, we will also show that for
some systems, an analytical expression can be derived.

The Python code to calculate the reaction order in CO at a spe-

cific temperature for the CO oxidation reaction can be found using
this link:  https://www.mkmcxx.nl/downloads/input/solve_
mkm_ode_calc_order.py

In Listing 4.5, the numerical procedure to calculate the reaction order is shown. To
calculate the derivative of the logarithm of the rate towards the partial pressure of CO, we
calculate the rate at five different partial pressures of CO. These five partial pressures are
generated by multiplying the coefficients 0.95, 0.98, 1.0, 1.02 and 1.05 with the partial
pressure at the working we are interested in (in our example, this is 13.33 bar). The derivative
is then calculated by performing a linear fit using the polyfit function. The order is finally
calculated by multiplying the slope (as found in the linear fit) by the partial pressure. At
T = 800 K, this gives a reaction order of about unity.
 4.3 Reaction orders 95

Listing 4.5: Python functions to calculate the reaction order of CO at temperature T.

C HAPTER 4

1 def main():
2 """


3 Calculate the reaction order in CO

4 """
5 T = 800 # temperature in K
6 p,r = calc_order(T)
7 m,b = np.polyfit(p, r, 1)
8 plt.plot(p, r, 'o', label='Data points')
9 plt.plot(p, m*p+b, '--', label='Linear fit')
10 plt.legend()
11 plt.xlabel('Pressure in Pa')
12 plt.ylabel('log(rate)')
13 plt.title('Reaction order CO = %f' % (m*p[2]))
14 plt.show()
15
16 def calc_order(T):
17 """
18 Calculate reaction order at temperature T
19 """
20 pt = 20
21 pa = 2.0/3.0 * pt * 1e5 # pressure of CO in Pa
22 pb = 1.0/3.0 * pt * 1e5 # pressure of O2 in Pa
23 pc = 0
24 mc = 80 * 1.66054e-27
25
26 # set series of factors to expend pressure in
27 diffs = [0.95, 0.98, 1.0, 1.02, 1.05]
28 rates = []
29 for diff in diffs:
30 x, y = solve_odes(T, pa * diff, pb, pc)
31 r_co2 = calc_kdes(T, 1e-20, mc, 1, 0.561, 10e3) * y[-1,2]
32 rates.append(r_co2)
33
34 return np.multiply(diffs, pa), np.log(rates)
 96 Chapter 4. Microkinetic Modeling

The reaction orders of CO and O2 as function of temperature is found in Figure 4.5. From
this Figure, we can see that at low temperature, the reaction order in CO is strongly negative.
A value of -1 indicates that if we would double the partial pressure of CO, than the rate would
decrease by a factor 2. With increasing temperature, we see that the reaction order in CO
increases. This can be related to the surface coverage of CO. At sufficient temperature, CO
no longer poisons the surface and a consequence, the reaction order increases to 0. Further
increasing the temperature leads to a nearly empty surface. Such a surface is lacking in CO and
increasing the partial pressure of CO would benefit the reaction rate. Thus, a positive reaction
order is seen. In contrast, the reaction order in O2 is nearly independent of temperature and
has a constant value of 0.5. The value of 0.5 originates from the fact that O2 dissociatively
adsorbs on the surface. For the reaction to occur, only half a molecule of O2 is needed. Thus
doubling the partial pressure of O2 only results in an increase of the overall reaction rate by a
√
factor of 2.

C HAPTER 4

1.00 CO
O2
0.75
Reaction order (dimensionless)

0.50
0.25
0.00
0.25
0.50
0.75
1.00
400 500 600 700 800 900 1000
Temperate T in K

Figure 4.5: Reaction orders in CO and O2 as a function of temperature.

Another way of looking at these reaction orders is from the perspective of a power law.
Assume that we can model our reaction by the following expression

νO
r = cpνCO 2
CO pO , (4.22)
2

then the reaction orders in CO and O2 would match the exponents νi in this expression.
Performing a series of experiments wherein the partial pressures of CO and O2 followed by a
fitting procedure would then provide the values for the variables in this equation. Important
to realize is that the above expression is limited to a fairly small region in temperature as
is evident from Figure 4.5. A significant variation of the temperature leads to a completely
different value for the reaction order. Another way of looking at this is by saying that this
approach is only valid locally. In contrast, constructing a microkinetic model provides the
possibility of modeling a reaction over a very broad range of temperatures and pressures and
is in a way a more global approach.

4.4 Apparent activation energy

Besides analyzing the dependence of the partial pressures of the reactants on the overall
reaction rate by means of calculating the reaction orders, one can also investigate the influence
 4.4 Apparent activation energy 97

of temperature on the activity by probing the apparent activation energy. The apparent
activation energy is given by the following equation (see also section 1.5.4 on page 24):

app ∂ ln r+
∆Eact = RT 2 . (4.23)
∂T

The procedure is very similar as to the one showed earlier for the calculation of the
reaction orders. A linear fit is performed on the basis of equation 4.23 as shown in Listing
4.6. The only differences are that the coefficient interval is much smaller as the rate depends
much stronger on the temperature as compared to the pressure and that we multiply the first
derivative towards temperature by RT 2 to obtain an answer in units of energy.
C HAPTER 4

Listing 4.6: Python functions to calculate the apparent activation energy at temperature T.

1 def main():


2 """
3 Calculate the apparent activation energy
4 """
5 T = 500 # temperature in K
6 R = 8.3144598 # gas constant
7
8 p,r = calc_eapp(T)
9 m,b = np.polyfit(p, r, 1)
10 plt.plot(p, r, 'o', label='Data points')
11 plt.plot(p, m*p+b, '--', label='Linear fit')
12 plt.legend()
13 plt.title('Eapp = %f kJ/mol' % (R*T**2*m/1e3))
14 plt.xlabel('Temperature in K')
15 plt.ylabel('log(rate)')
16 plt.show()
17
18 def calc_eapp(T):
19 """
20 Calculate the apparent activation energy at temperature T
21 """
22 pt = 20
23 pa = 2.0/3.0 * pt * 1e5 # pressure of CO in Pa
24 pb = 1.0/3.0 * pt * 1e5 # pressure of O2 in Pa
25 pc = 0
26 mc = 80 * 1.66054e-27
27
28 # set series of factors to expend pressure in
29 diffs = [0.998, 0.999, 1.0, 1.001, 1.002]
30 rates = []
31 for diff in diffs:
32 x, y = solve_odes(T * diff, pa, pb, pc)
33 r_co2 = calc_kdes(T * diff, 1e-20, mc, 1, 0.561, 10e3) * y[-1,2]
34 rates.append(r_co2)
35
36 return np.multiply(diffs, T), np.log(rates)

The Python code to calculate the apparent activation energy at a spe-

cific temperature for the CO oxidation reaction can be found using
this link:  https://www.mkmcxx.nl/downloads/input/solve_
mkm_ode_calc_eact.py

This derivative probes the effect on the overall rate as a result of a change in temperature.
A positive value of the apparent activation energy indicates that the overall reaction rate
increases when the temperature in increased. In contrast, a negative energy indicates that the
overall reaction rate would be decreased as a result of a decrease in temperature. Consequently,
98 Chapter 4. Microkinetic Modeling

if the apparent activation energy is zero, this means that the reaction is at an optimum with
respect to temperature.

Apparent activation energy Ea in J/mol

150000

100000

50000

C HAPTER 4
0


400 500 600 700 800 900 1000
Temperate T in K

Figure 4.6: Apparent activation energy as a function of temperature.

In Figure 4.6, the apparent activation energy as a function of temperature is shown. From
this Figure, it can be seen that at low temperature, the apparent activation energy is positive,
whereas at high temperature, it is negative. This result can be directly interpreted in terms
of the Sabatier’s principle. At low temperature, the surface is partially blocked by CO (see
Figure 4.4) and increasing the temperature results in more available sites by which the overall
reaction rate can increase. In contrast, at high temperature the apparent activation energy is
found to be negative. Here, lowering the temperature would result in more gaseous species
to fixate on the surface. Thus, lowering the temperature would result in an increase of the
overall reaction rate. Finally, around T=575K, the apparent activation energy is zero. Looking
at Figure 4.3, we see that at this particular temperature, the overall reaction rate is at an
optimum. This was to be expected, as finding the extremum of a function can be done by
equating the first derivative of a function to zero.
Interestingly, a lot of chemical processes are run under certain conditions where the
apparent activation energy is positive. Given the above explanation, one could reason that
increasing the temperature would result in a higher activity and wonder why this is not
done. The argument for not increasing the temperature is that the apparent activation energy
only gives us an indication whether the overall reaction rate would increase or decrease with
respect to temperature. It does not convey anything about the selectivity of the reaction.
Typically, increasing the temperature of the reaction results in the production of unfavorable
side products that eventually have to be removed from the production stream. As such, overall
activity is rarely the most important criterion in deciding at which temperature the reaction
should be operated.
The above discussion already shows that the rate r in equation 4.23 is not in a sense
unique and has to be chosen with care. Rates are calculated with respect to a so-called key
component. Here, we have always chosen the key component to be one of the reactants. This
is of course a sensible choice, but if you are looking at a reaction wherein multiple products
can be produced, it might be more sensible to pick one of these products that is of interest.
Thus, the apparent activation energy is not some universal value for an overall reaction, but
depends on the specific key component that has been chosen. As such, in the literature, the
formula that is used to fit the apparent activation energy to the experiment is shown to make
the matter clear. Often, the fitting of the apparent activation energy is combined with the
fitting of the reaction orders and a formula such as the following is used:
 4.5 Degree of rate control 99

app
!
∆Eact Y ν
rj = ν exp pi i , (4.24)
RT
i

where ri is the rate of production or consumption of key component j , pi is the partial

app
pressure of reactant i, νi is the stoechiometric coefficient of reactant i, Eact is the apparent
activation energy and ν is a pre-exponential factor.
In the results as shown in Figure 4.6, we have chosen CO to be the key component.
It should be clear though that, due to the stoichiometry, if we had chosen any other key
component for our analysis, we would have obtained exactly the same result.
C HAPTER 4

4.5 Degree of rate control

From the previous two sections, we have shown how to calculate the reaction orders and the


apparent activation energy. Another step further is to investigate the effect of the reaction
barrier of each of the elementary reaction steps on the overall reaction rate. To do this, we
use the method of as introduced by Campbell and coworkers. Herein, a degree of rate control
coefficient of a single elementary reaction step is defined as

!
∂ ln r
χi = , (4.25)
∂ ln ki
kj 6=ki ,Ki

where χi is the degree of rate control coefficient, r the overall reaction rate, ki the reaction
rate constant for elementary reaction step i and Ki the equilibrium constant of elementary
reaction step i as defined by

ki+
Ki = . (4.26)
ki−

Loosely speaking, the effect of lowering or increasing the reaction barrier of an elementary
reaction step on the overall reaction rate is probed. Importantly, only the barrier is varied. All
equilibrium constants including the reaction Gibbs free energy is kept constant. Schematically,
this is depicted in Figure 4.7.

initial state
final state

Figure 4.7: Schematic representation of varying the reaction barrier of an elementary reaction step while keeping
the Gibbs free energy of the reaction constant.

A positive value of χi indicates that the elementary reaction step is rate-controlling. In

other words, decreasing the reaction barrier for this elementary reaction step results in
an increase of the overall reaction rate. In contrast, a negative value of χi means that the
elementary reaction step is rate-inhibiting. Lowering the apparent activation energy then
100 Chapter 4. Microkinetic Modeling

results in a decrease of the overall reaction rate. Finally, the sum of all DRC coefficients of all
elementary reaction steps should be unity8 as given by

X
χi = 1. (4.27)
i
The DRC analysis for CO oxidation is given in Figure 4.8. From this Figure, we can see
that independent of temperature, CO+O recombination on the catalytic surface is the rate-
limiting step. This was to be expected as adsorption-desorption steps are rarely rate-limiting
and this is the only elementary reaction step that occurs on the surface.

C HAPTER 4
1.0 CO + * CO*
Degree of rate control coefficient (dimensionless)

O2 + 2* 2O*
CO* + O* CO2* + *


0.8 CO2 + * CO2*

0.6

0.4

0.2

0.0
400 500 600 700 800 900 1000
Temperate T in K

Figure 4.8: Degree of rate control coefficient as function of temperature.

Nevertheless, this result is very important as will become clearer in the next section. In the
case that the DRC analysis shows that only one elementary reaction step has a DRC coefficient
χ = 1 and that all other elementary reaction steps have χ = 0, then this means that this
elementary reaction step is rate-determining. In other words, the overall reaction rate r only
depends on the rate of this elementary reaction step. All other elementary reaction steps prior
to the rate-determining elementary reaction step are then at pseudo-equilibrium. We will use
this valuable approximation to create an analytic model of our kinetic system.
Finally, we should make an important note here. In the calculation of the degree of rate
control, we have used the overall reaction rate r. In principle, one can also use the rate for
any kind of particular reactant or product. The rate for this compound does not necessarily
have to be the same as the overall reaction rate, though for reactions where there is only a
single product, this is the case. But consider the situation wherein one reactant can form two
different product. In that case, the degree of rate control of one particular product will not
be the same as the degree of rate control of the reactant. It is up to the researcher to decide
which rate is important and relevant in your analysis and it might be that several degree of
rate control simulations have to be performed in order to fully convey the sensitivity on the
elementary reaction steps of the reaction mechanism.
8
The proof for this equality is given in section B.8 of the Appendix.
 4.6 Degree of selectivity control 101

4.6 Degree of selectivity control

Besides the degree of rate control, it can also be of interest what the influence of a particular
elementary reaction step is on the selectivity rather than the activity of a reaction. In this
evaluation, one solves the following differential

!
∂ηc
i,c = , (4.28)
∂ ln ki
kj 6=ki ,Ki

where i,c is the degree of selectivity coefficient of compound c due to a change in

elementary reaction step i and ηc is the selectivity of compound c.
Once the degree of rate control coefficients are known, it is fairly straightforward to
calculate the i,c coefficients as can be seen from the following derivation:
C HAPTER 4

!


∂ηc
i,c = (4.29)
∂ ln ki
kj 6=ki ,Ki
!
∂ ln ηc
= ηc (4.30)
∂ ln ki
kj 6=ki ,Ki
!
∂ ln rc /rr
= ηc (4.31)
∂ ln ki
kj 6=ki ,Ki
 
! !
 ∂ ln rc ∂ ln rr
= ηc  − (4.32)

∂ ln ki ∂ ln ki

kj 6=ki ,Ki kj 6=ki ,Ki
 
= ηc χc,i − χr,i (4.33)

In other words, the degree of selectivity control is the degree of rate control using the
rate of the compound you are interested in minus the degree of rate control using the rate of
the overall reaction (i.e. of a specific reactant), multiplied by the selectivity of the compound
wherein you are interested. A positive degree of selectivity coefficient indicates that the
selectivity of this compound will increase when the barrier of the corresponding elementary
reaction step is lowered and a negative value indicates that the selectivity decreases if the
barrier is lowered.
From the sum-rule for the degree of rate control coefficients, another sum rule can be
constructed. It is fairly straightforward to see that

X
i,c = 0. (4.34)
i,c

4.7 Comparison with Langmuir-Hinshelwood kinetics

In the previous section we have seen that some kinetic networks have only a single elementary
reaction step that is rate-controlling. In other words, that elementary reaction step is auto-
matically the rate-determining elementary reaction step. Such a system has the requirements
by which the Langmuir-Hinshelwood-Hougen-Watson approximation applies. Here, we will
demonstrate this by solving the system for CO oxidation in an analytic fashion.
If CO ∗ +O∗ → CO2 ∗ +∗ is the rate-determining step and we operate at zero conversion,
then we can assume that all other steps that precede this step are at pseudo-equilibrium.
Thus,
102 Chapter 4. Microkinetic Modeling

CO + ∗  CO∗ (4.35)
O2 + 2∗  2O∗ (4.36)

and consequently

θCO
KCO = (4.37)
pCO θ∗
2
θO
K O2 = , (4.38)
pO2 θ∗2

C HAPTER 4
where Ki is the equilibrium constant for adsorption of compound i, θi is the surface
coverage of compound i and θ∗ is the fraction of free sites on the surface.
By introducing a mass balance, 9


θCO + θO + θ∗ = 1, (4.39)

we can calculate the fraction of free sites as a function of the equilibrium constants and
the partial pressures.

1 = θCO + θO + θ∗ (4.40)
q
= KCO pCO θ∗ + KO2 pO2 θ∗ + θ∗ (4.41)
 q 
= 1 + KCO pCO + KO2 pO2 θ∗ (4.42)
1
θ∗ = q (4.43)
1 + KCO pCO + KO2 pO2

We can use this result to explicitly calculate the coverages for θCO and θO2 ,

KCO pCO
θCO = q (4.44)
1 + KCO pCO + KO2 pO2
q
KO2 pO2
θO = q . (4.45)
1 + KCO pCO + KO2 pO2

At zero coverage, the overall rate is given by the forward rate of the rate-determining
elementary reaction step as

r = krds θCO θO2 (4.46)

q
krds KCO pCO KO2 pO2
=  2 . (4.47)
q
1 + KCO pCO + KO2 pO2

9
Note that we have neglected to take the surface coverage of CO2 . The motivation for this assumption is that we
operate at zero conversion. This means that the partial pressure of CO2 is zero so that any CO2 that is formed on the
surface, immediately desorbs to the gas phase.
 4.7 Comparison with Langmuir-Hinshelwood kinetics 103

At this point, we have an analytic expression for the overall rate. From this expression, we
can derive analytic expressions for the reaction orders and the apparent activation energy.

∂ ln r
nCO = pCO (4.48)
∂pCO
 
q
 krds KCO pCO KO2 pO2 
∂ ln 
 q 2 

1+KCO pCO + KO2 pO2
= pCO (4.49)
∂pCO
1
= D1 + D2 − 2D3 , (4.50)
2
C HAPTER 4

where


∂ ln krds KCO pCO

D1 = pCO (4.51)
∂pCO
 
∂ ln KO2 pO2
D2 = pCO (4.52)
∂pCO
 q 
∂ ln 1 + KCO pCO + KO2 pO2
D3 = pCO . (4.53)
∂pCO

The terms Di can be readily solved, which give

D1 = 1 (4.54)
D2 = 0 (4.55)
KCO pCO
D3 = q (4.56)
1 + KCO pCO + KO2 pO2

and combining these three terms results in

nCO = 1 − 2θCO . (4.57)

In a similar fashion, we obtain

1
nO2 = − θO . (4.58)
2
These results are completely in line with the numerical results as shown in Figure 4.5,
though here we have gained the insight that the reaction order can be directly related to the
surface coverages. At a high surface coverage of CO, we see that the second term in the
equation becomes 1, which gives a reaction order of nCO = 1 − 2 · 1 = −1. With decreasing
surface coverage, the reaction order increases and finally at low surface coverage of CO, the
reaction order becomes 1.
For nO2 , we can see from Figure 4.4 that the surface coverage of O* is negligible at all
temperatures (θO ≈ 0). From our analytical derivation we can see that the reaction order
in O2 should then be nO2 = 12 − 1 · 0 = 12 . This confirms our results as seen in Figure 4.5
where indeed a reaction order of 12 is seen for O2 .
104 Chapter 4. Microkinetic Modeling

The apparent activation energy can also be readily derived.

app ∂ ln r
∆Eact = RT 2 (4.59)
∂T
 
q
 krds KCO pCO KO2 pO2 
∂ ln 
 q 2 

1+KCO pCO + KO2 pO2
= RT 2 (4.60)
∂T
1
= D1 + D2 + D3 + 2D4 (4.61)
2
where

C HAPTER 4
ln krds
D1 = RT 2 (4.62)


∂T
2 ln KCO
D2 = RT (4.63)
∂T
2 ln KO
D3 = RT (4.64)
∂T
 q 
∂ ln 1 + KCO pCO + KO2 pO2
D4 = RT 2 . (4.65)
∂T
which resolve to

D1 = ∆Erds (4.66)
D2 = ∆HCO (4.67)
D3 = ∆HO (4.68)
q
KCO pCO 1 KO2 pO2
D4 = ∆HCO q + ∆HO2 q .
1 + KCO pCO + KO2 pO2 2 1 + KCO pCO + KO2 pO2
(4.69)

Combining these terms gives

 
app 1
(4.70)


∆Eact = ∆Erds + ∆HCO 1 − 2θCO + ∆HO2 − θO .
2

Note that in the above expression, the part between the round brackets are equal to the
reaction orders:

app
 
∆Eact = ∆Erds + ∆HCO (nCO ) + ∆HO2 nO2 . (4.71)

In other words, there is a clear relation between the apparent activation energy and the
reaction order. As these reaction orders in turn depend on the surface coverages, the apparent
activation energy also strongly depends on these surface coverages.
Again, the analytic expression provides us with a valuable insight what terms contribute to
the apparent activation energy. Importantly, the dominant term of in the apparent activation
energy is the barrier of the rate-determining step. All other terms relate to the energetics of all
elementary reaction steps that precede the rate-determining step. Here, the adsorption of CO
 4.7 Comparison with Langmuir-Hinshelwood kinetics 105

and O2 are the two elementary reaction steps that precede the recombination of surface CO
with O. Thus, the energetics of these steps (∆Hi ) are seen in the expression of the apparent
activation energy.
The apparent activation energy in our example for CO dissociation can be interpreted as
follows. For the overall reaction to proceed, the system should have enough energy (tempera-
ture) to overcome the barrier of the rate-determining step. When the surface is completely
covered with CO, there are no free sites for O2 to adsorb on. As such, first a CO needs to
be desorbed before O2 can adsorb. This desorption costs energy and therefore the apparent
activation energy increases. Once CO is desorbed, oxygen can adsorb. As the adsorption
process is exothermic, some energy is released which decreases the apparent activation energy.
Let us assume that the surface coverage of CO is unity (completely covered), then the apparent
activation energy becomes
C HAPTER 4

app
 
∆Eact = ∆Erds + ∆HCO (nCO ) + ∆HO2 nO2 (4.72)


= 120kJ/mol + 80kJ/mol − 0.5 · 40kJ/mol (4.73)

= 180kJ/mol (4.74)

From Figure 4.6 we can see that around 400K, the apparent activation energy indeed
has a value of around 180 kJ/mol. Furthermore, from Figure 4.4 we see that this situation
corresponds to a surface nearly fully covered with CO. Indeed, from the coverages as obtained
from Figure 4.4 and using the formula for the apparent activation energy, Figure 4.6 can be
completely reproduced.
 106 Chapter 4. Microkinetic Modeling

4.8 Questions and Exercises

The answers to the questions and exercises are given at the end of this book in Appendix C
(page 135).

4.8.1 Questions
1. What is the motivation for constructing a rate expression using a power law? Why are
we allowed to use the stoichiometric coefficient of the reactant as the exponent in these
power laws for elementary reaction steps, but not (necessarily) for overall reactions?

2. Propose a series of elementary reaction steps which constitute the methanation reaction
(CO + 3H2 → CH4 + H2 O).

3. Why does methane immediately desorb after hydrogenation of CH3 ∗, but ammonia

C HAPTER 4
does not after hydrogenation of NH2 ∗?


4. For an overall elementary reaction steps with a clear stoichiometry and only one type of
product (i.e. having no side products), why does it not matter which key component is
used in the analysis of the reaction order or the apparent activation energy? Hint: Why
does the stoichiometric coefficient drop out?

5. Why is the surface coverage of O so low in the example of CO oxidation? (see Figure
4.4)

4.8.2 Exercises
The exercises are marked by a number of gears to indicate their difficulty levels.

 Exercise 4.1  

a) Calculate the degree of rate control parameters for all elementary reaction steps in the CO
oxidation reaction from the analytical expression as given in equation 4.47.

b) Derive the reaction order for O2 in the CO oxidation reaction and show that your result
matches that of equation 4.58. Hint: You need to use the chain-rule.

 Exercise 4.2  

a) Show that equation 4.34 holds using the proof for the sum rule of the degree of rate control
coefficients (eq. 4.27).


 5

CHAPTER
PERFORMING MICROKINETICS WITH
MKMCXX
C HAPTER 5

This chapter is based on joint work with Tom van den Berg.


Contents
5.1 Introduction 107
5.2 Installation 108
5.3 The basic input 108
5.3.1 Compounds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
5.3.2 Reactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
5.3.3 Settings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
5.3.4 Runs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
5.4 Kinetic analysis tools 114
5.4.1 Reaction order . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
5.4.2 Apparent activation energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
5.4.3 Degree of rate control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
5.5 Data interpretation 116
5.5.1 Operating temperature and production . . . . . . . . . . . . . . . . . . . . . . . . . 118
5.5.2 Reaction orders . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
5.5.3 Apparent activation energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
5.5.4 Degree of rate control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
5.6 Convergence 121
5.7 Numerical versus analytical 122
5.8 Exercises 123

5.1 Introduction
In the previous chapter we explained how to set up a simple microkinetic model for a unimolec-
ular and a bimolecular catalytic reaction. The reader was introduced to the stepwise approach
to construct such a model. We were able to calculate the reaction rate by time-integration of
 108 Chapter 5. Performing microkinetics with MKMCXX

the ordinary differential equations (ODE) underlying the kinetic system. Moreover, it was
explained how one could numerically calculate the reaction order and the apparent activation
energy for the reaction. Finally, the methodology behind the degree of rate control analysis
and the degree of selectivity control analysis was explained.
In this chapter we will expand on the previous chapter by providing a hands-on tutorial how
to perform microkinetics simulations using the MKMCXX tool. We will discuss the results
of typical simulations and explain how to interpret these results. The whole procedure is
explained using a couple of simple examples. The reader is invited to download the MKMCXX
software and work alongside the book.

5.2 Installation
The program MKMCXX is a microkinetic modeling software suite for educational and research
purposes. The program can be found on its website http://www.mkmcxx.nl and is free
to use. The program is originally written for the Linux operating system but has since been
ported to Windows. This chapter will focus on the use of MKMCXX in Windows as we
anticipate that most readers of this book will be using Windows. It is advised to always use
the latest stable version for Windows.

C HAPTER 5
Download the zip package from the website and extract MKMCXX. The extracted folder
will contain two directories. The subfolder bin consists of the executable of MKMCXX
together with the necessary supporting files (i.e. the .dll files). The MKMCXX executable


requires a set of command-line arguments in order to perform a simulation, hence you
cannot start a simulation by simply opening the program by double-clicking on it. The second
subfolder place_different_runs_here is used for the simulations. The example folder
contains an input file (typically input.mkm) and a run file (run.bat). The run file is used to
start the MKMCXX simulation using the input file in that specific folder. Alternatively, you
can use the Command Prompt of Windows to start the simulation. A short tutorial how to
install the program and how to use the Command Prompt of Windows to start a calculation
is given in the YouTube video as found below.

Watch the short installation tutorial on YouTube using this link:

 https://www.youtube.com/watch?v=_edo3JvAOAA

 Important

We advise to create a new folder for every simulation to keep everything tidy and
organized. Do not forget to add the input and run files into each folder.

5.3 The basic input

The input file of MKMCXX is structured into sections. The start of each section is indicated
using the syntax as found in Table 5.1.
The input file consists of four mandatory sections1 which are used to define the microki-
netic model. These four sections are

1. &compounds: A list of all compounds in the system, for both gases as well as surface
intermediates, and their corresponding initial concentration.
1
MKMCXX offers several non-mandatory sections. These instructions can be found in Appendix F on 211.
 5.3 The basic input 109

Table 5.1: Symbols recognized by MKMCXX to define of separate certain areas of the input.mkm file.

Symbol Function
& The symbol & followed by a keyword will define a new section in the
input file.
# Placing # in the text will convert the text to comments. MKMCXX will
not use comments when executed therefore this give the user the option
to leave remarks in the input file.
; Most sections in MKMCXX consist of several columns. These columns
are separated by ;.

2. &reactions: All elementary reaction steps and their relevant kinetic parameters (e.g.
activation barriers and pre-exponential factors).

3. &settings: Specifies the type of the simulation, the pressure and what kind of post-
analytics need to be performed.
C HAPTER 5

4. &run: A list of temperatures and total simulation times.



Here, we will explain how to construct a basic input file to perform an MKMCXX simu-
lation utilizing a simple example question. The different sections of the input file and the
interpretation of the results will be explained. It is strongly recommended to follow along
with this example as this gives a much better understanding of the input file and MKMCXX.

 Example question

Consider the catalytic oxidation of carbon monoxide to carbon dioxide. Assume that
O2 adsorbs dissociatively, whereas CO adsorbs molecularly. The overall reaction is
given by

2 CO + O2 −−→ 2 CO2 . (5.1)

Use the kinetic data as provided in Tables 5.2 and 5.3.

1. Calculate the steady-state production for catalytic CO oxidation. What is the

optimal operating temperature for the oxidation with respect to the production
of CO2 between the temperatures 400-1400 K? Take steps of 20 K at a time and
simulate for 1000 seconds. Assume ideal stoichiometry and a total pressure
of 20 bar.

2. What are the reaction orders of CO and O2 and what is the effect of partial
pressure of the compounds on the reaction rate?

3. The apparent activation energy is a measure for the influence of temperature

on the reaction rate. What is the apparent activation energy for the reaction at
700 K. At what temperature is the apparent activation energy 0?

4. Determine the rate controlling step. Does the rate controlling step vary with
the temperature?
110 Chapter 5. Performing microkinetics with MKMCXX

A demonstration video on how to set-up a MKMCXX input file can be

found on YouTube using the link below. Please note that the chemo-kinetic
network in this video is different from the one used in the example.
 https://www.youtube.com/watch?v=TQ1EychXXc8.

Table 5.2: Parameters for the adsorption and desorption steps in CO oxidation.

Compound A [m2 ] m (a.u.) σ [-] θrot [K] Eads [kJ/mol]

CO 1 · 10−20 28 1 2.8 80 2
O2 1 · 10−20 32 2 2.08 40
CO2 1 · 10−20 80 1 0.561 10

To set up the microkinetic model for catalytic CO oxidation we will partially use the steps
as previously shown in Chapter 4 on page 84.

1. Construct the set of elementary reaction steps that describes the reaction mechanism.

C HAPTER 5
2. Define boundary conditions, initial values and data parameters for the system (e.g.
temperature, pressure, concentrations).


3. Construct the input file for MKMCXX.

4. Interpret the results of the simulation using our chemical intuition.

We start with constructing the elementary reaction steps for CO oxidation. The overall
reaction of CO oxidation (Equation 5.1) constitutes the following four elementary reaction
steps:

CO + ∗ CO∗ (5.2)
O2 + 2∗ 2 O∗ (5.3)
CO ∗ +O∗ CO2 ∗ +∗ (5.4)
CO2 + ∗ CO2 ∗ (5.5)

Next, we define the boundary conditions of the microkinetic model. From the description
in the example question, we can obtain the following boundary conditions:

1. The pressure is 20 bar with a stoichiometric ratio of 1:2 for the partial pressures of CO
and O2 , respectively.

2. The temperatures of the reaction are between 400 and 1400 K and modeled every 20
K.

3. The kinetic data for the rate expressions are given in Tables 5.2 and 5.3.

Table 5.3: Parameters for the surface reaction of CO oxidation.

f
Reaction νf [s−1 ] νb [s−1 ] Ea [kJ/mol] Eab [kJ/mol]
CO* + O* −−→ CO2 * + * 1013 1013 120 180
 5.3 The basic input 111

5.3.1 Compounds
We start constructing the input file for the calculation by defining all compounds in our
system. In MKMCXX a compound is defined by three arguments, separated by semicolons.
These three arguments are:

1. The name of the compound, which can in principle be anything but it is advised to use
the molecule structure for clarity. Please avoid special characters such as ’#’ as they
have a different meaning in the input file.

2. Indicate whether the compound is a gas phase species or a surface intermediate. A

value of 0 designates a gas phase species and a value of 1 a surface intermediate. From
here on, we will use the term isSite to refer to this value.

3. The initial partial pressure as fraction of the total pressure. The latter is specified in
the &settings section.

The list of compounds has to contain every compound in the system. That is, the input file
must contain a complete list of reactants, products, catalyst sites and adsorbed intermediates
of the reaction. In the next section, we will refer to these compounds when defining the
C HAPTER 5

elementary reaction steps. Neglecting to specify a compound will result in an error message
of the program.
Let us start by first defining all gas phase compounds. In Listing 5.1, a snippet of the


input file corresponding to the gas phase compounds can be found. We start by opening
the &compounds section. The compounds are specified using three semicolon separated
columns, wherein the first column sets the name of compounds, the second column whether
the compound is a surface intermediate and the final column the starting concentration.

Listing 5.1: Gaseous compounds

1 #The line below tells MKMCXX to start indexing the compounds.

2 &compounds
3
4 #Name; isSite; Concentration
5 CO; 0; 1.0
6 O2; 0; 2.0
7 CO2; 0; 0.0

Note that the gaseous compounds are not adsorbed onto the catalyst surface, thus isSite
is set to 0. The concentration is partial pressure of each compounds with respect to the total
pressure defined in the &settings section. Since the mixture of gaseous compounds is
stoichiometric, the partial pressure of O2 is twice as high as the partial pressure of CO.

Listing 5.2: Adsorbed compounds

1 #Adsorbed compounds
2 #Name; isSite; Concentration
3 CO*; 1; 0.0
4 O*; 1; 0.0
5 CO2*; 1; 0.0

The input settings for the surface intermediates (adsorbed compounds) are provided in
Listing 5.2. To clearly list the adsorbed compounds, it is advised to append a symbol to the
compound names. Here, we have used a ”*”. Note that O2 adsorbs dissociatively onto the
catalyst surface, hence we only have O* and not O2 *.
Next to adsorbed species on the catalytic surface, an empty catalytic site is also a surface
intermediate. The empty catalytic site is shown in Listing 5.3 and is named ’*’. At the start of
 112 Chapter 5. Performing microkinetics with MKMCXX

the reaction, we assume that the catalytic surface is empty (this is a common assumption in
microkinetics) and hence we set the starting concentration to 1.0.

Listing 5.3: Catalyst compounds

1 #Catalyst
2 #Name; isSite; Concentration
3 *; 1; 1.0

5.3.2 Reactions
The &reactions session specifies all elementary reaction steps in the system. The section
is opened using the &reactions keyword. MKMCXX differentiates between two types of
elementary reaction steps: Arrhenius-type (see Equation 3.12 on page 79) and Hertz-Knudsen
type (see Equation 3.7 on page 77 and Equation 3.10 on page 78). The former is used for
surface reaction, whereas the latter is used for adsorption/desorption reactions. Arrhenius-
type reactions require 6 arguments and Hertz-Knudsen reactions require 8 arguments. For
the adsorption and desorption the Hertz-Knudsen equations are a better approximation
compared to the Arrhenius equations as they more adequately describe the entropy changes

C HAPTER 5
upon adsorption/desorption. The arguments for the Hertz-Knudsen equations are as follows:

1. Specify which set of equations need to be used by MKMCXX. The Hertz-Knudsen


equations are specified by HK and the Arrhenius equations are specified by AR.

2. The adsorption or desorption reaction. The compounds need to be within curly brackets
{}. Make sure that the compounds match the compounds in the &compound section.

3. The surface area of the active site in m2

4. The molecular mass of the molecule in Dalton.

5. The characteristic temperature for rotation θrot in K .

6. The symmetry number, σ , which represent the number of indistinguishable orienta-

tions as result of rotation.

7. The sticking coefficient of the molecule. (if unknown, use a value of 1 here)

8. The desorption energy in J/mol. (it is defined in such a way that for exothermic adsorp-
tion, the value here should be positive positive)

For the elementary reaction steps that occur completely on the catalyst surface, the
Arrhenius-type equation is best. Herein, it is assumed that the entropy between the initial,
transition and final state does not change significantly. For the Arrhenius equations, the 6
arguments are as follows:

1. Specify the Arrhenius equation with AR.

2. The surface reaction (see above).

3. The pre-exponential factor for the forward reaction in s−1 .

4. The pre-exponential factor for the backward reaction in s−1 .

5. The activation energy for the forward reaction in J/mol.

6. The activation energy for the backward reaction in J/mol.

 5.3 The basic input 113

By using the data as given in Tables 5.2 and 5.3, the adsorption/desorption reactions are
specified as given in Listing 5.4.

 Important

Always specify Hertz-Knudsen-type elementary reaction steps as adsorptions. In

other words, place the gas phase components on the left-hand-side of the equation
and the adsorbed species on the right-hand-side.

Listing 5.4: Adsorption and desorption reaction steps

1 #The line below tells MKMCXX to index all reaction steps.

2 &reactions
3
4 #Adsorptions/desorption's
5 #Type Reaction m^2 amu Theta sigma stick J/mol
6 HK; {CO} + {*}=> {CO*}; 1e-19; 28; 2.8; 1; 1; 80e3
7 HK; {O2} + 2{*}=> 2{O*}; 1e-19; 32; 2.08; 2; 1; 40e3
HK; {CO2} + {*}=> {CO2*}; 1e-19; 80; 0.561; 1; 1; 10e3
C HAPTER 5

In a similar fashion, the input for the Arrhenius-type of equations is given in Listing 5.5.


Listing 5.5: Surface reaction

1 #Surface reactions
2 #Type Reaction vf vb Eaf Eab
3 AR; {CO*} + {O*} => {CO2*} + {*}; 10e13; 10e13; 120e3; 180e3

5.3.3 Settings
The settings section of the input file is started using the &settings keyword. In the settings
sections, the program determines what analytical routines are used to evaluate the specified
system. The settings are specified using the well-known keyword-value pairing as represented
by
Keyword = new value.
For a basic simulation without using any of the analytical tools, only the type of simulation
and the overall pressure of the system need to be specified. The simulation type is a so-called
sequence run and the overall pressure, as given in the example exercise, is 20 bar. The code
snippet for these settings is provided in Listing 5.6.

Listing 5.6: Settings CO

1 #The line below tells MKMCXX what settings to use

2
3 &settings
4
5 TYPE = SEQUENCERUN
6 PRESSURE = 20

MKMCXX is highly tunable and offer a much broader set of keywords. In the upcoming
sections, the most important analytical tools will be discussed. For a more thorough overview
of the possible keywords, we refer to Appendix F on page 211.
 114 Chapter 5. Performing microkinetics with MKMCXX

5.3.4 Runs
The final (mandatory) section is &runs. Herein, the set of simulations are provided. A
simulation is the time-integration of the ordinary differential equations underlying the chemo-
kinetic system. To perform this time-integration, the following parameters are required:

1. The temperature in K.

2. The time in s.

3. The absolute tolerance.

4. The relative tolerance.

The example exercise asks us to evaluate the reaction between 400 and 1400 K in steps
of 20 K. The simulation time for each of the simulations is set to 1000 s. The sample exercise
does not provide any specifications for the tolerances and hence we opt for using the default
values. If no values are given, the program automatically resorts to using the default values,
which are 10−8 for the absolute tolerance and 10−12 for the relative tolerances. Here, we
are going to use the default values and explicitly provide these as input. This results in the

C HAPTER 5
following &runs section as provided in Listing 5.7.

Listing 5.7: Temperature time and tolerance variables for the model.


1 #The line below tells MKMCXX on what temperatures and how to simulate the catalytic
reaction
2
3 &runs
4
5 #Temp; Time; AbsTol; RelTol
6 400; 1e3; 1e-8; 1e-12
7 420; 1e3; 1e-8; 1e-12
8 440; 1e3; 1e-8; 1e-12
9 # ...
10 # More desired temperatures.
11 # ...
12 1360; 1e3; 1e-8; 1e-12
13 1380; 1e3; 1e-8; 1e-12
14 1400; 1e3; 1e-8; 1e-12

This concludes the basic input data that is needed for the oxidation reaction of CO to
CO2 . The input file is complete and MKMCXX is ready to perform the simulation. We
warmly recommend the reader to copy the input above and perform a single simulation before
continuing to the data interpretation section. If any errors occur, look into the Common
Problems section in Appendix F on page 211.

5.4 Kinetic analysis tools

In this section, we introduce a few kinetic analysis tools to investigate the chemokinetic
network. If the reader is curious about interpreting the results of the simulation, he/she is
invited to skip this section and immediately go to section 5.5 on page 116.

5.4.1 Reaction order

MKMCXX is able to calculate the reaction order by means of numerical approximation as
shown in the previous chapter (see page 94). This is done by performing four additional
calculations, each at a slightly different partial pressure of one of the key reactants and finally
calculating the derivative using a linear fit. In this section, we will elaborate on how a reaction
order analysis is performed using the CO oxidation example.
 5.4 Kinetic analysis tools 115

Listing 5.8: Code snippet to enable a reaction order analysis.

1 #The line below tells MKMCXX what settings to use

2 &settings
3 TYPE = SEQUENCERUN
4 PRESSURE = 20
5
6 # Specify REAGENTS here
7 REAGENTS = {CO},{O2}
8
9 # Calculate reaction order
10 ORDER = 1

By default, MKMCXX does not calculate the reaction order of a reaction. Evaluation of the
reaction orders is specified in the &settings section by setting ORDER = 1. Furthermore,
the user needs to set for which reactants the reaction order should be evaluated. This is shown
in Listing 5.8.
Note that the settings section has been slightly altered. First the REAGENTS were specified.
C HAPTER 5

This tells MKMCXX which compounds should be evaluated when calculating the reaction
order. Multiple gas phase compounds can be added to this section and are delimited by a
comma. Remember to use the curly brackets to sandwich the compounds!


5.4.2 Apparent activation energy

The effect of temperature on the overall rate of the reaction can be determined by calculating
the apparent activation energy as discussed in chapter 4.4 on page 96.
The apparent activation energy can be calculated in MKMCXX by setting the EAPP keyword
to unity. The program simulates four additional points at slightly different temperatures.
After calculating these 5 points, the apparent activation energy is found by performing a linear
fit on the following equation:

app ∂ ln r ∂ ln r+
Eact = −R = RT 2 (5.6)
∂ T1 ∂T

In equation 5.6, the rate r refers to the rate of a key component. One can easily imagine
that in parallel reaction networks, the apparent activation energy is not necessarily the same
for each reactant or each product. In principle, there is no such thing as a unique apparent
activation energy for a reaction. Thus, to calculate the apparent activation energy, we need
to specify the key component for which the apparent activation energy is evaluated. The
corresponding &settings section of the input file is shown in Listing 5.9.

Listing 5.9: Code snippet to enable apparent activation energy analysis.

1 #The line below tells MKMCXX what settings to use

2 &settings
3 TYPE = SEQUENCERUN
4 PRESSURE = 20
5
6 # Specify key components
7 KEYCOMPONENTS = {CO}
8
9 # Calculate apparent activation energy
10 EACT = 1
 116 Chapter 5. Performing microkinetics with MKMCXX

Please note the inclusion of the KEYCOMPONENTS keyword. Multiple key components can
be specified and these components are delimited by using a comma. By setting EACT = 1,
MKMCXX will calculate the apparent activation energy for each of the key components.

5.4.3 Degree of rate control

To investigate the influence each elementary reaction step has on the overall reaction, we can
perform a degree of rate control analysis. The degree of rate control, as explained in chapter
4, probes the effect of increasing or decreasing the barrier of the elementary reaction step on
the overall reaction rate (of a designated key component).
MKMCXX evaluates the degree of rate control by performing 4 additional calculations
with slightly different prefactors (νf and νb ) while keeping the equilibrium (Ki ) of all reactions
constant. Because the DRC analysis is relatively expensive, you need to specifically set for
which elementary reaction steps the analysis needs to be performed. This is done by appending
a 0 or a 1 after the elementary reaction step. The procedure is shown in Listing 5.10.

Listing 5.10: Enabling degree of rate control analysis for the elementary reaction steps. Note that ”;1” after each
elementary reaction step.

C HAPTER 5
1 #The line below tells MKMCXX to index all reaction steps.
2 &reactions
3 #Adsorptions/desorption's
4 #Type Reaction m^2 amu K sigma


sticking J/mol DRC
5 HK; {CO} + {*} => {CO*}; 1e-19; 28; 2.8; 1; 1;
80e3; 1
6 HK; {O2} + 2{*} => 2{O*}; 1e-19; 32; 2.08; 2; 1;
40e3; 1
7 HK; {CO2} + {*} => {CO2*}; 1e-19; 80; 0.561; 1; 1;
10e3; 1
8
9 #Surface reactions
10 #Type Reaction vf vb Eaf Eab
DRC
11 AR; {CO*} + {O*} => {CO2*} + {*}; 10e13; 10e13; 120e3; 180e3;
1

In the &settings section of the input file, the keyword DRC needs to be set to one and
the KEYCOMPONENTS need to be set. This is shown in Listing 5.11.

Listing 5.11: Degree of rate control settings

1 #The line below tells MKMCXX what settings to use

2 &settings
3 TYPE = SEQUENCERUN
4 PRESSURE = 20
5
6 # Specify key components
7 KEYCOMPONENTS = {CO}
8
9 # Perform degree of control analysis
10 DRC = 1

5.5 Data interpretation

If you completed the procedure as shown in section 5.3, your calculation should now be up
and running (if not, consult the Troubleshooting section in the Appendix on page 211). If
you have also added the kinetic analysis tools of the previous section to your simulation, you
should now have calculated the steady-state rates of adsorption, desorption and production
 5.5 Data interpretation 117

at a series of temperatures, the reaction order for the compounds CO and O2 , the apparent
activation energy and the degree of rate control for the oxidation of CO to CO2 . In this section,
we are going to investigate the results from the simulation.
MKMCXX will create a folder called TYPE_OF_RUN_'date'_'time' to store all output
files in, where ’date’ and ’time’ have been replaced by the current system date and time. In
the example above, this will results in a folder called SEQUENCERUN_'date'_'time'. In
this folder, a subfolder is created for every temperature as specified in the &runs section.
Furthermore, there is a folder called graphs, a folder called networkplots and a folder
called range. Lastly, there are four text files which are: compounds.log, input.log,
reactions.log and warnings.log.
Each temperature folder contains the fractional surface coverages at that temperature
as a function of the simulation time as well as the consumption and production of gas
phase reactants and products. These files are tab-delimited text files and can be opened in a
Spreadsheet program such as Microsoft Excel.
The graphs folder contains graphs which can be used to quickly evaluate the calculated
data. These graphs are automatically generated using the Cairo library3 and are specifically
not intended for reproduction in papers or reports, although this up to discretion of the user.
We always recommend remaking the graphs to suit your (or the journal’s) style using a proper
C HAPTER 5

tool such as Origin, Excel or the matplotlib module in Python.4 In Figure 5.1, an example
is shown of a typical graph generated by MKMCXX depicting the production of CO2 and the
consumption of CO and O2 as function of the temperature. Figure 5.2 shows the same data,


but then remade using matplotlib.5 Arguably, there is quite some difference in quality
between Figures 5.1 and 5.2 and the latter Figure can furthermore be altered or tuned to your
specifications. Throughout this chapter, we will opt for using matplotlib to generate our
figures.

SEQUENCERUN_2018-10-30_1420/graphs/pdf/derivative.pdf
VERSION 2.9.7@2d13d60
MKMCXX / (C) 2012 / WWW.MKMCXX.NL

8.0e-02
6.0e-02
Production [mol/s]

4.0e-02
2.0e-02
7.5e-10
-2.0e-02
-4.0e-02
-6.0e-02
-8.0e-02
400 500 600 700 800 900 1000

Temperature [K]

CO CO2 O2

Figure 5.1: Production of CO2 as function of temperature, the MKMCXX figure.

The networkplots folder contains files which can be visualized using the program
GraphViz6 and these files visualize the network of elementary reaction steps and their reaction
rates. Unfortunately, the procedure only works (at the moment) on Linux and hence we will
not further discuss it here. The user is advised to consult the online Wiki.7
The range folder contains the fractional coverage of surface intermediates and the
consumption and production rates of gas phase reactants and products as a function of
temperature (i.e. the range refers here to the range of temperatures). Furthermore, this
subfolder holds the calculated values of the reaction rate constants k, the prefactors and the
rates of the elementary reaction steps. If the reaction orders, apparent activation energy and
degree of rate control are calculated, each will get their own sub folder respectively.
3
See: https://www.cairographics.org/
4
The main reason for this is that the plotting tool is not exactly smart and it uses some default settings for the
visualization. These settings are not universal and we anticipate that for production data some fine-tuning is necessary.
5
See https://matplotlib.org/ for more information about using matplotlib to generate graphs.
6
See https://www.graphviz.org/ for more information about GraphViz.
7
See https://wiki.mkmcxx.nl.
 118 Chapter 5. Performing microkinetics with MKMCXX

CO
0.010 O2
CO2

0.005
Production in (mol/s)
0.000

0.005

0.010
400 500 600 700 800 900 1000
Temperature in K

Figure 5.2: Production of CO2 as function of temperature plotted with Spyder3.

The file compounds.log contains an overview of the compounds in the system and

C HAPTER 5
whether they occupy a site. The file input.log contains an exact copy of the input.mkm
file which functions as a backup for future reference. Please do not see this as an excuse to
neglect performing every individual simulation in a separate folder. The reactions.log


file contains the elementary reaction steps and the rate constants as a function of temperature.
Finally, the warnings.log file reports problems or errors from the simulation. Always
consult this file carefully, but be mindful that sometimes false positives are triggered. This
means that if a warning is generated, it does not necessarily imply that your calculation is
erroneous.

5.5.1 Operating temperature and production

The adsorption equilibria of the compounds of a system depend on the temperature. In
Figure 5.3, the coverage of the compounds is plotted as a function of the temperature.

1.0

0.8

0.6 CO
Coverage

O2
CO2
0.4 *

0.2

0.0
400 500 600 700 800 900 1000
Temperature in K

Figure 5.3: Surface coverage of the compound CO as a function of temperature.

Figure 5.3 shows that with temperatures below 550 K, CO is the most abundant reaction
intermediate (MARI). With increasing temperature, the equilibrium of CO adsorption shifts
to the gas phase side and hence the amount of free catalyst sites increases. Comparing these
results with Figure 5.4, we see that with decreasing surface coverage of CO, the production of
 5.5 Data interpretation 119

CO2 increases. This result can be rationalized as with an increasing number of free surface
sites, there are more available sites for dissociative adsorption of O2 . Further increasing the
temperature above 600 K reduces the coverage of CO to nearly 0 and hence negatively impacts
the production of CO2 .

CO2
0.010 Maximum is at 584.3 Kelvin

0.008

0.006
1
Rate s

0.004

0.002

0.000
400 500 600 700 800 900 1000
C HAPTER 5

Temperature in K

Figure 5.4: Optimal operation temperature with the respect to the production of CO2 .


Using Figure 5.4, we are able to answer the first question of the example exercise. From
this Figure, it can be seen that the optimal operating temperature of our catalyst (i.e. the
temperature at which the activity is highest) is at approximately 570 K.

5.5.2 Reaction orders

The reaction order in CO and O2 as function of temperature is shown in Figure 5.5. The
graph generated by MKMCXX can be found in the graphs folder after the simulation. The
corresponding numeric data can be found in the graphs/order subfolder.

1.00 CO*
O2
0.75
0.50
0.25
Reaction order

0.00
0.25
0.50
0.75
1.00
400 500 600 700 800 900 1000
Temperate T in K

Figure 5.5: Reaction order of CO and O2 as function of temperature.

From the Figure 5.5, it can be seen that the reaction order in CO is around -1 at very low
temperature. From the definition of the reaction order, this implies that a doubling of the
partial pressure of CO will result in halving the rate for CO2 production. Furthermore, it can
be seen that the reaction order of O2 is independent of the temperature. We can now readily
 120 Chapter 5. Performing microkinetics with MKMCXX

answer the second part of the example exercise. The reaction order for O2 is a constant value
of 0.5 and therefore doubling the partial pressure of O2 will increase the rate of the reaction
√
by a factor 2. The reaction order for CO is not constant and increases with increasing
temperature. For low temperatures doubling the partial pressure will halve the rate of the
reaction as the surface is already poisoned by CO as can also be readily seen in Figure 5.3. For
higher temperatures, doubling the partial pressure of CO will double the rate of the reaction.
Another way of interpreting these results is as follows. At elevated temperatures, the
catalytic surface is nearly empty. When the surface is empty, there is insufficient amount
of reactant on the catalytic surface. Increasing the partial pressure of CO will thus increase
the impingement rate of CO with the surface and hence increase the overall reaction rate.
A similar reasoning applies to O2 , but with the notable difference that only a single oxygen
atom of the oxygen molecule is required. This leads to a reaction order of a half.

5.5.3 Apparent activation energy

The graph for the apparent activation energy, as generated by MKMCXX, can be found in the
graphs/eact subfolder of the simulation. The corresponding numeric data calculated can
be found in the subfolder range/eact. The apparent activation energy for CO oxidation as
function of temperature is plotted in Figure 5.6.

C HAPTER 5

Apparent activation energy Ea in J/mol

150000

100000

50000

400 500 600 700 800 900 1000

Temperate T in K

Figure 5.6: Apparent activation energy for CO as function of temperature.

Careful examination of Figure 5.6 shows that the apparent activation energy for CO
oxidation is negative above 570 K. As the highest activity is observed around 570 K, this
implies (form the definition of the apparent activation energy) that the apparent activation
energy at that point is 0. Above 570 K, this leads to negative apparent activation energies as
one needs to decrease the temperature in order to increase the rate. Vice versa, below 570 K,
the apparent activation energy is positive as one needs to increase the temperature to increase
the rate.

5.5.4 Degree of rate control

The graph for the degree of rate control analysis can be found in the drc subfolder. The
results for the degree of rate control analysis are shown in Figure 5.7.
In the example exercise, we were asked to determine the rate controlling step. As expected,
the dominant rate controlling step for the oxidation of CO is the recombination of CO and O
on the surface. This elementary reaction step has a DRC coefficient of unity, independent
of the temperature. In other words, it is the rate-determining step (as opposed to having
multiple rate-controlling steps).
 5.6 Convergence 121

1.0

Degree of rate control coefficient (dimensionless)

0.8

0.6 CO + * CO*
O2 + 2* 2O*
CO* + O* CO2* + *
0.4 CO2 + * CO2*

0.2

0.0
400 500 600 700 800 900 1000
Temperate T in K

Figure 5.7: Degree of rate control as function of temperature.

5.6 Convergence
C HAPTER 5

The program MKMCXX is capable of performing microkinetic simulations for large complex
catalytic reactions, nevertheless it is important that the compounds of the reaction have
converged at the end of the simulation time. In other words, we are typically seeking the


steady-state solution thus we integrate in time sufficiently long that we expect this solution to be
reached. This is of course not guaranteed and it can be the case, especially at low temperatures,
that the calculation is not yet converged with respect to time-integration. MKMCXX will issue
a warning for most convergence problems, however it is important to note that the generated
data should always be scrutinized by the user to check for convergence.

1.0 A*
B*
*
Surface coverage (dimensionless)

0.8

0.6

0.4

0.2

0.0
10 11 10 9 10 7 10 5 10 3 10 1 101
Time t in s

Figure 5.8: Surface coverage versus time at 450 K.

Figure 5.8 shows the surface coverage of the catalyst with respect to the time for the
simulation at 450 K. Evaluating this Figure shows that the components of the CO oxidation
are converged after 1 · 10−7 seconds.
Due to the significant difference of the rates of the individual elementary reaction steps, the
time-integration of the chemokinetic networks typically shows three characteristic regimes:

1. Reactant adsorption onto the surface site.

2. Surface reaction initiation.

 122 Chapter 5. Performing microkinetics with MKMCXX

3. Reaction propagation.

The last two step may occur simultaneously if the initial reaction steps are rate controlling
or the reaction network consists of no serial reaction steps.
 Recall

Recall from section 1.5.2 on page 19 that this is exactly the reason why we are allowed
to use the quasi-equilibrium assumption. The rates for adsorption/desorption are
much faster that the rates for the surface reactions. Thus, adsorption/desorption is
already pseudo-equilibrated with respect to the other elementary reaction steps.

 Important

Be careful with assuming a reaction has converged just because the concentration
does not change anymore. Always carefully check your data and perform additional
calculations at somewhat longer time scales.

C HAPTER 5
5.7 Numerical versus analytical


In the previous sections we explained how to model a microkinetic network using the program
MKMCXX. We showed how to model a heterogeneous catalytic reaction and how to interpret
the numerical output data. In this section we will derive an analytical solution for same the
chemokinetic network and compare the results with the numerical solution of MKMCXX. We
will highlight when such a procedure is possible and what additional insights can be gained
from deriving an analytical solution.
Assuming that the surface reaction from CO to CO2 is the rate determining step8 , and
furthermore using the pseudo-equilibrium assumption and the zero conversion assumption,
we obtain the following equations for the site-balance and the surface coverage of CO:

θCO + θO + θ∗ = 1 (5.7)
KCO pCO
θCO = q (5.8)
1 + KCO pCO + KO2 pO2

Because zero conversion was assumed, the overall rate is equal to the forward rate of the
rate-determining step.

q
krds KCO pCOKO2 pO2
rrds = q (5.9)
(1 + KCO pCO + KO2 pO2 )2

In chapter 3, the general rate of adsorption and desorption known as the Hertz-Knudsen
equations was derived. Here, we repeat the formula for convenience.

p A
kads = p i i (5.10)
2πmi kb T
!
k T 3 Ai (2πmi kb ) −Edes,i
kdes = b 3 exp , (5.11)
h σi θrot,i kb T
8
This is already confirmed by the simulation, but it is good to mention it here again as an assumption.
 5.8 Exercises 123

wherein i represents the molecule adsorbing onto the catalyst site. From equations 5.10
and 5.11, we can readily calculate the equilibrium constant for adsorption

kads
K= (5.12)
kdes
√ pi Ai
2πmi kb T
=   (5.13)
kb T 3 Ai (2πmi kb ) −Edes,i
h3 σθrot exp kb T

pi h3 σi θrot,i
!
Edes,i
= exp (5.14)
kb T 2 (2πmi kb T )1.5 kb T

Using the obtained expression for the equilibrium constant of the adsorption/desorption
reactions, the overall rate of the reaction can be calculated. In Figure 5.9 the analytical solutions
of the overall rate using the Hertz-Knudsen equations are plotted against the numerically
calculated solutions of MKMCXX.
C HAPTER 5

Numerical solution (mkmcxx)

0.010 Analytical solution


0.008

0.006
1
Rate s

0.004

0.002

0.000
400 500 600 700 800 900 1000
Temperature in K

Figure 5.9: Analytical solution versus numerical solution

Figure 5.9 shows that the numerical and analytical solution perfectly match. The reason
for this match is that the fundamental assumptions behind the analytical derivation remain
valid over the whole temperature range. This is more of an exception than a rule.
The same procedure can be repeated for the reaction order and the apparent activation
energy. This is basically already covered in section 4.7 on page 101.

5.8 Exercises
The answers to the exercises are given at the end of this book in Appendix C (page 135).
The exercises are marked by a number of gears to indicate their difficulty levels.

 Exercise 5.1 

Consider the CO oxidation reaction as discussed in Chapter 5 on 108.

The input file for this simulation can be found using the link be-
low:  https://www.mkmcxx.nl/downloads/input/input_ex1.
mkm
124 Chapter 5. Performing microkinetics with MKMCXX

The input file contains everything needed for this exercise and in principle needs no
altering. However, we always encourage to alter some values to get some more feeling for
performing microkinetic simulations.

a) Run the simulation using the provided input file in MKMCXX. What is the optimal
temperature for CO oxidation?

b) Which of the elementary reaction steps is the rate-determining step?

 Exercise 5.2 

Consider the following overall reaction which converts reactants A and B to product C, as
given by

A + B → C. (5.15)

C HAPTER 5
The chemokinetic network consists of the following four elementary reaction steps


A + ∗ A∗
B + ∗ B∗
A ∗ +B∗ C ∗ +∗
C + ∗ C∗

In Table 5.4 and 5.5, the kinetic parameters for the elementary reaction steps are given.

Table 5.4: Kinetic parameters for the adsorption and desorption steps

Compound A [m2 ] m (a.u.) σ [-] θrot [K] Eads [kJ/mol]

A 1 · 10−20 40 1 2.92 50
B 1 · 10−20 16 1 1.66 100
C 1 · 10−20 63 1 0.601 5

a) The total gas pressure is 30 bar and the reactants are fed stoichiometrically. Perform a
microkinetic simulation using MKMCXX between 500 K and 1500 K in steps of 50 K. What
is the surface composition of where the production of C is the highest? At what temperature
does this occur?

b) Calculate the forward reaction rate constant of the adsorption reaction of compound A
using the expression generated by MKMCXX at T = 750 K.

Table 5.5: Kinetic parameters for the surface reaction.

f
Reaction νf [s−1 ] νb [s−1 ] Ea [kJ/mol] Eab [kJ/mol]
A* + B* −−→ C* + * 2013 1013 90 320
 5.8 Exercises 125

 Exercise 5.3  

Consider the following overall reaction

A2 + B → E (5.16)

The relevant kinetic parameters for the rate expressions can be found in Tables 5.6 and
5.7.

Table 5.6: Parameters for the adsorption and desorption steps

Compound A [m2 ] m (a.u.) σ [-] θrot [K] Eads [kJ/mol]

A2 1 · 10−20 32 1 1 100
B 1 · 10−20 15 1 1 110
1 · 10−20
C HAPTER 5

E 10 1 1 50


a) Construct the set of elementary reactions steps that describes the reaction mechanism as
given by the overall reaction.

b) What is the expected rate-determining step on the basis of the kinetic parameters?

c) Construct an input file for MKMCXX to perform a microkinetic simulation between

T =500-1500 K. What is the rate-determining step and how does this change with respect to
temperature?

 Exercise 5.4   

Consider the following parallel reaction network, wherein A can be converted to both E2
as well as F as given by

A → E2 (5.17)
A→F (5.18)

The reaction proceeds via the following set of elementary reaction steps

Table 5.7: Parameters for the surface reaction.

f
Reaction νf [s−1 ] νb [s−1 ] Ea [kJ/mol] Eab [kJ/mol]
A* + B* −−→ C* + * 1013 1013 100 50
A* −−→ D* 1013 1013 75 120
C* + D* −−→ E* 1013 1013 50 50
126 Chapter 5. Performing microkinetics with MKMCXX

A + ∗  A∗
A∗  2B∗
B∗  C∗
C∗  E∗
2E∗  E2 + 2∗
B∗  D∗
2D∗  F∗
F∗  F
(5.19)

The input file for MKMCXX containing the relevant parameters for the
rate expressions is found using the following link:
 https://www.mkmcxx.nl/downloads/input/input_ex4.mkm

C HAPTER 5
a) Assume that the dissociation of A into 2B is rate controlling at 1400 K. Calculate the rate


and compare this with the rate as calculated by MKMCXX.

b) What is the most abundant reaction intermediate between 600 and 800 K?

c) Evaluate the results of the DRC analysis. At lower temperature, the DRC (with respect to
component A) of B∗  D∗ is negative. How should this result be interpreted?

d) The current simulation favors the production of E2 . What would you change in the
parameters to increase the selectivity towards F?


 A

APPENDIX
FUNDAMENTAL CONSTANTS

Constant Symbol Value Units

Speed of light c 2.99792458 · 108 m · s −1
Planck’s constant h 6.2606957 · 10−34 J·s
Boltzmann’s constant kb 1.3806488 · 10−23 J · K −1
A PPENDIX A

Avogadro’s constant NA 6.02214129 · 1023 mol −1

Gas constant R 8.3144621 J · K −1 · mol −1
Atomic mass constant mu 1.660538921 · 10−27 kg

 A PPENDIX A
 B

APPENDIX
MATHEMATICS

Contents
B.1 Stirling Approximation 129
B.2 Gaussian integrals 130
B.3 Taylor expansion for the exponential function 130
B.4 Geometric series 131
B.5 Lagrange’s method of undetermined multipliers 131
B.6 l’Hôspital’s Rule 133
A PPENDIX B

B.7 Matrix diagonalization 133

B.8 Sum of degree of rate control coefficients 134


B.1 Stirling Approximation

Stirling’s theorem states:

√ √
 
1 1 1
2πnn+ 2 e−n < n! < 2πnn+ 2 e−n 1+ (B.1)
4n

Therefore,

1
ln(n!) = (n + ) ln(n) − n + C (B.2)
2
 
1 . For large values of n
where C is a number inbetween 0.9189 and 0.9189 + ln 1 + 4n
this reduces to the simpler form,

ln(n!) ≈ n ln(n) − n (B.3)

 130 Appendix B. Mathematics

B.2 Gaussian integrals

Z ∞
2 √
exp−x dx = π (B.4)
−∞

Z ∞  2n+1
2 2 √ (2n)! a
x2n exp−x /a dx = π (B.5)
0 n! 2

Z ∞
2 2 n! 2n+2
x2n+1 exp−x /a dx = a (B.6)
0 2

B.3 Taylor expansion for the exponential function

A Taylor series is a representation of a function as an infinite sum of terms that are calculated
from the values of the function’s derivatives at a single point. Mathematically, the Taylor series
is given by the following formula

∞
f (n) (a)
(x − a)n
X
f (x) = (B.7)
n!
n=0

where f (x) is the function we wish to approximate, f (n) is the nth order derivative of
f (x) and a is the point from which we wish to evaluate the derivatives. When a = 0 is chosen,
the series is also called a Maclaurin series. Here, we are going to apply the above formula to

A PPENDIX B
derive a Maclaurin series for the exponential function.


∞
f (n) (a)
(x − a)n
X
exp x = (B.8)
n!
n=0
∞ dexp x |
dx x=0 n
X
= x (B.9)
n!
n=0
∞
X xn
= (B.10)
n!
n=0
10
11 12 13
= ++ + + ··· (B.11)
0!
1! 2! 3!
x 2 x3
=1+x+ + + ··· (B.12)
2! 3!

From the factorial in the denominator, you can already guess that this series converges
quite rapidly. Furthermore, when x is much smaller than one, only the first few terms have
to be considered of this series. If we for instance only take the first term into consideration,
we say that we cut the series off at the linear term so that our Taylor approximation becomes

exp x ≈ 1 + x for x  1. (B.13)

 B.4 Geometric series 131

B.4 Geometric series

In mathematics, a geometric series is a series with a constant ratio between successive terms.
Here, we will derive an expression for a special case of an infinite geometric series.
Consider the geometric series

∞
a + ar + ar2 + ar3 + · · · = ark .
X
(B.14)
k=0

Such a series can converge when the absolute value of r is sufficiently small as

∞
a
ark =
X
, for |r| < 1. (B.15)
1−r
k=0

B.5 Lagrange’s method of undetermined multipliers

When optimizing a (multivariable) function f under a constraint, we can use Lagrange’s
method of undetermined multipliers. We introduce for each constraint a new variable (λi )
called a Lagrange multiplier and study the Lagrange function

X
L(x1 , x2 , · · · , xi , λ1 , λ2 , · · · , λi ) = f (x1 , x2 , · · · , xi ) − λi · gi (x1 , x2 , · · · , xi ), (B.16)
i

where gi is a function representing constraint i. The constrained extrema of f are then

the critical points of the Lagrangian L. Thus we solve
A PPENDIX B

∇L(x1 , x2 , · · · , xi , λ1 , λ2 , · · · , λi ) = 0. (B.17)

Let us illustrate the above procedure with an example as shown in Figure B.1.


Suppose we wish to maximize

f (x, y) = x + 3y (B.18)

subject to the constraint

x2 + y 2 = 1. (B.19)

The Lagrangian for this problem is

 
L(x, y, λ) = x + 3y − λ x2 + y 2 − 1 . (B.20)

This leads to


 
∂L 
∂x 
 ∂L 1 + 2λx
∇L(x, y, λ) = 
 ∂y  =  3 + 2λy  = 0.
  (B.21)

∂L x2 + y 2 − 1
∂λ
132 Appendix B. Mathematics

x2 + y2 = 1
z = x + 3y
Optima
8
4
0
4
8
2
1
2 0 y
1 1
0
x 1 2
2

Figure B.1: Illustration of the constrained optimization problem using Lagrange’s method of undetermined
multipliers. The objective function f is depicted by the dashed surface and the projected constraint function g is
depicted by the black curve.

This set of equations has two solutions as given by

A PPENDIX B
λ = ± 2, (B.22)

which implies that the stationary points are



!
1 3 √
L0,1 = − √ ,− √ , 2 (B.23)
2 2 2 2

and

!
1 3 √
L1,1 = √ , √ ,− 2 . (B.24)
2 2 2 2

Evaluating the objective function yields

!
1 3 5
f ± √ ,± √ = ±√ , (B.25)
2 2 2 2 2

where the positive solution corresponds to a maximum and the negative solution to a
minimum.
 B.6 l’Hôspital’s Rule 133

B.6 l’Hôspital’s Rule

When evaluating a limit that is the quotient of two functions and the limit of those functions
are either zero or infinity, then l’Hôspital’s Rule states that

f (x) f 0 (x)
lim = lim . (B.26)
x→a g(x) x→a g 0 (x)

The differentiation of the numerator and denominator often simplifies the quotient or
converts it to a limit that can be evaluated directly.
For example:

exp (x) − 1 exp (x)

lim = lim (B.27)
x→∞ x2 + x x→∞ 2x + 1
1
= (B.28)
2·0+1
=1 (B.29)

B.7 Matrix diagonalization

A square matrix A can be diagonalized if there exists an invertible matrix P such that

D = P −1 AP, (B.30)

where D is a diagonal matrix. The values on the diagonal are then the eigenvalues,
whereas the columns in P are the corresponding right-eigenvectors. For many applications,
the square matrix A is a real-symmetric matrix (i.e. all values are real and Ai,j = Aj,i ). In
A PPENDIX B

such a case, the inverse of the matrix P equals its transpose P −1 = P | .

For example, consider the matrix


 
−2 −2 1
A = −1 3 −1 . (B.31)
 
2 −4 3

This matrix can be diagonalized with a matrix

 
2 −1 1
P = 1 0 −1 (B.32)
 
0 1 2

and a diagonal matrix

 
1 0 0
D = 0 1 0 . (B.33)
 
0 0 6

The eigenvalues are thus λ1 = 1, λ2 = 1 and λ3 = 6 with corresponding eigenvectors as

columns in matrix P .
 134 Appendix B. Mathematics

B.8 Sum of degree of rate control coefficients

Assume that we can write the rate of an overall reaction as a sum of the rates of its constituting
elementary reaction steps multiplied by an (as yet unknown) contribution constant ni as

 

ni ki+ cij − ki−

X Y Y
roverall = cij  (B.34)
 
i j j
 

cij − Ki−1
X Y Y
= ni ki  cij  , (B.35)

i j j

where ni ∈ R.
To obtain the contribution of an elementary reaction step to the overall reaction, we can
use the DRC analysis as given by

∂ ln roverall
χi = (B.36)
∂ ln ki
∂roverall ki
= (B.37)
∂ki roverall
Q 
P −1 Q
∂ j nj kj k cjk − Kj k cjk ki
= Q 
∂ki P −1 Q
n k
j j j k jk − Kj
c k cjk
(B.38)
Q 
−1 Q
ni ki j cij − Ki j cij
= P Q . (B.39)
−1 Q
j nj kj k cjk − Kj k cjk

A PPENDIX B
Since ni ∈ R, this gives χi ∈ R. When χi > 0, the reaction contributes to the overall
reaction and is rate-limiting. When χi < 0, the reaction reduces the overall reaction and is
rate-limiting.

Summing over χi gives

Q 
P −1 Q
X i ni ki j cij − Ki j cij
χi = P Q  (B.40)
−1 Q
i j nj kj k cjk − Kj k cjk

=1 (B.41)
 C

APPENDIX
SOLUTIONS

Contents
C.1 Solutions of Chapter 1 135
C.1.1 Solutions to exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
C.2 Solutions of Chapter 2 166
C.2.1 Solutions to questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
C.2.2 Solutions to exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
C.2.3 Solution to challenges . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
C.2.4 Challenge: Maxwell-Boltzmann distribution . . . . . . . . . . . . . . . . . . . . . . . 183
C.3 Solutions of Chapter 3 186
C.3.1 Solutions to questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186
C.3.2 Solutions to exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186
C.4 Solutions of Chapter 4 189
C.4.1 Solutions to questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
C.4.2 Solutions to exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190
A PPENDIX C

C.5 Solutions of Chapter 5 192

C.5.1 Solutions to Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192


C.1 Solutions of Chapter 1

C.1.1 Solutions to exercises
The solution below pertain to the exercises of Chapter 1 on page 28 and further.

 Solution 1.1

d[CH4 ]
= k2 [CH3 *][CH3 CHO] (C.1)
dt
The target is to express short-lived intermediates (such as radicals) in terms of gas-phase
concentrations. Here, the unknown variable is [CH3 *] and can be found by using the steady
state approximation. In the steady-state approximation, the time-derivative of one or more
136 Appendix C. Solutions

compounds, typically the reaction intermediates, is set to zero. We apply this approximation
d[CH *] d[CH3 CO*]
to dt3 and dt
.

d[CH3 *]
=k1 [CH3 CHO] − k2 [CH3 *][CH3 CHO]
dt
+ k3 [CH3 CO*] − 2k4 [CH3 *]2 = 0 (C.2)

and

d[CH3 CO*]
= k2 [CH3 *][CH3 CHO] − k3 [CH3 CO*] = 0 (C.3)
dt
Combining equations C.2 and C.3 yields

s
k1
[CH3 *] = [CH3 CHO]1/2 . (C.4)
2k4

Inserting equation C.4 into equation C.1 gives

s
d[CH4 ] k1
= k2 [CH3 CHO]3/2 . (C.5)
dt 2k4

and the rate of formation for [C2 H6 ] becomes

d[C2 H6 ] k
= k4 [CH3 *]2 = 1 [CH3 CHO] (C.6)
dt 2


 Solution 1.2
The problem in the exercise can be solved by either applying the steady-state approximation
A PPENDIX C

or by assuming a pre-equilibrium. Here, we have applied the former as such an approxima-

tion is more general and also better exemplifies the mathematical procedure. Obviously, a
derivation based on the pre-equilibrium assumption is also valid.


d[O2 ]
= k2 [NO2 ][NO3 ] (C.7)
dt
Applying the steady-state approximation to NO and NO3 yield

d[NO]
= k2 [NO2 ][NO3 ] − k3 [NO][NO3 ] = 0 (C.8)
dt
and

d[NO3 ]
= k1+ [N2 O5 ] − k1− [NO2 ][NO3 ] − k2 [NO2 ][NO3 ] − k3 [NO][NO3 ] = 0. (C.9)
dt
By subtracting equation C.8 from C.9, we obtain
 C.1 Solutions of Chapter 1 137

k1+
[NO2 ][NO3 ] = −
[N2 O5 ] (C.10)
k1 + 2k2

which we can readily insert into the rate equation of O2 formation, resulting in

+
d[O2 ] k k2
= −1 [N2 O5 ]. (C.11)
dt k1 + 2k2

 Think deeper...

• The reason why the reaction order is unity, rather than two, is because the overall
reaction is not an elementary reaction step.

• For NO2 , the rate equation is

d[NO2 ]
= k1+ [N2 O5 ] − k1− [NO2 ][NO3 ] + 2k3 [NO][NO3 ]. (C.12)
dt

Inserting equations C.8 and C.10 to C.12 gives

+ −
d[NO2 ] k k 2k1+ k2
=k1+ [N2 O5 ] − −1 1 [N2 O5 ] + [N2 O5 ] (C.13)
dt k1 + 2k2 −
k1 + 2k2
 
k1+ k1− 2k1+ k2
= k1+ − + −  [N2 O5 ] (C.14)
k1− + 2k2 k1 + 2k2

k+  
= − 1 k1− + 2k2 − k1− + 2k2 [N2 O5 ] (C.15)
k1 + 2k2
k+
= − 1 4k2 [N2 O5 ] (C.16)
A PPENDIX C

k1 + 2k2
d[O2 ]
=4 . (C.17)
dt


 Solution 1.3

Applying the steady-state approximation to [N2 O2 ] gives

d[N2 O2 ]
=k1+ [NO]2 − k1− [N2 O2 ] − k2 [N2 O2 ][H2 ] = 0. (C.18)
dt
From this expression, we can equate the [N2 O2 ] to

k1+ [NO]2
[N2 O2 ] = −
. (C.19)
k1 + k2+ [H2 ]
138 Appendix C. Solutions

The rate of formation of [N2 O] is then found to be

d[N2 O] k2 k1+ [NO]2 [H2 ]

=k2 [N2 O2 ][H2 ] = (C.20)
dt k1− + k2+ [H2 ]
k1+ [NO]2
= . (C.21)
1 + k1− /k2 [H2 ]

 Solution 1.4

In the overall chain reaction, the elementary reaction steps are denoted as

Br2 → 2Br* initiation

CH4 + Br* → CH3 * + HBr propagation
CH3 * + Br2 → CH3 Br + Br* propagation
2Br* → Br2 termination

The steady-state approximation can be readily applied to the reaction rates of the interme-
diates.

d[Br*]
= 2k1 [Br2 ] − k2 [CH4 ][Br*] + k3 [CH3 *][Br2 ] − 2k4 [Br*]2 = 0 (C.22)
dt

d[CH3 *]
= k2 [CH4 ][Br*] − k3 [CH3 *][Br2 ] = 0 (C.23)
dt
Inserting equation C.22 in C.23 results in

!1/2
A PPENDIX C

k1
[Br*] = [Br2 ] . (C.24)
k4


Applying equation C.24 to C.23 yields

 1/2
k2 [CH4 ] kk [Br2 ]
1

[CH3 *] =
4

k3 [Br2 ]
1/2
k2 k1 [CH4 ]
= . (C.25)
1/2
k3 k4 [Br2 ]1/2

Finally substituting the result of equation C.25 for the concentration of [CH3 *] in the
formation rate of [CH3 Br] results in

1/2
d[CH3 Br] k2 k1
= k3 [CH3 *][Br2 ] = [CH4 ][Br2 ]1/2 . (C.26)
dt k
1/2
4
 C.1 Solutions of Chapter 1 139

 Solution 1.5

a) The elementary reaction step for CO adsorption is

CO + *  CO* (C.27)

Assuming an equilibrium of the above reaction and only one type of surface site yields

θCO
KCO = . (C.28)
pCO θ*

Furthermore, we can apply the site-balance defined as

θ* + θCO = 1. (C.29)

Hence,

θCO =KCO pCO θ* = KCO pCO (1 − θCO ) (C.30)

θCO (1 + KCO pCO ) =KCO pCO (C.31)
KCO pCO
θCO = . (C.32)
1 + KCO pCO

b) The elementary reaction step for dissociative CO adsorption is

CO + 2*  C* + O*. (C.33)
A PPENDIX C

Assuming equilibrium of the above reaction and only one type of surface site gives us


θC θO
KCO = . (C.34)
pCO θ*2

For the mass balance (or site balance), we get

θ* + θC + θO = 1. (C.35)

Because the surface fraction of C and O are equal

2
θC
KCO = . (C.36)
pCO θ*2
140 Appendix C. Solutions

Thus,

(C.37)
p p
θC = KCO pCO θ* = KCO pCO (1 − 2θC )
(C.38)
p p
θC (1 + 2 KCO pCO ) = KCO pCO
p
KCO pCO
θC = p . (C.39)
1 + 2 KCO pCO

c) The two elementary reaction steps for co-adsorption are

CO + *  CO* (C.40)
H2 + 2*  2H*. (C.41)

Assuming pre-equilibrium of these two elementary reaction steps gives

θCO
KCO = (C.42)
pCO θ*

and

2
θH
KH2 = (C.43)
pH2 θ*2

and thus for the fractional coverages in terms of the fraction of free sites

q
θH = KH2 pH2 θ* (C.44)
θCO =KCO pCO θ* (C.45)

Using the site balance,

θCO + θH + θ* = 1 (C.46)
A PPENDIX C

and inserting the equation for the surface fraction into the equation


q
(1 + KCO pCO + KH2 pH2 )θ* = 1 (C.47)

and rearranging for θ* finally yields

1
θ* = p . (C.48)
1 + KCO pCO + KH2 pH2

This equation can be reinserted into the equations for the surface fractions to obtain
p
KH2 pH2
θH = p (C.49)
1 + KCO pCO + KH2 pH2
KCO pCO
θCO = p (C.50)
1 + KCO pCO + KH2 pH2
(C.51)
 C.1 Solutions of Chapter 1 141

d)  Write down the mechanism.

The mechanism is given by

1. CO + *  CO*
2. H2 + 2*  2H*
3. CO* + H*  HCO* + *
4. HCO* + H*  H2 CO* + *
5. H2 CO* + H*  H3 CO* + *
6. H3 CO* + H*  H3 COH* + *
7. H3 COH + *  H3 COH*

 Reason why you only need to take the surface coverages of CO, H and CH3 COH into account.
For the construction of the site balance, we only need to take into account those compounds
of which we can reasonably expect that they have a non-negligible surface coverage. These are
always those compounds before the rate-determining step (i.e. CO and H2 ). In addition, the
question states that we can consider the RDS to be irreversible and that all steps after the RDS,
except the last step, are very fast. From this piece of information, we can assume that once CO
reacts with H on the surface, methanol is formed rapidly in the subsequent hydrogenation
steps. Desorption of methanol is considered to be in equilibrium though, thus we anticipate
that methanol has a non-negligible surface coverage. Conclusively, for the construction of the
site-balance, we only need to consider CO, H2 and CH3 OH.
 Derive an expression for the reaction rate.

The third step is considered to be the rate-determining step. This gives the following expres-
sion for the overall rate

r = k3+ θCO θH . (C.52)

We here use the result of question 5c and note that besides CO and H2 , methanol is also
equilibrated with the surface. Hence the Langmuir isotherms of CO and H2 are
A PPENDIX C

p
KH2 pH2


θH = p (C.53)
1 + KCO pCO + KH2 pH2 + KCH3 OH pCH3 OH
K p
θCO = p CO CO . (C.54)
1 + KCO pCO + KH2 pH2 + KCH3 OH pCH3 OH

And the overall rate is

k3+ KH2 pH2 KCO pCO

p
r= 
p 2 . (C.55)
1 + KCO pCO + KH2 pH2 + KCH3 OH pCH3 OH

 Give the limits (the domain) of the reaction orders for H2 , CO and methanol.

To obtain the reaction order for CO, H2 and methanol, one needs to solve the following
differential:
142 Appendix C. Solutions

∂ ln(r+ )
nH2 =pH2 (C.56)
∂pH2
=D1 − 2D2 (C.57)

Here, we are going to split up the complex differential into two smaller (and hopefully simpler)
differentials:

h i
∂ ln k3+ + 12 ln KH2 + 12 ln pH2 + ln KCO + ln pCO
D1 = pH2 (C.58)
∂pH2

 
∂ ln 1 + KCO pCO + KH2 pH2 + KCH3 OH pCH3 OH
p
D2 = pH2 (C.59)
∂pH2

D1 can be readily solved as all the terms except the ln pH2 cancel out giving

∂ 12 ln pH2
D1 = pH2 (C.60)
∂pH2
1 1
= pH2 (C.61)
2 pH2
1
= (C.62)
2

D2 is a bit more complex and requires applying the chain-rule

∂ ln(a) ∂a
D2 = pH2 (C.63)
∂a ∂pH2 A PPENDIX C

where


q
a = 1 + KCO pCO + KH2 pH2 + KCH3 OH pCH3 OH (C.64)

Thus

 p 
1 ∂ 1 + KCO pCO + KH2 pH2 + KCH3 OH pCH3 OH
D2 =pH2 (C.65)
a ∂pH2
 
1 1q −1/2
=pH2 KH2 pH (C.66)
a 2 2
p !
1 KH2 pH2
= p (C.67)
2 1 + KCO pCO + KH2 pH2 + KCH3 OH pCH3 OH
1
= θH (C.68)
2
 C.1 Solutions of Chapter 1 143

and the final result is

∂ ln(r+ )
nH2 =pH2 (C.69)
∂pH2
1
= − θH (C.70)
2

We can now easily establish the lower and upper limit of the reaction order in H2 :

1 1
θH ∈ [0, 1] → nH2 ∈ [− , ]. (C.71)
2 2

The derivations for the reaction order in CO and methanol are quite similar to the one for H2
and result in:

∂ ln(r+ )
nCO =pCO (C.72)
∂pCO
=1 − 2θCO (C.73)

and

∂ ln(r+ )
nCH3 OH =pCH3 OH (C.74)
∂pCH3 OH
= − 2θCH3 OH (C.75)

This gives for the lower and upper limits for CO and methanol:

θCO ∈ [0, 1] → nCO ∈ [−1, 1] (C.76)

and
A PPENDIX C

(C.77)


θCH3 OH ∈ [0, 1] → nCH3 OH ∈ [−2, 0]

 Solution 1.6

a) The three steps leading to oxygen removal are:

8. H2 O + *  H2 O*
9. H2 O* + *  OH* + H*
10. OH* + *  O* + H*
(C.78)
144 Appendix C. Solutions

b) The reaction order is determined by the rate-limiting step and all the steps that are in
equilibrium before it. This is most likely step (5) as it involves a hydrogenation step subsequent
to two other hydrogenation steps.

c) The overall rate equation, based on the rate-determining step, is given by

r = k5 θH2 COO θH (C.79)

We need to substitute the surface coverages by expressions that only contain gas-phase
pressures and reaction rate constants. To start, we assume quasi-equilibrium of all elementary
reaction steps before the rate-determining step.

q
θH = K1 pH2 θ* (C.80)
θCO2 =K2 pCO2 θ* (C.81)
K3 θCO2 θH
θHCOO = (C.82)
θ*
K4 θHCOO θH
θH2 COO = (C.83)
θ*

By substituting the expressions for θH and θCO2 in equation C.82, we get

q
θHCOO = K3 K1 pH2 K2 pCO2 θ* (C.84)

This result can be readily inserted into equation C.83 to obtain

θH2 COO = K4 K3 K2 pCO2 K1 pH2 θ* (C.85)

Now, we need to use the above expressions in combination with the site-balance to construct
the expression for the free sites
A PPENDIX C


θH + θCO2 + θHCOO + θH2 COO + θ* = 1 (C.86)

 q q 
θ* 1+ K1 pH2 + K2 pCO2 + K3 K1 pH2 K2 pCO2 + K4 K3 K2 pCO2 K1 pH2 = 1 (C.87)

This leads to

1
θ* = p p
1+ K1 pH2 + K2 pCO2 + K3 K1 pH2 K2 pCO2 + K4 K3 K2 pCO2 K1 pH2
(C.88)

Using equations C.80, C.85, and C.88, we can construct the overall rate equation
 C.1 Solutions of Chapter 1 145

r = k5 θH2 COO θH (C.89)

q
= k5 K4 K3 K2 pCO2 K1 pH2 K1 pH2 θ*2 (C.90)
p
k5 K4 K3 K2 pCO2 K1 pH2 K1 pH2
=  2
p p
1 + K1 pH2 + K2 pCO2 + K3 K1 pH2 K2 pCO2 + K4 K3 K2 pCO2 K1 pH2
(C.91)

3/2
k5 K4 K3 K2 pCO2 K1 pH2
=  2
p p
1 + K1 pH2 + K2 pCO2 + K3 K1 pH2 K2 pCO2 + K4 K3 K2 pCO2 K1 pH2
(C.92)

Let us now check whether our hypothesis that the order in H2 is 3/2 is correct. To make
things a bit easier, let us assume that the surface is nearly empty and that θ* ≈ 1, the rate
equation then is

3/2
r ≈ k5 K4 K3 K2 pCO2 K1 pH2 (C.93)

and

3/2
∂ ln pH
nH2 = pH2 2
= 3/2 (C.94)
∂pH2

 Think deeper...

In order to find the optimal fraction, we define the constant α which is the ratio between
the partial pressure of H2 and CO2 .
A PPENDIX C

pH2
α= (C.95)
pCO2

From this, the total pressure becomes



pT = pCO2 + pH2 = pCO2 + αpCO2 = (1 + α)pCO2 (C.96)

and we can relate the partial pressure of H2 and CO2 to the total pressure as
αpT
pH2 = (C.97)
1+α
pT
pCO2 = (C.98)
1+α
(C.99)

Filling out these new expressions and assuming that the surface is nearly empty results in

αpT 3/2
 
pT
r = k5 K4 K3 K2 K1 (C.100)
1+α 1+α
146 Appendix C. Solutions

To find out the optimal value in α, we need to take the first derivative towards α and
equate this to zero. Therefore, we can simplify the above equation by collecting all variables
that do not depend on α

α3/2
r=c (C.101)
(1 + α)5/2

where c is a factor that does not depend on α.

3 α1/2 (1 + α)5/2 − 5 α3/2 (1 + α)3/2

∂r
= 2 2 =0 (C.102)
∂α (1 + α)5

To find the value for α, only the numerator of the above expression is relevant and we can
ignore the denominator

3 1/2 5
α (1 + α)5/2 − α3/2 (1 + α)3/2 = 0 (C.103)
2 2
3 1/2 5/2 5
α (1 + α) = α3/2 (1 + α)3/2 (C.104)
2 2
5
1+α= α (C.105)
3
3
α= (C.106)
2

This result is to be expected, as the best ratio between the partial pressures of the reactants
is of course the ratio between the reaction orders of said reactants. The fraction of CO2 in
terms of the total pressure is then 25 and the fraction of H2 is then 35 .

 Solution 1.7 A PPENDIX C

a) We assume the steady-state approximation on θO . Furthermore, we assume that θCO  θO

and that CO is in quasi-equilibrium with the surface.


b) First, we apply the quasi-equilibrium assumption to CO giving

θCO = K1 pCO θ* (C.107)

Next, we apply the steady-state assumption to θO

dθO
= 2k2+ pO2 θ*2 − k3+ θCO θO = 0 (C.108)
dt

Note that the second and third elementary reaction step in our mechanism are irreversible,
significantly reducing the complexity of the equation. The above result gives us an expression
for the partial coverage of O as
 C.1 Solutions of Chapter 1 147

2k2+ pO2 2
θO = θ* (C.109)
k3+ θCO

To obtain the expression for the free site coverage we assume that θCO  θO giving

θCO + θ* = 1 (C.110)
θ* (1 + K1 pCO ) = 1 (C.111)
1
θ* = (C.112)
1 + K1 pCO

Plugging this result into the overall rate equation yields

r = k3+ θCO θO (C.113)

+
K1 pCO 2k2 pO2 2
= k3+ θ (C.114)
1 + K1 pCO k θCO *
+
3
+
!2 !−1
+ K1 pCO 2k2 pO2 1 K1 pCO
= k3 (C.115)
1 + K1 pCO k+ 1 + K1 pCO 1 + K1 pCO
3
+
2k2 pO2
= (C.116)
(1 + K1 pCO )2

Note that from the above expression, we can easily see that the order in O2 is +1 and the order
in CO is between -2 and 0.


A PPENDIX C

 Solution 1.8


a) Synthesis gas is a mixture of CO and H2 . It is used in Fischer-Tropsch catalysis and in

methanol formation. The former reaction is typically performed over a Fe or Co catalyst. The
latter reaction is typically performed over a Cu catalyst.

b) In order to further increase the production of H2 , CO can be mixed with H2 O to give CO2
and H2 . This reaction is called water-gas shift.

c) From the reaction equation (CH4 + H2 O  CO + 3H2 ) we can see that the reaction
produces more moles of gas than are consumed. Hence, in principle the reaction should be
performed at low pressure to drive the equilibrium to the right hand side of the equation.
However, in a typical reactor a moderate pressure is used as to reduce the reactor volume.
Furthermore, the reaction needs to be performed at high temperature, because this reaction
is strongly endothermic. Again, the high temperature pushes the equilibrium to the right
hand side of the equation.
148 Appendix C. Solutions

d) To start, we need to propose a reaction mechanism in terms of a set of elementary reaction

steps. To convert methane and water into synthesis gas, a set of 10 elementary reaction steps
is necessary.

1. CH4 + 2*  CH3 * + H*
2. CH3 * + *  CH2 * + H*
3. CH2 * + *  CH* + H*
4. CH* + *  C* + H*
5. H2 O + *  H2 O*
6. H2 O* + *  OH* + H*
7. OH* + *  O* + H*
8. H2 + 2*  2H*
9. C* + O*  CO* + *
10. CO + *  CO*

We assume that all the above steps, except the rate-limiting step (9) are in quasi-equilibrium.
As such, we can express all surface coverages by one or more equilibrium constants as follows

θCO = K10 pCO θ* (C.117)

q
θH = K8 pH2 θ* (C.118)
θH2 O = K5 pH2 O θ* (C.119)
K6 θH2 O θ* K5 K6 pH2 O
θOH = = p θ* (C.120)
θH K8 pH2
K5 K6 pH O
K7 p 2
θ2
K7 θOH θ* K8 pH2 * K K K pH2 O
θO = = p = 5 6 7 θ* (C.121)
θH K8 pH2 θ* K8 pH2

In the above expressions for O and OH, we have applied subsequent substitution of the surface
A PPENDIX C

coverages in such a way that we can express every surface coverage in terms of the partial
pressures of the reactants or products and the equilibrium constants. These equilibrium
constants correspond to the elementary reaction steps that connect the partial pressures


with the surface coverages. Below, the same is done for the surface coverages of the CHx
intermediates

K1 pCH4
θCH3 = p θ* (C.122)
K8 pH2
K1 K2 pCH4
θCH2 = θ* (C.123)
K8 pH2
K1 K2 K3 pCH4
θCH =
3/2 θ* (C.124)
K8 pH2
K1 K2 K3 K4 pCH4
θC =
2 θ* (C.125)
K8 pH2

Now we can construct the site-balance

 C.1 Solutions of Chapter 1 149

θCO + θH + θH2 O + θOH + θO + θCH3 + θCH2 + θCH + θC + θ* = 1 (C.126)

and plug the above equations for the surface coverages into it

q K5 K6 pH2 O K K K pH2 O K1 pCH4

θ* = 1 + K10 pCO + K8 pH2 + K5 pH2 O + p + 5 6 7 +p ···
K8 pH2 K8 pH2 K8 pH2
−1
K1 K2 pCH4 K1 K2 K3 pCH4 K1 K2 K3 K4 pCH4
··· + +
3/2 +
2  (C.127)
K8 pH2 K8 pH2 K8 pH2

Since the overall rate is determined by the rate of step 9 we can construct the following
expression

r = k9+ θC θO − k9− θCO θ* (C.128)

k+
= k9+ θC θO − 9 θCO θ* (C.129)
K9

If we assume that there is an equilibrium r = 0, we can express K9 as

θ θ pCO p3H K83 K10 pCO p3H 1

K9 = CO * = 2
= 2
(C.130)
θC θO pH2 O pCH4 K1 K2 K3 K4 K5 K6 K7 pH2 O pCH4 Keq

and using this result for K9 for our overall rate expression results in

r = k9+ θC θO − k9− θCO θ* (C.131)

!
1 θCO θ*
= k9+ θC θO 1− (C.132)
K9 θC θO
A PPENDIX C

 
K1 K2 K3 K4 K5 K6 K7 pCH4 pH2 O pCO p3H
= k9+
3 1 − 1 2 1  2
θ* (C.133)
K8 pH2 K9 pH2 pCH4 Keq


Further assuming that O is the MARI, we can use the following equation for the free sites

!−1
K K K pH2 O
θ* = 1+ 5 6 7 (C.134)
K8 pH2

giving

!
3
K1 K2 K3 K4 K5 K6 K7 pCH4 pH2 O pCO pH
 3 1 − K1 p p 2 K1
9 H2 CH4 eq
K 8 pH 2
r = k9+  2 (C.135)
p
1 + K5 K6 K7 H2 O
K8 pH2
150 Appendix C. Solutions

For low conversions, the second term between parentheses in the numerator can be neglected,
further simplifying the equation to

!−2
K1 K2 K3 K4 K5 K6 K7 pCH4 pH2 O K5 K6 K7 pH2 O
r = k9+
3 1+ (C.136)
K8 pH2 K8 pH2

e) Note that for CH4 and CO, there are no dependencies in the term between parentheses,
so we only need to evaluate the derivative for the part in front of the parentheses. Again, we
apply the same trick as in the previous exercise, so that we obtain:

∂ ln(r+ )
nCH4 = pCH4 =1 (C.137)
∂pCH4

∂ ln(r+ )
nCO = pCO =0 (C.138)
∂pCO

The reaction order in H2 O and H2 is a bit more complicated but with some rewriting the
derivatives equate to

∂ ln(r+ ) K K K ∂pH2 O /∂pH2 O

nH2 O = pH2 O = 1 − 2pH2 O 5 6 7 p (C.139)
∂pH2 O K8 1 + KK
5 K6 K7 H2 O
p p
8 H2 H2
!−1
K K K pH2 O K K K pH2 O
=1−2 5 6 7 1+ 5 6 7 (C.140)
K8 pH2 K8 pH2 pH2
= 1 − 2θO (C.141)

and similarly for H2

A PPENDIX C

∂ ln(r+ )
nH2 = pH2 = −3 + 2θO (C.142)
∂pH2


From the above expressions for the order, we note that the experimental results are not within
the limits of the reaction orders. Consequently, the proposed reaction mechanism is not in
agreement with the experiment and needs to be revised.

f ) Applying the same procedure as previously, the reaction rate now equates to

pCH4
r = k1+  2 (C.143)
5 K6 K7 pH2 O
1 + KK p p
8 H2 H2

Note that steps 2-4 do not appear in the rate equation, because only the first dissociative
methane adsorption step is kinetically relevant. The equilibrium constants for steps 5-8
appear in the denominator because they control the amount of O on the surface.
From the above rate equation, we can establish the reaction orders to be
 C.1 Solutions of Chapter 1 151

∂ ln(r+ )
nCH4 = pCH4 =1 (C.144)
∂pCH4

∂ ln(r+ )
nCO = pCO =0 (C.145)
∂pCO

∂ ln(r+ )
nH2 O = pH2 O = −2θO (C.146)
∂pH2 O

∂ ln(r+ )
nH2 = pH2 = 2θO (C.147)
∂pH2

These results are consistent with the experimental observation.

 Solution 1.9

a) C2 H6 adsorption coincides with D2 adsorption. Upon adsorption, the hydrogen in the

C2 H6 complex can be exchanged for a deuterium by successive dehydrogenation, deuteration
and a desorption step.

b)

r = k2+ θC2 H5 θH − k2− θCH

2
3
(C.148)

c) The steady state equation applied to elementary reaction steps 1,3 and 4 and considering
competitive adsorption, the Langmuir isotherms for ethyl, methyl and hydrogen become as
follows
A PPENDIX C

K1 pC2 H6 2


θC2 H5 = θ* (C.149)
θH

pCH4 2
θCH3 = θ (C.150)
K3 θH *

q
θH = K4 pH2 θ* (C.151)

Plugging equation C.151 into C.149 and into C.150, gives

K 1 pC H
θC2 H5 = p 2 6 θ* (C.152)
K4 pH2
152 Appendix C. Solutions

and

pCH
θCH3 = p 4 θ* (C.153)
K3 K4 pH2

Using the site balance, we obtain the following equations

1
θ* =
K 1 pC H p
(C.154)
pCH4
p
1+ p 2 6 + + K4 pH2
K4 pH2 K3 K4 pH2

K p
p1 C2 H6
K4 pH2
θC2 H5 =
K pC2 H6 p
(C.155)
1
pCH4
p
1+ p + + K4 pH2
K4 pH2 K3 K4 pH2

p
pCH4
K3 K4 pH2
θCH3 =
K pC2 H6 p
(C.156)
1
pCH4
p
1+ p + + K4 pH2
K4 pH2 K3 K4 pH2

p
K4 pH2
θH =
K pC2 H6 p
(C.157)
1
pCH4
p
1+ p + + K4 pH2
K4 pH2 K3 K4 pH2

d) Plugging the above equations into the overall rate equation yields

!
p2CH
k2+ K1 pC2 H6 1− 2
4
K1 K2 K3 K4 pC2 H6 pH2
r=  2 (C.158)
K 1 pC H p
pCH4
p
1+ p 2 6 + + K4 pH2
K4 pH2 K3 K4 pH2

e) Applying the MARI approximation and assuming low conversion simplifies the overall
A PPENDIX C

rate equation to

k+ K1 pC2 H6


r=  2 2 (C.159)
p
1 + K4 pH2

 Think deeper...

The apparent activation energy can be readily evaluated to

app ∂ ln(r+ )
∆Eact = RT 2 (C.160)
∂T
  
∂ ln 1 + K4 pH2
p
+
∂ ln k ∂ ln K 1
= RT 2  2 + −2 (C.161)
 

∂T ∂T ∂T

(2) (1) (4)

= Eact + ∆Hads − θH ∆Hads (C.162)
 C.1 Solutions of Chapter 1 153

 Solution 1.10

a) From the quasi-equilibrated steps (1), (2) and (4), it follows that

θNO = K1 pNO θ* (C.163)

θCO = K2 pCO θ* (C.164)

q
θN = K4 pN2 θ* (C.165)

To derive the surface coverage of O*, we apply the steady-state approximation to this surface
intermediate

2k3+ pO2 θ*2 − k6+ θCO θO = 0 (C.166)

Note that we have neglected the term k5+ θCO θO , because we assume that NO dissociation is
rate-determining. Therefore, this term is negligible compared to the other two terms.
It follows then that

2k3+ pO2 2
θO = θ* (C.167)
k6+ θCO

Combining this expression with the expression for θCO , we obtain

A PPENDIX C

2k3+ pO2 2k+ pO2

θO = +
θ*2 = + 3 θ* (C.168)
k6 K2 pCO θ* k6 K2 pCO


Applying the site balance for all surface intermediates yields

1
θ* = (C.169)
2k3+ pO2 p
1 + K1 pNO + K2 pCO + + K4 pN2
k6+ K2 pCO

Finally, we can evaluate the rate of N2 and CO2 production

1 k+ K p
1 1 2 5 1 NO
rN2 = r5 = k5+ θNO θ* = !2
2 2
2k3+ pO2 p
1 + K1 pNO + K2 pCO + + K4 pN2
k6+ K2 pCO
(C.170)
154 Appendix C. Solutions

2k3+ pO2
rCO2 = r6 = k6+ θCO θO = !2
2k+ pO p
1 + K1 pNO + K2 pCO + + 3 2 + K4 pN2
k6 K2 pCO
(C.171)

 Think deeper...

This catalytic reaction pertains to three-way catalytic convertors for environmental pollu-
tion control.

 Solution 1.11

a)

1.C2 H4 + ∗  C2 H4 * (C.172)
2.O2 + 2∗  2O* (C.173)
3.C2 H4 * + O* → C2 H4 O + 2* (C.174)
(C.175)

Applying a quasi-equilibrium approximation, we obtain the following expressions for the

partial coverages as a function of the partial pressures and equilibrium constants

θC2 H4 = K1 pC2 H4 θ* (C.176)

and
A PPENDIX C

q
θO = K 2 p O2 θ * (C.177)


Note that because ethylene-oxide desorbs rapidly, the surface coverage of ethylene-oxide is
negligible. Thus, we obtain by constructing a site-balance (not shown here), the following
Langmuir-Hinshelwood isotherms:

K1 pC2 H4
θC2 H4 = p (C.178)
1 + K1 pC2 H4 + K2 pO2

and

p
K 2 p O2
θO = p . (C.179)
1 + K1 pC2 H4 + K2 pO2
 C.1 Solutions of Chapter 1 155

b)

rC2 H4 O = k3 θC2 H4 θO (C.180)

p
k3 K1 pC2 H4
K2 pO2
= 
p 2 (C.181)
1 + K1 pC2 H4 + K2 pO2

c) If oxygen strongly adsorbs, we can assume that oxygen is the MARI and hence the overall
rate expression simplifies to

p
k3 K1 pC2 H4 K2 pO2
rC2 H4 O =  p 2 (C.182)
1 + K2 pO2

The reaction orders in oxygen and ethylene are

∂ ln(r+ )
nC2 H4 = pC2 H4 (C.183)
∂pC2 H4
=1 (C.184)

∂ ln(r+ )
nO2 = pO2 (C.185)
∂pO2
1
= − θO (C.186)
2

d) The surface is predominantly occupied with adsorbed O.

e)
A PPENDIX C

app ∂ ln(r+ )
∆Eact = RT 2 (C.187)
∂T
 
(3) (1) 1 (2)
(C.188)


= Eact + ∆Hads + − θO ∆Hads

2

 Solution 1.12

a)

1.SO2 + ∗  SO2 * (C.189)

2.O2 + 2∗  2O* (C.190)
3.SO2 * + O* → SO3 * + * (C.191)
4.SO3 + ∗  SO3 * (C.192)
156 Appendix C. Solutions

b)

rSO3 = k3 θSO2 θO (C.193)

Applying the pre-equilibrium, the irreversible step and the rate-limiting step approximation
we can derive the following expression for the free sites and for the overall reaction rate.

1
θ* = p (C.194)
1 + K1 pSO2 + K2 pO2 + K4 pSO3

p
k3 K1 pSO2 K2 pO2
rSO3 = 
p 2 (C.195)
1 + K1 pSO2 + K2 pO2 + K4 pSO3

c) From the form of the overall reaction equation, the limits of the reaction orders can be
established as

nSO2 ∈ [−1, 1] (C.196)

nSO3 ∈ [−2, 0] (C.197)
1 1
nO2 ∈ [− , ] (C.198)
2 2

d) If O2 strongly adsorbs, then we can apply the MARI approximation for O. This gives the
following expression for the rate and the reaction orders

p
k3 K1 pSO2 K2 pO2
rSO3 = 
p 2 (C.199)
1 + K 2 p O2

nSO2 = 1 (C.200)
A PPENDIX C

nSO3 = 0 (C.201)
1
nO2 = − θO (C.202)
2


e)

app ∂ ln(r+ )
∆Eact = RT 2 (C.203)
∂T
 
(3) (1) 1 (2)
= Eact + ∆Hads + − θO ∆Hads (C.204)
2

f ) From the form of the equation, it is clear that two different types of Langmuir isotherms are
present. This indicates that there are two different types of surface sites that have particular
adsorbates that adsorb on these sites. The rate-determining step proceeds between two
species that are adsorbed on these two different sites.


 C.1 Solutions of Chapter 1 157

 Solution 1.13

a)

1.C2 H5 OH + *  C2 H5 OH* (C.205)

2.O2 + 2∗  2O* (C.206)
3.C2 H5 OH* + O* → C2 H4 O + H2 O + 2∗ (C.207)

K1 pC2 H5 OH
θC2 H5 OH = p (C.208)
1 + K1 pC2 H5 OH + K 2 p O2

p
K 2 p O2
θO = p (C.209)
1 + K1 pC2 H5 OH + K2 pO2

b)
p
k3 K1 pC2 H5 OHK2 pO2
r = k3 θC2 H5 OH θO = 
p 2 (C.210)
1 + K1 pC2 H5 OH + K2 pO2

c)
p
k3 K1 pC2 H5 OH K2 pO2
r = k3 θC2 H5 OH θO =
2 (C.211)
1 + K1 pC2 H5 OH

nC2 H5 OH ∈ [−1, 1] (C.212)

1
nO2 = (C.213)
2

d)
A PPENDIX C

app ∂ ln(r+ )
∆Eact = RT 2 (C.214)
∂T
1


(3) (1) (2)

(C.215)


= Eact + 1 − 2θC2 H5 OH ∆Hads + ∆Hads
2

The apparent activation energy depends on the reaction barrier of the rate-determining step
and all kinetically relevant steps that proceed before the rate-determining step (which are
at equilibrium). In this particular case, this means that the apparent activation energy is
decreased due to the release of energy by the O2 adsorption and is increased in the case that
there is a high surface coverage of ethanol. When the surface coverage of ethanol is high, one
ethanol molecule needs to desorb in order to form a vacant site necessary for this reaction to
occur. If on the other hand the surface coverage of ethanol is very low (and a sufficiently large
amount of free sites are present), the apparent activation energy is decreased even more as
the adsorption heat of ethanol further decreases the apparent activation energy.

e) In the case of elevated temperatures, both ethanol and oxygen only sporadically adsorb on
the surface as these compounds have a higher entropy in the gas phase than adsorbed on the
surface. Consequently, the catalytic surface is predominantly empty.
158 Appendix C. Solutions

 Solution 1.14

a)

r = k3+ θO θH − k3− θOH θ* (C.216)

b) The surface concentrations of θH , θO , θOH and θ* can be found using the pre-equilibrium
approximation. In that way, we can express these surface concentrations as a function of the
partial pressures and the equilibrium constants.

q
θH = K1 pH2 θ* (C.217)
q
θO = K2 pO2 θ* (C.218)
θH2 O = K5 pH2 O θ* (C.219)
θH2 O θ*
θOH = (C.220)
K4 θH
(C.221)

Substituting these surface concentrations into the rate equation yields

q q θH O θ*
r = k3+ K1 pH2 θ* K2 pO2 θ* − k3− 2 θ* (C.222)
K4 θH
q K5 pH2 O θ*2
= k3+ K1 K2 pH2 pO2 θ*2 − k3− p θ* (C.223)
K4 K1 pH2 θ*
q K5 pH2 O
= k3+ K1 K2 pH2 pO2 θ*2 − k3− θ2 (C.224)
K4 K1 pH2 *
p
 
K5 pH2 O
A PPENDIX C
q
+ 2 −
= k3 K1 K2 pH2 pO2 θ* 1 − k3 +
 p  (C.225)
k3 K4 K1 pH2 K2 pO2
!
q pH2 O


= k3+ K1 K2 pH2 pO2 θ*2 1 − √ (C.226)

Keq pH2 pO2

From the last two expressions, it is clear that the equilbrium constant has to be

p
K1 K2 K3 K4
Keq = (C.227)
K5

c) The free site coverage can be readily obtained from the expressions of the surface coverages
in terms of equilibrium constants and the site balance:

1
θ* = p p K5 pH2 O
(C.228)
1 + K1 pH2 + K2 pO2 + p + K5 pH2 O
K4 K1 pH2
 C.1 Solutions of Chapter 1 159

This expression simplifies by applying the MARI approximation (alternatively, you can start
by defining a site balance only containing O* and then derive the expression below).

1
θ* = p (C.229)
1+ K2 pO2

d) To establish the reaction order, first fill out the expression for the free sites in the overall
rate equation

! !2
q pH2 O 1
r = k3+ K1 K2 pH2 pO2 1− √ · p (C.230)
Keq pH2 pO2 1 + K 2 p O2

Because we assume that the reaction is far from equilibrium, we can neglect the second term
between the first set of brackets giving

k3+ K1 K2 pH2 pO2

p
r= 
p 2 (C.231)
1 + K 2 p O2

From the above expression, the following reaction orders can be derived (please refer to the
previous answers for a more thorough derivation).

1
nH2 = (C.232)
2
1
nO2 = − θO (C.233)
2
nH2 O = 0 (C.234)

e) For very low surface coverage, the rate equals to

q
r = k3+ K1 K2 pH2 pO2 (C.235)
A PPENDIX C

Plugging in the following expressions for pH2 and pO2



pH2 = α · pT (C.236)
pO2 = (1 − α) · pT (C.237)

we obtain

r = k3+ (C.238)
p
K1 K2 α(1 − α)pT

∂r = 0):
Taking the first derivative and equating to zero yields ( ∂α

1
α= (C.239)
2

In other words: the partial pressure of hydrogen and of oxygen need to be equal.
160 Appendix C. Solutions

 Think deeper...

The apparent activation energy is

app ∂ ln(r+ ) (3) 1 (1) 1 (2)

∆Eact = RT 2 = Eact + ∆Hads + ∆Hads (C.240)
∂T 2 2
The reaction takes place on a nearly empty surface. For the reaction to take place we need
to have one H and one O to be adsorbed. Their adsorption facilitates the process (lowers the
apparent activation energy) by half the adsorption enthalpy of the corresponding molecule.
The only positive contribution to the apparent activation energy originates from the barrier of
elementary reaction step #3.

 Solution 1.15

a)
1
rN = r5 = k5+ θNO θ* − k5− θN θO (C.241)
2 2

Because all other steps are quasi-equilibrated, we can write

θNO = K1 pNO θ* (C.242)

θCO = K2 pCO θ* (C.243)
θN2 = K3 pN2 θ* (C.244)
q q q
θN = K4 θN2 θ* = K4 K3 pN2 θ* θ* = K3 K4 pN2 θ* (C.245)
θCO2 θ* K7 pCO2 θ* θ* K7 pCO2
θO = = = θ* (C.246)
K6 θCO K6 K2 pCO θ* K2 K6 pCO A PPENDIX C

θCO2 = K7 pCO2 θ* (C.247)

Using the site balance, this results in



1
θ* = p K7 pCO
1 + K1 pNO + K2 pCO + K3 pN2 + K3 K4 pN2 + K K p2 + K7 pCO2
2 6 CO
(C.248)

Plugging this into the overall rate expression and defining an overall reaction equilibrium
constant we obtain

 √ 
pN2 pCO2 1
k5+ K1 pNO 1− pCO pNO Keq
r5 =  2
p K7 pCO
1 + K1 pNO + K2 pCO + K3 pN2 + K3 K4 pN2 + K K p2 + K7 pCO2
2 6 CO

(C.249)
 C.1 Solutions of Chapter 1 161

where

K K2 K5 K6
Keq = p 1 p (C.250)
K4 K3 K7

b) If O is the MARI, then the overall rate equation simplifies to

 √ 
pN pCO
k5+ K1 pNO1 − p 2p 2 K1
CO NO eq
r5 =  2 (C.251)
K7 pCO2
1+ K K p
2 6 CO

Note that the reaction order is always defined by the derivative of the forward direction of the
rate. Hence we do not have to take the second term within the brackets in the nominator into
account. In other words

k5+ K1 pNO
r+ =  2 (C.252)
K7 pCO
1 + K K p2
2 6 CO

We obtain the following reaction orders

nN2 = 0 (C.253)
nNO = 1 (C.254)
nCO = 2θO (C.255)
nCO2 = −2θO (C.256)

The surface contains only O* and *. There is no N2 * on the surface, so lowering or increasing
the partial pressure of N2 has no effect on the overall rate. Hence the reaction order in N2 is
A PPENDIX C

0. The rate scales linearly with the partial pressure in NO because its partial pressure directly
controls the amount of surface NO and accordingly, the overall rate. If we increase the CO
partial pressure, the CO coverage increases which lowers the amount of O* on the surface and


results in more sites for NO adsorption. In constrast, adding CO2 will increase the surface
coverage of CO2 and in turn the O* coverage. This leads to a lower NO surface coverage and
hence the rate decreases. Hence, the order in CO2 is negative.

c) The apparent activation energy for the nearly empty surface yields

app ∂ ln(r+ ) (5) (1)

∆Eact = RT 2 = Eact + ∆Hads (C.257)
∂T

The surface is nearly empty and one way to increase the rate, is to adsorb more NO. The
latter depends on its adsorption energy, hence the negative contribution of the adsorption
energy to the apparent activation energy. (note that adsorption energies are always negative,
so the (+)-sign in the equation in conjunction with the negative energy results in a negative
contribution)
162 Appendix C. Solutions

Alternatively, the barrier for the rate-determining elementary reaction step could be lowered.
Note that this is purely hypothetical. In practice this could perhaps be done using promotors,
but it turns out that lowering the barrier of an elementary reaction step in practive is very dif-
ficult to say the least. Nevertheless, if possible, then a lowering of the barrier of the elementary
reaction step results in a lowering of the apparent activation energy and hence in an increase
of the overall rate.

 Think deeper...

The apparent activation energy in this particular situation would be

∂ ln(r+ )
 
app (5) (1) (2) (6) (7)
∆Eact = RT 2 = Eact + ∆Hads + 2θO ∆Hads + ∆Hads − ∆Hads
∂T
(C.258)

Note that the complete derivation of the above expression is more a test of mathematical
stamina than of chemical understanding.

 Solution 1.16

a) If the third elementary reaction step is the RDS, the overall rate towards H2 O2 is given by
the following equation:

rH2 O2 = k3+ θO2 θH − k3− θOOH θ* . (C.259)

We assume that all other steps are quasi-equilibrated, hence

A PPENDIX C
q
θH = K1 pH2 θ* (C.260)
θO2 = K2 pO2 θ* (C.261)


θH2 O2
θOOH = θ* (C.262)
K4 θH
K5 pH2 O2
= p θ* (C.263)
K4 K1 pH2
θH2 O2 = K5 pH2 O2 θ* (C.264)

By applying the site balance, we can derive the following expression for the free sites

1
θ* = p K5 pH2 O2
(C.265)
1+ K1 pH2 + K2 pO2 + p + K5 pH2 O2
K4 K 1 pH 2

Plugging this into the overall rate expression and introducing an equilibrium constant for the
reverse reaction yields
 C.1 Solutions of Chapter 1 163

 
pH O
k3+ K2 pO2 1 − K1 p 2p 2
p
K1 pH2
eq O2 H2
r = rH2 O2 =  2 (C.266)
p K5 pH2 O2
1+ K1 pH2 + K2 pO2 + p + K5 pH2 O2
K4 K1 pH2

b) If O* is the MARI and we assume a low conversion, the above expression simplifies to

k+ K2 pO2 K1 pH2
p
r= 3
2 (C.267)
1 + K 2 p O2

The reaction orders are

1
nH2 = (C.268)
2
nO2 = 1 − 2θO (C.269)
nH2 O2 = 0 (C.270)

c) The apparent activation energy is given by

app ∂ ln(r+ ) (3) (2) 1 (1)

∆Eact = RT 2 = Eact + (1 − 2θO2 )∆Hads + ∆Hads (C.271)
∂T 2

For a more thorough description how this answer is obtained, please look at the results of the
previous questions.

d) In the high temperature regime, the surface is nearly empty and the rate is then given by

q
r = k3+ K2 pO2 K1 pH2 (C.272)
A PPENDIX C

and the apparent activation energy becomes



app ∂ ln(r+ ) (3) (2) 1 (1)

∆Eact = RT 2 = Eact + ∆Hads + ∆Hads (C.273)
∂T 2

The apparent activation energy constitutes all kinetically relevant steps, which is the rate-
limiting elementary reaction step and the two adsorption steps that precede the rate-limiting
step. The apparent activation energy is set by the barrier of the rate-limiting step and is lowered
(in the case of an empty surface) by the adsorption energy of O2 and half the adsorption energy
of H2 . Note that in this particular case, if the absolute value of the sum of the adsorption
terms is larger than the barrier of the elementary reaction step, that the apparent activation
energy becomes negative. This essentially means that the reaction rate is increased with
decreasing temperature. This is readily seen if one considers the Sabatier’s Principle. At
the high temperature limit, we have an empty surface. By decreasing the temperature, more
adsorbates will stick to the surface, hence enhancing the overall rate. In other words, by
decreasing the temperature, we are moving our reaction towards the Sabatier’s optimum.
164 Appendix C. Solutions

 Solution 1.17

a) The set of elementary reaction steps that describe the dual-site kinetic network is:

1. CO + ∗  CO∗ (C.274)
2. H2 + 2∗  2H∗ (C.275)
3. H2 + 2#  2H# (C.276)
4. CO ∗ +H#  HCO ∗ +# (C.277)
5. HCO ∗ +H#  H2 CO ∗ +# (C.278)
6. H2 CO ∗ +H#  H3 CO ∗ +# (C.279)
7. H3 CO ∗ +H#  H3 COH ∗ +# (C.280)
8. H3 COH + ∗  H3 COH∗ (C.281)

b) From elementary reaction step (3) and the site balance for τ , we obtain in a similar fashion
as shown in previous exercises the following Langmuir isotherm for H on τ sites.

q
K3 pH2
τH = q (C.282)
1 + K3 pH2

c) The rate-determining step approximation allows us to construct the following rate expres-
sion:

r = k5 θHCO τH . (C.283) A PPENDIX C

By applying a zero-conversion and irreversible step approximation, we only need to derive

Langmuir expressions for H, CO, and HCO on θ sites. Using the pseudo-equilibrium approx-


imation, we obtain the following expressions:

θCO = K1 pCO θ∗ (C.284)

q
θH = K2 pH2 θ∗ (C.285)

(C.286)
p
θHCO = K1 K3 K4 pCO pH2 θ∗
p

(C.287)

Using the site balance θCO + θH + θHCO + θ∗ = 1, gives

1
θ∗ = . (C.288)
K3 K4 pCO √pH2
q p
1 + K1 pCO + K2 pH2 + K1
 C.1 Solutions of Chapter 1 165

Using these expressions, we obtain the following final expression for the rate

  q 
K1 K3 K4 pCO √pH2 K3 pH2
p
r = k5   (C.289)
  
1 + K1 pCO + K2 pH2 + K1 K3 K4 pCO √pH2
q p  q
1 + K3 pH2

d) The reaction order in H2 is given by

1 1 1
nH2 = 1 − θH − θHCO − τH (C.290)
2 2 2

and the reaction order for CO is given by

nCO = 1 − θCO − θHCO (C.291)

Note that in contrast to single-site kinetic networks, in dual-site catalysis the effect of competi-
tive adsorptions is decreased leading to an increase of the lower limit in the reaction order.
This shows that such systems suffer to a lesser extend from poisoning conditions.

e) The apparent activation energy is given by

app (3)
∆Eact = ∆Eact + ∆H1 (1 − θCO − θHCO )
+ ∆H2 (−θH )
 
1 1
+ ∆H3 1 − τH − θHCO
2 2
+ ∆H4 (1 − θHCO ) . (C.292)


A PPENDIX C

 166 Appendix C. Solutions

C.2 Solutions of Chapter 2

C.2.1 Solutions to questions
a) For chemistry, the translational, rotational and vibrational partition functions are the most
relevant. These partition functions differ in the nature of the configurational freedom that
the complex has given a particular energy. A translational degree of freedom indicates that
the system can freely translate in one dimension without the energy of the complex changing.
Similarly, a rotational degree of freedom indicates the complex can rotate without its energy
changing. Finally, a vibrational degree of freedom can be seen as a hindered translation.
The molecule can perturb in a specific direction, but this results in an energy increase. As
such, vibrational degrees of freedom arise from bound states and the corresponding costs of
perturbing from the equilibrium position of these bound states.

b) The partition function is dimensionless and has no unit. This is both evident from the
meaning of the partition function (it represents the average number of possible configurations
given a particular energy, configuration and temperature) as well as from the various equations
wherein the partition function is used (i.e. it is used inside a logarithm).

C.2.2 Solutions to exercises

The solution below pertain to the exercises of Chapter 2 on page 67 and further.

 Solution 2.1
c1

a2 b1 2∆E

∆E ∆E

a1 b0 c0 A PPENDIX C

∆E ∆E ∆E


a0

Figure C.1: Energy level diagram of A, B, and C.

a) The equilibrium constant in terms of the partition functions is given by

q q
qi i = B C
Y ν
K= (C.293)
qA
i
b)
2
!
X j
qA = exp − (C.294)
kb T
j=0
! !
∆E 2∆E
= 1 + 2 exp − + exp − (C.295)
kb T kb T
 C.2 Solutions of Chapter 2 167

! !
∆E 2∆E
qB = exp − + exp − (C.296)
kb T kb T

! !
∆E 3∆E
qC = exp − + 2 exp − (C.297)
kb T kb T

         
exp − k∆E T + exp − 2∆E
k T exp − k∆E T + 2 exp − 3∆E
k T
b b b b
K=     (C.298)
1 + 2 exp − k∆ET + exp − 2∆E
k T
b b

c)

0·0
lim K = =0 (C.299)
T →0 1

2·3 3
lim K = = (C.300)
T →∞ 4 2

 Solution 2.2

a1

∆E
a0
a)
The partition function for molecule A is
A PPENDIX C

1
!

− i
X
qA = exp (C.301)
kb T


i=0
!
∆E
= 1 + exp − (C.302)
kb T

The partition function at T → 0 gives

lim qA = 1 (C.303)
T →0

The partition function at T → ∞ gives

lim qA = 2 (C.304)
T →∞
168 Appendix C. Solutions

a1 b2
∆E
4 b1
∆E
4 b0
∆E
∆E
2
a0
b)
The partition function for molecule B is

2
!
X i
qB = exp − (C.305)
kb T
i=0
! ! !
∆E 3∆E ∆E
= exp − + exp − + exp − (C.306)
2kb T 4kb T kb T

The equilibrium between A and B is given by

1
qiνi
Y
K= (C.307)
i=0
qB
= (C.308)
qA
   
 
∆E
exp − 2k 3∆E + exp − ∆E
+ exp − 4k
bT T k T
=  b  b
(C.309)
1 + exp − k∆ET
b

The equilibrium constant for T → 0 yields

0
lim K = =0 (C.310)
A PPENDIX C

T →0 1

The equilibrium constant for T → ∞ yields



3
lim K = (C.311)
T →∞ 2

 Solution 2.3

Note that the spectroscopically determined energy levels of A and B are an infinite series
of energy levels. The energy levels of A and B have the same energetic separation (∆E ). The
difference though is that the series of molecule B starts somewhat higher (∆E/2)
 C.2 Solutions of Chapter 2 169

∆E

ai
∆E
2
∆E bi
∆E
2
a2 b2
∆E
2
∆E b1
∆E
2
a1 b0

∆E ∆E

a0

Figure C.2: Energy level diagram of the spectroscopically determined energy levels of A and B.
The partition function for A yields

∞
!

− i
X
qA = exp (C.312)
kb T
i=0
∞
!
X ∆E
qA = exp −i (C.313)
kb T
i=0
∞
!i
X ∆E
= exp − (C.314)
kb T
i=0

The partition function for B yields

A PPENDIX C

∞
!
X j
qB = exp − (C.315)


kb T
j=0
∞
!
X j ∆E
= exp −( + 1) (C.316)
2 kb T
j=0
! ∞ !j
∆E X ∆E
= exp − exp − (C.317)
kb T 2kb T
j=0

The above sums are geometric series (see Appendix B.4 on page 131). These series can be
simplified using the formula below

∞
a
ark =
X
for |r| < 1 (C.318)
1−r
k=0

Thus we can write

170 Appendix C. Solutions

1
qA =   (C.319)
1 − exp − k∆E
bT

and

 
exp − k∆E
bT
qB =   (C.320)
∆E
1 − exp − 2k
bT

This gives for the equilibrium constant

   
exp − k∆E 1 − exp − k∆E
q bT bT
K= B =   (C.321)
qA ∆E
1 − exp − 2k T
b

The values for the equilibrium constant K are then:

K(T = 0.1∆E/kb ) = 0.000045 (C.322)

K(T = 2∆E/kb ) = 1.079 (C.323)
K(T = 10∆E/kb ) = 1.766 (C.324)
K(T → ∞) = 2 (C.325)
A PPENDIX C

In other words, at very low temperature, only levels of A are occupied. Hence, at low temper-


ature the equilibrium condition states that nearly all species will be in the A configuration.
With increasing temperature, more and more levels of B are occupied and the equilibrium
shifts to the situation where A and B states have same occupational fraction. Due to the fact
that there are twice as many levels of B (because of its separation which is half the separation
of the energy levels of A); the equilibrium constant yields 2 at infinite temperature.
To solve for the limiting case (T → ∞), it can help to employ l’Hôspital’s rule (see Appendix
B.6 on 133), which states

f (x) f 0 (x)
lim = lim (C.326)
x→a g(x) x→a g 0 (x)

Applying this rule, yields

 C.2 Solutions of Chapter 2 171

     
∆E 1 − exp − k∆E
 exp − kb T bT

K(T → ∞) = lim  (C.327)
 
  
T →∞  ∆E
1 − exp − 2k T 
b

 !
     
 ∆E2 exp − k∆E 1 − exp − k∆E − exp − 2∆E
 kb T bT bT kb T


= lim 
 
  
T →∞  ∆E ∆E
− 2k T 2 exp − 2k T 
b b
 

(C.328)
 
∆E (1 · 0 − 1)
k T2
= lim  b
−∆E
 (C.329)
T →∞
2kb T 2
 
− k∆ET 2
= lim  b
−∆E
 (C.330)
T →∞
2kb T 2
=2 (C.331)

 Solution 2.4

a) The translational partition function in three dimensions using the ideal gas law is:

3/2
2πmkb T
qT = V (C.332)
h3

3/2
k T 2πmkb T
= b (C.333)
P h3
A PPENDIX C

Plugging in the values gives



1.3806488 · 10−23 J · K−1 · 298 K

qT =
1.01325 · 105 Pa
 3/2
2π28 · 10−3 kg/6.0221409 · 1023 · 1.3806488 · 10−23 J · K−1 · 298 K
(C.334)
(6.62607004 · 10−34 m2 · kg · s−1 )3
= 5.8163 · 106 (C.335)

This answer means that an N2 molecule, given these temperature and pressure conditions,
has almost 6 million different translational configurations.

b) qT increases to the power of 5/2 with temperature. At higher temperature, more states
can be populated, hence increasing the partition function. qT also increases with volume, but
only linearly. A higher volume gives more space per molecule, hence more accessible states.
In contrast, qT decreases with increasing pressure as the volume per particle decreases with
pressure.
172 Appendix C. Solutions

c)
1
qV =   (C.336)
1 − exp − khνT
b
1
=   (C.337)
6.62607004·10−34 m2 ·kg·s−1 ·299792458·m·s−1 ·233000 m−1
1 − exp − 1.3806488·10 −23 J·K ·298 K
−1

≈1 (C.338)

Note that the vibrational partition function under typical atmospheric conditions usually
equals unity.

d) The rotational partition function is given by:

8π 2 Ikb T
qR = (C.339)
σh2

Please note that this partition function represents rotation for a diatomic molecule in two
dimensions. In other words, this partition function represents two degrees of freedom!
Plugging in the values yields

8π 2 · 1.407 · 10−46 kg · m2 1.3806488 · 10−23 J · K−1 · 298 K

qR = (C.340)
2(6.62607004 · 10−34 m2 · kg · s−1 )2
= 52.05 (C.341)

e)

Q = qT qV qR (C.342)
= 5.8163 · 106 · 1 · 52.05 (C.343)
= 3.03 · 108 (C.344) A PPENDIX C

Note that the translational partition function represents 3 degrees of freedom, the rotational
partition function 2 degrees of freedom and the vibrational partition function 1 degree of


freedom. Summing up these numbers gives a total of 6 degrees of freedom, exactly what we
would expect from a diatomic molecule!

 Solution 2.5

a) The Maxwell-Boltzmann distribution expresses the partial fraction of a set of species given
their velocity and is given by the following expression

!3/2 !
m −mv 2
f = 4π v 2 exp (C.345)
2πkb T 2kb T
 C.2 Solutions of Chapter 2 173

To calculate the average velocity of an ensemble, you simply have to multiply the Maxwell-
Boltzmann distribution with the velocity v and integrate over all possible velocities

Z ∞ !3/2 !
m −mv 2
v̄ = 4π v 2 exp v dv (C.346)
0 2πkb T 2kb T
!3/2 Z !
m ∞ −mv 2
= 4π v 3 exp dv (C.347)
2πkb T 0 2kb T
!3/2 
1 2kb T 2

m
= 4π (C.348)
2πkb T 2 m
 1/2
8kb T
= (C.349)
πm

Note that in the above formula, the mass m is the mass for a single particle. If you want to use
the molecular mass in kg/mol, you have to substitute the Boltzmann constant kb for the gas
constant R. Furthermore, to solve the improper integral, we have used the standard integral
as defined below. Alternatively, you can use the method of integration by parts.

Z ∞ !
x2 n! 2n+2
x2n+1 exp − 2 dx = a (C.350)
0 a 2

Calculating the average speed for a set of N2 molecules at room temperature gives

 1/2
8RT
v̄ = (C.351)
πM
!1/2
8 · 8.3145 J mol−1 K−1 · 298K
= (C.352)
π 28 · 10−3 kg · mol−1
A PPENDIX C

= 475 m · s−1 (C.353)



Note that you can remember as a rule of thumb that the speed of sound is roughly 70% of the
average velocity of the molecules the medium is made of.

b)
 1/2
8kb T1
πmN2
v̄N2 /v̄He = 1 =  (C.354)
8kb T2 1/2

πmHe

Rearranging yields

mN2
T1 = T2 · = 2090K (C.355)
mHe
174 Appendix C. Solutions

c) In a similar manner as in a, we can find the translational energy by integrating the following
expression

Z ∞ !3/2 !
m 2 −mv 2 1
¯t = 4π v exp mv 2 dv (C.356)
0 2πkb T 2kb T 2
!3/2 Z !
m ∞ −mv 2
= 2mπ v 4 exp dv (C.357)
2πkb T 0 2kb T
!3/2
√ kb T 5/2
 
m
= 2mπ · π · 12 · (C.358)
2πkb T 2m
3
= k T (C.359)
2 b

Note that in the above expression, t is the average translational energy per particle, not
per mole of particles. To get the average energy per mole, we need to substitute kb for R.
Furthermore, to solve the improper integral, we have used the standard integral as defined
below (this one differs from the one proposed above!). Alternatively, you can use the method
of integration by parts.

Z ∞  2n+1
2 2 √ (2n)! a
x2n exp−x /a dx = π (C.360)
0 n! 2

Plugging in the values for 1 mol of N2 at 100, 298 and 1000 K yields

¯t (T = 100 K) = 1.247 kJ · mol−1 (C.361)

−1
¯t (T = 298 K) = 3.71 kJ · mol (C.362)
−1
¯t (T = 1000 K) = 12.47 kJ · mol (C.363)
A PPENDIX C




 Solution 2.6

a) Using the ideal gas law, we can write for the particle density

n P
ρ= = (C.364)
V kb T

Plugging in the numbers yields

P 1 · 105 Pa
ρ= = (C.365)
kb T 1.3806488 · 10−23 J · K−1 · 298.15K
= 2.43 · 1025 particles · m−3 (C.366)
 C.2 Solutions of Chapter 2 175

Because the partial pressure of N2 is 14 of the total pressure and the partial pressure of H2 is
3 of the partial pressure, their respective number densities are
4

1
ρN2 = · 2.43 · 1025 = 6.1 · 1024 particles · m−3 (C.367)
4
3
ρH2 = · 2.43 · 1025 = 1.8 · 1025 particles · m−3 (C.368)
4

b) The collision density is given by equation 2.113 on 57:

 1/2  1/2
πkB T πRT
Z=2 σ 2 n2 = 2 σ 2 n2 . (C.369)
m M

(note that in the above formula, σ is the collision radius)

Plugging in the values yields

Z(H2 , H2 ) (C.370)
!1/2 !2
π · 8.3145 · 298 K 2.71 · 10−10  2
=2 −1
· m · 1.8 · 1025 particles · m−3
2 · 10−3 kg · mol 2
(C.371)
= 2.35 · 1034 collision · m−3 s−1 (C.372)

Note that due to the large difference in the smallest and largest numbers (10−23 and 1025 ),
some calculators have numerical problems (that do not raise an error...) resulting in wrongly
calculated numbers.
A PPENDIX C

Perform this calculation on Wolfram Alpha: https:

//www.wolframalpha.com/input/?i=2+*+(pi+*+8.


3145+*+298+%2F+(2e-3))%5E0.5+*+(2.71e-10+%2F+2)
%5E2+*(1.8e25)%5E2

c) Similar to the above results, we can calculate for the collision between N2 and N2

Z(N2 , N2 ) (C.373)
!1/2 !2
π · 8.3145 · 298 K 3.73 · 10−10  2
=2 −1
· m · 6.1 · 1024 particles · m−3
28 · 10−3 kg · mol 2
(C.374)
= 1.36 · 1033 collision · m−3 s−1 (C.375)
176 Appendix C. Solutions

Perform this calculation on Wolfram Alpha: https:

//www.wolframalpha.com/input/?i=2+*+(pi+*+8.
3145+*+298+%2F+(28e-3))%5E0.5+*+(3.73e-10+%2F+
2)%5E2+*(6.1e24)%5E2

d) We can calculate the number of collisions between N2 and H2 by plugging in the effective
diameter d = (d1 + d2 )/2, we then obtain:

 1/2
8πkB T 2 N N
Z(N2 , H2 ) = σAB A B (C.376)
µ
!1/2 !2
8 · π · 8.3145 · 298 K (2.71 + 3.73) · 10−10
= −1
· m ·
1.86 · 10−3 kg · mol 4
   
1.8 · 1025 particles · m−3 · 6.1 · 1024 particles · m−3 (C.377)
= 1.64 · 1034 collision · m−3 s−1 (C.378)

Perform this calculation on Wolfram Alpha: https:

//www.wolframalpha.com/input/?i=(8+*+pi+*+8.
3145+*+298+%2F+(1.86e-3))%5E0.5+*+((2.71+%2B+3.
73)*1e-10+%2F+4)%5E2+*1.8e25+*+6.1e24

e) The number of collision is just the sum of all possible sets of collisions yielding
A PPENDIX C

Ztotal = 4.13 · 1034 collision · m−3 s−1 (C.379)



 Solution 2.7

a) From the formula, the pre-exponential factor and the activation energy can be easily
extracted

ν = 2 · 1015 (C.380)
−1
∆Eact = 274 kJ · mol (C.381)

b) Draw an reaction energy diagram with three states (initial, transition and final state). Make
the height of the transition state higher than the initial and final states.
 C.2 Solutions of Chapter 2 177

c) A cyclopropane molecule can be thermally excited by collission with another cyclopropane

molecule or another molecule in the same gaseous phase. If the pressure would be signif-
icantly lowered, the reaction rate would be lower because there are fewer collisions.

 Solution 2.8

R#

∆Eact

R
∆ER
P
a) The general reaction equation being used is

k1+
R−
(−
−− R# −−→ P
+ (C.382)
k1− k2

This gives the following rate expression

A PPENDIX C

r = k2 K1 [R] (C.383)

b) The general expression is



k T Q##
!
∆Ea
r= b exp − [R] (C.384)
h Q kb T

Note that in the above expression, Q## denotes the total partition function of the transition
state without the partition function corresponding to the imaginary frequency.

c) Transition State theory takes all degrees of freedom (rotational, translational as well as
vibrational) into account and not just the translational degrees of freedom in collision theory
(rendering the latter theory inconsistent with thermodynamics).


178 Appendix C. Solutions

 Solution 2.9

a) The main assumptions of transition state theory are

• There exists an equilibrium between the initial and the transition state
• The crossing of a species from the transition state to the final state is for the calculation
of the rate constant assumed to be irreversible and corresponds to a particular frequency
• The frequency of transmission or crossing can be modeled as an ultrasoft vibrational
mode (other alternatives are possible, but are mathematically more complicated)

b) The general expression is

k T Q##
!
∆Ea
k= b exp − (C.385)
h Q kb T

Note that in the above expression, Q## denotes the total partition function of the transition
state without the partition function corresponding to the imaginary frequency.

c) The first term in the equation ( kbhT ) is a frequency factor (its dimension is in reciprocal
seconds), corresponding to the number of species crossing the barrier in the transition state
in a given time frame. The second term is the quotient of the total partition functions in the
transition and initial state (please note the comment above regarding the transition state). The
last part relates to the difference in electronic energy between the initial and transition state,
commonly expressed as the activation energy. Loosely speaking, this term is the obtained
result when dividing the electronic partition functions of the transition and initial state.

d) Assuming that all entropy is lost in the transition state, we can formulate the following
identities

k1+
! ! !
∆G ∆H ∆S
K1 = exp − = exp − · exp = (C.386)
A PPENDIX C

kb T kb T kb k1−
(C.387)


The forward and backward rate constants are defined as

forward k T Q## forward

! !
∆Eact ∆Eact
k1+ = νforward exp − = b exp − (C.388)
kb T h Q kb T

and

backward backward
! !
∆Eact k T ∆Eact
k1− = νbackward exp − = b exp − (C.389)
kb T h kb T

Dividing these two yields

 C.2 Solutions of Chapter 2 179

forward
 
kb T Q## ∆Eact
h Q exp −
! !
∆H ∆S kb T
exp − · exp = 
backward
 (C.390)
kb T kb kb T ∆Eact
h exp − kb T

If we assume that all entropy is lost in the transition state, than the difference between forward
and backward activation energies is equal to the reaction enthalpy.

Q##
!
∆S
exp = (C.391)
kb Q

!
k T −130 J/mol
νforward = b exp = 2.701 · 106 s−1 (C.392)
h R

Note that I have replaced the Boltzmann constant for the gas constant, because the energy
was given in a per mole basis. Furthermore, note that the result is quite a low value for a
chemical reaction. Typically, chemical reactions have a pre-exponential factor around 1013 .
The reason is the significant reduction in the number of degrees of freedom upon dissociative
adsorption.

e)
! !
k T −130 J/mol 50 · 103 J/mol
r = b exp exp = 1469s−1 (C.393)
h R R · 800 K

 Solution 2.10

a)
 ##
(2) (3)
qR qT
Q##
! !
A PPENDIX C

k T ∆Ea k T ∆Ea
r= b exp − = b exp − (C.394)
h Q kb T h q (1) q (2) q (3) kb T
V R T


b) To write down the above expression in Arrhenius-form, normally we are allowed to neglect
the vibrational partition functions. Here, such an assumption is not given, thus we have to
evaluate this expression. Luckily, because the number of translational and rotational partition
functions is similar for the transition state as for the initial state, we do not have to evaluate
these.

arrhenius = k T 2 ln ∂r
∆Eact (C.395)
b
∂T
  
∂ ln 1 − exp − khνT
∂ ln kbhT b
= ∆Ea + kb T 2 + kb T 2 (C.396)
∂T ∂T
 
hν exp − khνT
= ∆Ea + kb T +  
b
 (C.397)
1 − exp − khνT
b
180 Appendix C. Solutions

Applying a Taylor approximation to the last term and neglecting all quadratic terms gives

 
hν exp − khνT hν(1 − khνT )
 b
 ≈
b
hν
(C.398)
−

hν
1 − exp − k T 1 1 + kb T
b

This we can further simplify this equation by assuming that khνT  1.

b

hν(1 − khνT ) hν
b
= kb T (1 − ) ≈ kb T (C.399)
1 − 1 + khνT kb T
b

And thus finally obtaining

arrhenius = ∆E + 2k T
∆Eact (C.400)
a b

 Solution 2.11

a) The second derivative of the energy (potential) can be calculated using a numerical deriva-
tive as follows

∂2V Vx=h − 2Vx=0 + Vx=−h

k= = (C.401)
∂x2 h2

The direction in which we pertubate the system is given by

A PPENDIX C

q1 = rBC − rAB (C.402)

q2 = rBC + rAB (C.403)


where q1 is a vector in the direction of the reaction coordinate and q2 is a vector perpendicular
to the direction of the reaction coordinate.
From Figure 2.4, we see that the energy in the direction of the reaction coordinate at Vx=−h =
Vx=h = 11.00 a.u., where h = 0.152 Å. The energy at the transition state is Vx=0 = 11.35
a.u.
Thus,

∂2V
kk = = −49.38 a.u. / Å2 (C.404)
∂x2
= −2125 J / m2 (C.405)

In the direction perpendicular to the reaction coordinate, we find V (x = −h) = V (x = h) =

12.00 a.u., where h = 0.239 Å, thus
 C.2 Solutions of Chapter 2 181

∂2V
k⊥ = = 341.1 a.u. / Å2 (C.406)
∂x2
= 14870 J / m2 (C.407)

b)
r
k
ωk = (C.408)
m
√
= 1.127 · 1015 −1s−1 (C.409)
= 33.59i cm−1 (C.410)

r
k
ω⊥ = (C.411)
m
= 2.982 · 1015 s−1 (C.412)
= 99.47 cm−1 (C.413)

Note that the wavenumber in the direction of the reaction coordinate is very low and imaginary.
In contrast, the wavenumber perpendicular to the direction of the reaction coordinate is real
and has a higher force constant.

c) A transition state on the multi-dimensional potential energy surface is characterized by

one imaginary frequency in the direction of the reaction coordinate and real frequencies in all
other directions. In other words, a transition state is a maximum in energy in one direction
and a minimum in energy in all other directions. Such a point is considered stable as the
forces (the first derivative of the energy) is zero as given by

∂V
A PPENDIX C

= 0 for all i. (C.414)

∂qi


Note though that such a state is meta stable in the sense that a small pertubation in the
direction of the reaction coordinate would propagate this system either towards the initial or
the final state due to the particular shape of the potential energy curve around this point.

 Solution 2.12

a) The vibrational frequency correlates with the strength of a bond. So a stronger bond has a
higher vibrational frequency.

b) The bond between two atoms in the transition state is relatively weak, as such, the frequency
of this bond is weak (or loose) as well.
182 Appendix C. Solutions

c) From statistical thermodynamics, we can express K as the quotient of the molecular

partition function in the transition and initial state:

Z†
K= (C.415)
Z

where Z † is the molecular partition function in the transition state and Z is the molecular
partition function in the initial state. The molecular partition function is the product of
the electronic, vibrational, translational and rotational partition functions. We can extract
the electronic partition function from the above equation and use it to define the electronic
reaction barrier as follows

 ! !
Q†  −† −
K= exp / exp  (C.416)
Q kb T kb T
!
Q† −(† − )
K= exp (C.417)
Q kb T
!
Q† −∆Ea
K= exp (C.418)
Q kb T

where where Q†† is the configurational partition function in the transition state and Q the
configurational partition function in the initial state. The configurational partition function
is the product of the rotational, translational and vibrational partition functions. From the
configurational partition function in the transition state, we are going to extract the partition
function corresponding to the loose vibration, this yields

!
Q† −∆Ea
K= exp (C.419)
Q kb T
!
1 Q†† −∆Ea
K=   exp (C.420)
hω
1 − exp − k T Q kb T
A PPENDIX C

b
Q !
1 i fi −∆Ea
K=   Q exp (C.421)
1 − exp − khωT j fj kb T


d) The vibrational partition function for a loose vibration (i.e. the vibration in the direction
of the reaction coordinate at the transition state) is

† 1
fvib =   (C.422)
1 − exp − khωT
b
1
= (C.423)
1 − 1 + khωT
b
1
=
hω
(C.424)
kb T
k T
= b (C.425)
hω
 C.2 Solutions of Chapter 2 183

Note that in the above expression, i runs overall the configurational degrees of freedom of
the transition state minus the degree of freedom corresponding to the loose vibration in the
direction of the reaction coordinate and j runs over all the degrees of freedom of the initial
state.

e) Plugging the result for the loose vibration into the formula for the rate constant and using
ν = ω gives

k = νK (C.426)
Q !
kb T i fi Ea
= νQ exp − (C.427)
hω j fj kb T
Q !
kb T i fi Ea
= Q exp − (C.428)
h j fj kb T
!
kb T Q†† Ea
= exp − (C.429)
h Q kb T

where Q†† is the configurational partition function in the transition state without the partition
function corresponding to the loose vibration (hence the double †) and Q the configurational
partition function in the initial state. Here, the configurational partition function is the
product of the rotational, translational and vibrational partition functions.

C.2.3 Solution to challenges

C.2.4 Challenge: Maxwell-Boltzmann distribution
The total set of macrostates and the corresponding number of microstates for each macrostate
is schematically represented in Figure C.3. The number of microstates were calculated using
equation 2.162 on page 72. For example, the number of microstates for macrostate 4 is given
by:
A PPENDIX C

5! 120
Ω4 = = = 30. (C.430)


0! · 0! · 1! · 0! · 0! · 2! · 2! 4

We calculated the average number of particles per energy state using equation 2.164 on
page 73. We obtained the following numbers as given in Table C.1.

Table C.1: Average number of particles ni per energy level i.

i ni
0 2.000
1 1.333
2 0.833
3 0.476
4 0.238
5 0.095
6 0.023
184 Appendix C. Solutions

5 20 20 30 10
6 6 6 6 6

5 5 5 5 5

4 4 4 4 4

3 3 3 3 3

2 2 2 2 2

1 1 1 1 1

0 0 0 0 0

60 20 10 30 5
6 6 6 6 6

5 5 5 5 5

4 4 4 4 4

3 3 3 3 3

2 2 2 2 2

1 1 1 1 1

0 0 0 0 0

Figure C.3: Schematic depiction of all macrostates for distributing 6 units of energy over 5 particles. Above each
subfigure is the number of microstates for that particular macrostate given.

The average number of particles in energy state i is shown in Figure C.4. We used a
simple curve fitting procedure to fit the data as shown in Table C.1 to equation 2.165 on page
73. The result of this fit is shown in the same Figure. Note that the distribution we derived
using only five particles and six energy units approximates the exact Maxwell-Boltzmann
distribution to a very large extend. Increasing the number of particles would result in a better
match with respect to the exact solution. Typically, you would consider a number of particles
in the order of 1023 , which would give the exact solution.1
A PPENDIX C

We obtained a value of A = 2.05028583 and c = 0.488230. From this, the temperature is

easily calculated as


∆E
T = · c ≈ 587 K. (C.431)
R

1
Although figuring out all the macrostates and number of microstates would become nearly impossible, but that is
exactly why we employ statistics.
 C.2 Solutions of Chapter 2 185

2.00
1.75
Average population at energy level i

1.50
1.25
1.00
0.75
0.50
0.25
0.00
0 1 2 3 4 5 6
Energy level i
A PPENDIX C

Figure C.4: Average number of particles as a function of the energy level. The dotted line represents the curve fit.

 186 Appendix C. Solutions

C.3 Solutions of Chapter 3

C.3.1 Solutions to questions
1. The difference between a rotational degree of freedom and a hindered rotor is the
extend by which both can freely rotate. If a complex can freely rotate without the energy
of the system increasing or decreasing, then the complex has a rotational degree of
freedom. In contrast, if such a rotation is hindered, then the degree of freedom can
be approximated as a hindered rotor. In fact, if the rotation is severely hindered (for
instance by a harmonic potential), then the degree of freedom can be modeled by a
vibration.

2. The zero point energy can be calculated by

k
hm
EZPE = (C.432)
2kb T

from which it can be seen that the mass m is in the denominator and thus is inversely
proportional to the zero point energy. Thus, atoms with a low mass have a more
pronounced effect on the zero point energy than atoms with a high mass. For this
reason, substitution of a hydrogen atom by a deuterium atom has a very strong effect
in the evaluation of many kinetic properties.

C.3.2 Solutions to exercises

The solution below pertain to the exercises of Chapter 3 on page 81 and further.

 Solution 3.1

a) We assume that this reaction occurs over a surface and that the corresponding partition
functions are vibrational. Furthermore, we assume that the separation of the vibrational
levels is thus high that only the ground state is occupied. Furthermore, we assume that we
can neglect the ZPE correction. In that case, the reaction rate in the forward and backward
direction are
A PPENDIX C

!
k T −∆Eact
kforward = b exp (C.433)


h kb T
 
k T −∆Eact
= b exp (C.434)
h RT
= 4.84 · 104 s−1 (C.435)

and

!
k T −∆Eact
kbackward = b exp (C.436)
h kb T
 
k T −∆Eact
= b exp (C.437)
h RT
= 3.21s−1 . (C.438)
 C.3 Solutions of Chapter 3 187

b) The surface area A of the adsorption site is calculated by

1 2
A= l (C.439)
2
1
= (2.71 Å)2 (C.440)
2
2
= 3.67205 Å (C.441)
= 3.67205 · 10−20 m2 (C.442)

The rate constant for CO desorption is calculated as

!
k T 3 A 2πmkb


−Edes
k = b3 exp (C.443)
h σθrot kb T
= 1.08 · 105 s−1 (C.444)

c) The rate constant for the adsorption of CO is calculated by

PA
kads = p (C.445)
2πmkb T
= 6.16 · 10−3 s−1 (C.446)

d) Despite that the adsorption reaction has no activation energy, for the desorption reaction the
transition state has more entropy than the initial state, whereas for adsorption the transition
state has less entropy than the initial state. As such, at relatively high temperature, the
desorption becomes several orders of magnitude faster than the adsorption rate.


A PPENDIX C

 Solution 3.2


a) The forward rate is given by

k T Q††
 
−∆Eact
kforward = b exp (C.447)
h Q RT
k T (2π)3/2 l p
 
−∆Eact
= b 3kb mH T exp (C.448)
h h RT
= 1.16717 · 109 s−1 . (C.449)

Here, we have assumed that all vibrational partition functions have a value of unity. To avoid
confusion, we have used l, rather than R for the C-H bond length.

b) From equation C.449, we can calculate the apparent activation energy by

188 Appendix C. Solutions

app ∂ ln r
Eact = RT 2 (C.450)
∂T
2 ∂ ln k
= kb T (C.451)
∂T
1
= kb T + kb T + ∆Eact (C.452)
2

Thus, the apparent activation energy is higher by 32 kb T , which is about 6.2 kJ/mol at 500K. In
other words, the correction of the zero point energy correction is relatively small as compared
to the electronic activation energy. Nevertheless, for reactions with small barriers (e.g. for
some hydrogenation reactions), the correction can be significant.

A PPENDIX C

 C.4 Solutions of Chapter 4 189

C.4 Solutions of Chapter 4

C.4.1 Solutions to questions
1. A power law describing a reaction has the following form r = k i ci νi , where k is
Q

the rate constant of the reaction, ci are the concentrations of compounds i and νi are
the stoichiometric coefficients of the reaction. If the reaction is an elementary reaction
step, then k represents the rate by which the reactants are converted to the product state
when they meet. The product i then represents the chance that these reactants meet.
Q

Logically, that chance is correlated to the concentration of the reactants. Furthermore, if

for the reaction two or more of the same reactants are required, then this concentration
is raised to a power equal to the stoichiometric coefficient.

2. A valid set of elementary reaction steps is

CO + ∗  CO∗ (C.453)
CO ∗ +∗  C ∗ +O∗ (C.454)
C ∗ +H∗  CH ∗ +∗ (C.455)
CH ∗ +H∗  CH2 ∗ +∗ (C.456)
CH2 ∗ +H∗  CH3 ∗ +∗ (C.457)
CH3 ∗ +H∗  CH4 + 2∗ (C.458)
O ∗ +H∗  OH ∗ +∗ (C.459)
OH ∗ +H∗  H2 O ∗ +∗ (C.460)
H2 O + ∗  H2 O∗, (C.461)

though more sets can be valid. For instance, the CO dissociation can be hydrogen-
assisted (where first CO is hydrogenated prior to C-O bond scission). Alternatively,
water can be formed by migration of a hydrogen atom between two hydroxyl species.

3. Hydrogenation of CH3 to methane forms a stable molecule (one that fulfills the octet
rule for C) and consequently such a stable molecule is no longer covalently interacts
with the surface. In fact, such a complex has only a weak interaction with the surface
A PPENDIX C

due to Van der Waals interactions. In contrast, ammonia has a lone pair which can still
interact with the surface to form a metal-nitrogen bond.

4. The reaction order or apparent activation energy are obtained by calculating the first


derivative of the logarithm of the rate towards the pressure or the temperature, re-
spectively. For an overall reaction wherein there is only one type of product, the rate
of consumption of any of the reactants can be expressed by the rate of the product
multiplied by some constant that can be derived from the stoichiometric coefficients.
For example, in the reaction A + 2B + 3C → D, the rate of consumption and production
of all the compounds are related by the following expression

rD = −rA = −2rB = −3rC (C.462)

Thus, we can express the rate of consumption of any of the rate constant by calculating
the quotient of the corresponding stoichiometric coefficient. As this constant has no
dependency on either the pressure or on the temperature, it drops out of the logarithm
as follows
 190 Appendix C. Solutions

∂ ln r ∂ ln (cr)
= (C.463)
∂x ∂x
∂ ln (r) ∂ ln (r)
= + (C.464)
∂x ∂x
∂ ln (r)
= , (C.465)
∂x

where x denotes either the partial pressure for the reaction order or the temperature
for the apparent activation energy.

5. The surface coverage of O is determined by adsorption energies of CO, O and CO2 .

Because the adsorption energy of CO is so high, the surface is predominantly covered
with CO by which there are no free sites available for O or CO2 to adsorb.

C.4.2 Solutions to exercises

The solution below pertain to the exercises of Chapter 4 on page 106 and further.

 Solution 4.1

a) As evident by the assumption used in the derivation, the DRC coefficient voor CO dissocia-
tion is unity, whereas all other DRC coefficients are zero.

b)

∂ ln r
nO2 = pO2 (C.466)
∂pO2
 
q
 krds KCO pCO KO2 pO2 
∂ ln 
 q 2 

1+KCO pCO + KO2 pO2
= p O2 (C.467)
∂pCO
A PPENDIX C

1
= D1 + D2 − 2D3 , (C.468)
2


where

∂ ln krds KCO pCO

D 1 = p O2 (C.469)
∂pCO
q 
∂ ln KO2 pO2
D 2 = p O2 (C.470)
∂pO2
 q 
∂ ln 1 + KCO pCO + KO2 pO2
D 3 = p O2 . (C.471)
∂pO2

The terms Di can be readily solved, which give

 C.4 Solutions of Chapter 4 191

D1 = 0 (C.472)
1
D2 = (C.473)
2
q
1 K O2 p O2
D3 = q (C.474)
2 1 + KCO pCO + KO pO
2 2

and combining these three terms results in

1
nCO = − θO (C.475)
2

 Solution 4.2

To proof that the sum-rule for the degree of selectivity holds, we need to convert the sum
of degree of selectivity coefficients to a sum of degree of rate control coefficients.

X X ∂ηc
i,c = (C.476)
∂ ln ki
i,c i,c
X ∂ ln ηc
= ηc (C.477)
∂ ln ki
i,c
X ∂ ln rc /rr
= ηc (C.478)
∂ ln ki
i,c
!
X ∂ ln rc ∂ ln rr
= ηc − (C.479)
∂ ln ki ∂ ln ki
i,c
X
= ηc (χc − χr ) (C.480)
A PPENDIX C

i,c
X X 
= ηc χc,i − χr,i (C.481)
c i


 
X X X
= ηc  χc,i − χr,i  (C.482)
c i i
X
= ηc (1 − 1) (C.483)
c
=0 (C.484)

From a conceptual point of view, the above can be rationalized. Due to conservation of
mass, you can only generate more of one particular product in expense of another product.
So if one particular elementary reaction steps favors the product of a compound, there are
other elementary reaction steps that favor another product by the same magnitude.


 192 Appendix C. Solutions

C.5 Solutions of Chapter 5

C.5.1 Solutions to Exercises
 Solution 5.1

The input file for this simulation can be found using the link below:
 https://www.mkmcxx.nl/downloads/input/input_ex1.mkm

a) The optimal temperature for the production of CO2 can be found, for instance, by investigat-
ing derivative.dat or derivative.png. The results are visualized using matplotlib
in Figure C.5a. From this Figure, it can be seen that the optimal temperature is around
T = 580 K.

b) The rate determining step is evaluated using a degree of rate control (DRC) analysis. This
analysis is visualized in Figure C.5b. From this Figure, it can be seen that the rate determining
step is the surface reaction: CO ∗ +O∗  CO2 ∗ +∗.

0.08 CO2 1.0 CO + * CO*

Maximum is at 569.3 Kelvin O2 + 2* 2O*
0.07 CO* + O* CO2* + *
0.8 CO2 + * CO2*
Degree of rate control coefficient

0.06
0.05 0.6
1
Rate s

0.04
0.4
0.03
0.02 0.2
0.01
0.0
0.00
600 800 1000 1200 1400 600 800 1000 1200 1400
Temperature in K Temperate T in K

(a) Rate of production of CO2 (b) Degree of rate control for CO oxidation


A PPENDIX C

 Solution 5.2


The input file for this simulation can be found using the link below:
 https://www.mkmcxx.nl/downloads/input/input_ex2.mkm

a) The optimal coverage is found by comparing the derivative graph with the coverage graph.
The optimal coverage for the reaction is when θB = 0.5 corresponding with an optimal
temperature around 600 K.

b) The reaction.log file contains the equations of the rate constants and their dependence
on the temperature. This results in

4.16595 · 109
+
kA = √ = 1.5212 · 108 (s−1 ) (C.485)
T
 C.5 Solutions of Chapter 5 193

1.0

0.8

Degree of rate control coefficient

0.6
A*+B* C*+*
C*+D* E*+*
0.4

0.2

0.0
600 800 1000 1200 1400
Temperate T in K

Figure C.6: Degree of rate control

 Solution 5.3

The input file for this simulation can be found using the link below:
 https://www.mkmcxx.nl/downloads/input/input_ex3.mkm

a) The set of elementary reaction steps is as follows:

A2 + 2∗  2A∗
B + ∗  B∗
A ∗ +B∗  C ∗ +∗
A PPENDIX C

A∗  D∗
C ∗ +D∗  E ∗ +∗
E∗  E + ∗


Note that compound A2 adsorbs dissociatively onto the catalyst surface. If component A2
would not have adsorbed dissociatively, an extra surface reaction for the dissociation of A2
would have been added to Table 5.7.

b) From Tables ?? and ??, the following can be observed. (1) The forward reaction of A ∗+B∗ 
C ∗ +∗ has a high barrier whereas the backward reaction has a low barrier. (2) The adsorption
rates of compounds A2 and B are comparable. (3) The backward reaction of A∗  D∗ has a
high activation barrier, which could lead to surface poisoning by compound D∗.
From the above, we conclude that elementary reaction step A ∗ +B∗  C ∗ +∗ will, most likely,
be significantly rate-controlling.

c) Figure C.6 shows that for the lower temperature regime the rate controlling step is A ∗
+B∗  C ∗+∗, however for higher temperatures the elementary reaction step C ∗+D∗  E ∗+∗
becomes rate controlling.
194 Appendix C. Solutions

 Solution 5.4

The input file for this simulation can be found using the link below:
 https://www.mkmcxx.nl/downloads/input/input_ex4.mkm

a) If elementary reaction step A∗  2B∗ is rate-controlling and we assume zero conversion,

the site balance becomes

θA + θ∗ = 1 (C.486)

with θA being

θA = KA pA θ∗ (C.487)

Note that for calculating the equilibrium constant of A, the Hertz-Knudsen equations are used.
This results in the following overall reaction rate

+
kB K A pA
rRDS = = 9.44 · 105 (mol/s) (C.488)
1 + K A pA

This rate is consistent with the result as obtained using MKMCXX. This confirms that the
analytic expression is valid as long as the underlying assumptions are valid.

b) In coverage.png it can be seen that the most abundant reaction intermediate is D∗.

c) A negative DRC coefficient implies that the elementary reaction step is rate-inhibiting and
A PPENDIX C

therefore that increasing the reaction barrier will increase the rate of the overall reaction. The
forward activation barrier for B∗  D∗ is low compared to the backward activation barrier.
Furthermore the forward activation barrier for 2D∗  F ∗ +∗ is also relatively high. As such,


for lower temperatures component D∗ poisons the catalyst surface and slows down the overall
reaction rate.

d) There are several options to increase the selectivity towards component F. Some are:

(a) Decreasing the forward activation barrier for 2D∗  F ∗ +∗.

(b) Increasing the forward activation barrier for B∗  D∗.
(c) Decreasing the backward activation barrier for B∗  D∗.
(d) Increasing the forward activation barrier for B∗  C∗.
(e) Increasing the forward activation barrier for C∗  E∗.


 D

APPENDIX
EXAM PRACTICE QUESTIONS

D.1 Explanation
The set of questions introduced here can be used as practice material for the examination
of the course 6A5X0. The difficulty of these questions is expressed by the number of stars,
where 3 stars (***) is considered average and 5 stars (*****) is the highest difficulty level.
Solutions to these questions are provided in the next chapter on page 201.

D.2 Kinetics
 Exam Practice Question D.1 Kinetics 1
A novel route for the selective oxidation of methane towards methanol is to use sulfur-
trioxide (SO3 ) as an oxidant. In this reaction, adsorbed SO3 is in equilibrium with adsorbed
oxygen and SO2 in the gas phase.

SO3 ∗  SO2 + O∗ (D.1)

Methane adsorbs molecularly after which it can react with adsorbed oxygen to form
methanol on the catalytic surface. This reaction is considered to be the rate-determining step.
Adsorbed methanol is in equilibrium with methanol in the gas phase.

a) Provide all elementary reaction steps for the overall reaction CH4 + SO3 → CH3 OH + SO2

b) Derive an expression for the surface coverage of methane, methanol, sulfur-trioxide and
A PPENDIX D

oxygen as a function of the relevant equilibrium constants and partial pressures.

c) Derive an expression for the rate of formation of methanol as function of the partial


pressures of methane, methanol, sulfur-trioxide, sulfur-dioxide and oxygen.

For the next subquestions, assume that at low temperature methanol adsorbs significantly
stronger on the catalytic surface than any of the other surface intermediates.

d) Describe how the surface looks given the above assumption. Which species is the MARI
(most abundant reaction intermediate)?

e) Derive an expression for the rate of formation of methanol. Deduce the reaction orders in
methanol and methane.
196 Appendix D. Exam Practice Questions

f ) Derive an expression for the apparent activation energy and connect the terms to the
physical process. In other words, rationalize your obtained expression.

 Exam Practice Question D.2 Kinetics 2

Ethylene (CH2 CH2 ) can be oxidized to epoxide (CH2 CH2 O) using N2 O. This process
proceeds over an Fe catalyst. In this process, ethylene adsorbs molecularly whereas N2 O
adsorbs dissociatively according to the following reaction equation

N2 O + ∗  N2 + O∗ (D.2)

Two elementary reaction steps occur on the catalytic surface, which is the selective oxida-
tion of ethylene towards epoxide

C2 H4 ∗ +O∗ → C2 H4 O ∗ +∗ (D.3)

and the recombination of two surface oxygen atoms to molecular oxygen in the gas phase

2O∗ → O2 + 2∗ (D.4)

Assume that these processes occur on a catalytic surface with only one type of surface
sites. Further assume that adsorbed ethylene is in equilibrium with gas phase ethylene and
adsorbed epoxide is in equilibrium with gas phase epoxide. Finally, assume that both reactions
D.3 and D.4 are rate-determining for the formation of epoxide and oxygen, respectively.

a) Write down all elementary reaction steps for the above mechanism.

b) Derive an expression for the rate of formation of gas phase epoxide.

c) Derive an expression for the rate of formation of gas phase molecular oxygen.

d) Deduce the lower and upper limit for the reaction orders in oxygen, nitrogen, nitrous
oxide, ethylene and epoxide for both rate expressions.
A PPENDIX D

Assume that for the next subquestions, ethylene and epoxide adsorb very weak as compared
to dissociative adsorption of nitrous oxide.


e) Derive an expression for the rate of formation of epoxide and molecular oxygen and relate
the reaction orders in oxygen, nitrogen, nitrous oxide, ethylene and epoxide to the surface
coverages.

f ) Derive an expression for the apparent activation energy for the formation of epoxide and
molecular oxygen.


 D.3 Statistical Thermodynamics 197

D.3 Statistical Thermodynamics

 Exam Practice Question D.3 Statistical Thermodynamics 1

Consider the following reaction which is at equilibrium:

A+BC+D (D.5)

• Molecule A has three energetic states. The ground state of A lies ∆E higher than the
ground state of B and the excited states are separated by 2∆E .

• Molecule B also has three energetic states. The two excited states are degenerate (i.e.
they have the same energy) and lie 3∆E above the ground state.

• Molecule C only has a single energetic state which lies equal in energy as compared to
the ground state of A.

• Finally, Molecule D has three energetic states. Its ground state equals that of the ground
state of C in energy. Its two excited states lie 4∆E above the ground state.

a) Draw a schematic representation of the distribution of the energetic states over the four
molecules. In this drawing, the separation in energy between the states within the same
molecule as well as the separation in energy between the ground states of the different
molecules should be clearly conveyed.

b) Construct the partition function for each of the molecules A, B, C, D.

c) Derive the equilibrium constant K as function of the previously defined partition functions.

d) Calculate the limits of the equilibrium constant for T → 0 and T → ∞. Provide a physical
interpretation of your results (i.e. rationalize the obtained values).

 Exam Practice Question D.4 Statistical Thermodynamics 2

Consider the following reaction which is at equilibrium:
A PPENDIX D

A+BC+D (D.6)

• Molecule A has five energetic states. The ground state of A lies ∆E higher than the


ground state of B and the excited states are separated by 2∆E .

• Molecule B also has three energetic states. The two excited states are degenerate (i.e.
they have the same energy) and lie 3∆E above the ground state.

• Molecule C only has a single energetic state which lies equal in energy as compared to
the ground state of A.

• Finally, Molecule D has three energetic states. Its ground state equals that of the ground
state of C in energy. Its two excited states lie 4∆E above the ground state.
 198 Appendix D. Exam Practice Questions

a) Draw a schematic representation of the distribution of the energetic states over the four
molecules. In this drawing, the separation in energy between the states within the same
molecule as well as the separation in energy between the ground states of the different
molecules should be clearly conveyed.

b) Construct the partition function for each of the molecules A, B, C, D.

c) Calculate the average energy of molecule B at T → 0 and T → ∞. Rationalize this results

on the basis of the partial occupation of the states of A at these two temperature extremes.

d) Derive the equilibrium constant K as function of the previously defined partition functions.

e) Calculate the limits of the equilibrium constant for T → 0 and T → ∞. Provide a physical
interpretation of your results (i.e. rationalize the obtained values).

D.4 Collision theory

 Exam Practice Question D.5 Collision theory 1
Consider the following reaction

H2 + Br2 → 2HBr (D.7)

which occurs at T = 450K . Consider this reaction to be an elementary reaction step. The
rate equation is hence given by

r = kreaction [H2 ][Br2 ] (D.8)

Collision theory can be utilized to calculate the collision frequency. The frequency is
given by

 1/2
8kb T
kcollisions = πd2 (D.9)
πµ

The masses of atomic hydrogen and bromine are 1.008 Da and 79.904 Da, respectively.
A Dalton equals 1.66054 · 10−27 kg. The effective collision diameter of H2 and Br2 are 1.5 Å
and 2.8 Å, respectively. The Boltzmann constant is given by kb = 1.38064852 · 10−23 m2 kg
A PPENDIX D

s−2 K−1 .

a) Calculate the value for the collision frequency between H2 and Br2 using collision theory


given the above conditions.

b) What are the dimensions of kcollision ?

c) Explain in the context of collision theory why the number of collisions that results in a
reaction is very low and why this number increases rapidly with temperature.

d) How does the number of effective collisions (i.e. that lead to a reaction) scale with respect
to temperature? Choose between constant, linear, quadratic, exponential and logarithmic
scaling and rationalize, on the basis of statistical thermodynamics, your choice.
 D.5 Transition state theory 199

e) Provide an expression for the rate constant kreaction and calculate its value at T = 450K .
Assume that the activation energy is ∆Ea = 200 kJ/mol.

f ) Besides collision theory, there also exists transition state theory. Why do these two theories
not give the same equilibrium constant when applying these to the same reaction? Consider
the nature of the degrees of freedom involved in both theories.


D.5 Transition state theory

 Exam Practice Question D.6 Transition State Theory 1
Ozone (O3 ) can adsorb on a metal surface. Upon adsorption, the molecule can neither
translate nor rotate on the surface. Assume that all vibrational partition functions equal
unity, with exception of the one corresponding to the reaction coordinate. We consider two
situations:
1. O3 dissociates to O and O2 , where the O2 immediately desorbs from the surface:
O3 ∗ → O2 + O∗

2. O3 dissociates to adsorbed O and adsorbed O2 :

O3 ∗ → O2 ∗ +O∗

a) Provide a schematic depiction of the reaction for situation (1) (i.e. make a drawing of the
reacting fragments on the catalytic surface). Clearly indicate the reaction coordinate. The O2
fragment cannot translate in the transition state, but it is able to rotate. Derive an expression
for the reaction rate of this dissociation reaction within the framework of transition state
theory. Pay careful attention to the number and nature of the involved partition functions.

b) Use the expression obtained in the previous subquestion and compare it to the Arrhenius
expression. Assuming that both expressions give the same rate, what would be the activation
energy and the pre-exponential factor of the Arrhenius equation if the previously obtained
equation is cast to the Arrhenius formulation?

c) Provide a schematic depiction of the reaction for situation (2). In the transition state, O2
can neither rotate, nor translate. Derive an expression for the reaction rate of this dissociation
reaction within the framework of transition state theory. Pay careful attention to the number
and nature of the involved partition functions.

d) What is in situation (2) the activation energy and pre-exponential factor if the rate expres-
sion of the previous subquestion is cast to Arrhenius form?
A PPENDIX D

e) In which of the two situations will the reaction rate for dissociation be the largest if we
assume that the activation energy for situation (1) and (2) is identical? Rationalize your result


using the concept of entropy and relate this to the partition functions.


 Exam Practice Question D.7 Transition State Theory 2

In this question, you are going to derive the Eyring expression from basic principles. The
Eyring expression describes the rate of change between two stable states (i.e. initial and final
state) via a transition state that connects these two states.
200 Appendix D. Exam Practice Questions

a) What are the fundamental assumptions of the Eyring equation and use these assumptions
to construct the following equation:

k = κK, (D.10)

where k is the rate constant, κ is a prefactor with dimensionality s−1 and K is an equilibrium
constant (dimensionless). Explain which equilibrium the equilibrium constant K represents.

b) Express the equilibrium constant as a function of the partition functions of the involved
states. For the time being, do not make any assumptions about the number and nature of the
degrees of freedom.

c) We assume that the motion over the transition state can be modeled as a very weak vibration.
Show (i.e. provide a clear and detailed derivation) that using this assumption we are able to
derive the following expression for the reaction rate constant:

† !
1 Q −∆Eact
k=κ   TS exp , (D.11)
1 − exp −hν
k T
QIS kb T
b

where Q is the product of the partition functions of one or more degrees of freedom, the
subscripts IS and TS refer to initial and final state, respectively, and ν is the frequency of the
vibration (in s−1 ). Please note the dagger superscript (†) in Q†TS !

d) Provide a supporting motivation for the assumption that the vibration over the transition
state is weak. Use in your explanation the concept that a vibrational degree of freedom
corresponds to a chemical bond.
 
e) When  −hν
kb T   1, the expression for the vibrational partition function can be simplified.
 

Use the Taylor-expansion of the exponential function:

x2 x3 x4
exp x = 1 + x + + + ··· (D.12)
2! 3! 4!

to show that the vibrational partition function can be simplified to

k T
qv = b (D.13)
hν
A PPENDIX D

by excluding all terms in the Taylor expression beyond the linear term (i.e. discard all quadratic
and higher order terms).

f ) Finally, which two variables need to be equal to obtain the general Eyring equation as given


by

† !
k T QTS −∆Eact
k= b exp (D.14)
h QIS kb T


 E

APPENDIX
EXAM QUESTION SOLUTIONS

E.1 Introduction

Below, the solutions to the exam practice questions are provided. These solutions are written
in a concise fashion, sometimes omitting trivial steps in the derivation. If you are unable to
follow some or more of the derivations, it is recommended that you look back at one of the
earlier exercises of the corresponding chapter. More elaborate solutions are provided there
which might help you in understanding.

E.2 Kinetics

 Exam Solution E.1 Kinetics 1

a) The set of elementary reaction steps is given below. Herein, equation E.4 is the rate-
determining step.

1. SO3 ∗  SO2 + O ∗ (E.1)

2. CH4 + ∗  CH4 ∗ (E.2)
3. SO3 + ∗  SO3 ∗ (E.3)
A PPENDIX E

4. CH4 ∗ +O∗  CH3 OH ∗ + ∗ (E.4)

5. CH3 OH + ∗  CH3 OH∗ (E.5)


b) Assume a quasi-equilibrium for all the elementary reaction steps before the rate-determining
step. From these, the following Langmuir adsorption isotherms can be derived.
202 Appendix E. Exam Question Solutions

K2 pCH4
θCH4 =
K1 K3 pSO3
(E.6)
1 + K2 pCH4 + K3 pSO3 + K5 pCH3 OH + pSO2
K3 pSO3
θSO3 =
K1 K3 pSO3
(E.7)
1 + K2 pCH4 + K3 pSO3 + K5 pCH3 OH + pSO2
K5 pCH3 OH
θCH3 OH =
K1 K3 pSO3
(E.8)
1 + K2 pCH4 + K3 pSO3 + K5 pCH3 OH + pSO2
K1 K3 pSO3
pSO2
θO =
K1 K3 pSO3
(E.9)
1 + K2 pCH4 + K3 pSO3 + K5 pCH3 OH + pSO2
(E.10)

c) The rate of formation for methanol can be directly found by plugging in the relevant
Langmuir isotherms in the rate expression for elementary reaction step (4)

K1 K3 pSO3
k4 K2 pCH4 pSO2
rCH3 OH = k4 θCH4 θO =  2 (E.11)
K1 K3 pSO3
1 + K2 pCH4 + K3 pSO3 + K5 pCH3 OH + pSO2

d) Adsorbed methanol is the MARI, hence the surface mainly contains methanol and vacant
sites.

e) Applying the MARI enables us to neglect those coverage terms in the denominator which do
not correspond to the MARI. Next, the differential for the reaction is solved for the simplified
rate expression to obtain the reaction order in methanol and methane.

K1 K3 pSO
3
k4 K2 pCH4 pSO2
rCH3 OH = k4 θCH4 θO =  2 (E.12)
1 + K5 pCH3 OH

∂ ln r
nCH3 OH = pCH3 OH = −2θCH3 OH (E.13)
∂pCH3 OH

∂ ln r
nCH4 = pCH4 =1 (E.14)
∂pCH4

f ) The apparent activation energy is given by

A PPENDIX E


app ∂ ln r (4)
∆Eact = RT 2 = ∆Eact + ∆H1 + ∆H2 + ∆H3 − 2θCH3 OH ∆H5 (E.15)
∂T
The apparent activation energy is the amount of energy which needs to invested for the
overall reaction to occur. This energy is equal to the activation energy of the rate-determining
 E.2 Kinetics 203

step and is reduced by the adsorption energy of particular components when the surface
is vacant but increased when the surface is poisoned. The latter can be interpreted in the
sense that some energy needs to be invested to remove a component from the surface. In
this particular case, the reaction can be inhibited by methanol as is seen from the (negative)
dependence on the methanol surface coverage.

 Exam Solution E.2 Kinetics 2

a)

1. N2 O + ∗  N2 + O∗ (E.16)
2. C2 H4 + ∗  C2 H4 ∗ (E.17)
3. C2 H4 ∗ +O∗  C2 H4 O∗ (E.18)
4. O2 + 2∗  2O∗ (E.19)
5. C2 H4 O + ∗  C2 H4 O∗ (E.20)

b)

rC2 H4 O =k3 θC2 H4 θO (E.21)

pN O
K1 p 2 K2 pC2 H4
N2
=k3  2 (E.22)
pN O
1 + K1 p 2 + K2 pC2 H4 + K5 pC2 H4 O
N2

(E.23)

c)
2
rO2 =k4 θO (E.24)
 2
pN O
K1 p 2
N2
=k4  2 (E.25)
pN O
1 + K1 p 2 + K2 pC2 H4 + K5 pC2 H4 O
N2

d)

reaction orders O2 N2 N2 O C2 H 4 C2 H 4 O
rC2 H4 O 0 -1,1 -1,1 -1,1 -2,0
r O2 0 -2,0 0,2 -2,0 -2,0
e)
pN O
K1 p 2 K2 pC2 H4
N2
(E.26)
A PPENDIX E

rC2 H4 O =k3  2
pN O
1 + K1 p 2
N2

2



pN O
K1 p 2
N2
rO2 =k4  2 (E.27)
pN O
1 + K1 p 2
N2
 204 Appendix E. Exam Question Solutions

We obtain the the following reaction order for the epoxide

nO2 =0 (E.28)
nN2 = − 1 + 2θO (E.29)
nN2 O =1 − 2θO (E.30)
nC2 H4 =1 (E.31)
nC2 H4 O =0 (E.32)

and for molecular oxygen

nO2 =0 (E.33)
nN2 = − 2 + 2θO (E.34)
nN2 O =2 − 2θO (E.35)
nC2 H4 =0 (E.36)
nC2 H4 O =0 (E.37)

f ) The apparent activation energy for epoxide formation is

app
∆Eact = ∆E3 + ∆H1 (1 − 2θO ) + ∆H2 (E.38)

and for the formation of molecular oxygen

app
∆Eact = ∆E4 + ∆H1 (2 − 2θO ) (E.39)

E.3 Statistical Thermodynamics

 Exam Solution E.3 Statistical Thermodynamics 1

a)
a2 d1,2

2∆E

a1 b1,2 4∆E
A PPENDIX E

2∆E
3∆E
a0 c0 d0


b0

Figure E.1: Energy level diagram of A, B, C, and D.

 E.3 Statistical Thermodynamics 205

b) Set the energy level of b0 to the global ground state (i.e. zero):

! ! !
−∆E −3∆E −5∆E
qA = exp + exp + exp (E.40)
kb T kb T kb T
!
−3∆E
qB = 1 + 2 exp (E.41)
kb T
!
−∆E
qC = exp (E.42)
kb T
! !
−∆E −5∆E
qD = exp + 2 exp (E.43)
kb T kb T

c) The equilibrium constant K for the above reaction is

q q
K= C D (E.44)
qA qB

d) When T → 0, all exponentials go to zero, whereas at T → ∞, all exponentials go to unity.

Thus:

0·0
K(T → 0) = lim =0 (E.45)
1·0
1·3 1
K(T → ∞) = lim = (E.46)
3·3 3

At T → 0, only the lowest state is occupied, which is b0 . Hence, the equilibrium constant is
0. At T → ∞, all states are equally likely to be occupied. Hence, our equilibrium constant
should resemble the ratio of the products of the number of states of each component (which
is basically the definition of the partition function at infinitely high temperature).


A PPENDIX E


 Exam Solution E.4 Statistical Thermodynamics 2

a)
206 Appendix E. Exam Question Solutions

a4

2∆E

a3

2∆E

a2 d1,2

2∆E

a1 b1,2 4∆E

2∆E
3∆E
a0 c0 d0

b0

Figure E.2: Energy level diagram of A, B, C, and D.

b) Set the energy level of b0 to the global ground state (i.e. zero):

! ! !
−∆E −3∆E −5∆E
qA = exp + exp + exp ···
kb T kb T kb T
! !
−7∆E −9∆E
· · · + exp + exp (E.47)
kb T kb T
!
−3∆E
qB =1 + 2 exp (E.48)
kb T
!
−∆E
qC = exp (E.49)
kb T
! !
−∆E −5∆E
qD = exp + 2 exp (E.50)
kb T kb T

c) The equilibrium constant K for the above reaction is

q q
K= C D (E.51)
qA qB

d) The average energy of molecule B is given by

A PPENDIX E

∂ ln qB
B =kb T 2 (E.52)
∂T
 
2 · 3∆E · exp −3∆E


kb T
=   (E.53)
1 + 2 exp −3∆E
k T b

At T → 0, A = 0 as only the ground state is occupied which has an energy value of 0.

 E.4 Collision theory 207

At T → ∞, A = 2∆E , as all its states are equally occupied. Hence, we get the arithmetic
average of the energy values of all these states.

e) When T → 0, all exponentials go to zero, whereas at T → ∞, all exponentials go to unity.

Thus:

0·0
K(T → 0) = lim =0 (E.54)
1·0
1·3 1
K(T → ∞) = lim = (E.55)
5·3 5

At T → 0, only the lowest state is occupied, which is b0 . Hence, the equilibrium constant is
0. At T → ∞, all states are equally likely to be occupied. Hence, our equilibrium constant
should resemble the ratio of the products of the number of states of each component (which
is basically the definition of the partition function at infinitely high temperature).

E.4 Collision theory

 Exam Solution E.5 Collision theory 1

a)
!2
1.5 · 10−10 + 2.8 · 10−10
d2 = = 4.6 · 10−20 m2 (E.56)
2
1.008 · 2 · 79.904 · 2
µ= · 1.66 · 10−27 = 3.30 · 10−27 kg (E.57)
1.008 · 2 + 79.904 · 2
 1/2
8kb · 450
k =π · 4.6 · 10−20 = 3.16 · 10−16 collisions · m3 · s−1
π · 3.30 · 10−27
(E.58)

b) The unit is collisions · m3 · s−1 .

c) Not every collision immediately results in a reaction. Only those collisions where the
particles have sufficient kinetic energy, i.e. equal or larger than the activation energy, will result
in a reaction. The number of particles that meets this criterion increases with temperature.

d) The average kinetic energy of the particles in the ensemble increases exponentially with
temperature, as can be seen from the Maxwell-Boltzmann velocity distribution and the related
average kinetic energy expression.

e)
A PPENDIX E

!
−200 · 103
 
−∆Ea
kreaction =kcollision · exp = 3.16 · 10−16 · exp (E.59)
RT R · 450


=6.1880452 · 10−40 m3 · s−1 (E.60)

Note that this is the reaction rate constant on a per-particle basis. On a per molecule basis,
the number is much larger (by about 23 orders of magnitude).
 208 Appendix E. Exam Question Solutions

f ) In collision theory, only translational degrees of freedom are taken into account, whereas in
transition state theory, also rotational and vibrational degrees of freedom are considered. (Note
that despite this discrepancy in collision theory, since translational degrees of freedom give
the largest partition functions as compared to rotational and vibrational degrees of freedom,
collision theory remains fairly accurate.)

E.5 Transition state theory

 Exam Solution E.6 Transition State Theory 1

a)

O O O O O O

O O O

Figure E.3: Schematic depiction of the initial, transition and final state of O3 dissociation over a catalytic
surface. Note that this is a side view, where the viewing direction is parallel to the catalytic surface. The reaction
coordinate is shown as a dashed arrow in the image for the transition state.

(7) (1)
! !
k T qv qr −∆Eact k T −∆Eact
k= b exp = b qr exp (E.61)
h (9) kb T h kb T
qv

In the transition state, the complex has in total 9 degrees of freedom. One degree of freedom
(DOF) corresponds to the imaginary frequency and is in the direction of the reaction coordinate.
This DOF is represented by kbhT . Another DOF is the rotational DOF and all other DOFs are
vibrational. In the initial state, all DOFs are of a vibrational nature.

b) The Arrhenius-form activation energy is given by

arrhenius = RT 2 ∂ ln k
∆Eact (E.62)
∂T

Plugging the above rate expression into this formula yields for the activation energy:

arrhenius = ∆E
∆Eact (E.63)
act + 2kb T

If the two forms (Arrhenius and Eyring) are equal, this yields the following expression for the
A PPENDIX E

pre-exponential factor:


k T
νarrhenius = b · e2 · qr (E.64)
h

c)
 E.5 Transition state theory 209

O O O O O O

O O O

Figure E.4: Schematic depiction of the initial, transition and final state of O3 dissociation over a catalytic surface.
Note that this is a top view, where the viewing direction is perpendicular to the catalytic surface. The reaction
coordinate is shown as a dashed arrow in the image for the transition state.

(8)
! !
k T qv −∆Eact k T −∆Eact
k= b exp = b exp (E.65)
h q (9) kb T h kb T
v

In the transition state, the complex has in total 9 degrees of freedom. One degree of freedom
(DOF) corresponds to the imaginary frequency and is in the direction of the reaction coordinate.
This DOF is represented by kbhT . All other DOFs are vibrational. In the initial state, all DOFs
are of a vibrational nature.

d) The Arrhenius-form activation energy is:

arrhenius = ∆E
∆Eact (E.66)
act + kb T

And the corresponding pre-exponential factor becomes:

k T
νarrhenius = b · e (E.67)
h

e) The important difference between situation (1) and (2) is that in situation the O2 fragment
in the transition state has a rotational degree of freedom. Rotational degrees of freedom have
more configurational freedom as compared to vibrational degrees of freedom. Hence, the
transition state of situation (1) is higher in entropy and thus lower in Gibbs free energy (recall
that ∆G = ∆H − T ∆S ) as compared to situation (2). We can thus conclude that the reaction
of situation (1) will proceed faster than situation (2).

 Exam Solution E.7 Transition State Theory 2

a) We assume that the initial state is in thermal equilibrium with the transition state. Hence,
we introduce an equilibrium constant K . To determine the rate constant to go from the initial
to the final state, we introduce a crossing frequency κ that represents the number of species
that goes to the final state once that species is at the transition state. This requires us to define
another assumption, which states that once a species has crossed the transition state, it will
always go to the final state. (it can of course go back to the initial state, but that is captured in
A PPENDIX E

the rate expression for the backward reaction)

b) The equilibrium constant is given by



QTS
K= (E.68)
QIS
210 Appendix E. Exam Question Solutions

c) The total partition function Q is the product of the molecular partition functions repre-
senting vibrational, rotational and translations degrees of freedom and the electronic partition
function. If we extract the weak vibrational partition function and the electronic partition
function, we can rewrite the above equation to

†
QTS qTS,e
K = qv,weak · · , (E.69)
QIS qIS,e

wherein the † indicates that we have extracted the partition function corresponding to the
reaction coordinate and we have redefined Q to only consist of molecular degrees of freedom
(i.e. without the electronic partition function). The quotient of the electronic partition
functions can be rewritten using the Boltzmann formula and we can plug in the formula for
the vibrational partition function to obtain

† !
1 Q −∆Eact
k=κ   TS exp (E.70)
−hν
1 − exp k T QIS kb T
b

d) Upon bond-breaking or formation, there exist a transition state wherein the bond is
elongated with respect to the most stable state (i.e. the bonded state). In the transition state,
it is thus expected that the bond is weaker. A weaker bond is represented by a shallower
potential, hence elongation or shortening of the bond will not result in a significant change in
the energy. Hence, the force constant representing the vibrational degree of freedom will be
relatively small, thus rationalizing the assumption to model the transition as a weak vibration.

e)
1
qv =   (E.71)
1 − exp −hν
k T
b
1
≈   (E.72)
1 − 1 + −hν
k T
b

k T
≈ b (E.73)
hν

f ) To obtain the general expression for the Eyring equation, the identify ν = κ needs to be
true in order to cancel out the κ in the equation.


A PPENDIX E

 F

APPENDIX
MKMCXX

F.1 Settings
The settings section of the input file offers the option to change a lot of default values. In
Table F.1, the most relevant settings are given. Most settings are to turn certain functions on
or off. Use 0 to turn a feature off and 1 to turn a feature on.

F.2 Troubleshooting
If the input file contains an error, MKMCXX will return one of the following error messages.

• Terminal output: ’The system cannot find the path specified.’

Make sure the MKMCXX.exe file is present in the bin folder. Also check if the run.bat
is in the right sub folder.

• Terminal output: ’Error, could not allocate compound.’

MKMCXX can’t find all compounds specified in the reactions section. Check if all
compounds are listed in the compound section and that there are no mistakes in the
reaction section.

• Terminal output: ’No conservation of sites for run X: (Variables)’

Check the elementary reaction steps in the reaction section. Make sure every reaction
step is balanced.

• Terminal output: ’Caught signal seg 11. You broke MKMCXX!!!’

Check the input data in the run section for mistakes. Also make sure the tolerances are
reasonable, e.g. 1e − 10 and not 1e10.

• Terminal output: ’ERROR: RUN (line X): Temperature terminate called after throwing
an instance of ’std::runtime_error.’
Make sure that the temperature on the given line is correct en is a real number.
A PPENDIX F

The input file can also be incomplete or contain other problems, as listed below.

212 Appendix F. MKMCXX

Table F.1: Settings for MKMCXX

Keyword Value Explanation

TYPE - Type run used for the simulation. Default = sequence
run.
PRESSURE # bar Sets the overall pressure for the system.
ORDERS 0 or 1 Calculate reaction order. Default = 0.
EACT 0 or 1 Calculate apparent activation energy. Default = 0.
DRC 0 or 1 Calculate degree of rate control. Default = 0.
REACTANTS Compound Specify reactant for reaction order.
KEYCOMPONENT Compound Specify components for Eapp and DRC.
USETIMESTAMP 0 or 1 Should MKMCXX add a timestamps or not. Default =
1.
MAKEPLOTS 0 or 1 Should MKMCXX make preliminary plots. Default =
1.
GRAPHDATA 0 or 1 If MKMCXX does not need to create plots, MKMCXX
can still create the data for the plots. Default = 0
GRAPHFILTER 0 or 1 If MKMCXX should remove small value from prelimi-
nary plots. Default = 1
NETWORK 0 or 1 Should MKMCXX add network graphs. Default = 1.
NETWORK_RATES 0 or 1 Should MKMCXX incorporate forward and backward
rates. Default = 0.
COLORBLIND 0 or 1 MKMCXX can use improved colors to counter color-
blindness. Default = 0.
DEBUG 0 or 1 Generate more convergence data. Default = 0.
NPAR 1-8 Amount threads of the processor MKMCXX is aloud to
use. Default = 1.
SOLSTOPTIME 0 or 1 The solver is forced to evaluate dydt/jac at t out or al-
lowed to extrapolate backwards. Default = 1.
SOLMAXSTEP 1-∞ Amount internal steps solver may take before tout is
reached. Default = 5000.
SOLTESTFAIL 1-∞ Maximum amount of error test failures per step are
aloud. Default = 70.
SOLCONVFAIL 1-∞ Maximum amount of convergence test failure per step
are aloud. Default = 100.
PRECISION 0-∞ Amount of decimals in the output files. Default = 10

• The simulation runs, but the output shows no production.

Make sure all elementary reaction steps are listed in the reactions sections. Also check
if the chosen reaction network is complete, in other words check if you can fully go
from reactant to product by following your network.

• The simulation runs, but for some temperature one or multiple compounds have not
converged.
Convergence problems can be solved using the following steps.

– First of all, increase the simulation time for the specific run. It is possible that the
system has not achieved steady state yet, and this can be achieved by increasing
the simulation time up until t = 1e12.
– If the simulation time does not solve the problem, try to tighten the tolerances.
A PPENDIX F

Tightening the absolute tolerance up until AbsT ol = 1e − 45 can solve the


 F.2 Troubleshooting 213

convergence problem. The relevant tolerances can be tightened up until RelT ol =

1e − 16.
– If the specific temperature is not important for the simulation, try removing it.
– If the specific temperature is important for the simulation, try adjusting the
temperature with a few degrees. This may fix the problem if you happen to be at
a bifurcation point.

• The reaction order is not calculated by MKMCXX.

Check the reactants specified by the reactants in the settings, make sure all the reac-
tants are specified. Also make sure the key components, often products, are correctly
specified.

• The apparent activation energy is not calculated by MKMCXX, or not correctly.

If the apparent activation energy is not calculated correctly by MKMCXX, check the
specified compound by the keycompounds in the settings section.

• The degree of rate control is not calculated by MKMCXX.

The sum of the degree of rate control should always be one. It is possible to have a
negative degree of rate control. If the sum is not 1, make sure the right elementary
reaction steps are taken into account when calculating the degree of rate control. Also
make sure the right compounds are specified by the keycompounds.
The program MKMCXX offers a debug mode which offers messages in the terminal.
The debug mode produces more data on the relative derivatives (dydt/y) to check for the
convergence. The debug mode is enabled by

Listing F.1: Enabling debug mode.

1 &settings
2
3 # Enable debug mode.
4 DEBUG = 1
A PPENDIX F

Bibliography

[1] E. Kreyszig. Advanced Engineering Mathematics. Wiley, New York, 8th edition edition,
1999.

[2] J.E. Mayer and M.G. Mayer. Statistical Mechanics. Wiley, New York, 1940.

[3] T.L. Hill. An introduction to Statistical Thermodynamics. Dover Publications Inc., New
York, 1986.

[4] B.J. McClelland. Statistical Thermodynamics. Chapman and Hall Science Paperbacks,
London, 1973.

[5] R.H. Fowler and E.A. Guggenheim. Statistical Thermodynamics. The Macmillan Company,
New York City, 1940.

[6] P. Atkins, J. de Paula, and J. Keeler. Physical Chemistry. Oxford University Press, New
York, 11th edition edition, 2018.

[7] K.J. Laidler. Theories of Chemical Reaction Rates. McGraw-Hill Book Company, 1969.

[8] Eyring H. J. Chem.. Phys., 3:107–115, 1935.

[9] I. Shavitt, R.M. Stevens, F.L. Minn, and Karplus M. J. Chem.. Phys., 48(6):2700–2713,
1968.

[10] S. Glasstone, K.J Laidler, and H. Eyring. The Theory of Rate Processes. McGraw-Hill Book
Company, Inc., New York and London, 1940.

[11] A. Fernández-Ramos, Ellingson B.A., Meana-Pañeda R., J.M.C. Marques, and D.G. Thruh-
lar. Theor. Chem. Account, 118:813–826, 2007.

[12] A.P.J. Jansen. An Introduction to Kinetic Monte Carlo Simulations of Surface Reactions.
Springer, Heidelberg, New York, Dordrecht and London, 2012.
Index

adsorption, 76 trajectory, 40
apparent activation energy, 24 transition state theory, 59
translational partition function
collision theory, 55 derivation, 50
competitive adsorption, 12 formula, 51

Degree of Rate control, 99 vibrational partition function

degree of selectivity control, 101 derivation, 51
desorption, 78 formula, 53

electronic energy, 50 Zero-conversion approximation, 21

elementary reaction step, 3 Zustandssumme, 45
energy
harmonic oscillator, 51
kinetic, 50
nuclear, 50
translational, 50
ergodicity, 41

Hilbert space, 40

Irreversible step approximation, 21

Langmuir adsorption isotherm, 12

MARI, see Most Abundant Reaction Inter-

mediate
Matrix diagonalization, 133
Maxwell-Boltzmann distribution, 41
formula, 44
mean-field approximation, 20
Most Abundant Reaction Intermediate, 22

partition function, 45
electronic, 50
nuclear, 50
translational formula, 51
vibrational formula, 53
phase space, 40, 89

rate expression, 3
reaction order, 22

Sabatiers principle, 90
Schrödinger equation, 40
steady state assumption, 8
stoichiometric coefficient, 2
INDEX 217