Catalytic reactions of mixtures of carbon dioxide, ethene, and hydrogen on cobalt surfaces

Article abstract of DOI:10.1006/jcat.1999.2623

A study on the hydrogenation of mixtures of CO₂ and ethylene over unsupported cobalt using a plug-flow and a batch reactor showed that the hydrogenation of CO₂ is blocked as long as ethylene is present in the gas phase. In contrast, CO₂ did not influence the hydrogenation of ethylene that led to a complex mixture of hydrocarbons. The relation can be explained as a copolymerization proceeding via two different monomers (C₁ and C₂ fragments). A numerical model can quantitatively interpret the observed oscillatory molar product distribution in dependence on the number of C atoms.

Full text of DOI:10.1006/jcat.1999.2623

Journal of Catalysis 187, 348–357 (1999)
Article ID jcat.1999.2623, available online at http://www.idealibrary.com on
Catalytic Reactions of Mixtures of Carbon Dioxide, Ethene,
and Hydrogen on Cobalt Surfaces
M. Voß, G. Fro¨hlich, D. Borgmann,1 and G. Wedler
Institute of Physical and Theoretical Chemistry, University of Erlangen-Nu¨rnberg, Egerlandstr. 3, D-91058 Erlangen, Germany
Received February 16, 1999; revised June 29, 1999; accepted June 29, 1999
and the mechanism of the hydrogenation of CO2 on silica
supported Ni (and other group VIII metals) are still con-
troversely discussed (16).
Using a plug-flow and a batch reactor the hydrogenation of mix-
tures ofcarbon dioxide and ethene over unsupported cobalt has been
studied. The results show that the hydrogenation of carbon dioxide
is totally blocked as long as ethene is present in the gas phase. In
contrast carbon dioxide does not at all influence the hydrogenation
of ethene, which leads to a complex mixture of hydrocarbons. This
reaction can be explained as a copolymerisation proceeding via two
different monomers (C1 and C2 fragments). The observed oscilla-
tory molar product distribution in dependence on the number of
C atoms can exactly be described by means of a numerical model.
The present paper is part of a detailed model study of the
hydrogenation of carbon dioxide over pure metallic cobalt
catalysts and cobalt catalysts supported on Al2O3, SiO2, and
TiO2. The study comprises the characterisation of the cata-
lysts by means of various analytical methods (10), their acti-
vation and deactivation (17, 18), and the investigation of the
reaction kinetics, as well as of the reaction mechanism (19).
Since the addition of carbon monoxide to the CO2/H2
feed led to interesting results concerning the activation and
deactivation of the catalysts (20) the question concerning
the action of other additives arose. It is well known from
the literature (e.g., 21, 22) that ethene is able to act as a
chain initiator and to influence both yield and selectivity of
the hydrogenation of carbon monoxide (Fischer–Tropsch
reactions). Thus it seemed to be worth investigating also
the hydrogenation of carbon dioxide–ethene mixtures.
c
1999 Academic Press
Key Words: cobalt; carbon dioxide; ethene; hydrogen; plug-flow
reactor; batch reactor; Auger electron spectroscopy.
1. INTRODUCTION
For several decades the Fischer–Tropsch synthesis on
group VIII metals has been an intensely studied field in
catalysis research and has been well described in the litera-
ture (1–6). The mechanism of CO hydrogenation on various
Co catalysts plays a dominant role. Product distribution and
paraffin/olefin ratios over polycrystalline Co foils have been
measured by Kuipers et al. (7, 8). Bartholomew et al. (9)
have studied the adsorption of CO on alumina supported
cobalt catalysts. They report that at 523 K dissociation of
CO leads to the deposition of carbon which can be hydro-
genated to methane. These results correlate very well with
our own measurements (10).
2. METHODS
In the course of the studies two types of apparatuses have
been used which differ not only in the type of the reac-
tor (plug-flow reactor and batch reactor) but also in the
whole experimental equipment. They shall be separately
presented. The catalyst used, however, was in all kinds of
experiments a cobalt foil (Goodfellow), 0.05 mm thick and
of purity 99.9% .
As compared with the hydrogenation of CO there are
relatively few publications on the hydrogenation of CO2.
Reaction rates, product distributions, activities, and selec-
tivities were determined on silica supported Co, Fe, and
Ru catalysts (11), on polycrystalline Rh foils (12), and on
alumina supported Rh and Ru catalysts (13, 14), as well as
on silica and titania supported Rh (14). The high activity
of the supported Rh catalysts was attributed to a strong
metal-support interaction (SMSI effect). This effect could
not be confirmed in the case of CO (15) and CO2 (10) hy-
drogenation on Co/Al2O3, Co/SiO2, and Co/TiO2. Kinetics
2.1. Plug-Flow Reactor
The plug-flow apparatus comprises a gas supplying sys-
tem, a plug-flow reactor equipped with a programmable
heating device, and a gas analysing system. The gas supply-
ing system consists of a gas handling system which allows
further purification of the high purity gases introduced from
gas flasks, gas dosing by means of flow controllers (Brooks,
Bronkhorst Hi-Tec), and gas mixing. The gas leaving the re-
actor can be dosed into a gas chromatograph (Carlo Erba,
VEGA Series GC 6000) supplied with a Carbowax–Porasil
column and a flame ionisation detector (for hydrocarbons)
1 To whom correspondence should be addressed.
348
0021-9517/99 $30.00
c
Copyright
1999 by Academic Press
All rights of reproduction in any form reserved.

INTERACTION OF CO2 AND C2H4 IN HYDROGENATION REACTIONS
349
as well as a Porapack QS column and a thermal conductivity dioxide and vice versa. The use of a plug-flow reactor has
detector (for carbon monoxide and carbon dioxide). Details proven to be the best way to answer this general question.
of this apparatus have been published elsewhere (18).
3.1.1. Influence of admission of ethene on the hydrogena-
tion of carbon dioxide. Figure 1 refers to the hydrogena-
tion of carbon dioxide on cobalt foils at T = 493 K. The
feed consisted of 800 mbar hydrogen and 200 mbar carbon
dioxide. The volumetric flow rate dV/dt was 60 mlN/min.
The turnover frequency of the product gases (carbon mono-
xide, methane, and ethane) is plotted in logarithmic scale
against the time on stream.
Since the surface of the cobalt foils had been oxidised
before the start of the experiment there is an induction
period at the beginning, in which the surface is reduced.
Steady-state conditions are achieved after about 60 min.
Then the composition of the product gas is 54% CH4, 40%
CO, and 6% C2H6. Under these conditions the conversion
of carbon dioxide amounts to less than 2% .
The cobalt foil was cut into 7 mm 1 mm 0.05 mm
pieces. Twenty-five pieces were introduced into the reactor.
The catalyst was activated by at least 10 reduction and oxi-
dation steps (15 mlN/min H2, 873 K, 10 h; 15 mlN/min 0.1%
O2 in N2, 823 K, 30 min) until its activity concerning the
hydrogenation of carbon dioxide was constant. BET mea-
surements showed that the surface of the activated cobalt
foils exceeds the geometric surface area by a factor of 400.
Further details concerning activation and deactivation of
the catalyst have been reported elsewhere (18).
Measurements with the plug-flow reactor were carried
out in the temperature range 433 K
T
533 K, since at
lower temperatures the activation of the catalyst by the feed
was too slow and at higher temperatures the deactivation
by deposition of carbon was too fast.
At t = 150 min 16 mbar ethene were admitted to the feed.
This results in a drastic change of the composition of the
product stream. The turnover frequencies of methane and
ethane increase by a factor of 6 and 450, respectively. That
of carbon monoxide decreases by 25% . The conversion of
carbon dioxide decreases and that of ethene amounts to
>95% . In addition significant amounts of both linear and
branched alkanes and alkenes (up to C7) are detected. The
concentration of the saturated hydrocarbons exceeds that
of the corresponding unsaturated hydrocarbons by one or-
der of magnitude. The latter are reactive intermediates as
will be shown later on.
2.2. Batch Reactor
The other apparatus is equipped with a continuously
stirred batch reactor (V = 275 ml), a gas chromatograph
(Carlo Erba, MEGA Series HRGC 5300), an ultrahigh vac-
uum (UHV) chamber with an Auger electron spectrometer
(AES), and facilities for cleaning or preparing the cobalt
foil, as well as a mass spectrometer in order to take thermal
desorption spectra. A differentially pumped ultrahigh vac-
uum lock system connects the reactor and UHV chamber.
The catalyst is fixed in a cut-out of a transfer rod and is
heated by passing an electrical current through the cobalt
foil. It can be transferred from the reactor into the UHV
chamber and back by means of the transfer rod within a few
minutes. Columns and detectors of the gas chromatograph
are equal to those used in the plug-flow apparatus. Details
of the whole device have been published in preceding pa-
pers (20, 23).
This experiment indicates that the reaction is dominated
by the hydrogenation of the admitted ethene and points to
a hindrance of the hydrogenation of the carbon dioxide by
the admitted ethene.
The cleaning procedure of a new foil started with re-
peated heating up to 1273 K, sputtering with neon ions
(INe = 2.5 A, ENe = 750 eV at 573 K) followed by reduc-
tion in 1 bar H2. Between the different experiments both
sides of the foil were sputtered with neon ions for 15 min.
Oxidation in 100 mbar O2 at 873 K for 30 min and reduction
in 1000 mbar H2 at 873 K for 12 h completed the procedure.
Auger electron spectroscopy has been used to control
the cleaning process as well as to analyse the composition
of the catalyst surface during the activation and the catalytic
reaction.
3. RESULTS AND DISCUSSION
3.1. Hydrogenation of Carbon Dioxide–Ethene Mixtures
in the Plug-Flow Reactor
The first question that should be answered is whether the
FIG. 1. Hydrogenation of carbon dioxide on cobalt foils in a plug-flow
presence of ethene influences the hydrogenation of carbon reactor at T = 493 K. At t = 150 min ethene was admitted to the feed.

350
VOß ET AL.
3.1.2. Influence of the partial pressures of carbon diox- considered. The reaction is dominated by the direct hydro-
ide and hydrogen on the selectivity of the hydrogenation genation to ethane (selectivity 95% ; note that the logarith-
of ethene. These experiments were performed at 433 K. mic scale is enlarged in the upper part of Fig. 2a). However,
Since the partial pressure of one of the components of the methane, propane, and propene, as well as n-butane and
feed was changed in the course of the experiments, he- 1-butene are formed too. It is interesting to note that under
lium was admitted to adjust the total pressure always to the given conditions the hydrocarbons with an even num-
1000 mbar. The volumetric flow rate was again 60 mlN/min. ber of C atoms exhibit a higher selectivity than those with
In the first case the partial pressures of the feed were an odd number of carbon atoms. This observation will be
p(H2) = 800 mbar, p(C2H4) = 16 mbar, p(CO2) = variable. picked up later in this paper.
Figure 2a shows the selectivity of the hydrogenation of
In the second case the influence of the hydrogen pres-
ethene in logarithmic scale at varying carbon dioxide pres- sure on the selectivity of the hydrogenation of ethene was
sures. There is clear evidence that carbon dioxide does not studied. The partial pressures were p(CO2) = 200 mbar,
influence the selectivity observed with the hydrogenation p(C2H4) = 16 mbar, p(H2) = variable. Figure 2b shows
of pure ethene ( p(CO2) = 0 mbar), even when the partial the result. An increase in hydrogen pressure leads to
pressure of CO2 exceeds that of ethene by a factor of 12. an increase of the direct hydrogenation of ethene. Si-
The same holds when instead of the selectivity the yield is multaneously the probability of chain growth diminishes.
FIG. 2. (a) Selectivity of the hydrogenation of ethene as a function of the carbon dioxide pressure at T = 433 K. (b) Selectivity of the hydrogenation
of ethene as a function of the hydrogen pressure at T = 433 K. (c) Turnover frequency of the product gases as a function of the partial pressure of
ethene at T = 433 K.

INTERACTION OF CO2 AND C2H4 IN HYDROGENATION REACTIONS
351
This affects the formation of unsaturated hydrocarbons are independent of the partial pressure of carbon dioxide.
(propene and 1-butene) more than the formation of satu- They are even valid in the absence of carbon dioxide.
rated hydrocarbons (propane and n-butane), since the un-
3.2. Hydrogenation of Carbon Dioxide–Ethene Mixtures
in a Batch Reactor
saturated hydrocarbons are reactive intermediates, which
are easily hydrogenated to the saturated compounds. In this
way the sequence in the selectivity can be changed in depen-
The situation changes when the catalytic reaction is not
dence on the hydrogen pressure (see propane and butane).
carried out in a plug-flow reactor under steady state condi-
The formation of methane, i.e., the fragmentation reaction,
tions, but in a batch reactor, i.e., in a closed system. Then, ac-
is much less influenced by the hydrogen pressure.
cording to the results dealt with in the preceding paragraph,
after some time the store of ethene will be consumed and
the hydrogenation of carbon dioxide will follow. Therefore
studies by means of a batch reactor will be suitable to inves-
tigate the mutual influences of carbon dioxide and ethene
3.1.3. Influence of the partial pressure of ethene on
the turnover frequencies of the hydrogenation reactions.
These experiments were also performed at T = 433 K.
The partial pressures were p(CO2) = 200 mbar, p(H2) =
in hydrogenation reactions in more detail.
800 mbar, p(C2H4) = variable. The volumetric feed rate was
60 mlN/min.
3.2.1. The induction period. The reaction described in
Fig. 3a was performed in the batch reactor at 523 K un-
der the initial conditions p(H2) = 800 mbar, p(CO2) =
200 mbar, p(C2H4) = 80 mbar. In this case it is expedient to
plot the concentrations of the substances in the gas phase
as a function of reaction time. It is evident that there is
a rather long induction period before the reactions start.
This is due to the circumstance that the cleaning process
of the cobalt foil ended with an oxidation step, i.e., that at
the beginning the catalyst surface consists of cobalt oxide.
Figure 3a shows that neither carbon dioxide nor ethene can
be hydrogenated over the oxidised surface. Between 30 and
60 min the hydrogenation of ethene starts accompanied by
fragmentation under the formation of methane. Only when
most of the ethene has been consumed the hydrogenation
of carbon dioxide sets in. At a reaction time of 90 min the
conversion of ethene amounts to 89% and that of carbon
dioxide to 1% . At 120 min nearly no ethene has been left.
From this time onward no significant increase in the con-
centration of ethane can be observed. There is, however, a
strong formation of methane, the main product with the hy-
drogenation of carbon dioxide. As soon as carbon dioxide
begins to react carbon monoxide can be detected, however,
only in a very small concentration.
The verticaldotted linesin Fig. 3a mark the reaction times
at which the catalyst has been transferred via the differen-
tially pumped lock system into the UHV chamber in order
to take Auger electron spectra. For this time (about 15 min)
the catalytic reaction is interrupted. As soon as the catalyst
is brought back into the batch reactor the reaction goes on
without any disturbance, as follows from the fact that the
values of the concentrations determined before and after
the interruption vary only within the limit of error.
Figure 3b shows the Auger electron spectra. Just at the
beginning of the experiment the oxygen signal is dominat-
ing and the peak-to-peak height of the L3M45M45 cobalt
signal at 775 eV relative to the corresponding values of the
L3M23M23 and L3M23M45 cobalt signals at 655 and 715 eV,
respectively, indicate that the cobalt at the surface is not
metallic.
In Fig. 2c the logarithm of the turnover frequency of the
product gases has been plotted against the partial pres-
sure of ethene. As expected the turnover frequencies in-
crease with increasing ethene pressure, since under the
given conditions the active sites are not saturated with
ethene. The conversion of ethene amounts to >90% ; i.e.
the gas leaving the reactor still contains ethene. This sit-
uation will prove to be the cause for the observation that
carbon dioxide is not at all hydrogenated in these experi-
ments.
3.1.4. Influence of temperature on the selectivity of the hy-
drogenation reactions. The influence of the reaction tem-
perature on the hydrogenation reactions has been com-
piled in Table 1 for the temperature range from 433 to
523 K. The composition of the feed was p(H2) = 800 mbar,
p(CO2) = 200 mbar, p(C2H4) = 16 mbar, and the volumetric
feed rate amounted to 60 mlN/min.
The total selectivity of all the hydrocarbons not men-
tioned in Table 1 was less than 1% . At all temperatures the
reactions which occur are dominated by the direct hydro-
genation of ethene. However, the fragmentation of ethene
strongly increases with temperature. This influences not
only the formation of methane but also the formation of
propane, n-butane, and n-pentane. The observation that
the selectivity for the latter two products passes through
a maximum is due to the increase of hydrogenolysis as will
be discussed later on. The trends following from Table 1
TABLE 1
Influence of Temperature on the Selectivity of the
Hydrogenation Reactions
Temperature (K) Methane Ethane Propane n-Butane n-Pentane
433
463
493
523
1.0
4.6
11.5
27.7
94.5
85.3
75.8
63.1
2.3
5.8
8.0
6.3
1.2
3.2
3.4
2.0
0.1
0.3
0.6
0.3

352
VOß ET AL.
The reduction of the oxidic surface needs atomic hydro-
gen which can only be formed on the metallic surface. The
same holdsfor the hydrogenation ofethene which isalso ad-
sorbed only on the metallic surface. Hydrogen and ethene
compete for the initially small number of metallic sites. In
this way ethene hinders the reduction of the cobalt oxide.
The same effect has been observed with carbon monoxide,
which prolongs the induction period, when carbon dioxide
is hydrogenated over initially oxidised cobalt (20).
When the cleaning procedure of the cobalt foil is not
finished with the oxidation step but with reduction no in-
duction period exists, the reaction between ethene, carbon
dioxide, and hydrogen starts immediately with the hydro-
genation of ethene. Already after 4 min 70% of the initial
amount of ethene have been consumed. When after 30 min
the second sample is taken for gas chromatographic analy-
sis no ethene can be detected and already 20% of the initial
amount of carbon dioxide have been hydrogenated. At a
reaction time of 560 min the composition of the gas phase
is identical with that shown in Fig. 3a for the same reaction
time.
3.2.2. Byproducts formed during the induction period.
The rather slow rate of reaction during the induction pe-
riod allows a detailed analysis of the composition of the gas
phase. In Fig. 4a the logarithms of the concentrations of
the saturated hydrocarbons are plotted in dependence on
the reaction time. The approximately constant slope of the
curves between 30 and 90 min points to an approximately
exponential increase of their concentrations with time. It
is evident that the higher hydrocarbons are only formed as
long as ethene is available. That means that they are not a
product of the hydrogenation of carbon dioxide. The selec-
tivities determined after a reaction time of 90 min are in the
cases of CH4 10.8% , C2H6 70.2% , C3H8 7.9% , C4H10 4.5% ,
and C5H12 0.6% .
Figure 4b shows that during the induction period also
the corresponding unsaturated hydrocarbons are formed.
However, as soon as the concentration of ethene has
dropped to only some percent of its inital value they dis-
appear completely. They seem to be active intermediates,
which are replaced from the surface by ethene before they
can be hydrogenated. However, when the ethene concen-
tration is low enough, they are readsorbed and hydro-
genated.
The correctness of this argumentation is supported by
the fact that up to 200 min the carbon content of all hydro-
carbons is balanced within an error of 2.5% when carbon
dioxide is not taken into account.
FIG. 3. (a) Hydrogenation of carbon dioxide and ethene on cobalt
foils in a batch reactor at T = 523 K. The dashed lines represent where the
Auger electron spectra of Fig. 3b were taken. (b) Auger electron spectra
of the cobalt foil taken during the hydrogenation reaction.
The ratio of the peak-to-peak heights of the signals of
oxygen and cobalt at 515 and 775 eV, respectively, remain
nearly constant up to t = 90 min. Between 90 and 120 min
3.2.3. Hydrogenation of ethene in the batch reactor in ab-
this ratio diminishes by a factor of 6 10 2. Simultaneously sence of carbon dioxide. The results discussed in the pre-
the cobalt signal at 775 eV increases relative to the cobalt ceding paragraphs seem not to show any influence of car-
signals at 655 and 716 eV. This observation clearly indicates bon dioxide on the reactions as long as ethene is present.
that the reduction of the surface occurs mainly between 90 In order to check whether this statement is correct, the hy-
and 120 min, just when the ethene has been consumed.
drogenation of ethene in the batch reactor should also be

INTERACTION OF CO2 AND C2H4 IN HYDROGENATION REACTIONS
353
FIG. 4. (a) Concentration of the saturated hydrocarbons in dependence on the reaction time. (b) Concentration of the unsaturated hydrocarbons
in dependence on the reaction time.
studied in the absence of carbon dioxide as was done in constants have been compiled in Table 2. The rate constant
the case of the experiments with the plug-flow reactor (see of hydrogenolysis increases with increasing chain length;
Fig. 2a, p(CO2) = 0 mbar). For this purpose carbon dioxide linear hydrocarbons react faster than branched hydro-
was substituted by neon. The initial composition of the gas carbons.
phase was p(H2) = 800 mbar, p(Ne) = 200 mbar, p(C2H4) =
80 mbar; the reaction temperature was T = 523 K. The cata-
lyst cleaning was finished with the oxidation step.
In Fig. 5a the concentrations of ethene, ethane, and
methane have been plotted in dependence on the reaction
time. The induction period is considerably shorter than in
the presence of carbon dioxide (Fig. 3a). It is more pro-
nounced for the fragmentation reaction than for the hydro-
genation of ethene. The fact that carbon dioxide prolongs
the induction period without being involved in the hydro-
genation reactions points most likely to the formation of a
carbonate species on the oxidised cobalt surface, which is
responsible for the long induction period. The formation of
carbonate after coadsorption of oxygen and carbon dioxide
on transition metal surfaces (24, 25) or after adsorption of
carbon dioxide on metal oxide surfaces (26) is well known
from the literature.
Already after 60 min ethene has disappeared from the
gas phase. The following changes in the concentrations of
the products have to be traced back to consecutive reac-
tions. As can be seen from Fig. 5b the product spectrum
just after the consumption of ethene is very similar to that
observed in Figs. 3a, 4a, and 4b, when carbon dioxide had
not yet begun to react. The concentrations of all prod-
ucts with the exception of methane pass through a maxi-
mum at 90 min. Then they decrease, whereas the methane
concentration still increases. That means that hydrogenol-
ysis under formation of the thermodynamically more sta-
ble methane takes place. Since all curves in Fig. 5b (loga-
rithm of concentration against time) are straight lines this
hydrogenolysis occurs in first order kinetics (27). The rate cobalt foils in a batch reactor at T = 523 K.
FIG. 5. Hydrogenation of ethene in the absence of carbon dioxide on

354
VOß ET AL.
TABLE 2
pressures of hydrogen of 200 and 800 mbar, respectively.
The partial pressure of ethene was in both cases 16 mbar.
As already mentioned the formation of the hydrocarbons
with an even number of C atoms is favoured before the for-
mation of the molecules with an odd number of C atoms.
This partially leads to an oscillating behaviour of the curves
in Fig. 6.
In the following an attempt is made to describe the mech-
anism of the chain growth quantitatively. It is based on an
approach made by Novak et al. (28) and successfully ap-
plied by Jordan et al. (29) in order to explain oscillations in
the weight distribution for chain growth in Fischer–Tropsch
synthesis.
In the preceding paragraphs it has been discussed that
besides hydrogenation of ethene and chain growth, frag-
mentation of ethene also takes place. This means that chain
growth could occur by both C1 and independent C2 addi-
tion. This situation is explained in Fig. 7a. It refers to the
steady-state conditions in the plug-flow reactor.
The rate of desorption of the hydrocarbon with n car-
bon atoms (Cn,g) into the gas phase is given by the product
kt2n where 2n is the fraction of active sites occupied by
the adsorbate Cn and kt is the rate constant of desorption,
i.e., the rate constant of the termination reaction. The rate
of formation of Cn in the adsorbed layer is given on one
hand by the surface concentration of Cn 1 (2n 1), the sur-
face concentration of C1 (21), and the rate constant of this
propagation reaction (kp,1), i.e. by kp,12n 121 and on the
Rate Constants of Hydrogenolysis at 523 K
1
Hydrocarbons
k1 (min
)
5
4
3
4
3
2
Ethane
5.35 10
2.53 10
1.10 10
7.26 10
4.17 10
1.10 10
Propane
n-Butane
i-Butane
n-Pentane
Ethene
Hydrogenolysis cannot be observed in the presence of
carbon dioxide, i.e., when mixtures of carbon dioxide and
ethene or pure carbon dioxide are hydrogenated. When
carbon dioxide (or carbon monoxide) had been completely
hydrogenated and the water vapour had been removed by
freezing out in a cooling trap, hydrogenolysis of ethane and
propane was observed (19). The rate constants agreed with
those given in Table 2. This observation can be explained
by the assumption that carbon monoxide, carbon dioxide,
and water compete with the hydrocarbons for adsorption
sites and are more strongly bonded to the surface than the
latter ones.
3.3. Mechanism of the Formation of Higher Hydrocarbons
in Ethene–Hydrogen–Carbon Dioxide
Mixtures over Cobalt
As shown in Fig. 2a carbon dioxide does not at all influ-
ence the selectivity of the hydrogenation reactions, in con-
trast to the behaviour of hydrogen (Fig. 2b). An increase of
the partial pressure of hydrogen leads even to a change in
the sequence of the formed hydrocarbons in the selectivity
scale. From the data of Fig. 2b the mole fraction of the hy-
drocarbons can be calculated in dependence on the number
n of carbon atoms. This has been done in Fig. 6 for partial
other hand by the surface concentration of Cn (2n 2),
the surface concentration of C2 (22), and the rate constant
of the corresponding propagation reaction (kp,2), i.e., by
2
k
p,22n 222. 2n decreases not only by desorption of Cn but
also by the propagation reactions forming Cn+1 and Cn+2
,
respectively. The corresponding reaction rates are kp,1212n
and kp,2222n, respectively. Under steady-state conditions
2n does not change with time, i.e.,
2n = 0 = kp,1212n 1 + kp,2222n
kt2n
2
k
p,1212n
k
p,2222n.
[1]
Since also 21 and 22 do not change under steady-state con-
ditions Eq. [1] can be simplified by introducing
kp0 ,1 = kp,121 and kp0 ,2 = kp,222
[2]
resulting in
2n = 0 = kp0 ,12n 1 + kp0 ,22n
kt2n kp0 ,12n kp0 ,22n [3]
2
or
kp0 ,1
kp0 ,2
FIG. 6. Mole fraction of the hydrocarbons calculated from the data
of Fig. 2b in dependence on the number, n, of carbon atoms.
2n =
2n
+
2n 2. [4]
1
kt + kp0 ,1 + kp0 ,2
kt + kp0 ,1 + kp0 ,2

INTERACTION OF CO2 AND C2H4 IN HYDROGENATION REACTIONS
355
FIG. 7. (a) Reaction model of the hydrogenation of ethene at cobalt surfaces. See text for further explanations. (b) 22/21 as a function of the
hydrogen pressure at T = 433 K and an ethene pressure of 16 mbar. (c) Probabilities of chain growth 1 and 2 in dependence on the reciprocal of the
hydrogen pressure. (d) Ratio
/ 1 in dependence on the hydrogen pressure.
2
When
form
1
2n = xn
[8]
kp0 ,1
kp0 ,2
=
and
=
[5]
[6]
1
2
kt + kp0 ,1 + kp0 ,2
kt + kp0 ,1 + kp0 ,2
is tried by Novak et al. (28). Under these conditions Eq. [6]
is given by
are introduced Eq. [4] can be written as
2n = 12n 22n
1
2
3
xn
=
1xn
+
2xn
[9]
+
.
2
1
which is equivalent to
1 and 2 represent the probabilities of chain growth by C1
and C2 species, respectively. In order to be able to calculate
the mole fraction
x2
1x
= 0.
[10]
2
The solution of this equation is
2n
12n
+
22n
2
1
P∞
P
=
[7]
∞
n=1 2n
(
12n
+
22n )
2
1
n=1
q
1
2
2
1
x1,2
=
+ 4
.
2
[11]
1
2n has to be expressed as a function of n. A solution of the

356
VOß ET AL.
The general solution of Eq. [6] is therefore
cedures deliver the corresponding values of 22/21, 1, and
2 in dependence on pH and pC H , respectively.
2
2
4
Figure 7b shows that there is a linear relationship be-
1
1
2n = Ax1n + Bx2n
.
[12]
tween 22/21 and pH over the whole range of applied hy-
2
drogen pressures. The partial pressure of ethene amounted
to 16 mbar and the temperature was 433 K. 22, the cover-
age of C2 intermediates, exceeds that of C1 intermediates
(21) by about two orders of magnitude. This corresponds
rather well with the selectivities of the formation of ethane
and methane, respectively. The increase of 22/21 with in-
The coefficients A and B can be determined by the bound-
ary conditions for n = 1 and n = 2, respectively. It follows
that
22 21x2
21x1 22
A =
and B =
[13]
x1 x2
x1 x2
creasing pH may be due to an increase of 22, to a decrease
2
of 21, or to a superposition of both effects. Most likely the
direct hydrogenation of ethene is favoured with increasing
hydrogen pressure with the consequence that the fragmen-
tation reaction falls off. The same holds for the formation
of the higher hydrocarbons, and it is not surprising that the
unsaturated hydrocarbons are even more effected than the
saturated ones.
and
22 21x2
21x1 22
1
1
2n =
x1n
+
x2n
.
[14]
x1 x2
x1 x2
The mole fraction is given by
1
1
2n
Ax1n + Bx2n
n=1
The probabilities of chain growth,
and 2, exhibit a
1
P∞
=
.
[15]
[16]
P∞
1
1
Ax1n + Bx2n
linear relationship to the reciprocal of the hydrogen pres-
sure (Fig. 7c), at least in the range of the applied hydrogen
pressures. This becomes easy to understand, when Eqs. [5]
are taken into account. In the same way as kp0 ,1 = kp,121
(Eq. [2]) kt comprises the coverage of hydrogen or the
n=1 2n
∞
X
1
Ax1n = A x10 + x11 + x12 +
=
,
1
1
x1
n=1
hydrogen pressure, respectively. Both 2H and pH are
2
since x1 < 1, as follows from Eq. [11]. Inserting Eq. [16] into
constants under steady-state conditions. Independent of
whether the hydrogenation of the chemisorbed, probably
di- -bonded species occurs via a Langmuir–Hinshelwood
or an Eley–Rideal mechanism immediate0ly followed by
Eq. [15] leads to
1
1
2n
Ax1n + Bx2n
P∞
=
.
[17]
A
B
n=1 2n
desorption kt comprises pH . When kt
kp,1, kp,2, the de-
+
2
1
x1
1 x2
nominator in Eq. [5] is dominated by kt, which is probably
the case since
and
are in the order of magnitude of
1
2
This equation was used to fit the experimental data plotted
in Fig. 6. In order to reduce the fit parameters from four
2
10 to 10 1, i.e.,
and
are proportional to the reci-
1
2
procal of pH
.
2
(
,
2, 21, 22) to three (
, 2, r) the coverages 21 and 22
1
1
The ratio
/ 1, which does not contain kt, is on the order
2
were substituted by the ratio of the coverages
of magnitude of 1 and decreases linearly with pH (Fig. 7d),
2
as far as 200 mbar < pH < 800 mbar. Since
2
22
kp0 ,2
r =
,
[18]
k
k
p,222
p,121
2
1
21
=
=
[19]
kp0 ,1
the ratio of the rate constants kp,2 and kp,1 can be calculated,
because the ratio 22/21 is known. The ratio kp,2/kp,1 varies
from 0.02 ( pH = 200 mbar) to 0.003 ( pH = 800 mbar), i.e.,
so that A and B are finally given by
21(r x2)
2
2
21(x1 r)
the rate constant kp,2 of the chain growth via C2 is by 2 to 3
orders of magnitude smaller than the rate constant kp,1 of
chain growth via C1.
A =
and B =
.
[13a]
x1 x2
x1 x2
When A and B are defined in this way (Eq. [13a]) 21 no
longer appears in the mole fraction.
This section was so named because the results were ob-
tained with such gas mixtures. However, it has been clearly
The values of the mole fraction in dependence on n deter- demonstrated in the other paragraphs that under the given
mined on the basis of Eq. [17] by means of a least-square-fit conditions CO2 does not at all influence the reactions tak-
have been plotted in Fig. 6 as small symbols. The result is ing place. Therefore the discussed findings are characteris-
astonishingly good, not only in the case of partial pressures tic of the chain growth accompanying the hydrogenation of
of hydrogen of 200 and 800 mbar, but also for the whole ethene over Co and should not be compared with similar
range between these limiting pressure values. The fit pro- observations made with Fischer–Tropsch reactions.

INTERACTION OF CO2 AND C2H4 IN HYDROGENATION REACTIONS
4. CONCLUSIONS REFERENCES
357
1. Storch, H. H., Adv. Catal. 1, 115 (1948).
2. Pichler, H., Adv. Catal. 4, 271 (1952).
The experiments carried out with the plug-flow reactor
clearly show that the hydrogenation of carbon dioxide is
blocked as long as ethene is present in the gas phase. On
the other hand the presence of carbon dioxide does not at
all influence the hydrogenation of ethene. An increase in
the hydrogen pressure favours the hydrogenation of ethene
to ethane and detracts greatly from the chain growth. An
increase of the partial pressure of ethene leads to a pro-
nounced chain growth. The formation of methane (also in
the absence of carbon dioxide) proves that fragmentation
of ethene takes place. This observation is contrary to the
statements of Adesina et al. (21).
3. Pichler, H., and Schulz, H., Chem.-Ing.-Tech. 42, 1162 (1970).
4. Biloen, P., and Sachtler, W. M. H., Adv. Catal. 30, 165 (1981).
5. Schulz, H., Beck, K., and Erich, E., in “Proc. Methane Conv. Sympo-
sium, Auckland 1987” (D. Bibby, C. Chang, R. Howe, and S. Yurchak,
Eds.), Stud. Surf. Sci. Catal. 36, 457 (1988).
6. Adesina, A. A., Appl. Catal. A 138, 345 (1996).
7. Kuipers, E. W., Vinkenburg, I. H., and Oosterbeek, H., J. Catal. 152,
137 (1995).
8. Kuipers, E. W., Scheper, C., Wilson, J. H., Vinkenburg, I. H., and
Oosterbeek, H., J. Catal. 158, 288 (1996).
9. Lee, W. H., and Bartholomew, C. H., J. Catal. 120, 256 (1989).
10. Voß, M., Thesis, University of Erlangen-Nu¨rnberg, 1998. (Publication
in preparation.)
11. Weatherbee, G. D., and Bartholomew, C. H., J. Catal. 87, 352
(1984).
The use of the batch reactor allowed the correlation of
the gas phase measurements with the state of the catalyst
observed by means of Auger electron spectroscopy. When 12. Sexton, B. A., and Somorjai, G. A., J. Catal. 46, 167 (1977).
13. Solymosi, F., and Erdo¨helyi, A., J. Mol. Catal. 8, 471 (1980).
14. Solymosi, F., Erdo¨helyi, A., and Ba´nsa´gi, T., J. Catal. 68, 371 (1981).
15. Iglesia, E., Soled, S. L., and Fiato, R. A., J. Catal. 137, 212 (1992).
16. Weatherbee, G. D., and Bartholomew, C. H., J. Catal. 7, 460 (1982).
the experiment is started with the oxidised catalyst an in-
duction period is observed in the course of which the cata-
lyst is reduced. During this period the hydrogenation of
carbon dioxide is totally blocked, whereas ethene begins to
be hydrogenated. The oxygen signal in the Auger electron
spectrum disappears when nearly all the ethene has been
hydrogenated. Only then does carbon dioxide hydrogena-
tion start. It isinterestingto note that the presence ofcarbon
dioxide prolongs the induction period, possibly by the for-
17. Wetzel, H., Meyer, G., Borgmann, D., and Wedler, G., Chem.-Ing.
-Tech. 63, 613 (1991).
18. Fro¨hlich, G., Kestel, U., £ojewska, J., £ojewski, T., Meyer, G., Voß, M.,
Borgmann, D., Dziembaj, R., and Wedler, G., Appl. Catal. A: Gen.
134, 1 (1996).
19. Fro¨hlich, G., Thesis, University of Erlangen-Nu¨rnberg, 1996. (Publi-
cation in preparation.)
mation of carbonate. The hydrogenation of pure ethene in 20. Kestel, U., Fro¨hlich, G., Borgmann, D., and Wedler, G., Chem. Eng.
Technol. 17, 390 (1994).
21. Adesina, A. A., Hudgins, R. R., and Silveston, P. L., Appl. Catal. 62,
the batch reactor allows us to recognise that the olefines
are formed as intermediates and that at long reaction times
hydrogenolysis of the saturated hydrocarbons takes place.
295 (1990).
22. Percy, L. T., and Walter, R. I., J. Catal. 121, 228 (1990).
The oscillatory behaviour of the molar fraction of the
products in dependence on the number of C atoms, which
was observed with the hydrogenation of ethene, could
quantitatively be interpreted by a chain growth mechanism
regarding chain growth by incorporation of both C1 and C2
species.
23. Borgmann, D., Strattner, P., and Wedler, G., Chem.-Ing.-Tech. 58, 400
(1986).
24. Pirner, M., Borgmann, D., and Wedler, G., Surf. Sci. 211/212, 1091
(1989).
25. Wedler, G., Kießling, W., and Borgmann, D., Vacuum 41, 93 (1990).
26. Freund, H.-J., and Roberts, M. W., Surf. Sci. Rep. 25, 225 (1996).
27. Boudart, M., and Dje´ga-Mariadassou, G., “Kinetics of Heterogeneous
Catalytic Reactions.” Princeton University Press, Princeton, 1984.
(Further references therein.)
ACKNOWLEDGMENTS
28. Novak, S., Madon, R. J., and Suhl, H., J. Chem. Phys. 74, 6083
(1981).
29. Jordan, D. S., and Bell, A. T., J. Phys. Chem. 90, 4797 (1986).
This work was financially supported by the Fonds der Chemischen In-
dustrie and the Max-Buchner-Forschungsstiftung.