Transient Kinetic Studies On The Dehydrogenation PDF

Transient Kinetic Studies on the Dehydrogenation
of Ethylbenzene to Styrene over Potassium-

Promoted Iron Oxide
DISSERTATION
Submitted to the
Department of Chemistry
Ruhr-Universitt Bochum
for the degree of
Dr. rer. nat.
by
Xiaomeng Zhu
Shanghai
Ruhr-Universitt Bochum
Lehrstuhl fr Technische Chemie
2002
Die vorliegende Arbeit wurde in der Zeit vom April 1999 bis Oktober 2002
am Lehrstuhl fr Technische Chemie der Ruhr-Universitt Bochum unter der
Leitung von Prof. Dr. M. Muhler angefertigt.
Vorsitzender der Prfungskommission: Prof. Dr. Ch. Wll
Referent: Prof. Dr. M. Muhler
Korreferent: Prof. Dr. W. Grnert
Dritter Prfer: Prof. Dr. W.S. Sheldrick
Tag der mndlichen Prfung: 12. 12. 2002

Acknowledgements
First, I would like to express my deep gratitude to Prof. Dr. M. Muhler for giving me the
opportunity to work in his laboratory on the interesting theme and for providing me with
continuos support and advice in the scientific work and preparation of this manuscript.
Without his encouragement and help, I would not have succeeded in finishing this work.
I would like to thank Prof. Dr. W. Grnert not only for being the co-referee of my work but
also for the valuable advice and helpfulness during my stay in the Lehrstuhl.
My thanks also goes to Prof. Dr. W. S. Sheldrick for being the third examiner of my work.
Further, I would like to give my thanks to Dr. U. Bartmann for the fruitful scientific
discussion and providing the Opti-program for the parameter optimization, Dipl. M.
Schn for providing the method of preparation catalysts and Dr. A. C. van Veen for the
useful discussion and installing the MS analyzer. I would also thank Dr.-Ing. K.O.
Hinrichsen, Dr. H. Volker, Dipl. W. Hagen and Dipl. J. Amann for the useful advice and
helpfulness during my work.
My thanks goes further to Dipl.-Ing. Otto and H. Pfeiffer who provided always readily the
technical assistance during building and refitting the experimental equipments, and to
Mrs. A. Gomann for the XRD measurements and Mrs. S. Wiedemeyer for the BET
measurements.
Particularly, I would like to thank Prof. J.Y. Shen, Dr. Y. Wang, Dr. X.H. Mu and Dr. Y.X.
Wang for providing me the insightful comments and valuable advice, and for making my
life enjoyable in this country.
I am also very grateful to Mrs. S. Kalender who has helped me so much from the very
beginning until now to make my stay here much easier.
Furthermore, I want to give my great thanks to my family and friends. Without their
sincere encouragement and continuos support, I could not succeed.
Finally, I am deeply grateful to Friedrich-Ebert-Stiftung for providing the four-year
scholarship, and Mrs. M. Holona for giving the best of care to me over these years.
Abstract
The influence of the different experimental operating variables on the temperature-programmed
reduction ( TPR ) patterns has been studied. A modified model including the influence of H2O is
proposed. High resolution of TPR patterns could be achieved by decreasing the amount of
reducible sample. Suitable TPR experimental conditions are derived.
The transient kinetic experiments were carried out over the commercial and self-prepared
catalysts in a microreactor flow system equipped with MS and GC. In the initial periods of the
fresh and regenerated catalysts, much higher initial activities of the catalysts at the very
beginning were observed. The reason is the coupling of the dehydrogenation of ethylbenzene
with the reduction of Fe2O3 leading to a much higher equilibrium conversion. With the reduction of
Fe2O3 to Fe3O4, the activities of the catalysts decrease simultaneously, and reach finally steady
state.
A microreactor flow system equipped with Binos ( for H2 on-line measurement ), Magnos ( for O2 )
and Uras ( for CO and CO2 ) is built up for the TPR ( dry and wet series ) and TPO investigations.
In combination with the results of the transient kinetic experiments, it is found that the steam
during the reaction stabilizes the iron oxide against total reduction to Fe, and reacts with the coke
deposit forming CO, and CO reacts further with steam yielding CO2. By steaming of the used
catalyst, the catalyst is partially reoxidized to Fe3+-containing oxide and the activity of the catalyst
is recovered. The high K-promoted catalysts have higher activities than that of the lower K-
promoted and unpromoted catalysts. It is considered that KFeO2 is the active phase, and CO2
formed during the reaction causes the decomposition of KFeO2. Quantitative analysis shows that
the decomposition of KFeO2 reaches an equilibrium when the CO2 concentration during the
reaction is constant. Ex situ and in situ investigations have shown that K2CO3 exists not only in
the fresh catalyst but also in the catalyst during the reaction. Based on the above investigations, a
modified schematic life cycle of the K-promoted styrene catalyst is proposed.
Furthermore, the simulation of this reaction was carried out over the commercial and 10 % K-
promoted catalysts, and a model with consecutive reaction from styrene to toluene is proposed.
The pre-coefficients and apparent activation energies are obtained through the data fitting in the
range of T: 823 898 K; LHSV: 0.43 3 h-1; and S / O : 6 10. The model gives good agreement
with the experimental data.
Content V
Contents
1 Preface 1
2 Kinetic Study on the Temperature-Programmed Reduction ( TPR ) 2
Abstract 3
2.1 Introduction . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.2 Theory . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .7
2.3 Experimental . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.4 Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . 10
2.4.1 Dilution of the sample with quartz . . . . . . . . . . . . . . . . . . . . . . . . . . . . .10
2.4.2 Effect of sample weight . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .14
2.4.3 Heating rate . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .16
2.4.4 Total flow rate of reducing gas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .19
2.4.5 Influence of H2O in the reducing gas . . . . . . . . . . . . . . . . . . . . . . . .. . . .20
2.4.6 Activation energy. . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . 23
2.5 Conclusions. . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .28
References. . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
3 Transient Kinetic Studies on the Commercial Catalyst 32
Abstract 32
3.1 Introduction. . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .32
3.2 Experimental. . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3.3 Results and discussion. . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .38

Content VI
3.3.1 The blank experiments. . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . .. . . . . 38
3.3.2 The initial activation period of the fresh catalyst. . . . . . . . . .. . . . . . . . . . 40
3.3.3 The regeneration of the catalyst with steam. . . .. . . . . . . . . . . .. . . . . . . 43
3.3.4 The transient behavior of the catalyst after regeneration. . . . .. . . .. . . .45
3.3.5 The investigation of deactivation phenomena. . . . . . . . .. . . . . . . . . . . . .48
3.4 Conclusions. . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
References. . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
4 Initial Transient Behavior of the Iron Oxide Catalysts 53
Abstract 53
4.1 Introduction. . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .53
4.2 Experimental . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
4.2.1 Catalyst preparation. . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .55
4.2.2 Characterization techniques. . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . .55
4.2.3 Transient experiments. . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . 57
4.3 Experimental Results . . . . .. . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .58
4.3.1 Concentration step-up on the fresh catalysts. . . . . . . . . .. . . . . . . . . . .. . 58
4.3.2 Steaming of the used catalysts. . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . 61
4.3.3 Concentration step-up after the steaming. . . . . . . . . .. . . . . . . . . . . . . . . . 64
4.3.4 X-ray diffraction . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
4.3.5 Temperature-programmed Reduction. . . . . . . . . .. . . . . . . . . . . . . . . .. . .70
4.3.6 Temperature-programmed Oxidation . . . . . . . . .. . . . . . . . . . . . . . . . . . .76
4.4 Discussion. . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
4.4.1 Reduction of Fe2O3. . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . 81

Content VII
4.4.2 Effect of the steaming. . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . .. . . . . . .85
4.4.3 K2CO3 in the K-promoted catalysts. . . . . . . . . .. . . . . . . . . . . . . . . . . . . .87
4.4.4 Decomposition of KFeO2. . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . .91
4.4.5 Schematic life cycle of the catalyst. . . . . . . . . .. . . . . . . . . . . . . . . . .. . . . 93
4.5 Conclusions. . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
References. . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
5 Kinetic Parameter Optimization and Simulation Process 99
Abstract 99
5.1 Introduction. . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
5.2 Experimental. . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
5.3 Results and discussion. . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
5.3.1 Kinetic data . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
5.3.2 Kinetic Model . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .106
5.3.3 Kinetic data fitting. . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .112
5.4 Conclusions. . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .120
Nomenclature . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .120
References. . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .121
6 Summary 123
Appendix 125
1. Chemical reagents and gases. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..125
2. Experimental apparatuses. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
2.1 The set-up for the studies of TPR and TPO . . . . . . . . . . . . . . . . . . . . . . . . . .126
2.1.1 The full view of the set-up . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .126

Content VIII
2.1.2 The reactor for the TPR and TPO . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
2.1.3 The isothermal period of the oven . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
2.1.4 The delay of the concentration of H2 . . . . . . . . . . . . . . . . . . . . . . . . . 129
2.1.5 The consumption of H2 for the study of TPR . . . . . . . . . . . . . . . . . . .131
2.1.6 The consumption of O2 for the study of TPO. . . . . . . . . . . . . . . . . . . 132
2.2 The set-up for the study of dehydrogenation of ethylbenzene. . . . . . . . . . . .133
2.2.1 The full view of the set-up . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
2.2.2 The reactor for the dehydrogenation of ethylbenzene to styrene. . . . 136
2.2.3 The determination of LHSV . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
2.2.4 The saturators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
3. Calibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
3.1 The mass flow controller ( MFC ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
3.1.1 The MFCs for the study of TPR and TPO . . . . . . . . . . . . . . . . . . . . . 142
3.1.2 The MFCs for the study of the dehydrogenation of ethylbenzene. . . .143
3.2 The calibrations of Binos, Magnos and Uras . . . . . . . . . . . . . . . . . . . . . . . . 145
3.3 Gas chromatography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
3.4 Mass spectrometer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149

Chapter 1: Preface 1
1 Preface
The dehydrogenation of ethylbenzene over iron oxide catalyst is a main industrial process
for the production of the important monomer styrene [ 1, 2 ]. Although, it has been
intensively studied by many authors with the aid of numerous methods, there are still
some questions left, such as, what causes fast decrease of the initial activity of the
catalyst; does steam play other roles besides the gasfication of the coke deposited on the
catalyst? How can the CO2 in the gas phase lead to the decrease of the activity? As to
the simulation of this process, should the consecutive side-reaction be considered in the
reaction net? etc.
It is well known that the transient response method proved to be useful in obtaining
quantitative information concerning the initial nature of the reaction. Until now, this
method has not been reported to be used in the study for the dehydrogenation of
ethylbenzene to styrene. In this work, the transient response method in combination with
TPR ( Temperature-Programmed Reduction ), XRD ( X-ray diffraction ), and TPO
( Temperature-Programmed Oxidation ) were used to study the reaction and catalysts
aiming to solve the questions mentioned above. This dissertation is organized as follows.
In chapter 2, the influence of the different experimental operating variable on the TPR
patterns has been studied. Specially, the dilution of sample and the wet TPR are
investigated in detail for the high resolution of the TPR pattern. The transient kinetic
method is used first in chapter 3 for the study of the commercial catalyst. The
experimental equipment and method are described in detail, and the concentration of
educt ethylbenzene step up, step down and step up after steaming of the catalyst are
Chapter 1: Preface 2
carried out. The hysteresis study of the catalyst is also performed in this chapter. In
chapter 4, with the aid of transient kinetic method, TPR, XRD and TPO, the investigations
over self-prepared catalysts ( unpromoted, low K-promoted, high K-promoted, etc. ) are
carried out. The results of the transient kinetic experiments are in good combination with
those of phase investigation ( TPR, XRD, TPO ) to achieve a comprehensive
understanding of the initial behavior of the catalysts. A suitable reaction model is
developed in Chapter 5. The kinetic parameters of the pre-coefficients and activation
energies are optimized by the data fitting of the steady experimental results. A method of
random search is used for the parameter optimization. Finally, a summary follows in
chapter 6.
References
[ 1 ] G. Ertl, H. Knzinger and J. Weitkamp, Handbook of Heterogeneous Catalysis,
Wiley-VCH, Vol. 5, 2151-2158 ( 1997 )
[ 2 ] D.H. James and W. M. Castor, Ullmanns Encyclopedia of Industrial Chemistry,
Wiley-VCH, Vol. A25, 329-339 ( 1994 )

Chapter 2: Kinetic Study on TPR 3
2 Kinetic Study on the Temperature-Programmed Reduction
Abstract: The influence of the different experimental operating variables ( the dilution of sample,
amount of reducible sample, heating rate, total flow rate, H2O concentration of reducing gas ) on
the temperature-programmed reduction ( TPR ) patterns has been studied using the reduction of
CuO and Fe2O3 as examples. A modified model including the influence of H2O is proposed.
Different methods for estimating the kinetic parameters from TPR experiments are compared.
High resolution of TPR patterns can be achieved by decreasing the amount of sample. Then the
activation energies for the two step reduction of CuO via Cu2O are determined from the shift of
Tmax ( the temperature at the TPR peak maximum ) as a function of the heating rate with the same
amount of sample. Suitable TPR experimental conditions are derived.
Keywords: TPR, kinetics, copper oxide, iron oxide.
2.1 Introduction
TPR is a useful technique for characterizing reducible materials. The reduction is
measured by monitoring H2 consumption or H2 concentration while linearly increasing the
temperature of the sample. In this way, one or more reduction peaks occur at different
temperatures and the reduction pattern characterizing the material can be obtained. The
TPR technique has been applied widely to study the influence of support material [ 1, 2 ],
pretreatments [ 3,4 ] and promoters [ 5 ] on the catalysts. It is also used together with
other techniques, e.g. XRD ( X-ray diffraction ), TPO ( temperature-programmed
oxidation ) and Mssbauer Spectroscopy for the characterization of various catalysts [ 6
9 ]. Hurst et al. [ 10 ] have given a comprehensive review of both the theory and
applications of the technique. It is known [ 11,12 ] that the number of peaks and the peak
temperature of a TPR profile are influenced by the heating rate , the flow rate of the
reducing gas, the hydrogen concentration, sample mass, grain size and the presence of
water. These reported conditions differ widely, and for this reason, the TPR patterns
obtained by different investigators are merely comparative and the comparison can be
made normally only at a qualitative level. Some criteria were derived to be used for
determination of suitable TPR experimental conditions. Bosch et al. [ 13 ] proposed an
equation which relates the maximum allowable sample weight and the amount of oxygen
removed during a particular reduction step:
T1/2 M MO n
G MO n < H 2 (1)
O/M
where G MO n is the total sample weight, H 2 the flow rate of H2, T1/2 the peak width at
half height, O/M the molar ratio of oxygen removed and metal present, and M MO n is the
molar weight of MOn.
Monti and Baiker [ 14 ] considered that the heating rate and the hydrogen concentration
in the reducing gas had a more pronounced effect on the temperature of the maximum
reduction rate than the sample weight and total flow rate. For an appropriate selection of
operating variables, a characteristic number K was proposed:
S0
K= (2)
V * C0
where S0 is the amount of sample ( mol ), V* the total flow rate of reducing gas ( ml ( NTP
) s-1 ), and C0 the initial H2 concentration ( mol ml-1 ). For commonly used heating rates
between 0.1 and 0.3 K s-1, the limiting values are 55 < K < 140 s.
With regard to the kinetic study of TPR, Gentry et al. [ 15 ] have given a method for
calculating the apparent activation energy of the reduction. Assuming zero-order behavior
of hydrogen and that there is plug flow, they got the following equation:
2
ln ( Tmax / ) = E / R Tmax + constant (3)
where Tmax is the temperature of the peak maximum ( K ), the heating rate ( K s-1 ) and
E the apparent activation energy ( kJ mol-1 ). E can be determined by using Arrhnius
plots with the Tmax at different heating rate . However, some authors believed that the
apparent activation energy E is not constant during the TPR process, and a modified rate
equation was proposed, considering the compensation effect [ 16 ] and an activation
energy distribution [ 17 ].
With the obtained activation energy, a procedure for the calculation of TPR patterns is
described by Wimmers and Moulijn [ 18 ], which shows that the TPR peak shape is
different for the various reduction mechanisms. Therefore, it is possible to determine the
reduction mechanism by comparing measured TPR pattern with the calculated ones
based on different reduction mechanisms. For example, the reduction of small Fe3O4
particles under the dry condition was shown to proceed with the three-dimensional
nucleation according to Avrami-Erofeev ( E = 111 kJ mol-1 ), which represented an
uniform internal reduction. In addition, they found that in the presence of 3% H2O, the
reduction of small Fe3O4 particles was retarded significantly and could not be described
by regular solid-state kinetics.
Another method of kinetic analysis based on curve fitting was developed by Fierro et al. [
19, 20 ]. The mathematical model contained four kinetic parameters: the two reaction
orders ( m and q ), the pre-exponential factor ( k0 ), and the activation energy ( E ). A

preliminary estimation of these parameters was carried out by minimizing the objective
function ( Q ) for each TPR pattern. This function is expressed as the summation of the
square deviations between the experimental and calculated hydrogen concentrations.
Considering that the two reaction orders varied between 0.89 1.12 and the pre-
exponential factor did not change very much, in order to reduce the number of
parameters to be determined and to improve the reliability of the estimated activation
energies, they re-analyzed the experimental TPR profiles by constraining m = 1, q = 1
and k0 = 6 105 cm3 mol-1 s-1. For the two series of CuO-ZnO catalysts, the confidence
limits associated with the estimates of the activation energies were within 2 kJ mol-1 and
Tmax within 3 degrees.
The method based on curve fitting was also used by Munteanu and Andreeva [ 21, 22 ]
for the kinetic study of TPR of Fe2O3. Three stages were assumed for the reduction of
Fe2O3:
Fe2 O3
k1
Fe3 O4
k2
FeO
k3
Fe (4)
In their simulations, the pre-exponential factors k02 and k03 were restricted to be equal
and E2 was assumed smaller than E3. The kinetic parameters characterizing the
reduction process were evaluated by curve fitting for each TPR pattern. The result
showed that the activation energy for the reduction of Fe2O3 to Fe3O4 ( 139.2 kJ mol-1 ) is
significantly higher than those of the two subsequent reduction steps: Fe3O4 to FeO (
77.3 kJ mol-1 ) and FeO to Fe ( 85.7 kJ mol-1 ). Shinokawabe et al. [ 23 ] have also
reported a high value of the activation energy for the reduction of hematite to metallic iron
for a Fe2O3 sample prepared from Fe(OH)(CH3COO)2.
It should be mentioned that the mechanism figured out by comparing the measured TPR
pattern with the calculated ones based on various reduction mechanisms is not reliable,
since the measured TPR pattern is significantly affected by the TPR conditions. As to the
another method, it should be kept in mind that some assumptions were made to
increase the reliability of the simulation, but the adaptability of such assumptions has not
been proved in detail.
We think that for the kinetic study of TPR, it may be reasonable to carry out the work in
two parts. First, the activation energies are determined using the Arrhenius plot with Tmax
at different heating rates, this process is independent of the reaction mechanism.
Second, the pre-exponential factors are obtained by the curve fitting of different TPR
patterns. In this paper, a modified model including the role of H2O for the simulation of
the TPR pattern is derived. The parameters affecting the TPR pattern, such as the mass
of sample, heating rate, flow rate of the reducing gas and the presence of water are
studied in detail. The suitable TPR experimental conditions for the high resolution of TPR
patterns are suggested. Under these conditions, the activation energies for the two
reduction peaks in a TPR pattern are obtained.
2.2 Theory
Generally the reduction of metal oxide can be described by the equation
p S1 + H2 q S2 + H2O (5)
where S1 is the metal oxide, S2 the metal oxide or metal, p and q the stoichiometric
coefficients of reaction. Normally, a plug-flow reactor ( PFR ) is operated at steady state
so that properties at any position are constant with respect to time [ 24 ]. Hence, the
reduction of metal oxide cannot be regarded as a reaction taking place in a steady-state

PFR, since the reaction rate is changing. However, under the assumption that the
concentrations and temperature vary along the length of the reactor, not in the radical
direction, the equation which describes the differential mass-balance for hydrogen is
given by Eq. ( 6 ) and the reaction describing a finite mass balance for the metal oxide is
written by Eq. ( 7 ):
V * C H 2 ( z, t ) C H 2 ( z, t )
+ s r ( CH2 , T ) + =0 (6)
A z t
p s ( z, T )
r ( CH2 , T ) + = 0 (7)
M S1 t
where r is the reaction rate of hydrogen, V* the volumetric flow-rate, A the area of the
normal reactor section, C H 2 the hydrogen concentration and s the amount of sample
per unit volume of reactor. The reaction rate related to the concentrations and
temperature can be written with the power-law model equation:
1
r ( CH2 , T ) = k ( T ) ( CH2 C H 2O ) (8)
Kp
Where C H 2O is the concentration of H2O, Kp the equilibrium constant, which can be
obtained from the handbook of the thermochemical properties of inorganic substances
[ 25 ]. k(T) is a temperature-dependent rate constant, it can be expressed by the
Arrhenius equation :
Ea
k ( T ) = k 0 exp ( ) (9)
RT
The derivative s / t can be given as a function of the heating rate, according to the
following equation:
s ( z, T ) d T s ( z, T ) s ( z, T )
= = ( 10 )
t dt T T
When quasi-steady state conditions are assumed for the gas phase, and the Eqs. ( 8 ) -
( 10 ) are substituted in Eqs.( 6 ) - ( 7 ):
V * C H 2 ( z, t ) Ea 1
+ k 0 exp ( ) ( C H 2 ( z, T ) C H 2O ( z , T )) s ( z , T ) = 0 ( 11 )
A z RT Kp
Ea 1 q s ( z, T )
k 0 exp ( ) ( C H 2 ( z, T ) C H 2O ( z , T )) s ( z , T ) + = 0 ( 12 )
RT Kp M s1 T
The initial and boundary conditions associated with these equations are the following:
C H 2 ( 0, T ) = C H0 2 T T0 ( 13 )
s ( z, T0 ) = s0 0 z L ( 14 )
Where C H0 2 is the hydrogen concentration at the reactor inlet, s0 the initial amount of
metal oxide S1, T0 the initial temperature, and L the reactor length. The solution of Eqs.
( 11 ) ( 12 ) yields the two functions as follows:
C H 2 = C H 2 ( z, T ) ( 15 )
s = s ( z, T ) ( 16 )
The calculated H2 concentration C H 2 ( L, T ) with respect to reduction temperature T at z
= L can be used to evaluate the theoretical TPR pattern.
2.3 Experimental
TPR experiments were carried out in a microreactor flow system, and the flow scheme of
the experimental set up is shown in Fig.2.1. The concentration of hydrogen before and
Chapter 2: Kinetic Study of TPR 10
Rosemount ). As reducing gas 5.2 % H2 in Argon was used, and the gas mixture was led
through the reactor to a cold trap and dryer filled with silica to eliminate the water forming
during the reduction. The flow rate was 77 ml ( STP ) min-1 regulated by mass flow
controllers ( Brooks ), and the temperature was ramped up mostly at 15 K min-1. The
consumption of H2 could be calculated according to the change of the H2 concentration
and the flow rate of the reducing gas. With respect to the wet TPR investigation,
the reducing gas passed first through a saturator resulting in ca. 3% H2O in the reducing
gas, and then to the reactor. Samples of CuO, Cu2O and Fe2O3 powder ( Merck ) were
used for the TPR investigations. Quartz particles from the sieve fraction in the range from
160 to 250 m were employed for the dilution of the sample.
Waste gas
PI
FIC
He
H2O Waste
gas
PI
FIC
FIC
H2 in Ar
Saturater Reactor
Fig.2.1. The experimental set up for the temperature-programmed reduction

2.1 Results and discussion
2.4.1 Dilution of the sample with quartz
For the kinetic analysis of the TPR, the activation energy is important for the
characterization of the reduction and simulation of the patterns. According to equation
( 3 ) E can be determined by using Arrhnius plots with the peak temperature Tmax at
different heating rate . In equation ( 3 ) the influence of the sample mass is neglected,
however, as Tmax is shifted to higher value as the sample mass increases. When the
amount of the sample is decreased to very low level, the influence of sample mass on
Tmax might be neglected. The reduction of 5.0 mg Fe2O3 was carried out. With such a
small sample mass, the TPR measurement could not be well reproduced as shown in
Fig.2.2.
-7 Fe2O3 840 K
2.0x10
4.9 mg
5.0 mg 823 K
-1
H2-Consumption / mol s
-7
1.5x10
-7
1.0x10
-8
5.0x10
0.0
400 500 600 700 800 900
Temperature / K
Fig.2.2. TPR patterns of the Fe2O3 powder in H2 / argon at the heating rate of 15 K min-1.
The Tmax difference of the two TPR patterns is 17 K. This deviation of the measurement
might be caused by weighing and loading the sample. A solution to increase the
reproducibility of the measurement is to dilute the sample with quartz. The stability of the
quartz was tested, and the result is shown in Fig.2.3.
-8
8,0x10
-1
-8
6,0x10
-8
4,0x10
-8
2,0x10
0,0
-8
-2,0x10
300 400 500 600 700 800 900 1000 1100
Temperature / K
Fig.2.3. The stability of the quartz in H2 / argon at the heating rate of 15 K min-1.
It is seen that the quartz is stable in the reducing gas H2 in Argon. Thus, it can be used
for the dilution of the sample for the TPR measurement. Two samples of Fe2O3 diluted
with quartz at ratio of 1:18 were prepared, and the TPR measurements were performed
as shown in Fig.2.4.
In addition, two samples of Cu2O diluted at ratios of 1:118 and 1:57 were prepared,
respectively. The results of TPR experiments are displayed in Fig.2.5.

-7
2.5x10
Fe2O3 803 K
5.5 mg 807 K
-7
5.4 mg
-1
2.0x10
-7
1.5x10
-7
1.0x10
-8
5.0x10
0.0
400 500 600 700 800 900
Temperature / K
Fig.2.4. TPR patterns of Fe2O3 in H2 / argon at the heating rate of 15 K min-1, mixing ratio of
Fe2O3 to quartz: 1:18.
-8
4.0x10
Cu2O
a 1.0 mg 723 K
726 K
-1
b 1.0 mg
-8
3.0x10
-8
2.0x10
-8
1.0x10
0.0
300 400 500 600 700 800 900 1000
Temperature / K
Fig.2.5. TPR patterns of Cu2O in H2 / argon at the heating rate of 15 K min-1, mixing ratio of
Cu2O to quartz: ( a ) 1 : 57, ( b ) 1 : 118.
The effect of diluting the sample with quartz is significant. The reproducibility of TPR
measurements with small amounts of sample is increased, and the deviation of the Tmax
is reduced to 3~4 K.
2.4.2 Effect of sample weight
The TPR measurements of CuO were carried out at the same rate of 10 K min-1 with
samples of different weights. The Tmax was shifted by 42 K, as the sample weights were
increased from 9.9 to 80.2 mg. The results are shown in Fig.2.6.
-6
3.0x10
CuO
9.9 mg 541 K
-6
2.4x10 40.3 mg
-1
80.2 mg
-6
535 K
1.8x10
-6
1.2x10
499 K
-7
6.0x10
0.0
300 400 500 600
Temperature / K
Fig.2.6. TPR patterns of CuO powder at the same heating rate of 10 K min-1 with increasing
sample weight.
The sample weight affects not only the Tmax but also the TPR pattern. Gentry et al. [ 15 ]
2+
weight, and the similar observation was also obtained in our study. The reduction of
Fe2O3 was carried out at the heating rate 15 K min-1 by decreasing the sample weight
from 4.1 to 0.5 mg. The results are displayed in Fig.2.7. When the TPR experiment was
performed with 2.1 mg Fe2O3, the reduction profile was resolved into two peaks. With
smaller amount of samples ( 1.0 and 0.5 mg ), the resolution of the TPR profiles became
more distinct.
Fe2O3 794 K
-7
2.5x10
4.1 mg
2.1 mg
-1
H2-Consumptioin / mol s
-7
2.0x10
1.0 mg
0.5 mg
-7
1.5x10
791K
767 K
-7
1.0x10
756 K 787 K
-8
5.0x10 788K
754K
0.0
500 600 700 800 900 1000
Temperature / K
Fig.2.7. TPR patterns of Fe2O3 ( mixing ratio 1:58 ) obtained at the same heating rate of 15 K
min-1 by reducing the amount of sample.
The reduction of CuO at the heating rate 15 K min-1 with reducing the amount of sample
from 6.8 to 1.0 mg yielded the same results as shown in Fig.2.8.

522 K
CuO
-7
4.0x10 6.8 mg
1.5 mg
-1
1.0 mg
-7
3.0x10
-7
2.0x10
494 K
-7
1.0x10 529 K
490 K
528 K
0.0
350 400 450 500 550 600 650
Temperature / K
Fig.2.8. TPR patterns of CuO ( mixing ratio 1:60 ) obtained at the same heating rate of 15 K min-1
by reducing the amount of sample.
2.4.3 Heating rate
The influence of the heating rate on the Tmax was investigated with the sample of Fe2O3
( mixing ratio 1:18 ). The heating rate was increased from 2 to 15 K min-1 with the same
amount of Fe2O3, and the Tmax was shifted from 691 to 791 K. The results are presented
in Fig.2.9.
-7
1.0x10
Fe2O3
-1
15 K min 1.7 mg 791 K
-8 -1
8.0x10 10 K min 1.6 mg
-1
-1
5 K min 1.6 mg
-1
2 K min 1.6 mg 756 K
-8
6.0x10
703 K
-8
4.0x10
671 K
-8
2.0x10
0.0
500 600 700 800 900 1000
Temperature / K
Fig.2.9. TPR patterns of Fe2O3 ( mixing ratio 1:18 ) using the same sample weight with
increasing heating rate.
The influence of the heating rate on the Tmax with the two-peak TPR pattern was
investigated with the sample amount of Fe2O3 from 0.5 to 0.8 mg. The whole TPR
pattern was shifted with increasing the heating rate, thus, Tmax of the first peak was
increased by 11 K and Tmax of the second peak by 12 K as shown in Fig.2.10.

Fe2O3
-8
6.0x10 15 K/min 0.5 mg
-1
10 K/min 0.8 mg
788 K
733 K
-8
4.0x10
753K 762K
-8
2.0x10
0.0
650 700 750 800 850 900
Temperature / K
Fig.2.10. TPR patterns of Fe2O3 ( mixing ratio 1:58 ) in dry H2 / argon with increasing
heating rate.
If the sample amount varies reciprocally with the heating rate, how will Tmax be shifted?
Such an investigation was carried out with a sample of CuO powder. The heating rate
was increased from 1 to 15 K min-1, however, the sample weight was decreased from
99.9 to 6.8 mg. Tmax was still shifted from 437 to 521 K as shown in Fig.2.11.
-6
2.0x10 CuO
-1 483 K
15 K min 26.7 mg 535 K
-1 513 K
10 K min 40.3 mg
-1
568 K
-1
5 K min 79.6 mg
-6 -1
1.5x10 2 K min 201 mg
-6
1.0x10
-7
5.0x10
0.0
300 400 500 600
Temperature / K
Fig.2.11. TPR patterns of CuO powder in dry H2 / argon obtained by varying the heating
rate and the sample weight.
2.4.4 Total flow rate of reducing gas
The influence of the total flow rate was investigated over 1.0 mg Fe2O3 ( mixing ratio 1:58
) at the heating rate of 15 K min-1. The results are displayed in Fig.2.12. When the total
flow rate was increased from 20 to 50 ml ( STP ) min-1, Tmax decreased by 92 K and a
shoulder appeared. Further, at a higher flow rate of 76.7 ml ( STP ) mol-1, the TPR
pattern was resolved into a two-peak pattern, and the Tmax decreased to 756 K and 787
K, respectively. These results confirm that the total flow rate affects not only the Tmax but
also the TPR pattern. The higher resolution of the TPR pattern can be achieved at high
flow rate of the reducing gas.

-1
-7
1.0x10
76.7 ml min 801 K
-1
50.0 ml min
20.0 ml min
-1 893 K
-1
-8
8.0x10 756 K
787 K
-8
6.0x10
-8
4.0x10
-8
2.0x10
0.0
600 700 800 900 1000
Temperature / K
Fig.2.12. TPR patterns of Fe2O3 obtained by increasing the flow rate at the heating rate of 15 K
min-1 and 1.0 mg Fe2O3 ( mixing ratio 1:58 ).
2.4.5 Influence of H2O in the reducing gas
It is known that the TPR pattern shifts to higher temperature with increasing the sample
weight suggesting that the reduction rate is partially retarded by steam produced during
the reduction. In agreement with this, if the reducing gas is saturated with water, a
higher Tmax should be expected. Two wet TPR experiments were carried out with the
sample BA1 ( Cu : Al = 70 : 30 ) and B5 ( Cu : Zn = 70 : 30 ), respectively. In the wet
reducing gas, Tmax was increased by 20 K over BA1, and 23 K over B5. The results are
presented in Fig.2.13 and 2.14.

-6
1.6x10
a 497 K
b 474 K
-1
-6
1.2x10
-7
8.0x10
-7
4.0x10
0.0
300 350 400 450 500 550 600 650
Temperature / K
Fig.2.13. TPR patterns of sample BA1 at the same heating rate of 6 K min-1, sample
weight 28.6 mg: ( a ) in dry H2 / argon, ( b ) in wet H2 / argon.
-7
8.0x10
498 K
a
b
-1
-7 490 K
6.0x10
-7
4.0x10
-7
2.0x10
0.0
300 350 400 450 500 550 600 650
Temperature / K
Fig.2.14. TPR patterns of sample B5 at the same heating rate of 6 K min-1, sample
weight 28.7 mg: ( a ) in dry H2 / argon, ( b ) in wet H2 / argon.

The Fig.2.14 shows that the shoulder of the TPR pattern in wet H2 / argon becomes
greater than in dry H2 / argon. It is proposed that the water added to the reducing gas
affects not only the shift of Tmax but also the resolution of the TPR pattern. A comparison
of the TPR pattern over Fe2O3 in dry H2 / argon with that in wet H2 / argon is illustrated in
Fig.2.15.
-7
3.2x10
a 834 K
b
-1
-7
2.4x10
968 K
-7
1.6x10
-8 694 K 715 K
8.0x10
0.0
400 500 600 700 800 900 1000 1100 1200
Temperature / K
Fig.2.15. TPR patterns of Fe2O3 powder at the heating rate of 15 K min-1, 10.3 mg, a ) in dry
H2 / argon, ( b ) in wet H2 / argon.
In dry H2 / argon the hydrogen consumption ratio in the first and second reduction peak is
ca. 1.037 : 9, in good agreement with the theoretical value of 1 : 9, inferred from the
sequential reduction of Fe2O3 via Fe3O4 to Fe,
3 Fe2O3 + H2 = 2 Fe3O4 + H2O ( 17 )
Fe3O4+ 4 H2 = 3 Fe + 4 H2O ( 18 )
In wet H / argon the TPR pattern was completely resolved into two peaks and shifted to
higher temperature values. The quantitative analysis shows that the consumption of H2
under the two peaks is only 1/3 of the total theoretical consumption of H2 relating to the
reduction of Fe3+ to Fe0. This infers that in wet H2 / argon Fe2O3 was not completely
reduced to Fe. The hydrogen consumption ratio in the first and second reduction peak is
ca. 1 : 1.6, in comparison to the theoretical value of 1 : 2, which may suggest the
sequential reduction of Fe2O3 via Fe3O4 to FeO,
3 Fe2O3 + H2 = 2 Fe3O4 + H2O ( 19 )
Fe3O4 + H2 = 3 FeO + H2O ( 20 )
The experimental and theoretical results of H2 consumption with respect to different
reduction step in dry and wet H2 / argon are summarized in Table 2.1.
Table 2.1 The experimental H2 consumptions compared with calculations at different
reduction steps
H2 consumption ( in dry H2 / argon ) H2 consumption ( in wet H2 / argon )
First peak Second peak First peak Second peak

mol mol mol Mol
Experimental 2.200 . 10-5 1.689 . 10-4 2.162 . 10-5 3.451 . 10-5
Theoretical 2.146 . 10-5 1.716 . 10-4 2.146 . 10-5 4.292 . 10-5
The experimental H2 consumption of the second peak in the wet TPR is smaller than the
theoretical one. The reason may be that the reduction of Fe3O4 to FeO in 3% H2O does
not occur completely.

2.4.6 Activation energy
For the evaluation of the activation energy, the Tm- method can be applied in two
ways: ( 1 ) is changed, while the sample weight, total flow rate and hydrogen content
of the reducing gas remain constant; ( 2 ) is varied, together with another experimental
parameter ( e.g. the sample weight ), in such a way that the value of P remains constant
throughout the different TPR experiments, where P = S0 / F C0, is the heating rate ( K
s-1 ), S0 the amount of metal oxide ( mol ), F the total flow rate of reducing gas ( ml s-1 ),
and C0 the concentration of H2 in argon ( mol m-3 ). Malet et al. [ 27 ] thought that P
should be kept as low as possible within the limit of experimental sensitivity, and P lower
than 20 was appropriate to obtain both a good peak resolution and accurate activation
energy. Hence, two series of the TPR experiments were carried out with constant P of 7
and 29, respectively. The TPR patterns of CuO powder in dry H2 / argon with P of 7 are
displayed in Fig.2.16.
-7
8.0x10
CuO 499 K
15 K min
-1
6.8 mg 477 K
451 K
-1
10 K min 9.9 mg 437 K
-1
-1
-7
6.0x10 5 K min 19.5 mg
-1
2 K min 50.1 mg
-1
1 K min 99.9 mg
521 K
-7
4.0x10
-7
2.0x10
0.0
300 400 500 600
Temperature / K
Fig.2.16. TPR patterns of CuO powder in dry H2 / argon with P = 7.

2
According to the equation ( 3 ), ln ( / Tmax ) against 1 / Tmax are plotted for P of 7 and 29
as shown in Fig.2.17. It is seen that all the data fit to a straight line. By measuring the
slopes the activation energy E can be obtained, which are 53 kJ mol-1 for P equal to 7
and 46 kJ mol-1 for P equal to 14. These values of E are in good agreement with the E =
6710 kJ mol-1 obtained by Stephen et al. [ 26 ], however, deviate from the E of 102 kJ
mol-1 determined by Malet et al. [ 27 ]. The reason may be the various TPR experimental
conditions used by different authors. These conditions and corresponding factors P and K
are listed in Table 2.2.
-13,5
-14,0
-14,5
ln ( / Tmax )
2
-15,0
P7
-15,5
P 29
-16,0
-16,5
0,0018 0,0019 0,0020 0,0021 0,0022 0,0023
1 / Tmax ( 1/K )
Fig.2.17. Activation energy plots for the reduction of CuO with the constant value of P.
Table 2.2 Activation energies for the reduction of CuO under different conditions
Nr. S0 F C0 P K E Quote
-1
K min-1 mg cm3 min-1 % s kJ mol-1
1 0.51.75 5 18.4 66.7 0.060.2 7 88 [ 16 ]
2 0.7514 20 20 4 5 96 420 6710 [ 26 ]
3 2.520 530 50 5 520 34 202 102 [ 27 ]
4 115 7 100 76.7 5.17 7 29 423 53 this work
5 215 27 201 76.7 5.17 29 113851 46 this work
According to the results in Table 2.2 the activation energy of Nr.1 ( 88 kJ mol-1 ) is similar
to that of Nr.2 ( 67 10 kJ mol-1 ), however, P and K of Nr. 1 ( 0.2 and 7 s-1 ) are much
smaller than those of Nr.2 ( 96 and 420 s-1 ). Although the P and K of Nr. 3 ( 20 and 202 )
are between those of Nr.1 and 2, however, a higher apparent activation energy 102 kJ
mol-1 was obtained. Therefore, there may be not an obvious relation between the
activation energy and values of P and K.
It is known that the TPR pattern of CuO is resolved into a two-peak pattern at small
sample mass. Thus, a series of TPR experiments with 0.5 mg CuO at the heating rates of
5, 10 and 15 K min-1 were carried out, and the results are displayed in Fig. 2.18.
-8
3.0x10
-8
2.0x10 478 K
547 K
-8
-1
1.0x10
H2 Consumption / mol s
4.00E-008 a
0.0
200 300 400 500 600 700 800
474 K
0.00E+000
542 K
b
200 300 400 500 600 700 800
-8
2.0x10
451 K
1.0x10
-8 518 K
c
0.0
200 300 400 500 600 700 800
Temperature / K
Fig. 2.18. TPR patterns of CuO ( mixing ratio 1 : 60 ) using the same sample amount: ( a ) 0.503
mg, 15 K min-1, ( b ) 0.497 mg, 10 K min-1, ( c ) 0.497 mg, 5 K min-1.
2
ln ( / Tmax ) against 1 / Tmax is plotted as shown in Fig.2.19. By measuring the slopes the
activation energies E could be obtained, which are 57 kJ mol-1 for the first peak ( CuO
Cu2O ) and 77 kJ mol-1 for the second peak ( Cu2O Cu ), respectively.

-13.5
a
b
-14.0
ln ( / Tmax )
2
-14.5
-15.0
-15.5
0.00165 0.00180 0.00195 0.00210 0.00225
1 / Tmax ( 1/K )
Fig.2.19. Activation energy plots for the reduction of CuO ( mixing ratio 1 : 60 ): ( a ) low -
temperature peak ( CuO Cu2O ), E = 57 kJ mol-1 , ( b ) high temperature peak (
Cu2O Cu ), E = 77 kJ mol-1.
2.5 Conclusions
1) Tmax is highly affected by the TPR experimental conditions. Tmax is shifted to higher
values by increasing the heating rate and the amount of sample, and to lower values
by increasing the total flow rate of reducing gas.
2) The dilution of the sample with quartz may be an adaptable method to carry out the
TPR investigation with a small amount of sample.
3) The TPR patterns of Fe2O3 and CuO in dry H2 / argon are well resolved into two-peak
patterns using a small amount of sample. The reduction mechanisms can be
proposed as follows:
For the reduction of Fe2O3,
3 Fe2O3 + H2 = 2 Fe3O4 + H2O
Fe3O4 + 4 H2 = 3 Fe + 4 H2O
for the reduction of CuO,
2 CuO + H2 = Cu2O + H2O
Cu2O + H2 = 2 Cu + H2O
4) The addition of H2O in the reducing gas has significant influence on the Tmax and the
TPR pattern: Tmax is shifted to higher values and the TPR pattern is strongly changed.
The TPR pattern of Fe2O3 in wet H2 / argon is resolved completely into a two-peak
pattern. The quantitative analysis implies that the reduction of Fe2O3 in wet H2 / argon
occurs as follows:
3 Fe2O3 + H2 = 2 Fe3O4 + H2O
Fe3O4 + H2 = 3 FeO + H2O
5) The evaluations of activation energies E for the reduction of CuO are carried out for
one peak and two-peak TPR patterns, respectively. The value of the apparent
activation energy E for the one peak TPR pattern is 50 kJ mol-1. For the two-peak TPR
pattern the values of E are 57 kJ mol-1 for the first peak and 77 kJ mol-1 for the
second peak.
6) Up to now there seems to be not a suitable criterion of the experimental condition for
the TPR investigation. For the TPR studies of Fe2O3 and CuO at high resolution, the
following experimental conditions are suggested: a sample amount of 0.5 mg; mixing
ratio of 1 : 60; total flowing rate of reducing gas 80 ml ( STP ) min-1; H2 concentration
5 %; heating rate 1 20 K min-1.

References
[ 1 ] S. J. Gentry and P. T. Walsh, J. Chem. Soc., Faraday Trans. 78, 1515 ( 1982 )
[ 2 ] P.A. Jacobs, J. Linart, H. Nijs and J. B. Uytterhoeven, J. Chem. Soc., Faraday Trans.
73, 1745 ( 1977 )
[ 3 ] N. Wagstaff and R. Prins, J. Catal. 59, 434 ( 1979 )
[ 4 ] B. H. Isaacs and E. E. Petersen, J. Catal. 77, 43 ( 1982 )
[ 5 ] A. Lycourghiotis, C. Defosse, F. Delannay, J. lemaitre and B. Delmon, J. Chem.
Soc., Faraday Trans. 176, 1677 ( 1980 )
[ 6 ] H.F.J. Van, T. Blik and J.W. Niemantsverdriet, Appl. Catal. 10, 155 ( 1984 )
[ 7 ] X.T. Guo, J.Y. Shen, Y.F. Xia, Y. Chen, J. Chem. Soc., Faraday Trans. 89 ( 7 ), 1079
( 1993 )
[ 8 ] D. E. Stobbe and F. R. van Buren, J. Chem. Soc. Faraday Trans. 87 ( 10 ), 1631

( 1991 )
[ 9 ] Z. Gao, B. Zhang and J. Cui, Appl. Catal. 72 ( 2 ), 331 ( 1991 )
[ 10 ] N. W. Hurst, S.J. Gentry, A. Jones and B.D. McNicol, Catal. Rev. Sci. Eng. 24, 233
( 1982 )
[ 11 ] B.S. Sircar and R. Kumar, J. Chem. Soc., Faraday Trans.1 80, 2489 ( 1984 )
[ 12 ] G. Munteanu and M. Craiu, React. Kinet. Catal. Lett. 50 ( 1-2 ), 49 ( 1993 )
[ 13 ] B.H. Bosch, B.J. Kip, J.G. van Qwmmen and P.J. Gellings, J. Chem. Soc.,
Faraday Trans.1 80, 2479 ( 1984 )
[ 14 ] D. A. M. Montry and A. Baiker, J. Catal. 83, 323 ( 1983 )
[ 15 ] S.J. Gentry, N.W. Hurst and A. J. Jones, J. Chem. Soc., Faraday Trans.1 75, 1688
( 1979 )
[ 16 ] A. Andreini, E. K. Poels and A. Bliek, React. Kinet. Catal. Lett. 63 ( 2 ), 209 ( 1998 )
[ 17 ] L. Bin and Z. Liu, Chinese J. of Chem. Eng. 4 ( 3 ), 227 ( 1996 )
[ 18 ] O.J. Wimmers, P.Arnoldy and J.A. Moulijn, J. Phys. Chem. 90, 1331 ( 1986 )
[ 19 ] G. Fierro, M. L. Jacono, M. Inversi, P. Porta, R. Laveccina, and F. Cioci, J. Catal.
148, 709 ( 1994 )
[ 20 ] G.Fiero, M.L. Jacono, M. Inversi, P.Porta, F. Cioci and R. Lavecchia, Appl. Catal.
A: General 137, 327 ( 1996 )
[ 21 ] G. Munteanu, L. Ilieva and D. Andreeva, Thermochimica Acta 291, 171 ( 1997 )
[ 22 ] L.I. Ilieva, D. Andreeva, and A.A. Andreev, Thermochimica Acta 292, 169 ( 1997 )
[ 23 ] M. Shimokawabe, R. Furuichi and T. Ishii, Thermochimica Acta 28, 287 ( 1979 )
[ 24 ] J. M. Smith, Chemical Engineering Kinetics, McGraw-Hill, ( 1981 )
[ 25 ] O. Kancke et al., Thermochemical Properties of Inorganic Substances, Springer,

1992.
[ 26 ] B. Stephen, J. Gentry, N.W. Hurst and A. Jones, J. Chem. Soc., Faraday Trans. 1,
11, 603 ( 1981 )
[ 27 ] P. Malet and A. Caballero, J. Chem. Soc., Faraday Trans.1 84 ( 7 ), 2369 ( 1988 )

Chapter 3: Transient Kinetic Studies on the Commercial Catalyst 32
3 Transient Kinetic Studies on the Commercial Catalyst
Abstract: By performing the blank experiment it is found that Tonset without catalyst is 813 K,
contrarily, with catalyst Tonset decreases by 90 K, and much higher activity and selectivity can be
achieved. The coupling of the dehydrogenation of ethylbenzene and the reduction of Fe2O3 leads
to a higher equilibrium conversion, and a very high initial conversion of ethylbenzene was
obtained. During the regeneration with steam the catalyst is partly reoxidized to Fe3+-containing
oxide and a rather low concentration of CO and CO2 was observed. It is not the coke, but the
reduction of Fe2O3 by H2 produced during the dehydrogenation of ethylbenzene, which causes
the fast initial deactivation of the catalyst.
Keywords: Transient response, dehydrogenation of ethylbenzene, reduction of Fe2O3.
3.1 Introduction
The dehydrogenation of ethylbenzene to styrene is an endothermic reaction which is
performed at around 900 K in excess of steam. The main byproducts are benzene and
toluene. Since it is a very important industrial process, it has been studied by many
authors since 1948 up to now [ 1 ]. In the field of kinetics and simulation Sheel [ 2 ]
suggested a reaction scheme including 6 reactions which was accepted by most authors [
3-4 ]. In addition to the classical packed bed reactor a reverse flow operation [ 5-6 ] and a
composite palladium membrane reactor [ 7 ] were also suggested. The nature of the
active catalyst attracted the attentions of many researchers, and the active phase, the
role of the main promoter potassium and the possible reaction mechanism were
intensively studied with the aid of numerous methods. E. H. Lee [ 8 ] carried out X-ray
containing 7% K2O and 2% Cr2O3. The diffraction patterns of the fresh one were
DWWULEXWHGWR )H2O3, K2Fe22O34 and some K2O. In the used one, Fe3O4 was the major
crystalline component. XRD, X-ray photoelection spectroscopy ( XPS ) and the in situ
catalyst deactivation with CO2 were used by T. Hirano [ 9, 10 ] to probe the formation of
the active site. As a result, potassium ferrite ( KFeO2 ) formed on the surface of the
catalyst under reaction condition was assumed to take part in the formation of the active
state of the catalyst, and it was also confirmed that the steam treatment strongly
increased the activity for the formation of styrene, benzene, and toluene. Contrarily, the
CO2 treatment was found to decrease the activity. Using ex situ and in situ XRD,
extended X-ray absorption fine structure ( EXAFS ) and Mssbauer spectroscopy for the
analysis of the bulk, and XPS and UPS as surface sensitive methods Muhler et al. [
11,12 ] confirmed that the active phase is a thin layer of KFeO2 supported on a solid
solution of K2Fe22O34 in Fe3O4. Moreover, a small amount of potassium carbonate is
present on the surface. A detailed model considering the formation of the active phase
under reaction conditions, the deactivation and the regeneration with steam was
proposed. More recently, studies on the epitaxially grown iron oxide model catalyst films
were carried out aiming to understand the reaction on an atomic scale. M. Whn et al. [
13 ] investigated the adsorption of styrene on single crystal surfaces of FeO and Fe3O4
epitaxially grown on Pt ( 111 ) by NEXAFS and XPS. On Fe3O4 styrene formed a
chemisorbed monolayer stable up to room temperature, whereas on FeO a more weakly
bound physisorbed layer was formed. A corresponding adsorption study of ethylbenzene
and styrene on the Fe3O4DQG )H2O3 surface was made by Y. Joseph et al. [ 14
@ 7KH FKHPLVRUSWLRQ RI HWK\OEHQ]HQH RQ WKH LURQ VLWHV YLD WKH HOHFWURQ V\VWHP ZDV
supposed to be important for the dehydrogenation of ethylbenzene to styrene. A three-
chamber surface analysis system combined with a high-pressure reaction cell equipped
with a single crystal flow reactor was developed to study the dehydrogenation of
ethylbenzene on the different single crystals [ 15-17 ]. It was found that Fe3O4 ( 111 )
was catalytically inactive under both low pressure and atmospheric pressure, and that
WKH GHK\GURJHQDWLRQ RI HWK\OEHQ]HQH RQ WKH )H2O3 ( 001 ) surface started after an
activation period of about 45 minutes. The activation period on the potassium-promoted
film was even longer, and defects such as steps were considered as catalytically active
sites. Furthermore, W. Ranke et al. [ 18 ] determined the adsorption energies and kinetic
parameters of water, ethylbenzene and styrene on Fe ( 111 ) and Fe3O4 ( 111 ).
Many efforts were made to find new catalysts and industrial processes. One new
attractive process is to use CO2 instead of steam over the catalyst Fe2O3 / Al2O3 to save
energy and to reach higher conversion [ 19-20 ], and the other is the coupling of the
dehydrogenation of ethylbenzene with the decomposition of N2O over the -Al2O3-
supported transition metal oxide catalysts [ 21-22 ]. Reported new catalysts are MCM-41-
supported iron oxide [ 23 ] and manganese-substituted iron-potassium oxide [ 24 ]. Novel
materials used as the catalysts for this reaction are iron oxide-based honeycombs [ 25 ]
and carbonaceous materials [ 26-27 ].
Our study focuses on this process based on the iron oxide catalysts by transient kinetic
methods. The transient response method proved to be useful not only in obtaining
quantitative information concerning the adsorption characteristics of gases on a catalyst
but also in analyzing the detailed kinetic structure of multi-step reaction mechanisms. Any
kind of variable that defines the chemical kinetic state of the system may be used for the
perturbation method, i.e., concentration, temperature, pressure and flow rate. A well-
defined forcing function may be used for the perturbation of any one of these variables.
Possible functions are rectangular step functions, rectangular pulses, etc.
In addition to the study of the mass change on the catalyst surface during the transient
reaction, one can also follow the macroscopic response, for example, the responses of
concentration, temperature and pressure. C.O.Bennet [ 28 ], and Kobayaschi [ 29, 30 ]
are pioneers in this field. In order to follow the dynamic composition of the reactor
effluent the application of a mass spectrometer was suggested [ 31 ].
In the present study the initial activation period of the process and the steam regeneration
were investigated in detail by switching the ethylbenzene feed stream on and off sharply.
The temperature of the reaction was used as a variable, too. It was changed by applying
linear heating and cooling ramps under quasi isothermal steady-state conditions to obtain
more information on the reaction mechanism.
3. 2 Experimental
The study was carried out in a microreactor flow system. The concentration step-up and
step-down experiments were accomplished by switching a four-way valve ( Valco ). The
quantitative analysis of the educt and the products was performed on-line by a calibrated
mass spectrometer ( QMS 422 ). The flow scheme of the experimental set-up is shown in
Fig.3.1.
Waste gas
TC
TC
GC
Micropump
MS
PI
Gas Mixture 1 Valve1
FIC
He Gas Mixture 2
Valve2
Mixer
Waste gas
TC
TI
EtB H2O
PI
FIC
Ar
FIC
Reactor Cooler
Fig.3.1. The experimental set up for the transient kinetic study of dehydrogenation of ethylbenzene
An industrial potassium-promoted iron oxide catalyst ( S6-20, BASF ) was used with a
sieve fraction in the range from 250 to 355 m. 1.0 g was held between two layers of
quartz particles ( < 355 m ) in a quartz tubular flow reactor of 14 mm inner diameter.
The height of the catalyst bed was 7 mm, and a thermocouple ( diameter 0.5 mm ) in a
glass tube was placed in the middle of the catalyst bed through the upper quartz layer.
The reactor was immersed in a fluidized bed of sand in an oven to ensure isothermal
reaction conditions. The heating and cooling rates of the oven were programmed by a
PC. Water was pumped to the vaporizer, mixed with He, and vaporized ( mixture 1 ). All
gas flows were regulated by mass flow controllers. Argon was fed to two saturators with a
flow of 20 ml ( STP ) / min which were filled with ethylbenzene and water. The
concentrations of ethylbenzene and steam in the educt stream ( mixture 2 ) were
obtained by controlling the temperatures of the two saturators. With a educt concentration
of 2.5% ethylbenzene and 25% steam a liquid hourly space velocity ( LHSV ) of 0.43 h-1
and a H2O / etylbenzene ratio ( S/O ) of 10:1 were achieved. The step-up and step-down
experiments were accomplished by switching of the four-way valve No.1. By switching
four-way valve No. 2 the feed stream or the reactor effluent were analyzed on line by
the gas chromatograph ( GC ) and the mass spectrometer ( QMS ). The sensitivity
factors for the component matrix of the mass spectrometer analysis were obtained by the
calibration with reference mixtures containing CO, CO2, methane, ethylene or hydrogen.
Ethylbenzene ( EB ), styrene ( ST ), toluene ( TOL ) and benzene ( BZ ) were calibrated
by flowing Ar through a saturator filled with EB, ST, TOL or BZ, respectively. The
compositions of the calibration gases were analyzed by MS and GC at the same time,
and a comparison of the analysis by GC and MS was performed. The analytical GC
results under different reaction conditions are summarized in Table 3.1.
Table 3.1 The analytical results under different conditions by gas chromatography
Temp. XEB SST SBZ STOL Carbon balance
K % % % % %
823 30.4 96.2 2.5 1.3 98.0
848 49.7 95.5 2.6 1.9 94.1
873 66.7 94.4 2.6 2.9 98.5
898 82.7 92.2 2.9 4.9 99.2

The conversion of ethylbenzene XEB, the yield of styrene YST, and the selectivity to
styrene SST, to benzene SBZ , and to toluene STOL were calculated according to the
equations ( 1 ) to ( 5 ), where F is the effluent molar flow rate and F0 is the molar flow
rate into the reactor.
F0 C EB0 F C EB
X EB = 100% (1)
F0 C EB0
F C ST
YST = 100% (2)
F0 C EB0
F C ST F0 C ST0
SST = 100% (3)
( F C ST F0 C ST0 ) + ( F C BZ F0 C BZ0 ) + ( F C TOL F0 C TOL0 )
F C BZ F0 C BZ0
S BZ = 100% (4)
F C TOL F0 C TOL0
S TOL = 100% ( 5 )
There are few differences between the concentration results gotten from the two different
analytical methods. In the transient study the analytical results of MS would be taken,
and the analytical results of GC would be used in the kinetic simulation study.
3.3.1 The blank experiments
A temperature-programmed blank reaction experiment was carried out with quartz
particles loaded in the reactor. The temperature of the reactor was increased at a rate of
2 K / min from 523 K until 923 K, and the concentrations of all productions at the reactor
outlet were analyzed on line by the mass spectrometer. It was found that the onset
temperature was about 813 K ( 90 K higher compared to Tonset over the catalyst ), and
more benzene was formed than toluene. At 923 K a conversion of 23 % was reached
and a selectivity of only 65 %. Considering these results a temperature range of the
kinetic study was chosen from 823 K to 898 K.
For the step-up and step-down experiments the following blank experiments were carried
out at 873 K and at 898 K: first, helium-containing steam was introduced to the reactor;
after 30 minutes the flow was switched to argon containing 2.5% ethylbenzene and steam
as shown in Fig.3.2.
He+H2O Argon + H2O + EB

2.5
Concentration of EB / %
2.0
at 873 K
1.5
at 898 K
1.0
0.5
0.0
-20 -10 0 10 20 30 40
Relative time / min
Fig. 3. 2. The blank step-up experiment over quartz particles
The concentration of ethylbenzene was stable after about 2 minutes. During the first two
minutes of the step-up experiment the concentration of EB can be approximated as a
function of the time t by
C EB = 0.1409 + 8.336 t 11.232 t 2 + 6.698 t 3 1.469 t 4 ( % ) (6)
3.3.2 The initial activation period of the fresh catalyst
The investigation of the initial activation period of the reaction was carried out in the
following way: at 873 K first helium-containing steam was introduced to the reactor; after
30 minutes the flow was switched to argon containing ethylbenzene and steam thus
starting the reaction which reached steady state finally. The step-up experiment was
carried out at 898 K in the same way. The results of these initial step-up experiments
over the catalyst S6-20 at 873 K and 898 K are shown in Fig. 3.3.
The conversion of ethylbenzene was very high at the beginning, reaching even 95~96 %
at 898 K during the first three minutes. This degree of conversion was higher than the
equilibrium conversion under these reaction conditions ( X equilibrium = 94,2 % ).
Immediately it decreased very fast and after the first 20 minutes the degree of conversion
was about 50%. Then the conversion decreased more slowly, and after 3 hours the
conversion reached steady state at 45 % at 898 K and 34 % at 873 K, respectively.
The yield of styrene changed in the same way as the conversion of ethylbenzene. After
about 3 hours the yield of styrene stabilized at 39 % ( 898 K ) and 32 % ( 873 K ).

100
80 equilibrium con. at 898 K

Conversion of EB / %
equilibrium con. at 873 K

at 898 K
60 at 873 K
A
40
20
He+H2O Argon + H2O + Ethylbenzene
0 0.75
0 50 100 150
Concentration of H2 / %
80 0.60
Yield of Styrene / %
at 898 K
60 0.45
at 873 K
B
40 0.30
at 898 K
20 0.15
at 873 K
0 0.00
0 50 100 150
CO2 at 898 K
Concentrations of CO2 and CO / %
CO2 at 873 K
0.15 CO at 898 K
CO at 873 K
C 0.10
0.05
0.00
-40 0 40 80 120 160
Relative time / min
Fig. 3.3. The initial step-up experiments over the fresh catalyst S6-20 at 873 K and at 898 K,
LHSV = 0.43 h-1, S/O = 10:1 : ( A ) The conversion of ethylbenzene vs. relative time,
( B ) the yield of styrene and concentration of hydrogen vs. relative time, ( C ) the
concentrations of CO and CO2 vs. relative time.
It is interesting that in the initial period of the reaction the conversion of ethylbenzene was
higher than the equilibrium conversion. The second interesting observation is the time
dependence of the concentration of the hydrogen. As an accompanying product of the
dehydrogenation of ethylbenzene hydrogen should reach the highest concentration at the
very high conversion in the beginning of the experiment. Contrarily, the lowest
concentration was observed during the first minutes of the initial period. This implies that
there is another reaction which consumes hydrogen. At this point we must take the
catalyst into account, which is composed mainly of iron oxide. It is well known that before
the reaction the color of catalyst is red ( haematite ), whereas it is black ( magnetite ) after
the reaction:
3 Fe2O3 + H2 = 2 Fe3O4 + H2O (7)
This equation has to be combined with the main reaction, the dehydrogenation of
ethylbenzene:
C6H5C2H5 = C6H5C2H3 + H2 (8)
Thus, the dehydrogenation product hydrogen reduces the Fe3+-containing haematite,
and the consumption of hydrogen changes the equilibrium conversion of the reaction. At
the very beginning the concentration of hydrogen was only 0.1 %, and such a low
concentration of H2 leads to an increase of the equilibrium conversion up to 99,74 % at
898 K. In this way, the high conversion of 95~96 % during the first minutes can be
rationalized. In addition, the reduction of the catalyst is obviously correlated with the initial
decrease in activity.
The changes of the concentrations of CO and CO2 in the initial period are shown in the
Fig.3.3 C. At the beginning there was a great jump of the concentrations of CO2 at 873
K and 898 K, then the concentration of CO2 fell down very fast in the first 20 minutes to
very low concentration of about 0.05 %. The concentration of CO changed similarly at the
different temperatures.
3.3.3 The regeneration of the catalyst with steam
The regeneration experiment was carried out as follows: at 873 K the reaction was run for
several hours at steady state. Then, the Ar flow containing steam and ethylbenzene was
switched to the He flow containing only steam. The same experiment was carried out at
873 K and at 898 K. The concentrations of ethylbenzene, styrene, benzene, toluene are
shown in Fig.3.4 A, those of methane, ethylene and hydrogen are shown in Fig.3.4 B,
and the concentrations of CO and CO2 are shown in Fig.3.4 C.
It is found that the concentration of the educt ethylbenzene decreased very fast to zero,
and also the concentrations of the products styrene, benzene, toluene, methane, and
ethylene approached zero simultaneously. Only the concentration of the product
hydrogen changed in a different way, since it did not reach zero during the first 30 min.
This may indicate that magnetite is reoxidized :
2 Fe3O4 + H2O = 3 Fe2O3 + H2 (9)
The concentrations of CO and CO2 varied differently. At first, the concentration of CO
reached 0.065% very fast, and then it decreased. In contrast, the concentration of CO2
dropped sharply during the first minutes. During the regeneration with steam
carbonaceous deposits formed during the reaction are converted to CO and hydrogen
according to the following reaction:
C + H2O = CO + H2 ( 10 )
EB
1,35 ST
Tol
Concentration / %
Ben
A 0,90
0,45
Ar+H2O+EB He+H2O
0,00
-50 0 50 100 150
0,45 H2
Concentration / %
CH4
C2H4
B 0,30
0,15
0,00
-50 0 50 100 150
0,060 CO
CO2
Concentration / %
0,045
C
0,030
0,015
0,000
-40 0 40 80 120
Relative time / min
Fig.3.4. The regeneration of the catalyst S6-20 at 873 K and at 898 K, LHSV = 0.43 h-1, S/O =
10:1: ( A ) The concentrations of EB, ST, TOL and BZ vs. relative time, ( B ) the
concentrations of the C2H4, CH4 and H2 vs. relative time, ( C ) the concentrations of CO
and CO vs. relative time.

This can explain why the concentration of CO passed through a maximum during the first
minutes. CO reacts continuously with steam forming carbon dioxide and hydrogen
according to the following reaction ( water gas shift reaction ) :
CO + H2O = CO2 + H2 ( 11 )
The initial small drop of the concentration of CO2 is presumably an artifact induced by
the flow switching. Unfortunately, it is not possible within the experimental accuracy to
decide whether the ongoing production of H2 during the regeneration with steam is due to
the reoxidation of magnetite or the gasification of coke.
3.3.4 The transient behavior of the catalyst after regeneration
The investigation of the catalytic behavior after regeneration was carried out in the
following way: after 10 hours reaction at 873 K or at 898 K, the He flow containing steam
was introduced to the reactor; after 3 hours the flow was switched to Ar containing
ethylbenzene and steam, thus starting the reaction again. The experiments were carried
out at 873 K and 898 K, respectively. The conversion of ethylbenzene, the yield of
styrene and the concentrations of CO and CO2 at different reaction temperatures are
shown in Fig.3.5.
100
Conversion of EB / % 80 equilibrium con. at 898 K

equilibrium con. at 873 K
at 898 K
60 at 873 K
A
40
20
0 1.2
0 50 100 150
1.0
80
at 898 K
at 873 K
0.8
60
B 0.6
40
0.4
at 898 K
20
at 873 K 0.2
0 0.0
0 50 100 150
CO2 at 898 K
Concentrations of CO2 and CO / %
CO2 at 873 K
0.15 CO at 898 K
CO at 873 K
C 0.10
0.05
0.00
-40 0 40 80 120 160
Relative time /min
Fig.3.5. The step-up experiments after regeneration of the catalyst S6-20 at 873 K and at 898 K,
LHSV = 0.43 h-1, S/O = 10:1: ( A ) The conversion of ethylbenzene vs. relative time, ( B )
the yield of styrene and concentration of hydrogen vs. relative time, ( C ) the
The conversion of ethylbenzene was very high at the beginning, and then it decreased
very fast. During the first 20 minutes the conversion decreased by about 50 %. After 3
hours the conversion stabilized at about 39 % at 898 K or at 33 % at 873 K. The yield
of styrene decreased very fast in the first 20 minutes. After about 3 hours the yield of
styrene stabilized at 38% ( 898 K ) or 35% ( 873 K ). These results are in good
agreement with those obtained after the first activation ( Fig.3.3 ). Comparing the results
of the first step-up experiment with that after regeneration it can be concluded that rather
similar phenomena occur. During both experiments a very high initial activity of the
catalyst was observed, although the initial activity of the second step-up experiment was
somewhat lower than that of the first one. It has to be pointed out that the H2
concentration was again very low initially, and then it increased, passed through a
maximum and decreased.
In the initial period after regeneration there was no jump of the concentration of CO2,
which is different from the change of the concentration of CO2 in the initial period on the
fresh catalyst. It is obvious, too, that the concentrations of CO and CO2 were at a lower
level than those on the fresh catalyst indicating that there was little coke on the catalyst
after the regeneration.

3.3.5 The investigation of deactivation phenomena
To study hysteresis phenomena due to deactivation, the reaction was carried out in a
temperature-programmed way as follows: when the reaction was stable at 873 K, the
temperature of the reactor was decreased to 473 K at a rate of 2 K min-1. Then at the
same heating rate the temperature of the reactor was increased up to 873 K.
The conversion of the ethylbenzene, the yields of styrene, toluene and benzene under
ramping up and ramping down are displayed as a function of the reaction temperature. In
Fig.3.6 A the degrees of conversions at different reaction temperatures ( Table 3.1 ) are
displayed, too. The results obtained by GC are in good agreement with those obtained by
MS under ramping-up conditions. The conversion and yields under ramping-up conditions
are somewhat lower than those when ramping down the temperature. A possible reason
is that in the temperature range from 473 K to 723 K where no reaction occurs, a
regeneration of Fe3O4 to Fe2O3 might have taken place. In general, since there is no
pronounced hysteresis, coke deposits do not seem to play an important role with respect
to deactivation on a short time scale.

100
ramping down to 573 K
Conversion of ethylbenzene / % ramping up to 898 K
80 degrees of conversion from Table 1
60
A
40
20
0
450 500 550 600 650 700 750 800 850 900 950
ramping down to 573 K
ramping up to 898 K
80
Yield of styrene / %
60
B
40
20
0
600 650 700 750 800 850 900 950
ramping down to 573 K (TOL)
Yield of toluene and benzene / %
ramping up to 898 K (TOL)

8 ramping down to 573 K (BZ)
ramping up to 898 K (BZ)
6
C
4
0
600 650 700 750 800 850 900 950
Temperature / K
Fig. 3.6. The hysteresis experiment over catalyst S6-20, LHSV = 0.43 h-1, S/O = 10:1: ( A ) The
conversion of ethylbenzene vs. temperature, ( B ) the yield of styrene vs. temperature,
( C ) the yields of toluene and benzene vs. temperature.

3.4 Conclusions
1) By performing the blank experiment without catalyst it is found that there is reaction
occurring at temperatures higher than 813 K. With catalyst the Tonset is about 723 K.
Under the same reaction conditions much higher activity and selectivity can be
reached using a catalyst.
2) By means of transient kinetic experiments the initial periods of the dehydrogenation
of ethylbenzene over the fresh catalyst and after regeneration were studied. Although
there were some minor differences between the two initial periods, a much higher
initial activity of the catalyst at the very beginning was observed for both cases. It is
considered that the coupling of the dehydrogenation of ethylbenzene and the
reduction of Fe2O3 ( haematite ) led to a much higher equilibrium conversion.
Consequently, during the regeneration with steam the catalyst is partly reoxidized to
Fe3+-containing oxides such as haematite or KFeO2.
3) Another function of steam is remove coke deposits. It was confirmed that steam
reacts with the coke deposits on the catalyst forming CO, and CO reacts further with
steam yielding CO2. However, the rather low concentrations of CO and CO2 observed
during the regeneration imply that coking is not the main origin of the fast initial
deactivation. The special role of coke deposits was probed by performing
temperature-programmed reaction experiments. The results indicate that there is no
hysteresis using ramps of 2 K min-1. Thus, coking does not influence the catalytic
activity on a short time scale.

References
[ 1 ] R.R. Wenner and E.C. Dybdal, Chem. Eng. Pro. 44, 275 ( 1948 )
[ 2 ] J.G.P. Sheel and C.M. Crowe, The Canadian J. of Chem. Eng. 47, 183 ( 1969 )
[ 3 ] D. E. Clough and W. F. Ramirez, AIChE J. 22, 1097 ( 1976 )
[ 4 ] S.S.E.H. Elnashaie and R. Hughes etc., Ind. Eng. Chem. Res. 32, 2537 ( 1993 )
[ 5 ] J. D. Snyder and B. Subramaniam, Chem. Eng. Sci. 49, 5585 ( 1994 )
[ 6 ] G. Kolios and G. Eigenberger, Chem. Eng. Sci. 54, 2637 ( 1999 )
[ 7 ] C. Hermann, P. Quicker, R. Dittmeyer, J. of Membrane Science 136, 161 ( 1997 )
[ 8 ] E. H. Lee, Catal. Rev. 8 ( 2 ), 285 ( 1973 )
[ 9 ] T. Hirano, Appl. Catal. 26, 65 ( 1986 )
[ 11 ] M. Muhler, J. Schtze and G. Ertl, J. Catal. 126, 339 ( 1990 )
[ 12 ] M. Muhler, R.Schlgl and G. Ertl, J. Catal. 138, 413 ( 1992 )
[ 13 ] M. Whn, Y. Joseph, P.S. Bagus, A. Niklewski, R. Pttner, S. Reiss, W. Weiss, M.
Martins, G. Kaindl, and Ch. Wll, J. Phys. Chem. B 104, 7694 ( 2000 )
[ 14 ] Y. Joseph, M. Whn, A. Niklewski, W. Ranke, W. Weiss, Ch. Wll and R. Schlgl,
Phys. Chem. Chem. Phys. 2, 5314 ( 2000 )
[ 15 ] W. Weiss and R. Schlgl, Topics in Catalysis 13, 75 ( 2000 )
[ 16 ] C. Kuhrs, M. Swoboda and W. Weiss, Topics in Catalysis 15( 1 ), 13 ( 2001 )
[ 17 ] C. Kuhrs, Y.Arita, W.Weiss, W. Ranke and R. Schlgl, Topics in Catalysis 14 (1-4 ),
111 ( 2001 )
[ 18 ] W. Ranke und Y. Joseph, Phys. Chem. Chem. Phys. 4, 2483 ( 2002 )

[ 19 ] Naoki Mimura and Mashiro Saito, Catalysis Today 55, 173 ( 2000 )
[ 20 ] T. Badstube and P.Kustrowski, Appl. Catal. A: General 204, 153 ( 2000 )
[ 21 ] A.L. Sun, Z.F. Qin and J.G. Wang, Catal. Lett. 79, 33 ( 2002 )
[ 22 ] P. Ku trowski, M. Zbroja, R. Dziembaj and H. Papp, Catal. Lett. 80, 1 ( 2002 )
[ 23 ] She-Tin Wong and Chung-Yuan Mou, Appl. Catal. A: General 198, 115 ( 2000 )
[ 24 ] Akihiko Miyakoshi, Akifumi Ueno, Masaru Ichikawa, Appl. Catal. A: General 219,
249 ( 2001 )
[ 25 ] William P. Addiego, Wei Liu, Thorsten Boger, Catalysis Today 69, 25 ( 2001 )
[ 26 ] Gernard Mestl, Nadezhda I. maksimova, Nicolas Keller, Vladimir V. Roddatis, and
Robert Schlgl, Angew. Chem. Int. Ed. 40( 11 ), 2066 ( 2001 )
[ 27 ] M.F.R. Pereira, J.J.M. rfo, J.L. Figueiredo, Appl. Catal. A: General 218, 307
( 2001 )
[ 28 ] C. O. Bennett, Catal. Rev. Sci. Eng. 13 ( 2 ), 121 ( 1976 )
[ 29 ] M. Kobayashi and H. Kobayashi, J. Catal. 27, 114 ( 1972 )
[ 30 ] H. Kobayshi and M. Kobayashi, Catal. Rev.- Sci. Eng. 10 ( 2 ), 139 ( 1976 )
[ 31 ] E. Mueller and H. Hofmann, Chem. Eng. Sci. 42 ( 7 ), 1705 ( 1987 )

Chapter 4: Initial Transient Behavior of the Iron Oxide Catalyst 53
4 Initial Transient Behavior of the Iron Oxide Catalysts
Abstract: During the initial stage Fe2O3 is reduced to Fe3O4 by H2 and the active phase KFeO2 is
partially decomposed by CO2. The very high initial conversion of ethylbenzene is achieved by the
coupling of the dehydrogenation of ethylbenzene and the reduction of Fe2O3. With steaming of
the catalyst carbon deposition can be removed and Fe3O4 be partially reoxidized to Fe2O3.
Furthermore, some new KFeO2 is formed with the decomposition of K2CO3. The catalyst under
the steady reaction condition is considered to be composed of KFeO2, Fe3O4 and K2CO3. A
modified schematic life cycle of the K-promoted styrene catalyst is proposed.
Keywords: Transient response, dehydrogenation of ethylbenzene, TPR, TPO, XRD.
4.1 Introduction
The catalytic dehydrogenation of ethylbenzene to styrene is the main industrial process
of the production of the very important monomer styrene [ 1, 2 ]. As a typical chemical
process the dehydrogenation of ethylbenzene to styrene over the iron oxide catalysts
attracts a lot of researchers to probe many aspects of it, such as developing new
catalysts, novel design of reactor, the studies of kinetics, simulation, deactivation, and
active phase identification [ 3 - 9 ].
Herzog [ 10 ] studied the reaction of the catalyst with steam. It was concluded that the
steam treatment of the used catalyst increased the activity and selectivity of the catalysts.
The removal of coke with steam increased the conversion of ethylbenzene, while the
increase in the selectivity was caused by the redistribution of the potassium promoter in
the form of potassium hydroxide.
Hirano [ 11, 12 ] studied the effect of steam and CO2 treatment on the potassium-
promoted iron oxide, after 2 hours steaming at 893 K a very high initial activity ( almost 2
times higher than the activity at steady state ) was observed. The authors thought that
the change in crystal structure of the catalyst might cause the remarkable change in
activity. And they also found that the activity increased gradually at the initial stage and
reached the steady state within 20 hours. The similar results of the change of the activity
in the initial stage were also reported by Price [ 13 ] and Courty et al. [ 14 ].
In the study of MCM-41-supported iron oxide Wong et al. [ 15 ] observed also the high
initial activity over the fresh and regenerated catalysts. It was assumed that different
active centers are involved in the initial and steady stages, therefore a combination of
oxidative ( initial stages ) and non-oxidative ( steady stages ) mechanisms was proposed.
Kuhrs et al. [ 16, 17 ] studied the reaction in a single-crystal reactor. Fe3O4 was proved to
be inactive, and Fe2O3 became active after an induction period of about 45 min. Post-
reaction analysis showed that the surface was still Fe2O3 with only a small contamination
by carbon deposits. However, the defect concentration had increased considerably.
Although there are different explanations and mechanisms of the transient changes of the
catalyst, the unequivocal fact is that there is a very high initial activity over the fresh
catalyst and after steaming of the used catalyst the initial activity appears much higher
again. In this chapter, the transient behavior of the iron oxide catalysts in the initial period
is studied in detail with the aid of transient kinetic studies in combination with TPR (
Temperature-Programmed Reduction ), XRD ( X-ray Diffraction ) and TPO ( Temperature
-Programmed Oxidation ) aiming to gain a comprehensive and reasonable understanding
of the initial transient behavior of the catalyst.

4.2 Experimental
4.2.1 Catalyst preparation
All catalysts examined were prepared by wet mixing of *RHWKLWH FeOOH ), potassium
carbonate ( K2CO3 ) and suitable amounts of cerium nitrate ( Ce(NO3)3 +2O) and
chromium nitrate ( Cr(NO3)3 +2O ). Goethite was prepared by mixing of the 1M
Fe(NO3)3 and 5M KOH solutions at a ratio of 1:1.8. The resulting pastes were dried at first
in a vacuum freezer for 24 hours and then in an oven at 393 K for two hours. The
calcination was carried out at a heating rate of 5 K min-1 until 973 K for 2 hours in oxygen
( 5 ml ( STP ) min-1 ). The dried powder was pressed and crushed resulting in a sieve
fraction of 250 ~ 355 P ZKLFK ZDV XVHG IRU WKH IROORZLQJ SHUIRUPDQFH WHVW 7KH
compositions of the different catalysts are listed in Table 4.1.
Table 4.1 The compositions of the different catalysts
Catalyst Composition / wt %
Fe2O3 K2O CeO2 Cr2O3
Fe (A) 100 - - -
Fe-K (B) 99 1 - -
Fe-K (C) 90 10 - -
Fe-K-Ce ( D ) 87.3 9.7 3 -
Fe-K-Cr ( E ) 88.6 9.8 - 1.6
4.2.2 Characterization Techniques
BET surface areas of the fresh and used catalysts were determined by nitrogen
77.35 K. Before the experiment of physisorption, the samples were pretreated in nitrogen
at 873K for one hour. The results are summarized in Table 4.2.
Table 4.2 The BET results
Catalyst Surface area / m2 g-1

Fresh Used*
Fe (A) 9.7 5.9

Fe-K (B) 8.3 5.0
Fe-K (C) 6.0 2.5
Fe-K-Ce ( D ) 5.0 3.1
Fe-K-Cr ( E ) 4.8 3.1
*Reaction conditions: S/O = 10:1, LHSV = 0,43 h-1, Temp. Range: 823 ~ 898 K
X-ray diffraction ( XRD ) measurements were performed in a Siemens Diffraktometer
D500 mounted with a kristallo flex 810 X-ray generator by a rate of 0.015 / 5 s from 10
to 80 2.
The reduction behavior of the catalysts was studied using the method of temperature-
programmed reduction ( TPR ). The TPR investigations were carried out in a microreactor
flow system, which is described in detail in chapter 2. The concentration of hydrogen
before and after reaction was monitored continuously by a gas analyzer Binos 100 2M (
Fisher Rosemount ). As reducing gas 5.2 % H2 in Argon was used, and the gas mixture
was led through the reactor to a cold trap and dryer filled with silica to eliminate the water
forming during the reduction. The flow rate was 77 ml ( STP ) min-1 regulated by mass
flow controllers ( Brooks ), and the temperature was ramped up mostly at 15 K min-1. The
consumption of H2 could be calculated according to the change of the H2 concentration

and the flow rate of the reducing gas. With respect to the wet TPR investigation, the
reducing gas passed first through a saturator resulting in ca. 3 % H2O in the reducing
gas, and then to the reactor. Samples of different fresh and used catalysts with a sieve
fraction in the range from 25 to 355 m were used for the dry and wet series of TPR
investigations.
The temperature-programmed oxidation ( TPO ) experiments were carried out in the
same microreactor flow system described in the TPR investigation. The concentration of
oxygen was continuously detected by the gas analyzer Magnos 16 ( Hartmann & Braun ).
The concentrations of carbon dioxide and monoxide were detected by Uras 14
( Hartmann & Braun ). The oxidation gas was 20.5% O2 in Ar at a flow rate of 80 ml
( STP ) min-1. The oxidation temperature was usually raised at a rate of 15 K min-1.
4.2.3 Transient Experiments
The transient experiments were carried out in a microreactor flow system equipped with
four-way valve under high temperature and on-line analyzer of mass spectrometer (QMS
422). This transient reaction system has been described in detail in chapter 3. Five
different self-prepared catalysts with a sieve fraction in the range from 250 to 355 m
were investigated. Catalyst particles ( 1.0 g ) were held between two layers of quartz
particles ( < 355 m ) in a quartz tubular flow reactor of 14 mm inner diameter, and a
thermocouple ( 0.5 mm ) in a glass tube was placed in the middle of the catalyst bed. The
reactor was immersed in a fluidized bed of sand in an oven to ensure isothermal reaction
conditions. Generally, three different types of the transient experiments were carried out
in a run: the initial concentration step up experiments over the fresh catalysts, the
regenerations of the catalysts with steam at 898 K, and then the concentration step up
experiments over the used catalysts after the regeneration.
4.3 Experimental Results
4.3.1 Concentration step-up on the fresh catalyst
At 573 K helium ( 40 ml ( STP ) min-1 ) containing steam was introduced to the reactor.
Simultaneously, the temperature of the reactor was raised up at the heating rate of 15 K
min-1 from 298 K until 898 K. After 30 minutes the flow was switched to Ar ( 40 ml ( STP )
min-1 ) containing steam and ethylbenzene thus starting the concentration step-up
experiment, which reached steady state finally. The composition analysis of the educt
and product was accomplished by a mass spectrometer ( QMS 422 ) with an analyzing
interval of 6.3 s. The conversions of ethylbenzene, yields of styrene and hydrogen on
the five different catalysts are shown in Fig. 4.1.
The conversions of ethylbenzene over the different catalysts were very high at the
beginning, reaching even over 98 % during the first three minutes except for catalyst B
( X = 96.3 % ). The degrees of conversion were higher than the equilibrium conversion
under these reaction conditions ( X equilibrium = 94,2 % ). Immediately, they decreased very
fast and the conversion of unpromoted catalyst A fell to 87% in 10 minutes, and the
conversions of the other catalysts decreased to ca. 87 % after 20 minutes. Then, the
conversions of the high K-promoted catalysts C- E decreased slowly. After 3 hours, they
reached steady state at 80 ~ 83 %. The conversions of unpromoted catalyst A and low K-
catalyst B broke down abruptly after about 10 and 20 minutes, respectively. After that
they reached also slowly steady state at 62 ~ 63%, which was much lower than that of
the other high K-promoted catalysts C - E.

100
80
60
A Equilibrium
Catalyst A
40 Catalyst B
Catalyst C
Catalyst D
20 Catalyst E
0
0 50 100 150
80
60
B
Catalyst A
40 Catalyst B
Catalyst C
Catalyst D
20 Catalyst E
0
0 50 100 150
40
Yield of Hydrogen / %
30
C
20 Catalyst A
Catalyst B
Catalyst C
10 Catalyst D
Catalyst E
0
-30 0 30 60 90 120 150 180
Relative Time / min
Fig.4.1. The initial step-up experiments over the fresh catalysts A-E at 898 K, LHSV = 0.42 h-1,
S/O = 10:1: ( A ) The conversion of ethylbenzene vs. relative time, ( B ) the yield of
styrene vs. relative time, ( C ) the concentration of hydrogen vs. relative time.
Over the high K-promoted catalysts C, D and E the yields of styrene reached the
maximal values 82 ~ 83 % after 6 ~ 8 minutes reaction. Then the yields of styrene
decreased a little. Finally, they reached steady state at 76.5 ~ 78 %. Over the
unpromoted catalyst A, the yield of styrene reached maximal value of 78,8% after 10
minutes, then began to fall down until 62,3 %. Over the low K-promoted catalyst B the
yield of styrene reached 83,5 % after 4 minutes, and then decreased to 62,5 % after ca.
1 hour. In general, all the catalysts had very high initial activities at the first several
minutes and then deactivated rapidly. Obviously, the unpromoted and low K-promoted
catalysts reached steady state at much lower conversions of ethylbenzene and yields of
styrene than those of the high K-promoted catalysts.
The yields of hydrogen of the different catalysts were almost zero at first, then they rose
abruptly. In ten minutes, most of them reached maximal values and immediately
decreased a little. In chapter 3, it was already suggested that the unique change of the
conversion of ethylbenzene and of the yields of styrene and H2 during the initial activation
period was due to the coupling of the dehydrogenation of ethylbenzene and the
reduction of Fe2O3,
C6H5C2H5 = C6H5C2H3 + H2 (1)
3Fe2O3 ( haematite ) + H2 = 2Fe3O4 ( magnetite )+ H2O (2)
It was considered that the dehydrogenation of ethlybenzene produced H2 , which reduced
haematite, and the consumption of H2 during the reduction of Fe2O3 led to a new
equilibrium conversion of the reaction. Furthermore, the reduction of haematite to
magnetite partly resulted in the initial activity decrease of the catalyst.
The changes of the concentrations of CO and CO2 in the initial period are shown in
Fig.4.2A and Fig.4.2B. At the beginning there was a great jump of the concentration of
CO2, then it decreased. Evidently, the concentration of CO2 over the unpromoted catalyst
A and the low K-promoted catalyst B decreased much more slowly than those over the
high K-promoted catalysts C - E.
0.15

Concentration of CO / %
0.10
A
Catalyst A
0.05 Catalyst B
Catalyst C
Catalyst D
Catalyst E
0
0 50 100 150
Concentration of CO2 / %
2
Catalyst A
B Catalyst B
Catalyst C
1 Catalyst D
Catalyst E
0
-40 0 40 80 120 160
Relative Time / min

Fig.4.2. The initial step-up experiments over the fresh catalysts A-E at 898 K, LHSV = 0.42 h-1,
S/O = 10:1: ( A ) The concentration of CO vs. relative time, ( B ) the concentration of CO2
vs. relative time.
4.3.2 The regeneration of the catalyst with steam
The regeneration experiment was carried out as follows: at 898 K the reaction was run for
several hours under steady state. Then, the Ar flow containing steam and ethylbenzene
was switched to the He flow containing only steam. The same experiments were carried
out over catalysts A E. The concentrations of hydrogen are shown in Fig.4.3A, and
those of CO and CO2 are shown in Fig.4.3B and Fig.4.3C, respectively. The
concentrations of the product hydrogen over catalysts A E did not reach zero during
the first 30 minutes. This may indicate that magnetite is reoxidized:
2 Fe3O4 + H2O = 3 Fe2O3 + H2 (3)
The concentrations of CO and CO2 varied differently. At first, the concentrations of CO
reached ca. 0.02 % very fast, and then they decreased. During the regeneration with
steam carbonaceous deposits formed during the reaction are converted to CO and
hydrogen according the following reaction:
C + H2O = CO + H2 (4)
Furthermore, carbon monoxide reacted with steam forming carbon dioxide and hydrogen
as follows:
CO + H2O = CO2 + H2 (5)
Over the catalysts A E, the concentrations of CO2 dropped sharply during the first
minutes, and then they varied in a different way. Over the unpromoted and low K-
promoted catalysts, the concentrations of CO2 decreased continuously, and reached
zero at 11 minute and 17 minute, respectively. Contrarily, the concentrations of CO2 over
the high K-promoted catalysts decreased to ca. 0.085 % during the first 30 minutes, then
they stabilized at this level for 50 minutes over catalyst C and E, and for 110 minutes over
catalyst D. After that, they began to decrease again, and reached zero in ca. 40
minutes.
1.2 catalyst A
catalyst B
catalyst C
catalyst D
0.9
catalyst E
A
0.6
0.3 He+H2O Argon + H2O + Ethylbenzene
0.0
-50 0 50 100 150
catalyst A
catalyst B
catalyst C
0.02 catalyst D
catalyst E
B
0.01
0.00
-50 0 50 100 150

catalyst A
catalyst B
0.20 catalyst C
catalyst D
catalyst E
0.15
C
0.10
0.05
0.00
-40 0 40 80 120 160
Relative Time / min
Fig.4.3. The regeneration of the used catalysts A-E at 898 K, S/O = 10:1: ( A ) The concentrations
of hydrogen vs. relative time, ( B ) the concentrations of CO vs. relative time, ( C ) the
concentrations of CO2 vs. relative time.

4.3.3 Concentration step-up after the regeneration of the used catalyst
The investigations of the used catalysts after regeneration were carried out in the
following way: after 10 hours reaction at 898 K, the He flow containing steam was
introduced to the reactor; after 3 hours the flow was switched to Ar containing
ethylbenzene and steam, thus starting the reaction again. These experiments were
carried out over catalysts A E. The conversions of ethylbenzene, the yields of styrene
and hydrogen are shown in Fig.4.4.
The conversions of ethylbenzene over the high K-promoted catalysts C, D and E were
high at the beginning, the degrees of conversion over catalysts D and E were higher than
Xequilibrium ( 94.2 % ). The conversions of ethylbenzene over the unpromoted and low
promoted catalysts A and B reached only 77 % at the beginning, much lower than those
obtained on the fresh catalysts. After that, they decreased very fast, after 10 minutes the
conversions of catalysts A B and catalysts C E decreased to 77 % and 81 %,
respectively, then the conversions decreased more slowly. After ca. 2 hours, the
conversions of catalysts C, D and E stabilized at 81 % and 77 %. The conversions of
catalysts A and B reached steady state at 69 %. Over the high K-promoted catalysts C
E, the yields of styrene reached the maximal value of 82 % at the beginning. Then,
they decreased a little, reaching steady state at ca. 77 %. The yields of styrene over the
100
80
60
A Equilibrium
Catalyst A
Catalyst B
40
Catalyst C
Catalyst D
Catalyst E
20
0
0 50 100 150
80
60
B
40 Catalyst A
Catalyst B
Catalyst C
20 Catalyst D
Catalyst E
0
0 50 100 150
50
Yield of Hydrogen / %
40
C
30
Catalyst A
20 Catalyst B
Catalyst C
10 Catalyst D
Catalyst E
0
-40 0 40 80 120 160
Relative Time / min
Fig.4.4. The step-up experiments after regeneration of the catalysts A-E at 898 K, LHSV = 0.43
h-1, S / O = 10:1: ( A ) The conversion of ethylbenzene vs. relative time, ( B ) the yield of
styrene vs. relative time, ( C ) the concentration of hydrogen vs. relative time.
unpromoted catalyst A and low promoted catalyst B reached the maximal value of 76 %
and 66 %, respectively. After 0.5 hour they stabilized at ca. 64 %. After the regeneration
of the catalysts with steam, all the catalysts had very high initial activities at the first
minutes again, and then they deactivated immediately. However, the initial high activities
of the used catalysts after steaming are lower than those of the fresh catalysts, and in
much shorter time, the conversions of ethylbenzene and the yields of styrene reached
steady state at the same levels as those obtained over the fresh catalysts.
The yields of hydrogen with the different catalysts were almost zero at first, then they
jumped up immediately. In a few minutes, most of them reached the maximal values and
then decreased a little. This means that the degrees of reduction of the catalysts are
lower than those during the concentration step-up experiments with the fresh catalysts.
Furthermore, in the initial period after the regeneration of the catalysts there were no
jumps of the concentrations of CO2, and these were different from the investigations on
the fresh catalysts. The results are shown in Fig.4.5.

0.16 He+H2O Argon + H2O + Ethylbenzene

0.12
Catalyst A
Catalyst B
A 0.08 Catalyst C
Catalyst D
Catalyst E
0.04
0.00
0 50 100 150
Catalyst A
1.6
Catalyst B
Catalyst C
Catalyst D
1.2
Catalyst E
B
0.8
0.4
0.0
-40 0 40 80 120 160
Relative Time / min
Fig.4.5. The step-up experiments after regeneration of the catalysts A-E at 898 K, LHSV = 0.43
h-1, S / O = 10:1: ( A ) The concentration of CO vs. relative time, ( B ) the concentration
of CO2 vs. relative time.
4.3.4 X-ray Diffraction
X-ray diffraction analysis was performed on the fresh catalysts A, C, D and E. The results
are displayed in Fig.4.6. The diffraction pattern of the unpromoted catalyst A is
assigned to -Fe2O3. The diffraction patterns of the high K-promoted catalysts C-E
exhibited three small additional peaks ( K ) in addition to the peaks belonging to -Fe2O3.
of these peaks are due to the amorphous character of KFeO2, since it was exposed to the
ambient air for some time [ 19 ]. Although there were small amounts of CeO2 and CrO2 in
the catalysts D and E, the diffraction patterns belonging to CeO2 and CrO2 were not
observed.
H
2500 H
H H
H
2000
a H H
H H
Intensity (a.u.)
1500 H H
H
c K K K H H
H
1000 H
H
H H
d K K K H
500 H H
H
H H
e KK K H H
0
30 40 50
Fig.4. 6. XRD patterns of the different fresh catalysts: (a) 0 % K2O, 100 % Fe2O3 , (c) 10 %
K2O, 90 % Fe2O3 , (d) 9.7 % K2O, 87.3 % Fe2O3 , 3 % CeO2 , (e) 9.8 % K2O, 88.6 %
Fe2O3 , 1.6 % Cr2O3.
After a run, a catalyst was cooled down to room temperature in helium and was
transferred immediately to a glove box ( O2 < 1 ppm, H2O < 1 ppm ). The XRD patterns
of the used catalysts are shown in Fig.4.7. The diffraction patterns of the four
catalysts indicate that Fe3O4 is the major crystalline component ( magnetite ). The high K-
promoted catalysts C-E have a peak at 32 ( CoK radiation ) which indicates the
presence of KFeO2 as a second phase. In the catalyst C, there is an additional peak at
28, which suggests the third phase CeO2.
M
2500
M
M
2000 M
a
M
Intensity (a.u.)
1500 M
M
K M
c
1000
M
M
Ce K M Ce
d
500 M
M
M
M
e K
0
30 40 50
2
Fig.4.7. XRD patterns of the different used catalysts: (a) 0 % K2O, 100 % Fe2O3 , (c) 10 %
K2O, 90 % Fe2O3 , (d) 9.7 % K2O, 87.3 % Fe2O3 , 3 % CeO2 , (e) 9.8 % K2O, 88.6 %
Fe2O3 , 1.6 % Cr2O3.
4.3.5 Temperature-programmed Reduction
The TPR experiments were carried out with the different fresh and used catalysts in dry
H2 / argon and wet H2 / argon. The patterns shown in Fig.4.8 were obtained for the
fresh catalysts in dry H2 / argon. Quantitative analysis of these TPR patterns revealed
that the catalysts could be completely reduced to metallic iron ( Fe0 ). Comparing
patterns a and b showed that the only one main peak appeared and the presence of
small amounts of potassium caused the pattern to shift to lower temperature. When the
catalysts were promoted with more potassium and other oxides such as CeO2 and Cr2O3,
two reduction peaks were displayed.
Figure 4.9 shows the TPR patterns of the used catalysts in dry H2/ Argon. Pattern a (
unpromoted ) and pattern b ( low K-promoted ) had a shoulder, respectively. Comparing
the patterns c-e in Fig.4.9 and 4.8, the Tmax of the first peaks were increased by 20-40K,
and Tmax of the second peaks remained almost the same.

-7
4.0x10
939 K
-7
2.0x10
a
H2 Consumption / mol s-1
0.0
0.0000004 400 800 1200
901 K
0.0000003
0.0000002
0.0000001 b
0.0000004
0.0000000 400 600 800 1000 1200 1400
0.0000003 873 K 1008 K

0.0000002
0.0000001 c
0.0000000 0.0000004
400 600 800 1000 1200 1400
0.0000003
0.0000002 858 K 1032 K

0.0000001
-7
d
4.0x10
0.0000000
400 600 800 1000 1200 1400
-7
866 K 1021 K
2.0x10
e
0.0
400 600 800 1000 1200 1400
Temperature / K
Fig.4.8. TPR patterns of the different catalysts in dry H2 / argon at the same heating rate of 15 K /
min : (a) 0 % K2O, 100 % Fe2O3 ( m = 10.6 mg ) (b) 1 % K2O, 99 % Fe2O3 ( m = 10.6
mg ), (c) 10 % K2O, 90 % Fe2O3 ( m = 10.0 mg ), (d) 9.7 % K2O, 87.3 % Fe2O3 , 3 %
CeO2 ( m = 10.2 mg ), (e) 9.8 % K2O, 88.6 % Fe2O3 , 1.6 % Cr2O3 ( m = 10.1 mg).
-7
4.0x10
860 k 969 k
-7
2.0x10
a
0.0000004
0.0 400 800 1200
-1
0.0000003 969 k
862 k
0.0000002
0.0000001 b
0.0000004
0.0000000 400 600 800 1000 1200 1400
889 k 736 k
0.0000003
0.0000002
0.0000001 c
0.0000000 0.0000004
400 600 800 1000 1200 1400
0.0000003
904 k
0.0000002 1031 k
0.0000001
d
0.0000000
-7
4.0x10 400 600 800 1000 1200 1400
891 k
2.0x10
-7 1010 k
e
0.0
400 600 800 1000 1200 1400
Temperature / K
Fig.4.9. TPR patterns of the different used catalysts in dry H2 / argon at the same heating rate of
15 K / min: (a) 0 % K2O, 100 % Fe2O3 ( m = 10.2 mg ), (b) 1 % K2O, 99 % Fe2O3 ( m =
10.3 mg ), (c) 10 % K2O, 90 % Fe2O3 ( m = 10.3 mg ), (d) 9.7 % K2O, 87.3 % Fe2O3 , 3
% CeO2 ( m = 9.9 mg ), (e) 9.8 % K2O, 88.6 % Fe2O3 , 1.6 % Cr2O3 ( m = 10.2 mg )
It is known that the dehydrogenation of ethylbenzene to styrene is carried out in excess of
steam and the TPR patterns are highly affected by H2O added in the reducing gas [ 20,
21 ]. Thus, two TPR series for the fresh and used catalysts in wet H2 / argon were
performed. Figure 4.10 shows the TPR patterns for the fresh catalysts at the rate of 15 K
min-1 from 423 K up to 1188 K. In comparison with the TPR series in dry H2 / argon, the
patterns a and b were resolved into two-peak patterns. The quantitative analysis showed
that the consumption of H2 under the two peaks was only 1/3 of the total theoretical
consumption of H2 for the reduction of Fe3+ to Fe0. Furthermore, the ratio of the hydrogen
consumption in the first and second reduction peak is ca. 1:1.6, in comparison with the
theoretical value of 1:2, which may indicate the sequential reduction of Fe2O3 via Fe3O4
to FeO,
3 Fe2O3 + H2 = 2 Fe3O4 + H2O (6)
Fe3O4 + H2 = 3 FeO + H2 (7)
Figure 4.11 shows the TPR patterns for the used catalysts in wet H2/argon, which display
one peak. Considering the XRD results, which demonstrated that the used catalysts
were mostly composed of Fe3O4, the reduction of the used catalysts in wet H2 / argon
may occur as follows:
Fe3O4 + H2 = 3 FeO + H2O. (8)

-7
2.0x10
937 K
815 K
-7
1.0x10
a
0.0
0.0000002 400 800931 K 1200
-1
812 K
0.0000001
b
0.0000002
0.0000000 400 600 800 919 K
1000 1200 1400
0.0000001 818 K
c
0.0000000 0.0000002
400 600 800 1000 1200 1400
942 K
828 K
0.0000001
d
-7
0.0000000
2.0x10 400 600 800922 K 1000 1200 1400
826 K
-7
1.0x10
e
0.0
400 600 800 1000 1200 1400
Temperature / K
Fig. 4.10. TPR patterns of the different fresh catalysts in wet H2 / argon at the same heating rate
of 15 K min-1 : (a) 0 % K2O, 100 % Fe2O3 ( m = 10.3 mg ), (b) 1 % K2O, 99 % Fe2O3 (
m = 10.1 mg ), (c) 10 % K2O, 90 % Fe2O3 ( m = 10.0 mg ), (d) 9.7 % K2O, 87.3 %
Fe2O3 , 3 % CeO2 ( m = 9.9 mg ), (e) 9.8 % K2O, 88.6 % Fe2O3 , 1.6 % Cr2O3 ( m = 9.9
mg ).
-7
2.0x10
928 K
-7
1.0x10
a
0.0
-1
0.0000002 400 800 1200

964 K
0.0000001
b
0.0000002
0.0000000 400 600 800 934 K
1000 1200 1400
0.0000001
c
0.0000000 0.0000002
400 600 800 1000 1200 1400
941 K
0.0000001
-7
d
2.0x10 0.0000000
400 600 800 1000 1200 1400
946 K
-7
1.0x10
e
0.0
400 600 800 1000 1200 1400
Temperature / K
Fig.4.11 TPR patterns of the different used catalysts in wet H2/Ar at the same heating rate of 15 K
min-1: (a) 0 % K2O, 100 % Fe2O3 ( m = 10.0 mg ), (b) 1 % K2O, 99 % Fe2O3 ( m = 10.1
mg ), (c) 10 % K2O, 90 % Fe2O3 ( m = 10.1 mg ), (d) 9.7 % K2O, 87.3 % Fe2O3 , 3 %
CeO2 ( m = 9.9 mg ), (e) 9.8 % K2O, 88.6 % Fe2O3 , 1.6 % Cr2O3 ( m = 10.2 mg ).
4.3.6 Temperature-programmed Oxidation
The temperature-programmed oxidation ( TPO ) was carried out to determine the nature
of the carbonaceous material deposited during the test reaction. The TPO patterns of the
O2 consumption for the used catalysts are shown in Fig.4.12. The patterns a (
unpromoted ) and b ( low K-promoted ) consisted of one peak at 587K and 596K,
respectively. The other patterns c-e ( high K-promoted ) had a main peak at ca. 589 K
and a small peak at ca. 905 K.
Figure 4.13 shows the TPO patterns of the CO2 formation for the used catalysts. Pattern
a exhibited no formation of CO2, and pattern b showed small amounts of CO2 formation.
Contrarily, the patterns c-e displayed large peaks of CO2 formation. If the CO2 was
produced through the following reaction,
C + O2 = CO2 (9)
then, the consumption rate of O2 should be the same as the formation rate of CO2. This
means that the O2 consumption and the CO2 formation should occur simultaneously. In
contrast, the temperature of the CO2 peak is higher by ca. 300K than that of the O2
consumption.
-7
1.2x10
587 K
-8
6.0x10
a
0.0
596 K
-1
400 800 1200
0.0000001
O2 Consumption / mol s
b
0.0000000 200 400 600
581 K 800 1000 1200
0.0000001
895 K
c
0.0000000 0.0000001
200 400 600 800 1000 1200
598 K
894 K
-7
d
1.2x10 0.0000000
200 400 600 800 1000 1200
589 K
-8
6.0x10 926 K
e
0.0
200 400 600 800 1000 1200
Temperature / K
Fig. 4.12. TPO patterns of the consumption of O2 on the different used catalysts at the heating
rate of 15 K / min: (a) 0 % K2O, 100 % Fe2O3 ( m = 10.0 mg ), (b) 1 % K2O, 99 %
Fe2O3 ( m = 10.1 mg ), (c) 10 % K2O, 90 % Fe2O3 ( m = 10.1 mg ), (d) 9.7 % K2O,
87.3 % Fe2O3 , 3 % CeO2 ( m = 9.9 mg ), (e) 9.8 % K2O, 88.6 % Fe2O3 , 1.6 % Cr2O3
( m = 10.2 mg ).
-7
1.2x10
-8
6.0x10
a
0.0
1.20E-007 400 800 1200
-1
CO2 formation / mol s
6.00E-008
749 K
b
1.20E-007
0.00E+000 200 400 600 800 1000 1200
895 K
6.00E-008
c
0.00E+000 1.20E-007
200 400 600 800 1000 1200
916 K
6.00E-008
d
-7
1.2x10 0.00E+000
200 400 600 800 1000 1200
931 K
-8
6.0x10
e
0.0
200 400 600 800 1000 1200
Temperature / K
Fig.4.13. TPO patterns of the formation of CO2 on the different used catalysts at the heating rate
of 15 K / min: (a) 0 % K2O, 100 % Fe2O3 ( m = 10.0 mg ), (b) 1 % K2O, 99 % Fe2O3 (
m = 10.1 mg ), (c) 10 % K2O, 90 % Fe2O3 ( m = 10.1 mg ), (d) 9.7 % K2O, 87.3 %
Fe2O3 , 3 % CeO2 ( m = 9.9 mg ), (e) 9.8 % K2O, 88.6 % Fe2O3 , 1.6 % Cr2O3 ( m =
10.2 mg ).
Furthermore, a quantitative analysis of the TPO patterns is performed. The peak areas of
the O2 consumption for the pattern a and b are 4.40410-5 mol and 5.17210-5 mol,
respectively. According to equation 9, the peak areas of the CO2 formation should have
similar values, however, they are zero for pattern a and 0.408 mol for pattern b. Now, one
question is if O2 is consumed not only by the reaction with the carbonaceous material
deposited on the used catalyst, then, what has consumed O2? Considering the results of
the steaming of the used catalysts, it may be proposed that in addition to the reacting with
C, O2 should take part in the oxidation of Fe3O4 to Fe2O3 as expressed as the following:
2 Fe3O4 + O2 = 3 Fe2O3 ( 10 )
According to the above reaction, the theoretical values of O2 consumption are 5.37710-5
mol for catalyst A and 5.30710-5 mol for catalyst B, which agree well with the
experimental results. The theoretical O2 consumptions for the high K-promoted catalysts
C-E are calculated, and they are in good agreement with the experimental values for the
first peak. The theoretical and experimental O2 consumption for the TPO over the used
catalysts are summarized in Table 4.4.

Table 4.4 The consumption of O2 during the TPO over the used catalysts
Composition Experiment (mol /10-5) Calculation (mol /10-5)

of K2O
(wt %) 1st Peak 2nd Peak
Catalyst A 0 5.99 5.38
Catalyst B 1 5.17 5.31
Catalyst C 10 4.63 0.033 4.61
Catalyst D 9.7 4.22 0.027 4.65
Catalyst E 9.8 4.76 0.16 4.61
Comparing Fig 4.12 and Fig.4.13, the CO2 formation and O2 consumption for the second
peak have the same Tmax over the high K-promoted catalysts C-E. It seems that the O2
consumption for the second peak is ascribed to the CO2 formation based on the reaction
9. However, the quantitative analysis of the TPO patterns showed that the amounts of the
O2 consumption for the second peak were much lower than that of CO2 formation. For
example, the O2 consumption over catalyst D was 0.0272, but the CO2 formation was
3.137. It was also observed that the amount of the CO2 formation over the low K-
promoted catalyst B was only 0.408 mol, ca 1 / 9 of that over high K-promoted catalysts.
This ratio of the CO2 formation between the low K-promoted and high K-promoted
catalysts is almost the same as the ratio of the K-content between them. Therefore, for
the formation of CO2, in addition to the oxidation of the carbonaceous material deposited
on the catalysts, the following reaction can be proposed:
K2CO3 = K2O +CO2 ( 11 )
Based on the above reaction, the amounts of the decomposition of K2CO3 in the catalysts
were calculated. The experimental and calculated CO2 formation for the TPO of the used
catalysts are listed in Table 4.5.
Table 4.5 The formation of CO2 during the TPO over the used catalysts
Composition Experiment Calculation

of K2O
mol /10-5 mol /10-5
wt %
Catalyst A 0 0 0
Catalyst B 1 0.41 0.54
Catalyst C 10 3.32 5.22
Catalyst D 9.7 3.14 5.13
Catalyst E 9.8 4.58 5.14
It is seen that the theoretical values of CO2 formations over the high K-promoted catalysts
are greater than those of experimental values. The reason may be that the active phase
KFeO2 was not completely converted into potassium carbonate during the
dehydrogenation of ethylbenzene to styrene.
4.4 Discussion
4.4.1 Reduction of Fe2O3
The fresh catalyst ( nominally Fe2O3, red ) turned to black after the dehydrogenation of
ethylbenzene to styrene. The XRD patterns of the used catalysts [ 11, 22, 23, 24 ]
showed the magnetite ( Fe3O4 ) phase. Lee considered that Fe3O4 was the expected
state of the iron oxide and an equilibrium of the iron oxide system with hydrogen and
steam was suggested:

+2 +2
)H2 ( 12 )
+ +
According to this equation, Fe3O4 could be further reduced to FeO. FeO is unstable and
would be oxidized to Fe3O4 when it is exposed to air, therefore this hypothesis is not
contradictory to the results obtained by XRD, which indicate that Fe3O4 is the main
component of the catalyst. A phase diagram of the iron-iron oxide system in the mixtures
of H2 and H2O is shown in Fig.4.14. The equilibrium points were calculated based on the
data in ref. [ 25 ].
Obviously, according to the phase diagram, the final state of the iron oxide in the catalyst
will be Fe3O4. The following results of the TPR confirmed the theoretical phase diagram
calculation. A wet TPR was carried out on the pure Fe2O3 ( Merck ). The result is shown
in Fig.4.15.
6
Fe2O3
Fe3O4
2
log PH O / PH
( T = 898 K, at Reaction )
2
2
( T = 898 K, TPR in
wet H2 in Argon ) FeO
0
-2 Fe
400 600 800 1000 1200 1400 1600
Temperature / K
-7
2,0x10
965 K
-1
-7
1,6x10
-7
1,2x10
714 K
-8
8,0x10
-8
4,0x10
0,0
400 500 600 700 800 900 1000 1100 1200
Temperature K
Fig.4.15. TPR patterns of pure Fe2O3 in wet H2 / argon at the heating rate of 15 K min-1.
The first peak is attributed to the reduction of Fe2O3 to Fe3O4 and the second to the
further reduction of Fe3O4 to FeO. Fig.4.15 also showed that the reduction of Fe3O4 to
FeO began at 913 K, which was higher than the reaction temperature, indicating that
under the reaction conditions Fe2O3 was only reduced to Fe3O4. The steam retarded the
further reduction of Fe3O4 to FeO and Fe. Now, one question about the reduction is
whether there is such a phase equilibrium between Fe2O3 and Fe3O4. All the XRD
patterns showed that Fe3O4 is the only iron phase present in the used catalysts. Thus,
there is the possibility, that Fe2O3 under this case is amorphous, and hence undetectable
by XRD test. Following evidence may support this hypothesis: A pure Fe3O4 sample was
calcined at 973 K under a flow of N2. After calcination for 4 hours, the surfaces of some
pellets were turned to light red because of the traces of oxygen in the flow, indicating that
Fe3O4 was partly oxidized to Fe2O3. However, the XRD pattern of such a sample was
identical to that of pure Fe3O4. To solve this question, a TPR experiment under the
conditions similar to the reaction was performed. The reduction temperature was raised
at a rate of 15 K min-1 up to 898 K, then the reduction continued at the same temperature
to insure the complete reduction of Fe2O3. The TPR result is shown in Fig.4.16. The
consumed amount of H2 was obtained by the integration of the TPR pattern. The
experimental value is 2.38810-5 mol, a little greater than the theoretical value of 2.10
10-5 mol for the reduction of Fe2O3 to Fe3O4. Based on this result, it is considered that the
reduction of Fe2O3 to Fe3O4 under the dehydrogenation of ethylbenzene to styrene is
complete.
-7
1.2x10
-1
-8
9.0x10
-8
6.0x10
-8
3.0x10
0.0
0 10 20 30 40 50 60
Time / min
Fig.4.16. TPR pattern of the fresh catalyst A in wet H2 / argon at the heating rate of 15 K min-1 up
to 898 K.
4.4.2 Effect of the steaming
The steam treatment of the used catalysts increased the activity of the catalysts [ 10, 12,
19 ]. We observed the same phenomena in the transient experiment after steaming of the
catalysts. As mentioned in the introduction, the interpretation of the effect of steaming is
not consistent for the same experimental investigation. Herzog et al. [ 10 ] considered
there was no change in the oxidation state or crystallite structure after steaming.
Contrarily, Hirano [ 12 ] suggested that the formation of crystalline KFeO2 accelerated by
steaming was responsible for the increase of the activity. The transient experimental
results in Fig.4.3 C showed the decomposition of K2CO3, which supported Hiranos
opinion.
Whether there is a change of the oxidation state of the catalyst should be further
discussed. The conclusion drawn by Herzog was based on the XRD results, which
showed that the used catalysts before and after steaming had the same XRD patterns
corresponding to Fe3O4. We observed the same results. However, we thought that only
based on the results of XRD, it can not be concluded that there is no change in the
oxidation state. As the additional evidence, a series of TPR experiments were carried out
as follows: In wet H2 / argon the reduction temperature was raised at a rate of 15 K min-1
up to 898 K ( the same reaction temperature of the dehydrogenation of ethylbenzene ).
Then the reduction continued at 898 K for 1 hour to ensure that the complete reduction of
Fe2O3. The result is displayed in Fig.4.17 A. After that, the reducing gas was switched to
argon containing 3 % H2O, thus starting the regeneration of the reduced catalyst. After 3
hour - steaming, the argon containing 3 % H2O was switched to wet H2 in argon again.
The reduction in wet H2 / argon was performed for 1 hour as shown in Fig. 4.17 B. If there
is no oxidation state change of the catalyst, there should be no reduction peak after 3
hours steaming of the reduced catalyst. Contrarily, Fig.4.17 B shows clearly the reduction
peaks, and such an observation was reproduced as shown in Fig.4.17 C. Therefore, we
assume that the steaming changed the oxidation state of the used catalysts, although it
can not completely regenerate the oxidation state of the fresh catalyst comparing
Fig.4.17 A and Fig.4.17 B. Furthermore, it is seen that steaming more than once seems
to result in the same oxidation state as shown in Fig. 4.17 B and 4.17 C.
-8
8.0x10
-8
-1
4.0x10
8.00E-008 A
0.0
0 5 10 15 20 25 30 35 40 45 50 55 60
4.00E-008
0.00E+000
B
0 10 20 30 40 50 60
-8
8.0x10
-8
4.0x10
C
0.0
0 10 20 30 40 50 60
Time / min
Fig.4.17. TPR patterns of the catalysts A in wet H2 / argon at the heating rate of 15 K min-1 up to
898 K: (A ) Fresh catalyst, ( B ) the reduced catalyst after 3 hour-steaming, ( C ) the
repeated experiment ( B ).
4.4.3 K2CO3 in the K-promoted catalyst
The used K-promoted catalyst exhibits such a diffraction pattern as shown in Fig.4.18B:
between two intensive peaks of Fe3O4 at 30 and 36, there is a peak at 32.5. In
comparison to the high-resolution diffraction pattern of K2CO3 as shown in Fig.4.18A, the
peak at 32.5 may be assigned to K2CO3.
160
140
120
100
80
60
60
40
50
A
20
400
30 35
30
Intensity ( a.u. )
20
10
0
B
1
6x10 28 30 32 34 36 38
1
5x10
1
4x10
1
3x10
1
2x10
1x10
1 C
0
28 30 32 34 36 38
2
Fig.4.18. High-resolution profiles of the diffraction patterns: ( A ) K2CO3, ( B ) used K-promoted
catalyst, ( C ) Fe3O4.
Since the diffraction pattern of KFeO2 is almost identical to that of K2CO3, it cannot be
simply confirmed that K2CO3 is present, because the XRD was carried out ex situ,
although the samples were kept in the glove box ( H2O < 1 ppm, O2< 1 ppm ).
As observed in the TPO experiments, the O2 consumption peak temperature is different
from the peak temperature of CO2 concentration. A possible interpretation is that CO2 is
formed not only by the gasfication of carbonaceous deposits but also by the
decomposition of K2CO3. Aiming at studying the decomposition of K2CO3, a
temperature-programmed experiment in argon was carried out over a fresh K-promoted
catalyst. The temperature of the reactor was raised at 15 K min-1 up to 898 K, and then
kept at this temperature until no CO2 was released. The result is shown in Fig.4.19.
CO
Concentrations of CO and CO2 / %
0.12 0.12
CO2
0.09
0.08
0.06
0.04
0.03
0.00 0.00
0 5 10 15 20 25 30 35 40 45 50 55 60
Time / min
Fig.4.19. Temperature-programmed experiment in argon at the heating rate of 15 K min-1 up to
898 K on the fresh K-promoted catalyst.
The release of CO2 began at 573 K. Two peaks for the formation of CO2 at 744 K and at
898 K were observed. Since there is no gasification of carbonaceous deposits, hence, the
formation of the large quantity of CO2 indicates the decomposition of K2CO3 that exists in
the fresh K-promoted catalyst.
Does K2CO3 also exist in the catalyst during the reaction? It seems necessary to analyze
the in situ results of the transient experiments. The CO2 concentration variations over the
unpromoted and K-promoted catalysts in the transient experiments are shown in
Fig.4.20.
In Fig.4.20 A, the unpromoted catalyst shows a much large formation of CO2 than the K-
promoted catalyst, which agrees well with the results reported by Stobb et al. [ 26 ]. The
relatively lower initial CO2 yield over the high K-promoted catalyst infers the reaction
between K2O and CO2. In Fig.4.20 B it is shown that a greater quantity of CO2 is formed
by the K-promoted catalyst than by the unpromoted one upon steaming. In Fig.4.20C,
the K-promoted catalyst shows a decreased initial CO2 yield, and the CO2 yield of the
unpromoted catalyst reaches steady state immediately after the switching of He flow
containing steam to the Ar flow containing EB and steam. It is due to the partial recovery
of the activity of the catalyst during the steaming, which results in low deposition of
carbonaceous material on the catalysts, and hence a low initial formation of CO2. In
summary, the transient experimental results of CO2 can be taken as an in situ
confirmation of K2CO3 existing in the catalysts under reaction conditions.

4
Concentraiton of CO2 / %
3
A 2
Catalyst A
Catalyst C
0
0 50 100 150
Reaction He+H2O
0.15
0.10
B Catalyst A
Catalyst C
0.05
0
0 50 100 150
0.15
C 0.10
Catalyst A
0.05
Catalyst C
0.00
-40 0 40 80 120 160
Relative Time / min
Fig.4.20. The concentration of carbon dioxide of the different transient experiments on
unpromoted catalyst A and K-promoted catalyst C vs relative time: ( A ) The new step-
up transient experiment, ( B ) the steaming of the used catalyst, ( C ) the second step-
up after the steaming.

4.4.4 Decomposition of KFeO2
The delay of the activity of catalysts by addition of CO2 in the feed was confirmed by
several authors [ 12, 27, 28 ]. Hirano interpreted this effect based on the decomposition
of KFeO2:
2 KFeO2 + CO2 = K2CO3 + Fe2O3 ( 13 )
This interpretation was further confirmed by our transient experiments. The conversions
of ethylbenzene and the CO2 yields of the unpromoted and K-promoted catalysts after the
steaming are shown in Fig.4.21.
The activity of the unpromoted catalyst decreased rapidly at the initial stage and reached
the steady state in a short time due to the formation of Fe3O4 from Fe2O3 regenerated
during the steaming. The decline of the activity of K-promoted catalyst occurred much
more slowly, which may be caused by the reduction of Fe2O3 and the decomposition of
the active phase KFeO2 by CO2. The question is whether KFeO2 decomposes completely
or partially. The integration of the profile of CO2 consumption shows that 1.2610-4 ml
CO2 are consumed during the initial stage according to the reaction 13. This amount of
CO2 consumption means that 12 % KFeO2 is decomposed in the initial stage. Moreover,
the CO2 concentration affected the delay of the activity of the K-promoted catalysts [ 20,
28 ]. The higher concentration of CO2 in the flow led to more significant delay of the
activity, implying that the decomposition of KFeO2 would finally reach an equilibrium
when the CO2 concentration was constant. It is therefore confirmed that KFeO2
decomposed partially due to the reaction with CO2, and the CO2 concentration
determines the decomposition degree of KFeO2.

Conversion of ethylbenzene / %
90
80
A 70
60
Catalyst A
50 Catalyst C
0
0 50 100 150
-6
CO2 Consumption / mol min 10
6
-1
B 4
Catalyst A
2
Catalyst C
0
-40 0 40 80 120 160
Relative Time / min
Fig.4.21. The conversions of ethylbenzene and CO2 consumption of the unpromoted catalyst A
and high K- promoted catalyst C after steaming of the catalyst vs. relative time.
4.4.5 Schematic life cycle of the catalyst
Based on the model suggested by Muhler et al.[ 19 ], a modified life cycle of the K-
promoted styrene catalyst is schematically displayed in Fig.4.22. All precursors contain -
Fe2O3 and X-ray crystalline K2Fe22O34. If the precursor is subjected to reaction conditions,
the K2Fe22O34 phase becomes unstable due to the partial reduction of the iron oxide to
magnetite. During this process of partial reduction, the synthesis of KFeO2 occurs at the
edges of the matrix crystals.
In the initial stage, the conversion of ethylbenzene is very high, even over 95 %. Fe2O3 is
reduced completely to Fe3O4 by the simultaneously produced H2 during the
dehydrogenation of ethylbenzene. The presence of water vapor stabilizes the iron oxide
against total reduction to iron. The high initial activity of the catalyst leads to the high
carbon deposition rate, which in turn increases the gasification producing much more
CO2. In this process the active phase KFeO2 is partially converted to K2CO3. The
reduction of Fe2O3 and decomposition of KFeO2 decrease the activity of the catalyst.
After some hours the reaction reaches steady state. The K-promoted catalyst under the
steady reaction condition is considered to be a mixture of KFeO2, Fe3O4 and K2CO3.
The catalyst can be regenerated with steam, during which CO is first produced by the
gasification of deposited carbon with H2O, and then CO reacts with H2O to produce CO2.
Meanwhile, Fe3O4 is partially reoxidized to Fe2O3 and K2CO3 is decomposed, forming
again the active phase KFeO2. The activities of the both unpromoted and K-promoted
catalysts can be recovered after the steaming of the used catalyst.

.)H2
3UHFXUVRU
+2(%
.)H2
&DWDO\VW .
.)H2
.
+2 +
&R
+

,QLWLDOVWDJH & +2 &2 +
& R &2 +2 &2 +
&DWDO\VW 5HJHQHUDWLRQ
XQGHUUHDFWLRQ ZLWK+2
5H
WO\
YH
HQ
UVL
DQ
EO
\
UP
3H
.)H2
&DUERQ
'HDFWLYDWLRQ
GHSRVLWLRQ
6SDWLDOGLVLQWHJUDWLRQ
,QDFWLYHVWDWH
Fig.4. 22. Schematic life cycle of a K-promoted styrene catalyst.

In contrast to the reversible deactivation caused by carbon deposition, the spatial
disintegration of the iron oxide and the potassium compound lead to the irreversible
deactivation. The irreversible consumption of all excess Fe3+ ions is the end of the
operation of K2Fe22O34 as storage phase for potassium. The volatility of KOH under
reaction conditions causes enrichment of KOH within the inner pores of the catalyst and
transport along with the stream of water and production mixtures through the reactor.
This potassium redistribution finally leads to the inactive state of the catalyst, which
consists of a potassium-free highly porous shell and a potassium-enriched nonporous
core.
4.5 Conclusions
1 ) The BET results show that the surface areas of the fresh catalysts are greater than
those of the used catalysts, and the surface areas of the catalysts decrease, when
the catalysts are highly promoted with potassium.
2 ) By performing the XRD investigations it is found that the main phase of the fresh
catalysts is Fe2O3 and it is Fe3O4 for the used catalysts. Moreover, the active phase
KFeO2 is observed in the K-promoted catalysts before and after the reaction.
3 ) Two series of dry TPR experiments over the fresh and used catalysts show that they
exhibit one peak for the unpromoted and low K-promoted catalysts, and two peaks
for the high K-promoted catalysts, which should be assigned to the reduction of the
active phase KFeO2. By carrying out the two series of wet TPR experiments over the
fresh and used catalysts, it is seen that the reductions are retarded due to the
addition of H2O in the reducing gas. All the TPR patterns over the fresh catalysts
display two peaks. The first peak indicates the reduction of Fe2O3 to Fe3O4, and the
second one the further reduction of Fe3O4 to FeO. The TPR patterns over the used
catalysts show only one peak, which points to the reduction of Fe3O4 to FeO. In the
wet TPR, the promoter potassium has less influence on the TPR pattern.
4 ) By means of TPO, the carbonaceous material deposited in the catalysts was studied.
It is found that the formation of CO2 during TPO displays two peaks: one is attributed
to the oxidation of the carbonaceous material deposited, and the other the
decomposition of the K2CO3 formed due to the presence of the K promoter in the
catalyst.
5) By carrying out the transient kinetic experiments, the initial periods of the
dehydrogenation of ethylbenzene over the fresh and used catalysts were
investigated. The used catalysts after steaming have the same high initial activities at
the very beginning as the fresh one. It is thought that the coupling of the
dehydrogenation of ethylbenzene and the reduction of Fe2O3 leads to a much higher
equilibrium conversion, and the initial activity of the used catalyst can be recovered to
some extent by means of the steaming of the catalyst, since the catalyst is partially
reoxidized to Fe2O3 or KFeO2.
6 ) The ex situ TPO investigations indicate that K2CO3 exists in the used K-promoted
catalysts. The ex situ temperature-programmed experiment in argon shows that there
is also K2CO3 in the fresh K-promoted catalyst. Furthermore, the results of the
transient experiments for the CO2 formation as in situ evidence confirm that K2CO3
exists during the dehydrogenation of ethylbenzene.
7 ) Combining the results from the TPO and transient experiments, it is seen that the
active phase KFeO2 decomposes due to the CO2 formed during the reaction. Further
quantitative analysis of the transient experiment results shows that KFeO2 only
partially decomposes, since there exists an equilibrium between the decomposition
and formation of KFeO2.
8 ) A modified schematic life cycle of the K-promoted styrene catalyst is proposed.
References
Wiley-VCH, Vol 5, 2151 ( 1997 )
[ 2 ] D.H. James and W. M. Castor, Ullmanns Encyclopedia of Industrial Chemistry,
Wiley-VCH, Vol. A25, 329 ( 1994 )
[ 3 ] David E. Clough and W. Fred Ramirez, AIChE J. 22, 1097 ( 1976 )
[ 4 ] Jon D. Snyder and Bala Subramaniam, Chem. Eng. Sci. 49, 5585 ( 1994 )
[ 5 ] Grigorios Kolios and Gerhart Eigenberger, Chem. Eng. Sci. 54, 2637 ( 1999 )
[ 6 ] A. Miyakoshi, A. Ueno and M. Ichikawa, Appl. Catal. A: General 212, 239 ( 2001 )
[ 7 ] W. P. Addiego, W. Liu and T. Boger, Catalysis Today 69, 25 ( 2001 )
[ 8 ] Q. L. Chen, X. Chen, L. S. mao and W. Cheng, Catalysis Today 51, 141 ( 1999 )
[ 9 ] T. Badstube and P.Kustrowski, Appl. Catal. A: General 204, 153 ( 2000 )
[ 10 ] B. D. Herzog and H. F. Rase, Ind. Eng. Chem. Prod. Res. Dev. 23, 187
( 1984 )
[ 13 ] W.R. Price, U.S. Patent 2,971,927 ( 1961 )

[ 14 ] P. Courty and J.F. Lepage, Preparation of Catalyst, Elsevier, 293 (1979 )
[ 15 ] S. T. Wong and C. Y. Mou, Appl. Catal. A: General 198, 115 ( 2000 )
[ 16 ] C. Kuhrs, M. Swoboda and W. Weiss, Top. Catal. 15, No.14, 13 ( 2001 )
[ 17 ] C. Kuhrs, Y.Arita, W.Weiss, W. Ranke and R. Schlgl, Top. Catal. 1-4 ( 14 ), 111
( 2001 )
[ 18 ] X.M. Zhu, M. Schn, U. Bartmann, A.C. van Veen and M. Muhler, submitted to
Catal. Lett.
[ 19 ] M. Muhler, J. Schtze and G. Ertl, J. Catal. 126, 339 ( 1990 )
[ 20 ] O. J. Wimmers, P. Arnoldy and J. A. Moulijn, J. Phys. Chem. 90, 1331 ( 1986 )
[ 21 ] W. Grnert, W. Saffert, R. Feldhaus and K. Anoders, J. of Catal. 99, 149 ( 1986 )
[ 22 ] E. H. Lee, Catal. Rew. 8 ( 2 ), 285 ( 1973 )
[ 23 ] D. E. Stobbe and F. R. van Buren, J. Chem. Soc. Faraday Trans. 87 ( 10 ), 1623
( 1991 )
[ 25 ] O. Kancke et al., Thermochemical Properties of Inorganic Substances, Springer,
1992.
[ 26 ] D. E. Stobbe et al., J. Catal. 135, 548 ( 1992 )
[ 27 ] M. Muhler, R.Schlgl and G. Ertl, J. Catal. 138, 413 ( 1992 )
[ 28 ] E. Matsui, T. Sodesawa and F. Nozaki, Appl. Catal. 67, 179 ( 1991 )

Chapter 5: Kinetic Parameter Optimization and Process Simulation 99
5 Kinetic Parameter Optimization and Process Simulation
Abstract: A kinetic model of the dehydrogenation of ethylbenzene to styrene was developed on
the basis of the transient and steady kinetic measurements in a microreactor flow system
equipped with GC and MS covering a wide range of reaction conditions ( 823 K < T < 898 K, 0.43
h-1< LHSV < 3 h-1, 6 < S / O < 10 ). Besides the main reaction, the reaction scheme contains the
parallel side-reaction from ethylbenzene to benzene and consecutive side-reaction from styrene
to toluene. The pre-coefficients and apparent activation energies were estimated by the data
fitting. The activation energies are in the range of 22 to 40 kcal mol-1. With the kinetic model, the
conversion of ethylbenzene as well as the selectivities of styrene, benzene and toluene could be
predicted with an average accuracy of 20 %.
Keywords: Styrene production, kinetic simulation, reaction scheme.
5.1 Introduction
The dehydrogenation of ethylbenzene to styrene is performed at around 900 K in excess
of steam and the main byproducts are benzene and toluene. Firstly, Wenner [ 1 ] studied
the process in the field of thermodynamics and kinetics, and suggested that the
following three primary reactions should be considered:
C6H5C2H5 (EB) C6H5C2H3 (ST) + H2 (1)
C6H5C2H5 (EB) C6H6 (BZ) + C2H4 (2)
C6H5C2H5 (EB) + H2 C6H5CH3 (TOL) + CH4 (3)
Furthermore, Sheel [ 2 ] considered that the all reactions could be expressed by 10
reactions. After examination of various combinations of those 10 reactions, it was

concluded that at least six reactions were necessary and the best set of six was the
equations ( 1 ) ( 3 ), and the other 3 reactions were given by
H2O + C2H4 CO + 2H2 (4)
H2O + CH4 CO + 3H2 (5)
H2O + CO CO2 + H2 (6)
This reaction scheme including 6 reactions was accepted by most authors [ 3-7 ] who
dealt with the kinetics and simulation of the process. However, in this reaction scheme
only the parallel side-reactions of ethylbenzene to toluene and benzene were considered.
Could the consecutive reactions from styrene forming benzene and toluene be
neglected? Hirano [ 8 ] carried out such an experiment, in which different amounts of
styrene were added to the educt to investigate the influence of styrene on the reaction.
The results showed that there was an inhibition effect on the rate of styrene formation,
but no significant inhibition effect on the rates of benzene and toluene formation. Thus, it
was considered that both byproducts benzene and toluene were mainly produced from
ethylbenzene. While the mentioned experiment was carried out at a relative low reaction
temperature of 853 K, compared with the generally used reaction temperature range of
823 923 K, it might be necessary to investigate the side-reactions at a wide reaction
temperature range. In this chapter, a modified reaction scheme is proposed based on the
transient investigation. And the steady-state kinetic experiments will be carried out over a
wide range of reaction conditions with a commercial and self-prepared catalyst. The pre-
coefficients and apparent activation energies can be obtained by the data fitting. Finally,
the obtained kinetic parameters are used for the prediction of the process.
5.2 Experimental
The steady-state and transient experiments were carried out in a microreactor flow
system equipped with a four-way valve ( Valco ) under high temperature and on-line
analyzers of a gas chromatograph ( GC ) and a calibrated mass spectrometer ( QMS
422 ). This reaction system has been described in detail in chapter 3. In this paper an
industrial potassium-promoted iron oxide catalyst ( S6-20, BASF ) and a self-prepared
10% K promoted catalyst were studied. Catalyst particles with a sieve fraction in the
range from 250 to 355 m were held between two layers of quartz particles in a quartz
tubular flow reactor of 14 mm inner diameter. A thermocouple ( diameter, 0.5 mm ) in a
glass tube was placed in the middle of the catalyst bed through the upper quartz layer.
The reactor was immersed in a fluidized bed of sand in an oven to ensure isothermal
reaction conditions. The heating and cooling rates of the oven were programmed by a
PC. With a educt concentration of 2.4% ethylbenzene and 24% steam a liquid hourly
space velocity ( LHSV ) of 0.43 h-1 and a H2O / etylbenzene ratio ( S/O ) of 10:1 were
achieved. The steady experiments were carried out at 823, 848, 873 and 898 K,
respectively. The LHSVs of 0.43, 0.75, 1.0, 1.5 and 3.0 h-1 were obtained by loading
different amounts of catalyst.
The conversion of ethylbenzene ( XEB ), and the selectivity to styrene ( SST ), to benzene
( SBZ ), and to toluene ( STOL ) were calculated according to the equations given in
chapter 3. The analytical results of MS were used for the transient study, and those of
GC were employed for the data fitting to unravel the kinetic parameters.
5.3.1 Kinetic data
Preliminary experiments were carried out to study the influences of internal and external
transport resistance. The catalysts with the sieve fractions in the range of 45 100, 100
250, and 250 355 m were used for the investigation of the internal transport
resistance. The results are shown in Fig.5.1. It is seen that the reaction rates with the
sieve fractions of 100 250 m and 250 355 m were almost the same. Therefore, the
catalysts with the sieve fraction in the range from 250 to 355 m were used for the kinetic
study, where the intra-particle transport could be neglected.
100
80
60
45 - 100 m
40
100 - 250 m
20 250 - 355 m
0
0 1 2 3 4 5
Time / h
Fig. 5.1. The influence of intra-particle transport on the conversion of ethylbenzene, LHSV =
0.75 h-1, S/O = 6 : 1, and T = 923 K.

With respect to the external transport resistance, the experiments at different gas flow
rates were performed. The results are displayed in Fig.5.2. It is shown that at the same
LHSV ( 0.43 h-1 ) and S/O ( 6:1 ), the reaction rates did not change, although the flow rate
of the inert gas N2 was increased from 40 to 60 ml ( STP ) min-1. Further kinetic
experiments were carried out at the gas flow rate of 40 N2 ml ( STP ) min-1 as well as the
same S/O or higher S/O up to 10:1, and lower temperatures ( 823 898 K ). Hence, the
effect of the external mass transport on the reaction is absent under these reaction
conditions.
100
80
60
40
-1
40 N2 ml min
20 -1
60 N2 ml min
0
0,0 0,5 1,0 1,5 2,0 2,5
Time / h
Fig. 5.2. The influence of external transport on the conversion of ethylbenzene, LHSV =
1.12 h-1, S/O = 6 : 1, and T = 923 K.
S/O is also an important operation variable for the reaction. A series of temperature-
programmed reaction experiments under different S/O were carried out to determine the
suitable S/O for the kinetic study. Argon containing steam and ethylbenzene passed
through the reactor, and the temperature of the reactor was increased at a rate of 2 K
min-1 from 573 to 923 K. The different S/O of 10, 6, and 3 were used as shown in Fig.
5.3.
100
SOR = 10
80 SOR = 6
SOR = 3
60
40
20
0
700 750 800 850 900
Temperature / K
Fig. 5.3. The influence of S/O on the conversion of ethylbenzene, LHSV = 0.75 h-1, ramping
rate of 2 K min-1.
It is seen that S/O has a strong effect on the reaction conversion, however, it became
less important as S/O exceeded 6.0. Thus, the S/O for the further experiments will be
chosen in the range from 6 to 10.
Based on the results of the preliminary experiments, the steady-state kinetic experiments
over the S6-20 and 10% K-promoted catalysts were carried out under the following
reaction conditions: T, 823 898 K; LHSV, 0.43 3.0 h-1; S/O, 6 10; the sieve fraction
of the catalyst: 250 355 m. The results under different reaction conditions are
summarized in Table 1 for the S6-20 catalyst and in Table 2 for the 10 % K-promoted
catalyst.
Table 1 The results of the kinetic experiments over S6-20 catalyst
No. S/O LHSV Temp. X(EB) S(ST) S(BZ) S(TOL) C-

-1
h K % % % % Balance
%
1 6 0.75 823 28.3 95.1 3.2 1.6 95.7
2 6 0.75 848 38.5 94.7 3.1 2.2 93.7
3 6 0.75 873 43.0 93.2 3.9 2.9 93.4
4 6 0.75 898 55.5 89.3 5.5 5.1 92.2
6 6 1.00 823 21.9 95.9 3.1 1.0 94.6
7 6 1.00 848 32.5 95.1 3.4 1.6 91.7
8 6 1.00 873 39.2 92.7 4.5 2.8 90.2
9 6 1.00 898 51.5 88.4 6.5 5.1 88.5
11 6 1.50 823 16.2 93.1 5.5 1.4 96.6

12 6 1.50 848 23.9 93.3 4.9 1.8 96.0
13 6 1.50 873 31.4 93.1 4.6 2.3 93.7
14 6 1.50 898 43.7 89.3 6.9 3.8 93.7
16 6 3.00 823 11.6 95.2 4.8 0.0 98.2
17 6 3.00 848 17.3 93.8 4.6 1.5 96.4
18 6 3.00 873 25.4 92.1 5.9 2.0 94.4
19 6 3.00 898 33.3 87.5 9.0 3.5 94.4
21 8 0.75 823 27.7 94.3 4.0 1.7 93.4

22 8 0.75 848 39.8 93.9 3.8 2.2 92.1
23 8 0.75 873 45.1 92.5 4.5 3.0 91.6
24 8 0.75 898 56.2 88.1 6.5 5.4 90.8
26 8 1.00 823 23.0 95.3 3.5 1.2 93.7
27 8 1.00 848 35.7 94.7 3.7 1.6 91.0
28 8 1.00 873 42.7 93.3 4.2 2.5 90.9
29 8 1.00 898 52.8 90.1 5.8 4.1 89.4
31 8 1.50 823 12.6 95.2 4.3 0.5 97.6

32 8 1.50 848 22.2 95.0 3.8 1.3 94.7
33 8 1.50 873 31.5 92.2 5.2 2.6 92.9
34 8 1.50 898 43.3 87.3 8.8 3.9 90.3
36 8 3.00 823 12.5 95.0 5.0 0.0 98.0
37 8 3.00 848 19.4 94.4 4.2 1.4 97.4
38 8 3.00 873 28.2 93.1 4.9 2.0 94.4
39 8 3.00 898 38.1 89.3 7.9 2.8 92.5
Table 2 The results of the kinetic experiments over 10% K-promoted catalyst
No. SOR LHSV Temp. X(EB) S(ST) S(BZ) S(TOL) C-

-1
h K % % % % Balance
%
1 10 0.43 823 30.4 96.2 2.5 1.3 98.0

2 10 0.43 848 49.7 95.6 2.6 1.9 94.1
3 10 0.43 873 66.7 94.5 2.6 2.9 98.5
4 10 0.43 898 82.7 92.3 2.9 4.9 99.2
5 10 0.75 823 29.5 96.6 1.9 1.1 93.2

6 10 0.75 848 41.5 96.6 2.0 1.4 93.5
7 10 0.75 873 58.6 95.9 2.2 2.0 90.0
8 10 0.75 898 76.5 93.4 2.7 4.0 96.4
9 10 1.00 823 18.5 96.7 2.4 0.9 98.5

10 10 1.00 848 34.7 96.3 2.3 1.4 96.2
11 10 1.00 873 51.0 95.8 2.3 1.9 96.0
12 10 1.00 898 69.4 94.5 2.5 3.0 93.5
13 10 1.50 823 6.6 96.6 2.5 1.0 104.2

14 10 1.50 848 15.8 96.9 2.0 1.1 106.2
15 10 1.50 873 34.5 96.9 2.2 1.4 100.3
16 10 1.50 898 50.0 95.6 2.4 2.0 99.6
5.3.2 Kinetic Model
Most previous models only considered the parallel side-reactions forming benzene and
toluene from ethylbenzene. Are there also the consecutive reactions? To address this
issue, we carried out a temperature-programmed reaction experiment ( TPRE ) with
styrene as educt at a ramping rate of 15 K min-1. In comparison with other TPREs for the
dehydrogenation of ethylbenzene, the concentration of toluene vs. temperature is
illustrated in Fig.5.4.
0,25 A ( Educt, ST )
B ( Educt, EB )
Concentration of TOL / %
C ( Blank Exp.)
0,20
0,15
0,10
0,05
0,00
650 700 750 800 850 900
Temperature / K
Fig.5.4. The concentration of toluene vs. temperature: ( A ) Styrene as educt, 3.2 %, over S6-20,
ramping rate of 15 K min-1, ( B ) ethylbenzene as educt, 5.3%, over S6-20, ramping rate
of 2 K min-1, ( C ) ethylbenzene as educt, 5.3%, over quartz, ramping rate of 2 K min-1.
Obviously, toluene is much more easily formed from styrene than from ethylbenzene.
Thus, the consecutive reaction should be considered, and it is suggested that toluene is
formed mainly from styrene.

0,20
A (Educt, ST)
B (Educt, EB)
C (Blank Exp.)
Concentration of BZ / %
0,15
0,10
0,05
0,00
650 700 750 800 850 900
Temperature / K
Fig.5.5. The concentration of benzene vs. temperature: ( A ) Styrene as educt, 3.2 %, over S6-
20, ramping rate of 15 K min-1, ( B ) ethylbenzene as educt, 5.3%, over S6-20, ramping
rate of 2 K min-1, ( C ) ethylbenzene as educt, 5.3%, over quartz, ramping rate of 2 K
min-1.
The concentration of benzene vs. temperature under different reaction conditions is
shown in Fig.5.5. By comparing Fig.5.4 and Fig.5.5, it is seen that the formation rate of
benzene from styrene is much lower than that of toluene, and similar to that of the blank
experiment. For this reason, it is proposed that benzene is mainly produced by a parallel
reaction from ethylbenzene. According to those results of the transient experiments, the
reaction scheme may be written as follows:

k1
C6H5C2H5 C6H5C2H3 + H2 (1)
k-1
C 6 H 5 C 2 H 5
k2
C6 H 6 + C2 H 4 (2)
C 6 H 5 C 2 H 3 + 2 H 2
k3
C 6 H 5 CH 3 + C H 4 (3)
1/2 C 2 H 4 + H 2 O
k4
CO + 2 H 2 (4)
CH 4 + H 2 O
k5
CO + 3 H 2 (5)
CO + H 2 O
k6
CO 2 + H 2 (6)
Based on this scheme, the reaction rate equations can be expressed by
1
r1 = k1 (PEB PST PH2 ) (7)
Kp
r2 =k 2 PEB (8)
r3 = k 3 PSTa PHb2 (9)
r4 = k 4 Pw PC2 H 4 ( 10 )
r5 = k 5 Pw PCH 4 ( 11 )
r6 = k 6 ( Pt / T 3 ) Pw PCO ( 12 )
We supposed different orders for a and b in the equation 9 as shown in Table 2. The data
fitting based on the experimental data collected of at the different reaction temperatures
over the S 6-20 catalyst was carried out. The simulation results are summarized in
Table 3.
Table 3 The results of the data fitting with different reaction rate equations of toluene
Reaction Rate Reaction Temperature Root of Mean Square

Equation K
823 7.89 . 10-2

r3 = k3 PST PH22 848 7.31 . 10-2
873 7.78 . 10-2
898 1.12 . 10-1
r3 = k 3 PST0.5 PH 2 823 7.17 . 10-2

848 6.61 . 10-2
873 7.15 . 10-2
898 1.10 . 10-1
r3 = k 3 PST0.1 PH02.2 823 6.93 . 10-2

848 6.30 . 10-2
873 6.64 . 10-2
898 1.08 . 10-1
The difference of the roots of mean square for the different reaction models is not great
enough to distinguish which model should be taken into consideration for the production
of toluene. Further analysis was carried out to investigate the change of the toluene
concentration versus space velocity of reactants. Evidently, the reaction rate equation of
toluene has great influence on the result of the calculation, especially at the high reaction
temperatures. The result is shown in Fig.5.6.
A comparison of calculations using different equations for the formation of toluene at 898
K is shown in Fig.5.7. It is seen that the toluene reaction rate equation r3 = k 3 PST0.1 PH02.2 is
more suitable than the others for the further data-fitting and simulation.
Exp. at 823 K
20 Cal. at 823 K
-4
Concentration of TOL / 10
Exp. at 898 K
Cal. at 898 K
16
12
0,5 1,0 1,5 2,0 2,5 3,0
-1
LHSV / h
Fig.5.6. The concentration of toluene vs. LHSV at 823 K and 898 K
24
Exp. at 898 K
20 Cal. with a
-4
Cal. with b
Concentration of TOL / x 10
Cal. with c
16
12
0,5 1,0 1,5 2,0 2,5 3,0
-1
LHSV / h
Fig.5.7. The toluene concentration vs. LHSV: ( a ) r3 = k3 PST PH22 , ( b ) r3 = k 3 PST0.5 PH 2 ,
( c ) r3 = k 3 PST0.1 PH02.2 .
5.3.3 Kinetic data fitting
The catalytic reactor was described by a one-dimensional, pseudo-homogeneous plug-
flow model under isothermal condition. The parameters of the kinetic model were
estimated applying the random search method. The object function for the parameter
optimization is given by
J Exp , i J Cal , i
( )2

J Exp , i
Obj = ( 13 )
J = C EB , C ST ... N J , Exp 1
i =1,16
The Obj was calculated on the basis of experimental data and model prediction for the
concentrations of ethylbenzene, styrene, benzene and toluene. The flow chart for the
parameter optimization is shown in Fig.5.8. The program code was written in Fortran.
The parameter optimizations were carried out at 823, 848, 873 and 898 K, respectively.
The reaction constants k1 for the dehydrogenation of ethylbenzene, k2 for the formation of
benzene and k3 for the formation of toluene were obtained, respectively. ln k vs. 1/T is
plotted for the S6-20 and 10% K-promoted catalysts as shown in Fig. 5.9 and Fig. 5.10.
Start
OPTI
MARQUARDT POWELL QUAD.-MIN RANDOM DOWNHILL
FUNC
FIX1
READ_TEMP Loop, over all the

experimental data
INPUT
KINERES
FLUID
HETREA
MODELEQ
CONVERG
WEGSTEIN
PERFORM
Caculation of
new parameters
End
Fig.5.8. Calculation flow chart.

-12
-14
-16
Ln k
-18
-20
ln k1
-22 ln k2
ln k3
-24
-3 -3 -3 -3
1,1x10 1,1x10 1,2x10 1,2x10
1 / T ( 1/K )
Fig. 5. 9. ln k1, k2 and k3 as a function of 1/T for the S6-20 catalyst.
-10
-12
-14
-16
Ln k
-18
-20
ln k1
-22
ln k2
ln k3
-24
-3 -3 -3 -3
1,1x10 1,1x10 1,2x10 1,2x10
1 / T ( 1/K )
Fig. 5.10. ln k1, k2 and k3 as a function of 1/T for the 10% K - promoted catalyst.
It is seen that all the data points are on straight lines. By measuring the slopes and
interceptions the activation energies and pre-coefficients of the k1, k2 and k3 can be
obtained. The results are summarized in Table 4.
Table 4 Kinetic parameters for the industrial and home prepared catalysts
S 6-20 10 % K - promoted
Reaction
k0i Ei k0i Ei
kcal mol-1 kcal mol-1
1 0.3 21.9 1.7 . 103 35.3
2 5.8 32.9 2.5 .102 38.3
3 4.7 35.9 65.4 40.4
4* 0.9 24.9 0.9 24.9
5* 5.3 . 10-2 15.7 5.3 . 10-2 15.7
6* 1.7 . 109 17.6 1.7 . 109 17.6
* Froment and Bischof [ 10 ]
The prediction of the process was carried out with the parameters in Table 2. Parts of
the results are plotted in Fig.5.11 and 5.12. Fig.5.11 shows the influence of LHSV on the
conversion and selectivities over 10 % K-promoted catalyst at 848 K. With increasing the
LHSV, the conversion of ethylbenzene decreased strongly. Contrarily, the selectivities of
styrene, benzene, and toluene increased a little. The calculated prediction is in good
agreement with the experimental observations.

100
Conversion or selectivity / %
80
XEB
SST
60 SBZ
STOL
simulation
40
20
0
0,50 0,75 1,00 1,25 1,50
-1
LHSV / h
Fig.5.11. The influence of LHSV on the conversion and selectivities over the 10% K-promoted
catalyst, T = 848 K, S / O = 10:1.
100
XEB
Conversion or selectivity / %
80 SST
SBZ
STOL
60 simulation
40
20
0
820 840 860 880 900
Temperature / K
Fig.5.12. The influence of reaction temperature on the conversion and selectivities over S6-20
catalyst, LHSV = 1.0 h-1, S / O = 6:1.
The influence of the reaction temperature on the conversion and selectivities over the S6-
20 catalyst at the LHSV of 1.0 h-1 is shown in Fig.5.12. The conversion of ethylbenzene
increased with raising the reaction temperature. In contrast, the selectivities of styrene,
benzene and toluene decreased somewhat. The results of the prediction are in good
agreement with those of the experiments.
The further model validation was carried out by comparing model predictions with
experimental data for all 48 experiments. The parity plots, which illustrated the accuracy
of the model for predicting conversion of ethylbenzene as well as the selectivities of
styrene, benzene, and toluene, are displayed in Fig.5.13 - 5.16.
Almost a symmetrical distribution of the data points on both sides of the diagonal was
obtained for the most variables. For the 10% K-promoted catalyst, the average relative
prediction errors for the conversion of ethylbenzene amounted to 17 % , the selectivity of
styrene was 1 %, of benzene 8 %, and of toluene 10 %, respectively. Over S6-20
catalyst, the average relative prediction errors for the conversion of ethylbenzene, and
selectivity of styrene were similar to those for the 10 % K-promoted catalyst amounting to
18% and 2%. The selectivities of benzene and toluene were predicted with somewhat
lower accuracy, which amounted to 27 % and 20 %, respectively.

100
XEB + 20 %
SST
XEB and SST ( Simulation ) / %
80
60
- 20 %
40
20
0
0 20 40 60 80 100
XEB and SST ( Experiment ) / %
Fig. 5.13. Evaluation of the simulation of the concentrations of EB and ST over the 10% K-
promoted catalyst. T: 823 898 K, LHSV: 0.43 1.5 h-1 and S / O: 10:1.
6
BZ
+ 20 %
TOL
SBZ and STOL ( Simulation ) / %
- 20 %
0
0 2 4 6
SBZ and STOL ( Experiment ) / %
Fig. 5.14. Evaluation of the simulation of the concentrations of BZ and TOL over the 10% K-
-1
100
XEB + 20 %
SST
XEB and SST ( Simulation ) / %
80
60
- 20 %
40
20
0
0 20 40 60 80 100
XEB and SST ( Experiment ) / %
Fig.5.15. Evaluation of the simulation of the conversion of EB and selectivity of ST over the
catalyst S6-20. Reaction temperature: 823 898 K, LHSV: 0.75 3.0 h-1 and S / O:
6:1 8:1.
10
BZ + 20 %
TOL
SBZ and STOL ( Simulation ) / %
6
- 20 %
0
0 2 4 6 8 10
SBZ and STOL ( Experiment ) / %
Fig. 5.16. Evaluation of the simulation of the selectivities of BZ and TOL over the catalyst S6-20.
-1

5.4 Conclusions
1 ) Based on the transient experimental studies, a model with consecutive reaction of
styrene to toluene is proposed. Such a model has not yet been used until now for the
prediction of the reactor performance for the dehydrogenation of ethylbenzene to
styrene.
2 ) The data fitting was carried out over S6-20 and 10% K-promoted catalysts in the
range of the reaction temperature: 823 898 K; LHSV: 0.43 3 h-1 ; S/O: 6 10.
Over S6-20 catalyst, the activation energy E for the dehydrogenation of ethylbenzene
amounts to 22 kcal mol-1, for the parallel reaction of ethylbenzene to benzene 33
kcal mol-1 and for the consecutive reaction of styrene to toluene 36 kcal mol-1. Over
the 10% K-promoted catalyst, 35 kcal mol-1 for the main reaction, 38 kcal mol-1 for the
parallel reaction and 40 kcal mol-1 for the consecutive reaction were obtained. The
optimized pre-coefficients of the self-prepared catalyst are greater than those of the
commercial catalyst due to its higher conversion.
3 ) The model gives good prediction of the experimental data. The average prediction
errors for the conversion of ethylbenzene is below 18 %, and for the selectivities of
styrene below 2 %, of toluene below 20 % as well as of benzene in the range of
8 27 %.
Nomenclature
Ai Frequency factor of reaction i
Ci Concentration of component i in the gas phase, mol %
Ei Activation energy of reaction i, kcal mol-1

ki Rate constant of reaction i, kmol kg-1s-1 barn ( for k1 and k2, kmol kg-1s-1 bar; for
k3, kmol kg-1s-1 bar0.3 ; for k4 and k5, kmol kg-1s-1 bar2; for k6, kmol K3 kg-1s-1bar3 )
Kp Equilibrium constant for the dehydrogenation of ethylbenzene to styrene, bar:
ln Kp = -15350 / T + 16.12 [ 11 ]
Nj Total number of concentrations used for data fitting.
PT, Pi Total pressure and partial pressure of component i respectively, bar
R Universal gas constant ( = 8.314 kJ kmol-1K-1 )
ri Rate of reaction i, kmol kg-1s-1
T Temperature of the reaction, K
Xi Conversion of reactant i
Si Selectivity of product i
EB Ethylbenzene
ST Styrene
BZ Benzene
TOL Toluene
W Water
References
[ 1 ] R.R. Wenner and E.C. Dybdal, Chem. Eng. Pro. 44, 275 ( 1948 )
[ 2 ] J.G.P. Sheel and C.M. Crowe, The Canadian J. of Chem. Eng. 47, 183 ( 1969 )
[ 3 ] D.E. Clough and W.F. Ramirez, AIChE J. 22, 1097 ( 1976 )
[ 4 ] S.S.E.H. Elnashaie and R. Hughes etc., Ind. Eng. Chem. Res. 32, 2537 ( 1993 )
[ 5 ] J.D. Snyder and B. Subramaniam, Chem. Eng. Sci. 49, 5585 ( 1994 )
[ 6 ] G. Kolios and G. Eigenberger, Chem. Eng. Sci. 54, 2637 ( 1999 )
[ 7 ] C. Hermann, P. Quicker, R. Dittmeyer, J. of Membrane Sci. 136, 161 ( 1997 )
[ 9 ] X.M. Zhu, M. Schn, U. Bartmann, A.C. van Veen and M. Muhler, submitted to
Catal. Lett.
[ 10 ] Gilbert F. Froment and Kenneth B. Bischoff, Chemical reactor analysis and design,
Wiley ( New York ), ( 1990 )
VCH ( Weinheim ), Vol 5, 2151 ( 1997 )

Chapter 6: Summary 123
6 Summary
TPR is a useful technique for the study of reducible materials. However, the Tmax and
TPR pattern are highly affected by the TPR experimental conditions. It is found that both
the small amount of sample and the H2O addition in the reducing gas can lead to highly
resolved TPR pattern, and the dilution of the sample with quartz is a useful method to
carry out the TPR investigation with small amounts of sample. Up to now there seems to
be no suitable criterion for the experimental conditions of the TPR investigation. For the
TPR studies of Fe2O3 and CuO at high resolution, appropriate experimental conditions
are suggested on the basis of the results of this work. Particularly, the results of wet TPR
of Fe2O3 is very significant for the study of dehydrogenation of ethylbenzene to styrene
with excess of H2O.
The transient kinetic experiments were carried out over the commercial and self-
prepared catalysts. In the initial periods of the fresh and regenerated catalysts, a much
higher initial activity of the catalyst at the very beginning was observed. It is thought that
the coupling of the dehydrogenation of ethylbenzene and the reducing of Fe2O3 led to a
much higher equilibrium conversion. With the reduction of Fe2O3 to Fe3O4, the activities
of the catalysts decrease greatly, and reach finally steady state. It should be pointed that
the reduction of Fe2O3 to Fe3O4 is the main cause of the activity decrease in the initial
period.
During the reaction steam stabilizes the iron oxide against total reduction to Fe,
meanwhile, it reacts with the coke deposit on the catalyst forming CO, and CO reacts
further with steam yielding CO2. By steaming of the used catalyst, the catalyst is partially
reoxidized to Fe3+-containing oxide such as Fe O or KFeO , and the activity of the

Chapter 6: Summary 124
catalyst is recovered.
The high K-promoted catalysts have higher activities than those of the lower K-promoted
and unpromoted catalysts. It is considered that KFeO2 is the active phase, and that CO2
formed during the reaction causes the decomposition of KFeO2. Quantitative analysis
shows that the decomposition of KFeO2 reaches an equilibrium when the CO2
concentration during the reaction is constant. Under our reaction conditions, the KFeO2
could not be completely decomposed. Ex situ and in situ investigations showed that
K2CO3 exists not only in the fresh catalyst but also in the catalyst during the reaction. By
steaming of the catalyst, K2CO3 is decomposed to K2O and CO2, and consequently, new
KFeO2 can be formed.
Based on those investigations, a modified schematic life cycle of the K-promoted styrene
catalyst is proposed. It is helpful to understand this catalytic process comprehensively.
Furthermore, the simulation of this reaction was carried out over the commercial and self-
prepared catalysts. By means of the transient kinetic study, a model with consecutive
reaction from styrene to toluene is proposed. The pre-coefficients and apparent activation
energies are obtained using the data fitting in the range of T: 823 898 K; LHSV: 0.43
3 h-1; and S / O : 6 10. The model shows good agreement with the experimental data.
Appendix 125
Appendix
1 ChemicaI reagents and gases
The chemical reagents and gases used in the experiments are summarized in Table1.
Table 1 Chemical reagents and gases.
Name Content Manufacture Application
Ethylbenzene 99.8 % ACROS ORGANICS The steady and
Styrene 99.0 % Riedel deHaen transient studies of

Toluene 99.5 % Baker
dehydrogenation of
Benzene 99.9 % Sigma Aldrich
ethylbenzene to
Distilled Water Bio-chemistry Lab.
Catalyst S 6-20 BASF styrene.
Argon 99.999% Messer Griesheim
Helium 99.98% Central gas
Fe(NO3)3 9H2O 99.0 % ACROS ORGANICS The preparations of
KOH 87.8 % Baker the catalysts.
K2CO3 99 % Riedel deHaen
Ce(NO3)3 6H2O 99.99 % ALFA Products
Cr(NO3)3 9H2O 99.0 % Riedel deHaen

Appendix 126
Tabel 1 ( continued )
Name Content Manufacture Application
Fe2O3 99.0 % MERCK The investigations
CuO 97.0 % Riedel deHaen of TPR and TPO.
Cu2O 87.5 % ( of Cu ) Riedel deHaen
H2 / Argon 4.2 % Messer Griesheim
O2 / Argon 21.5 % Messer Griesheim
2 Experimental apparatuses
2.1 The set-up for the studies of TPR and TPO
2.1.1 The full view of the set-up
The TPR and TPO experiments were carried out in a microreactor flow system, the flow
scheme of the experimental set-up has been already shown in Fig.2.1 ( chapter 2 ). A full
view of the set-up for the studies of TPR and TPO is displayed in Fig.1 as follows.
Appendix 127
Fig.1. The experimental set-up for the investigations of the TPR and TPO.
2.1.2 The reactor for the TPR and TPO
The drawing of the reactor is exhibited in Fig.2. The quartz fiber was placed between the
glass tube and the wall of the reactor to support the quartz particles and sample. The
sample was held between two layers of quartz particles ( 160 250 m ). With the aid of
specially designed funnel, the sample and quartz particles could be loaded easily into the
reactor. The thermocouple ( diameter, 0.5 mm ) was immersed from below into the glass
Appendix 128
tube. The reducing gas passed down through the fixed bed of the sample and quartz
particles.
15
20
6
35
40 10
110
6
12
90
5 Quartz
10 Sample Funnel
5 Quartz
115
Quartz fiber
Thermocouple
Fig. 2. The reactor for the investigations of the TPR and TPO.
Appendix 129
2.1.3 The isothermal period of the oven
The measurements of the temperature profile of the oven were carried out at 550, 575,
600, 625, 650C, respectively. The results of the measurements are displayed in Fig.3. It
is known that the isothermal zone ranges from 6.5 to 8.5 cm of the length of the oven.
700
650
600
Temperature / C
550
500
450
400
350
300
250
0 2 4 6 8 10 12 14
Length of oven / cm
Fig. 3. The temperature profiles of the oven.
2.1.4 The delay of the concentration of H2
Waste gas
Ar 1704 mm
H2 in Ar Reactor
Fig. 4. The flow chart of the set up for the TPR investigation.
Appendix 130
When the reduction of the sample began, the concentration of H2 of the reducing gas and
the reaction temperature were changing, simutaneously. However, the recorded
concentration of H2 with time was a delay value, because after the reaction the reducing
gas flows through the cold trap, dryer and then got the gas analyzer Binos, which is
shown in Fig.4. Aiming at deriving the delay factors, an investigation was performed as
follows: Argon passed through in first the reactor, and then it was switched to the
reducing gas. The time began to be recorded, simultaneously, until the H2 signal was
collected by the Binos detector. This recorded time was defined as the total delay time.
Such an investigation was repeated several times, and an average total delay time of
67.95 s was obtained. In order to achieve the absolute delay time, it is necessary to
calculate the volume between the four-way valve and reactor. The volume of the tube is
given by
VTube = 0.3175 2 170.4 = 13.49 ml

4
and the volume of the reactor could be expressed as follows:
VReactor = 0.6 2 3.5 + 1.0 2 14.5 = 12.38 ml

4 4
Hence, the total volume is obtained through
VTotal = VTube + VReactor = 13.49 + 12.38 = 25.87 ml
A series of measurements of the total delay time were carried out, and the average total
delay time was 67.95 s. At the measurement condition of the flow rate of 77 ml ( STP )
min-1, the room temperature of 25 C, and the atmospheric pressure of 747 mmHg, the
absolute delay time could be calculated with the following equation:

Appendix 131
25.87
t Delay = 67.95 60 = 67.95 18.15 = 49.8 s
25 + 273 760
77
273 747
The reduction temperature was increased at the heating rate of 15 C min-1. Therefore,
the difference of the temperature T can be given by
15
T = 49.8 = 12.45 C
60
For the other heating rates of 10, 5, 2 C min-1, the difference of temperature T could be
obtained by the same way. The results are listed in Table 2.
Table 2 The temperature difference at the different heating rates.
Heating rate T
C min-1 C
15 12.45
10 8.3
5 4.2
2 1.7
2.1.5 The consumption of H2 for the study of TPR
)LQ
5HDFWRU &RROHU 'U\HU
& + LQ
& + RXW
Fig. 5. The flow chart of the mass balance in the TPR investigation.
The reducing gas flows through the reactor, cooler, dryer, and reaches the gas analyzer
Appendix 132
Binos. In the reactor, the sample is reduced and the H2 in the reducing gas is consumed.
The consumed volume of H2 per time is defined as VH2. Considering that a molecule of H2
is consumed to form a molecule of H2O, the total flow rate after the reactor ( Fout ) should
be equal to the total flow rate before the reactor ( Fin ). However, when the reducing gas
flows through the cooler and dryer, the water formed during the reduction will be
eliminated. Thus, the total flow rate Fout can be expressed as follows:
Fout = Fin VH 2
Therefore, the equation which describes the mass-balance for H2 is given by
Fin C H 2 , in = VH 2 + Fout C H 2 , out
= VH 2 + ( Fin VH 2 ) C H 2, out
= VH 2 ( 1 C H 2 , out ) + Fin C H 2 , out
and,
( C H 2 , in C H 2 , out )
VH 2 = Fin ml min 1
1 C H 2 , out
Finally, the quantitative amount of the consumed H2 can be obtained by
10 3 1
n H 2 = VH 2 mol s 1
22.4 60
( C H 2 , in C H 2 , out ) 10 3 1
= Fin mol s 1
1 C H 2 , out 22.4 60
2.1.6 The consumption of O2 for the study of TPO

Appendix 133
)LQ
5HDFWRU &RROHU 'U\HU
&
&
2 LQ
2 RXW
Fig.6. The flow chart for the mass balance in the TPO investigation.
In the TPO investigation, oxygen reacts with the deposited carbon to form a CO2, thus it
can be assumed that the total flow rate of the gas mixture is kept equal before and after
the oxidation, which is expressed as follows:
Fin = Fout

= Fout
Therefore, the equation which describes the mass-balance for oxygen is given by
Fin C O2 , in = VO2 + Fout C O2 , out
then,
VO2 = Fin C O2, in Fout C O2 , out

= Fin ( C O2 , in - C O2 , out )
Finally, the quantitative amount of the consumed O2 can be achieved through
10 3 1
n O2 = VO2 ml min 1
22.4 60
10 3 1
= ( C O2 , in C O2, out ) Fin mol s 1
22.4 60
Appendix 134
2.2 The set-up for the study of dehydrogenation of ethylbenzene
2.2.1 The full view of the set-up
The flow scheme of the experimental set-up for the steady and transient studies of the
dehydrogenation of ethylbenzene to styrene is shown in Fig.3.1 ( chapter 3 ). A full view
photo of this set-up is displayed in Fig.6.
Fig. 6. The experimental set-up for the dehydrogenation of ethylbenzene to styrene
The transient step-up and step-down experiments were accomplished by switching the
four-way valve one as shown in Fig.7. By switching the four-way valve two, the educt or
Appendix 135
the products are led to the gas chromatography and the mass spectrometer. The
concentrations of the gas mixture before and after the reaction are analyzed, including
those of ethylbenzene, styrene, benzene, toluene, water, CO, CO2, methane, ethylene
and hydrogen.
Waste Gas
Gas Mixture 1
Valve 1
Gas Mixture 2
Valve 2
Analysis
Reactor
Fig. 7. The experimental switching part for the dehydrogenation of ethylbenzene to styrene
A typical procedure of a transient step-up and step-down experiment is described as
follows:
1) Valve 2 was set to analysis, hence, the gas mixture 1 ( helium and steam ) or the gas
mixture 2 ( argon, steam and ethylbenzene ) could be directly analyzed by gas
chromatography and mass spectrometer without flowing through the reactor.
2) After the analysis of the gas mixture 1 and 2, the Valve 1 should be kept on gas
Appendix 136
mixture 1, and then the Valve 2 is switched to Reactor. After that, the steam in
helium goes through the catalyst bed. Simultaneously, the temperature of the reactor
is increased at a heating rate of 15 C min-1 from 200 C until the suitable reaction
temperature. Normally, the reaction temperature ranges from 550C to 625 C.
3) When the temperature reaches the suitable value, for example 600C, the Valve 1 is
switched to the gas mixture 2 ( ethylbenzene, steam, argon ), then the step-up
experiment for the investigation of the initial activity of the fresh catalyst starts. The
gas mixture is analyzed at an interval of 6.3 s by MS, simultaneously.
4) After 3 hours, the reaction reaches steady state. The Valve 1 is switched again to the
gas mixture 1, and thus the steaming of the used catalyst begins.
5) After 3 hour steaming of the catalyst, the valve 1 is switched to the gas mixture 2, then
the transient experiment for the investigation of the initial activity of the used catalyst
is performed.
2.2.2 The reactor for the dehydrogenation of ethylbenzene to styrene
The picture of the reactor for the study of the dehydrogenation of ethylbenzene to styrene
is shown in Fig. 8. The quartz fiber was placed in first at the bottom of the reactor as a
support for the quartz particles and catalyst, and then the quartz particles ( < 355 m )
was filled into the reactor. The glass tube for the thermocouple was put on the top of the
quartz particle layer. After that, the catalyst and the upper layer of quartz particles were
loaded. The reactor was set in the oven and connected with the other parts of the set up.
Finally, the thermocouple was immersed in the glass tube. The gas mixture passed
through the fixed bed of the catalyst and quartz particles.

Appendix 137

4XDUW]

&DWDO\VW

4XDUW]
4XDUW]ILEHU
Fig.8. The reactor for the dehydrogenation of ethylbenzene to styrene
2.2.3. The determination of LHSV
In order to calculate the LHSV ( liquid hourly space velocity ), the liquid flow rate of
ethylbenzene should be determined first. FEB can be derived by the following equation:
Appendix 138
FAr 1
FEB = C EB M EB 60
22.4 C Ar EB
where FEB is the liquid flow rate of ethylbenenzene ( ml hr-1 ), FAr the volumetric flow rate
of argon ( ml min-1 ), CAr the concentration of argon, CEB the concentration of
ethylbezene, MEB the molecular weight of ethylbenzene and EB the liquid density of
ethylbenzene. Under the reaction conditions of 2.4 % for CEB, and of 24% for CH2O, the
concentration of argon CAr can be calculated by
CAr = ( 1 CEB CH2O ) = ( 1 2.4% -24% ) = 73.6 %
When the volumetric flow rate of the inert gas argon is 40 ml ( STP ) min-1, the liquid
flow rate of ethylbenzene can be derived from
40 10 3 1
FEtB = 2.4 % 106 60 = 0.43 ml h 1
22.4 73.6 % 0.867
The load of the catalyst is 1 g, and the load density of the catalyst equals to 1 g . ml-1.
Therefore, the LHSV can be obtained with the following equation:
0.43 ml h 1
LHSV = 1
= 0.43 h 1
1 g 1 g ml
The different LHSVs were calculated according to the different loads of the catalysts, and
the results are summarized in Table 3.
Table 3 The LHSVs for the different loads of the catalysts
LHSV Load of catalyst

h-1 g
0.43 1.00
0.75 0.57
1.00 0.43
Appendix 139
2.2.4 The saturators
Two saturators are used for the study of dehydrogenation of ethylbenzene to styrene.
The structure of the saturator is shown in detail in Fig. 9.
Fig. 9. The saturator for the study of dehydrogenation of ethylbenzene.

Appendix 140
The temperatures of the saturators are controlled by the thermostats. The flow chart for
the regulation of the saturators is displayed in Fig.10.
Gas Mixture
Mixer
EB H2O
Ar Ar
Thermostat 1 Thermostat 3
Fig. 10. The flow chart for the saturators
Normally, the temperature of the thermostat 2 was set at 92 C, and thus the temperature
of the lower part of the saturators could be controlled at about 90 C. The temperature of
the upper part of the EB saturator was controlled by the thermostat 1, and that of the H2O
saturator was controlled by thermostat 3. The suitable temperatures could be predicted

Appendix 141
based on the thermal data of the vapor pressures listed in Table 4.
Table 4 Vapor pressures and boiling points
Temperature Water Ethylbenzene Styrene Benzene Toluene
C mbar mbar mbar mbar Mbar
0 6.10 2.60 1.51 35.06 9.15
10 12.27 5.12 3.10 60.46 16.76
20 23.39 9.54 5.99 99.85 29.26
30 42.48 16.93 10.99 158.53 48.95
40 73.86 28.75 19.20 242.99 78.80
50 123.52 46.95 32.17 360.91 122.56
60 199.40 74.03 51.88 521.11 184.80
70 311.81 113.09 80.85 733.56 270.94
80 473.68 167.87 122.14 1009.24 387.31
90 700.92 242.79 179.43 1360.10 541.08
100 1012.61 342.96 256.99 1798.94 740.30
110 1431.24 474.15 359.69 2339.34 993.82
120 1982.85 642.84 493.02 2995.6 1311.27
130 2697.13 856.12 663.00 3782.64 1702.97
140 3607.45 1121.70 876.25 4715.99 2179.92
150 4750.89 1447.88 1139.86 5811.8 2753.76
160 6168.19 1843.50 1461.44 7086.83 3436.65
170 7903.72 2317.93 1849.08 8558.52 4241.33
180 10005.41 2881.04 2311.31 10245.05 5181.06
190 12524.68 3543.17 2857.16 12165.48 6269.61
200 15516.38 4315.18 3496.11 14339.85 7521.31

Appendix 142
3 Calibration
3.1 The mass flow controller ( MFC )
3.1.1 The MFCs for the study of TPR and TPO
The calibrations of the MFC were carried out in the gas mixture of H2 in argon and O2 in
argon, respectively. The results are shown in Fig.11 and 12.
140
MFC ( H2 in Ar )
120 PDay = 751.8 mmHg, TDay = 23.2 C

-1
VSTP / ml min
100 VSTP = -3.2667 + 1.3925 V%
80
60
40
20
0
0 20 40 60 80 100
V/%
Fig.11. The calibration of MFC ( H2 in Argon ).

Appendix 143
140
MFC ( O2 in Ar )
120
PDay = 751 mmHg, TDay = 24.5 C
-1
VSTP / ml min
100
VSTP = -3.7667 + 1.3132 V%
80
60
40
20
0
0 20 40 60 80 100
V/%
Fig. 12. The calibration of MFC ( O2 in Argon )
3.1.1 The MFCs for the study of the dehydrogenation of ethylbenzene
Three MFCs were used in the experimental set up for the study of the dehydrogenation of
ethylbenzene to styrene. The calibrations of these MFCs were carried out in argon and
helium, respectively. The results of the calibrations are exhibited in Fig. 13 14.
Appendix 144
80
MFC1 ( Ar )
PDay = 745.4 mmHg, TDay = 21 C

60
-1
VSTP = 6.0565 + 1.3894 V%

VSTP / ml min
40
20
0
0 10 20 30 40 50
V/%
Fig.13. The calibration of MFC for the Argon used for the EB saturator.
30
MFC2 ( Ar )
25
PDay = 745.4 mmHg, TDay = 21 C
-1
20 VSTP = 0.8597 + 0.4757 V%

VSTP / ml min
15
10
0
0 10 20 30 40 50
V/%
Fig.14. The calibration of MFC for the Argon used for the H2O saturator.
Appendix 145
140 MFC3 ( He )
120 PDay = 751.4 mmHg, TDay = 21 C

-1
100 VSTP = -7.2753 + 15.4749 V%

VSTP / ml min
80
60
40
20
0
1 2 3 4 5 6 7 8 9
V/%
Fig.15. The calibration of MFC for the Helium used for the H2O evaporator.
3.2 The calibrations of Binos, Magnos and Uras
The gas analyzer Binos was used for the monitoring of H2. The concentration of H2 was
displayed in the indicator and recorded in the computer simultaneously. After the H2
concentration of the reducing gas was input, the calibration could be performed by Binos
automatically. The verification of the set up for the study of TPR was carried out with a
sample of pure powder CuO. 20.1 mg sample was loaded in the reactor and the
reduction temperature was increased at a heating rate of 15 C min-1 from 25 C until 400
C. The TPR result is shown in Fig. 16.

Appendix 146
-6
1.4x10 Integration von A041001TPR15K_VerbH2 ab 0:
i = 1801 --> 6084
-6 x = 359,997 --> 1223,197
1.2x10
-1
Area Peak at Width Height

------------------------------------------------------------
2.688E-4 1095.99 153.4 1.429E-6
-6
1.0x10
-7
8.0x10
-7
6.0x10
-7
4.0x10
-7
2.0x10
0.0
0 500 1000 1500 2000 2500
Time / s
Fig.16 TPR pattern of the CuO powder at the heating rate of 15 C min-1.
After the integration of the TPR profile, a total H2 consumption amount of 2.68810-4 mol
was obtained. Assuming that the powder CuO of 20.1 mg was completely reduced to
Cu, the theoretical amount of H2 consumption could be given by
20.1
n consumed H 2 = 10 3 = 2.527 10 4 mol
63.55 + 16
This value is in good agreement with that obtained in the TPR investigation.
The Magnus detector was used for the monitoring the concentration of O2, and the Uras
detector for the concentrations of CO and CO2. The calibration of Magnus was carried out
with a standard gas mixture O2 in Argon of 20.5 %. After the O2 concentration of 20.5%
was input, the calibration was performed automatically. The standard gas mixtures of CO
( 4.93 % ) and CO2 ( 5.03 % ) were used for the calibration of Uras.
Appendix 147
3.3 Gas chromatography
The calibration factors of the components ethylbenzene, styrene, benzene and inert gas
argon were achieved by loading ethylbenzene, styrene, benzene or toluene in the
saturator, respectively. The following example is for the calibration of the component
ethylbenzene. Ethylbenzene was loaded in the saturator and the inert gas argon was led
through the saturator and reached the GC at the room temperature. The peak profiles of
ethylbenzene and argon were monitored and the peak areas were integrated. At the
atmospheric pressure of 755.4 mmHg and room temperature of 28.8 C, the partial
pressure of ethylbenzene could be obtained from the thermal data in Table 4.

16.93 9.54
0
PEtB = 8.8 + 9.54 = 16.04 mbar
10
Hence, the ethylbenzene concentration of the gas mixture was obtained by
16.04 10 3
C EB = = 1.614 %
755.4 / 760
The average peak area of the ethylbenzene could be calculated as following:
305680 + 304187 + 305821

SEB = = 305229
3
The concentration of ethylbenzene could be expressed as:
C EB = K EB 305229
Therefore, the calibration factor of ethylbenzene was achieved by

Appendix 148
1 1
K EB = C EB = 1.614 % = 5.29 10 -8
305229 305229
As to the calibration factor of argon, the concentration of argon should be determined
first
C Ar = 1 1.614 % = 98.386 %
and then the average peak area of the argon profile could be obtained by
1133296 + 1133632 + 1135534

S Ar = = 1134154
3
Finally, the calibration factor of argon was calculated out through
1 1
K Ar = C Ar = 98.386 % = 8.67 10 -7
1134154 1134154
The calibration factors of styrene, benzene and toluene could be obtained in the same
way. The results of the calculation factors are listed in Table 5.
Table 5 The calibration factors of the different components for GC
Component K KAr
.
EB 5.29 10-8 8.67 .
10-7
.
ST 5.52 10-8 8.62 .
10-7
.
BZ 5.89 10-8 8.76 .
10-7
.
TOL 5.51 10-8 8.62 . 10-7
8.62 . 10-7 ( K Ar )
Appendix 149
3.4 Mass spectrometer

The calibration of the mass spectrometer could be performed using the standard gas
mixture. The gas mixtures of ethylbenzene, styrene, benzene, toluene and water were
obtained by passing argon through the saturator loaded with ethylbenzene, benzene,
toluene or water. The following example is for the calibration of the MCD parameter for
the component ethylbenzene. At the atmospheric pressure of 755.4 mmHg and room
temperature of 26.5 C, the partial pressure of ethylbenzene could be obtained from the
thermal data in Table 4.

16.93 9.54
0
PEtB = 6.5 + 9.54 = 14.34 mbar
10
Then, the ethylbenzene concentration of the gas mixture could be expressed by
14.34 10 3
C EB = = 1.44 %
755.4 / 760
The program Gas Specific Calibration was carried out, and the calibration result of the
component ethylbenzene and argon is exhibited in Table 6.
Table 6 The gas calibration factors for ethylbenzene and argon

Appendix 150
The obtained calibration factors of ethylbenzene and argon were included into the MCD
matrix. The gas mixture of ethylbenzene and argon was analyzed by the mass
spectrometer, and the result of the measurement is shown in Fig.17. The measured
concentration of ethylbenzene is 1.434 %, which is in good agreement with the actual
concentration of the gas mixture of 1.440 %.
Fig.17. The measurement of gas mixture of ethylbenzene and argon.
The gas mixtures of CO, CO2, methane, ethylene and hydrogen were obtained from the
standard gas bottles. The results of the calibration factors of all the components are
summarized in the Table 7.

Appendix 151
Table 7 The calibration factors of all the components for the MCD matrix
Comp He Ar EB ST TOL BZ CO CO2 C2H4 CH4 H2 H2O
Mass
2 5.218
4 1.0
12 0.3081 0.5333
15 4.725
16 0.08489 0.8758 5.340 0.04795
17 0.9159
18 0.002712 3.9480
20 0.259900
26 0.2886
28 4.416 0.7877 4.5270
36 0.009631
40 2.090000
44 3.531
52 0.1892 0.2849 0.06728 0.6666
77 0.2995 0.4910 0.02792 0.4945
78 0.2515 0.9265 0.004997 2.069
91 2.4440 0.01065 1.9290 0.000813
92 0.1873 0.006286 1.1400 0.000479
103 0.0797 0.6051 0.001744 0.000427
104 0.0314 1.2300 0.001271 0.000360
106 0.6042 0.01097 0.001142 0.000404
Resume
Name Xiaomeng Zhu

Gender Male
Date of Birth Oct.14, 1964
Place of Birth Shanghai, P.R.China
Marital Status Married
July 1981 Graduated from high-school in Tongliao, P.R.China

Sep.1981 July1985 Department of Chemical Engineering, East China University of
Science and Technology ( Shanghai ),
as undergraduate student, B.E. degree, July 1985
Aug.1985 June1987 Institute of Chemical Reaction Engineering, East China University
of Science and Technology ( Shanghai ),
as research assistant
July 1987 July 1990 Institute of Chemical Reaction Engineering, East China University
as graduate student with the thesis:
Study on Benzene Catalytic Oxidation Kinetics by Transient
Response Method
Aug.1990 Nov.1998 Institute of Chemical Reaction Engineering, East China University
as lecturer with the following research directions:
Transient Response Method for the Reaction Study
Supercritical Reaction Technology
Electrochemical Synthesis with Packed-Bed Technology
Dec.1998 Mar.1999 Goethe Institute ( Mannheim ), learning German
April 1999 present Lehrstuhl fr Technische Chemie, Ruhr-Universitt Bochum,
as Ph.D. student supervised by Prof. Dr. M. Muhler with the
Dissertation:
Transient Kinetic Studies on the Dehydrogenation of Ethylbenzene
to Styrene over Potassium-Promoted Iron Oxide

Transient Kinetic Studies On The Dehydrogenation PDF

Загружено:

Сведения о документе

Оригинальное название

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

Transient Kinetic Studies On The Dehydrogenation PDF

Загружено:

Авторское право:

Доступные форматы

Transient Kinetic Studies on the Dehydrogenation

of Ethylbenzene to Styrene over Potassium-

am Lehrstuhl fr Technische Chemie der Ruhr-Universitt Bochum unter der

Leitung von Prof. Dr. M. Muhler angefertigt.

Vorsitzender der Prfungskommission: Prof. Dr. Ch. Wll

Referent: Prof. Dr. M. Muhler

Korreferent: Prof. Dr. W. Grnert

Dritter Prfer: Prof. Dr. W.S. Sheldrick

Tag der mndlichen Prfung: 12. 12. 2002

2 Kinetic Study on the Temperature-Programmed Reduction ( TPR ) 2

2.4 Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . 10

2.4.1 Dilution of the sample with quartz . . . . . . . . . . . . . . . . . . . . . . . . . . . . .10

2.4.2 Effect of sample weight . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .14

2.4.3 Heating rate . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .16

2.4.4 Total flow rate of reducing gas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .19

2.4.5 Influence of H2O in the reducing gas . . . . . . . . . . . . . . . . . . . . . . . .. . . .20

2.4.6 Activation energy. . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . 23

2.5 Conclusions. . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .28

3 Transient Kinetic Studies on the Commercial Catalyst 32

3.1 Introduction. . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .32

3.3 Results and discussion. . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .38

3.3.1 The blank experiments. . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . .. . . . . 38

3.3.2 The initial activation period of the fresh catalyst. . . . . . . . . .. . . . . . . . . . 40

3.3.3 The regeneration of the catalyst with steam. . . .. . . . . . . . . . . .. . . . . . . 43

3.3.4 The transient behavior of the catalyst after regeneration. . . . .. . . .. . . .45

3.3.5 The investigation of deactivation phenomena. . . . . . . . .. . . . . . . . . . . . .48

4 Initial Transient Behavior of the Iron Oxide Catalysts 53

4.1 Introduction. . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .53

4.2.1 Catalyst preparation. . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .55

4.2.2 Characterization techniques. . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . .55

4.2.3 Transient experiments. . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . 57

4.3 Experimental Results . . . . .. . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .58

4.3.1 Concentration step-up on the fresh catalysts. . . . . . . . . .. . . . . . . . . . .. . 58

4.3.2 Steaming of the used catalysts. . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . 61

4.3.3 Concentration step-up after the steaming. . . . . . . . . .. . . . . . . . . . . . . . . . 64

4.3.4 X-ray diffraction . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67

4.3.5 Temperature-programmed Reduction. . . . . . . . . .. . . . . . . . . . . . . . . .. . .70

4.3.6 Temperature-programmed Oxidation . . . . . . . . .. . . . . . . . . . . . . . . . . . .76

4.4.1 Reduction of Fe2O3. . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . 81

4.4.2 Effect of the steaming. . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . .. . . . . . .85

4.4.3 K2CO3 in the K-promoted catalysts. . . . . . . . . .. . . . . . . . . . . . . . . . . . . .87

4.4.4 Decomposition of KFeO2. . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . .91

4.4.5 Schematic life cycle of the catalyst. . . . . . . . . .. . . . . . . . . . . . . . . . .. . . . 93

5 Kinetic Parameter Optimization and Simulation Process 99

5.2 Experimental. . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101

5.3 Results and discussion. . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102

5.3.1 Kinetic data . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102

5.3.2 Kinetic Model . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .106

5.3.3 Kinetic data fitting. . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .112

5.4 Conclusions. . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .120

1. Chemical reagents and gases. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..125

2. Experimental apparatuses. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126

2.1.1 The full view of the set-up . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .126

2.1.2 The reactor for the TPR and TPO . . . . . . . . . . . . . . . . . . . . . . . . . . . 127

2.1.3 The isothermal period of the oven . . . . . . . . . . . . . . . . . . . . . . . . . . . 129

2.1.4 The delay of the concentration of H2 . . . . . . . . . . . . . . . . . . . . . . . . . 129

2.1.5 The consumption of H2 for the study of TPR . . . . . . . . . . . . . . . . . . .131

2.1.6 The consumption of O2 for the study of TPO. . . . . . . . . . . . . . . . . . . 132

2.2 The set-up for the study of dehydrogenation of ethylbenzene. . . . . . . . . . . .133

2.2.1 The full view of the set-up . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133

2.2.2 The reactor for the dehydrogenation of ethylbenzene to styrene. . . . 136

2.2.3 The determination of LHSV . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137

2.2.4 The saturators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139