The Role of Coke in Acetylene Hydrogenation on Pd/α-Al2O3

Article abstract of DOI:10.1006/jcat.1998.2128

The formation of coke and the influence of the coke on selectivity were investigated during hydrogenation of acetylene on supported palladium catalysts. It was found that the total amount of coke was not directly related to the increase in formation of undesired ethane. Instead, the surface coverage of hydrogen during the deactivation was found to be a crucial parameter. A catalyst deactivated at low hydrogen surface coverage showed a higher ethane selectivity than a sample deactivated at higher surface coverage of hydrogen when compared under the same reaction conditions. In contrast, the coke formation rate was found to increase with increased hydrogen surface coverage. The role of carbon monoxide was also investigated, and the impact on selectivity and coke formation was explained by the reduced surface coverage of hydrogen in the presence of carbon monoxide. The coke was characterized by temperature-programmed oxidation, and deconvolution of the obtained peaks was carried out using a power-law model.

Full text of DOI:10.1006/jcat.1998.2128

JOURNAL OF CATALYSIS 178, 49–57 (1998)
ARTICLE NO. CA982128
The Role of Coke in Acetylene Hydrogenation on Pd/ꢀ-Al2O3
Mikael Larsson,1 Jonas Jansson,2 and Staffan Asplund3
Department of Chemical Reaction Engineering, Chalmers University of Technology, S-412 96 Go¨teborg, Sweden
Received August 8, 1997; revised April 28, 1998; accepted April 29, 1998
(4), respectively, while additional kinetic work is available,
e.g., in Refs. (5–9).
The formation of coke and the influence of the coke on selec-
tivity were investigated during hydrogenation of acetylene on sup-
ported palladium catalysts. It was found that the total amount of
coke was not directly related to the increase in formation of unde-
sired ethane. Instead, the surface coverage of hydrogen during the
deactivation was found to be a crucial parameter. A catalyst deac-
tivated at low hydrogen surface coverage showed a higher ethane
selectivity than a sample deactivated at higher surface coverage of
hydrogen when compared under the same reaction conditions. In
contrast, the coke formation rate was found to increase with in-
creased hydrogen surface coverage. The role of carbon monoxide
was also investigated, and the impact on selectivity and coke for-
mation was explained by the reduced surface coverage of hydrogen
in the presence of carbon monoxide. The coke was characterized
by temperature-programmed oxidation, and deconvolution of the
c
In addition to hydrogenation, acetylene undergoes hy-
dropolymerisation reactions, leading to liquid polymer mix-
tures (“green oils”) and/or the deposition of carbonaceous
species on the catalyst surface. Such liquid fractions have
been found to consist of paraffins and olefins ranging from
about C8 to C24 with an H/C ratio of about 1.9 (10, 11). In
industrial operation the coke formation results in a reduced
overall activity as well as an increased selectivity for the un-
desired ethane (10). In laboratory experiments, normally
performed at lower pressure, several groups have observed
constant or even slightly increasing activity for about 100 h
of operation (3, 8, 12, 13). Sa´rka´ny et al. (3) have presented
a comprehensive study using different catalysts and condi-
tions. In summary, they found ethane selectivity to increase
with increasing coke formation on all supported catalysts.
On unsupported Pd black and in the presence of CO, this
effect was suppressed.
Several groups (2, 10, 20) report that the coke concen-
tration decreases with increasing hydrogen pressure. Bat-
tiston et al. (2) found that the concentration of carbon on
the catalyst increased with decreasing H2/C2H2 ratio. How-
ever, with no hydrogen in the feed the coke formation was
negligible. Sa´rka´ny et al. (3) varies the hydrogen surface
coverage by having different CO concentrations in the gas
feed but keeping the hydrogen partial pressure constant.
They note that the C4 formation goes through a maximum
upon increasing CO concentration.
Asplund (14) has studied the effect of coke formation
on the intraparticle mass transfer. Accumulation of large
amounts of coke was found to block the catalyst pores and
induce mass transfer limitations that could not be detected
on a fresh catalyst. This phenomenon was found to be im-
portant even on a commercial catalyst of the eggshell type.
Asplund et al. (15) also studied deactivation of a mono-
lithic catalyst in different gas mixtures and in the presence
of an organic liquid phase. Catalyst aging in a hydrogen-rich
atmosphere (H2/C2H2 = 4.5) gave rapid coke formation and
an increasing ethane selectivity. The catalyst performance
was easily restored by oxidation in air at 573 K. When
catalyst aging was performed in the presence of a liquid
obtained peaks was carried out using a power-law model.
ꢀ 1998
Academic Press
INTRODUCTION
In the thermal steam cracking process, producing ethene
and propene, acetylene is also formed. This impurity is re-
moved in a selective hydrogenation process, which isusually
performed on a supported palladium catalyst. An obvious
objective in this process is to avoid hydrogenation of the
ethene to ethane while reacting all but a few ppm of the
acetylene. Modern catalysts are indeed very selective, at
least in the presence of carbon monoxide, but the ethane
formation is well known to increase with time on stream
(1–3).
The mechanisms and kinetics involved, including the role
of the selectivity promoter carbon monoxide, have received
considerable attention in previous literature. Surveys of
commercial processes and kinetics can be found in (1) and
1 Currently at Volvo Technological Development Corporation, 06130/
¨
PVOA201, S-405 08 Go¨teborg, Sweden.
2 To whom correspondence should be addressed. Fax: +46 31 772 30 35.
E-mail: Jonas.Jansson@cre.chalmers.se. http://www.cre.chalmers.se/.
3 Currently at AKZO Nobel Surface Chemistry AB, S-444 85 Stenung-
sund, Sweden.
49
0021-9517/98 $25.00
c
Copyright ꢀ 1998 by Academic Press
All rights of reproduction in any form reserved.

50
LARSSON, JANSSON, AND ASPLUND
phase (n-heptane) the coke accumulation was almost com-
pletely stopped but the ethane selectivity still increased.
Further, catalyst performance could not be restored even
by oxidation at 743 K. It was concluded that the effect of
coke on selectivity is caused by highly unsaturated species,
whose formation is suppressed by a high enough hydrogen
excess.
The present work aims to clarify further the relation
between operating conditions, coke composition, and the
effect of coke on catalyst performance. The idea is to first
deactivate the catalyst in different gas-phase conditions
(different partial pressures of H2, C2H2, and CO). There-
after the catalyst performance (i.e., ethane selectivity) is
measured at a standard gas-phase composition. The influ-
ence on the selectivity of different reaction conditions is
thus eliminated and the deactivated catalysts are evaluated
under equal conditions. Temperature-programmed oxida-
tion (TPO) and desorption (TPD) are used to study the
coke accumulated on the catalyst surface (16–18).
FIG. 1. The experimental apparatus used in the different temper-
ature-programmed analyses.
METHODS
C2H2, and 7.0 kPa H2 with N2 to balance). The tempera-
ture and total pressure were kept unchanged. After 45 min
under the standard conditions the ethane selectivity was
recorded. Comparison of catalyst performance is thus made
under equal conditions.
The Catalyst
Three different catalysts were used in the study (Table 1).
A description of the preparation methods can be found
in (14). The pellets originally prepared were crushed and
sieved to avoid mass transfer limitations (14). The particle
size was 0.3–0.4 mm.
Determination of the Coke Concentration
Three different methods were applied in order to deter-
mine the coke concentration:
Acetylene Hydrogenation
1. Measurement of the weight increase during a run. The
reactor basket, including the catalyst, was weighed before
and after the deactivation run on a standard laboratory bal-
ance. Prior to use, the catalyst was kept in a desiccator to
avoid water contamination.
2. Measurement of the weight loss when the deposited
coke was burned off in air at 1073 K. Reference samples
of fresh catalyst were given the same treatment and the
difference in weight loss was attributed to coke.
The deactivation experiments were performed in an in-
ternal recycle (Berty) reactor from Autoclave Engineers.
Details about the apparatus can be found elsewhere (14).
The deactivation was performed in a gas mixture of C2H4,
C2H2, and H2 with N2 to balance. The total pressure was
1.0 MPa, the C2H4 partial pressure 0.3 MPa, and the tem-
perature 313 K in all experiments. The partial pressures of
C2H2 and H2 for each experiment during the deactivation
are given in Table 2. The duration of the deactivation was
usually around 24 h (Table 2). The reaction rate was con-
tinuously measured during the deactivation. Directly after
the prescribed deactivation time the gas composition was
changed to the standard conditions (0.3 MPa C2H4, 3.5 kPa
3. Temperature-programmed oxidation (see below).
Temperature-Programmed Oxidation
The experimental setup can be seen in Fig. 1. During the
TPO experiments, oxygen (1 ml/min) and argon (24 ml/
min) were used and the temperature in the main oven was
increased from 300 to 953 K at a heating rate of 10 K/min.
The temperature in the second oven was kept at 673 K
during the whole experiment. A Pt foil catalyzed the ox-
idation of uncombusted hydrocarbons, and only CO2 and
H2O were formed. The advantage of this technique, which
was previously used by our research group (21), is that it is
easy to estimate the amount of the desorbing hydrocarbons
without knowing the often very complex composition.
TABLE 1
Catalyst Properties
Dispersion
Pore
BET
Pd load (mol CO/ volume surface
Catalyst Carrier Pd source (wt% )
mol Pd)
(cm3/g) (m2/g)
A
B
ꢀ-Al2O3 Pd(NO3)2
ꢀ-Al2O3 PdCl2
0.05
0.04
0.035
0.05
0.23
0.20
8
8

COKE IN ACETYLENE HYDROGENATION ON Pd/ꢀ-Al2O3
51
TABLE 2
Experimental Plan with Reaction Conditions and Coke Concentrations
b
c
d
pC
(kPa)
pH
(kPa)
pC
(kPa)
TOSa
(h)
Cw1
Cw2
CTPO
Ethane
selectivity
Mean
TOFe(sꢁ1
H
H
4
2
2
2
2
Run
Catalyst
(wt% )
(wt% )
(wt% )
)
Remark
1
2
3
3b
4
5
6
7
8
9
A
A
A
A
A
A
A
A
A
A
A
B
B
7.0
3.5
7.0
7.0
7.0
7.0
14.0
3.5
3.5
7.0
7.0
7.0
7.0
0f
7.0
7.0
7.0
2.8
36.0
7.0
11.7
2.8
17.2
17.0
7.0
300
300
300
300
300
300
300
300
300
300
300
300
300
24
25
24
24
29
20
25
24
24
31
24
24
48
0.2
0.22
0.16
0.16
0.19
0.55
0.23
0.27
0.20
0.34
0.45
0.41
2.26
3.75
3.52
1.20
3.40
3.29
3.00
1.35
1.21
1.88
3.40
5.68
2.44
2.50
3.71
3.64
1.36
1.20
0.83
0.87
0.34
4.27
0.59
1.98
0.39
0.44
0.56
TPO
TPO
D
2 during deactivationg
TPO
90 ppm CO
50 ppm CO
TPO
10
11
12
1.62
3.51
5.59
7.0
5.72
TPO
Several TPO experiments were also performed on the catalyst deactivated in run 3:
Time of exposure to air at room temperature
before TPO or TPD (days)
13
14
15
16
17
See run 3
See run 3
See run 3
See run 3
See run 3
TPO with 3% O2
TPO with 6% O2
TPO with 8% O2
TPO without extra reactor
TPD
48
49
50
51
27
Note. Constant conditions in all experiments: T = 313 K, Ptot = 1.0 MPa, N2 to balance.
a Time on stream.
b Coke concentration, by the weight increase during the run.
c Coke concentration, by the weight loss by burning off the coke.
d Coke concentration, from TPO.
e The time average TOF during the deactivation. TOF is based on CO chemisorbtion measurement on fresh catalyst samples (Table 1).
f No H2 in feed. Small amounts were detected during the first hours, presumably from C2H2 dissociation.
g The experiment was carried out with deuterium instead of hydrogen (22).
Temperature-programmed oxidation was also perform- Temperature-Programmed Desorption
ed without the platinum foil in the second oven. This
Temperature-programmed desorption of the coke accu-
experiment was conducted in order to see how much coke
in the first peak of the TPO spectrum consisted of hy-
drocarbons desorbing. The hydrocarbons were not anal-
yzed.
mulated on the catalysts was performed in the same appara-
tus described earlier. In these experiments, however, only
argon (24 ml/min) was flowing through the catalyst bed.
Oxygen (1 ml/min) was introduced after the bed, immedi-
ately before the Pt foil (see Fig. 1). The desorbing species
reacted on the Pt foil and formed CO2 and H2O that were
identified in the MS.
A mass spectrometer (MS), Gaslab 300, Fisons Instru-
ments, was used mainly to follow M/e 44 (CO2), M/e 32 (O2),
and M/e 18 (H2O). The presence of unburned hydrocarbons
was checked for by using the MS. By burning different con-
centrations of acetylene, the CO2, H2O, and O2 responses
were calibrated. The hydrogen-to-carbon ratio in the coke
was estimated from the CO2 and H2O formation.
After the coking experiment, the catalysts were placed
in a container that was flushed with nitrogen, sealed, and
put in a freezer. Immediately before the TPO experiments,
the catalysts were taken out of the freezer and placed in the
reactor, except for runs 13–17 (see Table 2).
Deconvolution
The TPO peaks were deconvoluted using a kinetic model
formulated by Querini and Fung (18),
dxi
mi
O2
Ai /RT
= Ai eꢁE
ꢂ (1 ꢁ xi )ni p
,
[1]
dt
where xi is the conversion of coke, Ai is a pre-exponential
The amounts of deactivated catalyst analyzed in the tem- factor, EAi is the activation energy, ni is the reaction order
perature-programmed studies were varied (43–140 mg) so of coke, pO is the partial pressure of oxygen, and mi is the
2
that about the same amount of coke was burned in each oxygen reaction order. A few different types of coke are
experiment.
usually assumed, and the index i denotes the type of coke.

52
LARSSON, JANSSON, AND ASPLUND
transfer effects (19). In the next section we concentrate on
the characteristics of the coke by performing temperature-
programmed oxidation on the deactivated samples.
Temperature-Programmed Oxidation
Results from the TPO experiments on catalysts aged un-
der different reaction conditions are shown in Fig. 3. The
most striking feature is that, for a given catalyst, the coke
characteristics (peak shape and H/C ratio) are independent
of gas composition. In contrast, the total amount of de-
posited coke varies substantially. However, the location of
the TPO peaks is different on the two catalysts prepared
by different precursors. On the PdCl2 catalyst (cat. B) the
peaks obtained at the highest temperature are shifted to
lower temperature by about 40 K compared to the catalyst
prepared by impregnation with Pd(NO3)2 (cat. A).
In order to study the coke located in the first peak in
the TPO curves, a TPO experiment without the Pt foil in
the second reactor and a TPD experiment were conduct-
ed. The former experiment showed that only small amounts
of CO2 are formed on the catalyst during the TPO experi-
ment at temperatures lower than 530 K (curve iii in Fig. 4a).
Pure desorption of hydrocarbons, without any influence by
oxygen in the gas phase, cannot explain the whole first peak,
because a higher oxygen pressure will result in a larger peak
(compare curves i and ii in Fig. 4a and also Fig. 4b).
Before the TPD experiment; the TPO with 3, 6, and 8%
O2; and the TPO without Pt foil in the second reactor, the
catalyst was exposed to air at room temperature for several
weeks (Table 2). It is possible to see the effect of these
different treatments on the first peak in the TPO profiles.
The area of this peak is larger when the catalyst has not
been exposed to air at room temperature for a longer time.
The amount of coke measured by TPO also differs. With
only a short exposure to air at room temperature 3.71 wt.%
coke was detected (run 3, TPO with 4% O2), while after 48
FIG. 2. Ethane selectivity under standard conditions (0.3 MPa C2H4,
3.5 kPa C2H2, and 7.0 kPa H2 with N2 to balance) obtained with catalyst
samples coked at different reaction conditions. The numbers at each point
correspond to run numbers in Table 2.
RESULTS
The Acetylene Hydrogenation
In all experiments, the decrease in acetylene hydrogena-
tion rate during the deactivation was below 10% and the
discussion willconcern the selectivityand the coke only. The
amounts of coke determined by three different methods
can be found in Table 2. There is great consistency among
the methods. The values achieved by direct weighing of the
catalyst before and after the deactivation are more compre-
hensive, and will be used in the following analyses.
The ethane selectivity is plotted versus the coke content
in Fig. 2. If the ethane selectivity had been directly related
to the amount of coke, one would have expected a positive
slope in the figure. However, this is not the case, which
implies that the main fraction of the coke has a very limited
influence on the catalyst properties aside from mass
FIG. 3. TPO analyses of catalysts deactivated at different conditions (Table 2). The numbers at each curve correspond to run numbers in Table 2.
The solid lines mark the catalyst similar to catalyst A, but impregnated with PdCl2 instead of Pd(NO3)2 (cat. B). (a) The CO2 formation and (b) the
hydrogen-to-carbon ratio.

COKE IN ACETYLENE HYDROGENATION ON Pd/ꢀ-Al2O3
53
FIG. 4. (a) The CO2 response at (i) normal TPO (run 3), (ii) TPD (run 17), and (iii) TPO without Pt foil in the second oven (run 16). (b) TPO
profiles for experiments conducted at different partial pressures of oxygen and a TPD experiment. All experiments were carried out on a catalyst
deactivated during the same experiment (run 3, Table 2).
days exposure to air at room temperature 3.55 wt.% coke is illustrated in Fig. 5b. It should be noted that i = 1 does
was detected (run 13, TPO with 3% O2).
not mean the desorption peak discussed earlier, but is
Querini and Fung (18) obtained poor kinetic parameters mainly what is left after subtraction of the Gaussian curve,
when deconvoluting single TPO experiments. In order peak D.
to improve the results, they regressed two independent
sets of data, obtained at different heating rates. Because
of this, we used Eq. [1] to deconvolute TPO profiles
The Effect of the Hydrogen Partial Pressure
The analysis continued by studying the influence of the
partial pressure of hydrogen during the deactivation on the
selectivity measured at standard conditions. From Fig. 6a
the decisive role of the partial pressure of hydrogen is clear.
However, the catalysts deactivated in the presence of car-
bon monoxide showed a high ethane selectivity once the
carbon monoxide was removed. CO is known to hinder
hydrogen adsorption and may thus decrease the hydro-
gen coverage on the metal surface. No direct information
about the hydrogen coverage is available. Nevertheless, by
from experiments where the oxygen partial pressure was
changed (Fig. 4b). The nature of the first peak in the TPO
can be derived mainly from desorption, and the kinetic
equation above is not suitable to use for deconvolution.
Instead a Gaussian peak was estimated from the TPD
experiment and this area was subtracted from the TPO
profiles before performing the deconvolution (Fig. 5a).
The deconvolution was then performed, and the results
can be found in Fig. 5a and in Table 3. The parameters in
Table 3 are determined for each peak i. The numbering
FIG. 5. (a) The experimental TPO profiles (runs 13, 14, and 15) after subtracting the contribution from the desorption, and results from the model.
(b) Results from the deconvolution of the experiment conducted with an oxygen concentration of 6% (run 14, Table 2).

54
LARSSON, JANSSON, AND ASPLUND
TABLE 3
Parameters Estimated from the TPO Experiments Conducted at Different Partial Pressures of Oxygen Using Eq. [1]
i
Ais
EAi (kJ/mol)
ci (ꢁmol)
ni
mi
1
2
3
4
(3.06 ꢃ 0.43) ꢄ 105
(2.04 ꢃ 0.16) ꢄ 1010
(1.32 ꢃ 0.37) ꢄ 1010
(7.72 ꢃ 1.01) ꢄ 1012
47.75 ꢃ 0.04
112.73 ꢃ 4.04
113.94 ꢃ 9.56
177.64 ꢃ 5.57
27.65 ꢃ 1.38
51.13 ꢃ 3.06
119.13 ꢃ 2.91
52.23 ꢃ 1.30
1.120 ꢃ 0.087
0.5568 ꢃ 0.015
2.158 ꢃ 0.198
1.214 ꢃ 0.039
0.8108 ꢃ 0.046
0.4936 ꢃ 0.015
0.7773 ꢃ 0.061
0.9581 ꢃ 0.035
Note. Here i indicates the different peaks in the TPO spectra; 95% confidence intervals are given.
assuming that the acetylene hydrogenation rate is con-
trolled by the surface reaction, the following equation can
be formulated:
DISCUSSION
The Effect of the Partial Pressure of Hydrogen
on the Coke Formation
r = kꢂC H ꢂꢃ
.
Figures 7a and 7b indicate that the coke formation rate
increases with the surface coverage of hydrogen on the ac-
tive metal. This is not in line with some earlier studies (2,
10, 20). However, the present study has been performed
at relatively low partial pressures of hydrogen, and we may
consider the possibility that the coke formation rate reaches
a maximum and then decreases with increasing hydrogen
pressure. It is also possible that we have already reached the
maximum somewhere in the range 12–36 kPa hydrogen. In
a previousstudy(21) we suggested that a half-hydrogenated
surface intermediate (C2H3) is formed by hydrogenation of
acetylene. This intermediate can either form ethene (22,
23) or, if the hydrogen coverage is low, react with other
species on the surface and form coke or coke precursors
(21). This mechanism would explain the complex effect
of the hydrogen partial pressure on the coke deposition
rate.
2
2
H2
Except for very low concentrations of acetylene, the over-
all reaction order of acetylene is negative. The surface cov-
erage of acetylene is close to unity, and one may assume
that ꢂC H is constant with little error. This means that the
acetylene reaction rate is an explicit function of the hy-
drogen coverage ꢂH. Figure 6b is based on this concept. It
appears that the availability of surface hydrogen during de-
activation has a crucial effect on the selectivity of a used
catalyst.
2
2
The Effect of Hydrogen on the Coke Formation
The coke deposition after 24 h of operation is shown in
Fig. 7a as a function of partial pressure of hydrogen. Con-
trary to previous findings (2, 10, 20), the coke concentra-
tion increases with the hydrogen pressure, at least at low
values. The presence of CO reduces the coke deposition,
presumably by reducing the surface coverage of hydrogen.
In Fig. 7b the coke deposition has been plotted versus the
The Effect of Coke on the Selectivity
From this study it is also clear that the total amount of
rate of the acetylene formation, as discussed in the previous coke is of minor interest for the changes in selectivity dur-
section. The coke formation is found to increase with the ing acetylene hydrogenation. At first glance, this stands in
surface coverage of hydrogen.
contradiction to what is previously found. However, in most
FIG. 6. (a) The influence of hydrogen partial pressure during deactivation on ethane selectivity under standard conditions. (b) Ethane selectivity
under standard conditions versus mean reaction rate for acetylene consumption during deactivation. This rate is assumed to be proportional to the
surface coverage of hydrogen. The numbers at each point correspond to run numbers in Table 2.

COKE IN ACETYLENE HYDROGENATION ON Pd/ꢀ-Al2O3
55
FIG. 7. (a) The influence of the partial pressure of hydrogen during deactivation on coke concentration. (b) The coke concentration versus mean
reaction rate for acetylene consumption during deactivation. This rate is assumed to be proportional to the surface coverage of hydrogen. The numbers
at each point correspond to run numbers in Table 2.
earlier studies the selectivity changes and the coke forma- on the catalyst increases, the total amount of coke formed
tion rate have been measured simultaneously as a function also increases, but mainly the harmless type. At even higher
of time. The same results would have been the outcome also hydrogen coverage, coke formation is suppressed. The sug-
in the present study using the same approach. Instead we gested mechanism showed in Fig. 8 is not the only one pos-
let the catalyst deactivate under different conditions for a sible. Other mechanisms with a higher hydrogen reaction
given time period. After that the selectivity toward ethane order for the formation of harmless coke than for the for-
formation was estimated under the same reaction condition mation of harmful coke could also explain the observed
in all experiments. It turns out that the surface coverage of effect of hydrogen on the selectivity.
hydrogen during the deactivation is of great importance
in reducing the formation of the coke responsible for the
increased selectivity to ethane formation.
Coke
We found that the different reaction conditions resulted
Sa´rka´ny et al. (3) discussed a mechanism whereby ethene
is adsorbed and hydrogenated to a large extent on the sup-
port or the coke surface. Spillover hydrogen is required
in the latter reaction, and the role of coke is primarily to
promote hydrogen transfer from the metal surface to the
adsorbed ethene. Sermon et al. (24) used a special combined
fixed-fluidized bed reactor. They have shown that coke de-
posited on a silica–alumina catalyst is catalytically active
in the hydrogenation of cyclohexene at 343 K if dissociated
hydrogen is available by spillover. They also concluded that
polyaromatic compounds were responsible for the catalytic
activity.
In the previoussection about the effect ofthe partialpres-
sure of hydrogen on the coke formation, a mechanism was
proposed. We can extend this mechanism to also explain
the role of hydrogen in altering the selectivity. The half-
hydrogenated surface intermediate (C2H3) could at low hy-
drogen availability form a coke precursor (C4H6). We now
suggest that this coke precursor could form two types of
coke, one harmful type that causes unwanted formation
of ethane by, for example, a spillover mechanism as pre-
viously discussed (3, 24), and one harmless type. The two
types of coke are formed in parallel and the great majority
of the coke is harmless. A similar mechanism was proposed
by Larsson et al. (19) for propane dehydrogenation. Fig-
ure 8 shows one possible mechanism. The harmless coke is
here formed by a reaction with hydrogen. At very low sur-
face coverage of hydrogen, coke of the harmful type would
be formed almost exclusively. When the hydrogen available
in no significant differences in the TPO profiles other than
the amount of coke, although the shape of the profile dif-
fers between the catalysts impregnated with PdCl2 and
Pd(NO3)2. This implies that the composition and location
of the coke are independent of reaction conditions. We may
conclude that the key to how the different gas compositions
cause different deactivation cannot be found from the TPO
experiments.
The slightly different shape of the TPO profile for the
catalyst impregnated with PdCl2 can be explained by the
presence of Clꢁ ions in this catalyst. This might influence
the type or location of the coke formed, thus leading to
the peaks in the TPO spectra from the Clꢁ containing cata-
lyst shifts compared to the catalyst without Clꢁ. Augustine
et al. (34) report that the presence of Clꢁ on Pt/Al2O3 affects
the mobility of the coke so that more coke is retained on
the support. Although our catalyst is quite different we may
speculate that the same phenomenon may occur on our
catalyst. It is also possible that the presence of Clꢁ affects
the activation energy for the coke burn-off, thus leading to
different peak temperatures.
FIG. 8. A possible coke formation mechanism explaining the influ-
ence of the surface coverage of hydrogen on the formation of harmful and
harmless types of coke.

56
LARSSON, JANSSON, AND ASPLUND
From Fig. 4a, it is possible to infer that the coke in the For the next peak (i = 3) m = 0.78. This value is reasonable
major part of the first peak in the TPO spectrum consists of if a combination of the above reactions takes place. These
heavy hydrocarbons that are adsorbed on the catalyst sur- estimations of the oxygen reaction order are in good agree-
face or absorbed in the pore system. This is also consistent ment with results by Pieck et al. (29), who estimated it to be
–
with the decrease of this peak upon long exposures to air at 0.5 in burning coke on a commercially coked Pt Re/Al2O3
room temperature previously discussed. However, oxygen naphtha-reforming catalyst. Liu et al. (28) found that the re-
–
assists the desorption (Fig. 4b). Explanations could be that action order in oxygen partial pressure was 0.55 for a Pt Sn
oxygen cleans the metal surface from adsorbed hydrocar- reforming catalyst and 0.75 for zeolite cracking catalyst.
bons, or that the oxygen reacts with heavy hydrocarbons, For the last peak (i = 4) the activation energy is much
causing a cleavage and formation of lighter hydrocarbons higher, 178 kJ/mol. This value corresponds well to the burn-
that desorb. ing of graphite (35) and indicates a more graphite-like type
We can draw conclusions about the type of coke found of coke. The coke reaction order is 1.2, which is reasonable
on the catalyst by studying the parameters determined for a type of coke where all carbon atoms are exposed, i.e.,
through the deconvolution (Table 3). The peak i = 1 is the coke is well dispersed on the support. Furthermore, an-
difficult to evaluate because of the proximity of the des- other possibility is, of course, that a distribution of different
orption peak. For the next two peaks, i = 2 and i = 3, the coke types exists or that the structure of the coke changes
activation energies, 113 and 114 kJ/mol, are well in the during the heating. The oxygen reaction order is close to
range found by others for coke on supported metal cata- one and the coke combustion probably takes place through
lysts (25). The most important difference between the two direct oxidation of the coke by oxygen in the gas phase.
peaks is the reaction order of coke, n = 0.56 and n = 2.2 for
We agree with a model, earlier proposed by others (28,
i = 2 and i = 3, respectively. As has been discussed in de- 30–32), that a small part of the coke is located on the metal,
tail by Querini and Fung (18) and Butt and Petersen (26), while a larger part is placed close to the metal and is assisted
the coke reaction order depends upon the geometry and by the metal in the combustion. The coke identified in peaks
how coke is consumed during the oxidation process. If all i = 2 and 3 is coke on and in the vicinity of the palladium,
carbon atoms are exposed, then n = 1. Spherical coke par- and the peak i = 4 indicates coke deposited on the carrier
ticles will result in n = 2/3, and if the number of exposed combusted without any influence by the metal.
carbon atoms is unchanged during combustion, as could be
We previously performed experiments using deuterium
the case if multiple layers of coke exist, then n = 0. A reac- as a tracer during the acetylene hydrogenation (21). From
tion order of coke of 0.56 as for i = 2 is a reasonable value these results a mechanism for the coke formation was pro-
if all carbon atoms are not exposed, as will be the case if posed. We also found that the ratio between deuterium and
larger coke particles are formed or if pores are plugged. hydrogen in the coke formed on the catalyst was constant
The value n = 2.2 is not reasonable if only one type of coke for all types of coke, when studied during a TPO exper-
is assumed. However, Cider and Scho¨o¨n (27) have shown iment, and we can conclude that the same mechanism is
that a number of first-order reactions, having an exponen- valid for formation of all types of coke.
tial rate-constant distribution, will give an apparent second-
order reaction. Using the same motivation, we propose that
The Methods
the peak i = 3 consists of a distribution of different coke
types. Furthermore, another explanation may be that the
coke, while heated during the TPO, changes structure, thus
generating a different type of coke. This is not unlikely to
happen since the temperature where peak i = 3 occurs is
about 300 K higher than the reaction temperature.
The treatment of the coked catalyst before the TPO ex-
periment is important in order to give reproducible results.
Long exposure to air after the coking experiment was found
to reduce the first peak. It is recommended that catalyst be
kept in the freezer and exposure to air minimized after cok-
ing.
The oxygen reaction order has also been determined
individually for the different types of coke (Table 3). For
the peak i = 2 it is very close to 0.5, which would be the re-
action order if the rate-determining step were the reaction
From Table 2 it can be seen that the amounts of coke ob-
tained by TPO and weighing agree very well for the Pd/ꢀ-
Al2
O catalysts. However, if, for example, ꢄ -Al O is used
3
2
3
between coke and a single dissociated oxygen atom (28). as support, problems arise with water adsorbed on the car-
An oxygen reaction order of one, on the other hand, can rier, and the agreement would be less consistent. Asplund
be found either if a noncatalytic direct oxidation of the coke (14) reports that a fresh ꢄ -Al2O3 catalyst lost about 5%
occurs or if, in the catalytic reaction discussed above, the ad- of its weight when oxidized at 1073 K whereas an ꢀ-Al2O3-
sorption of oxygen is the rate-limiting step. Other possible supported sample lost no weight at all. As discussed earlier,
catalytic mechanisms resulting in an oxygen reaction order the treatment between the coking reaction and the TPO ex-
of one may occur if two dissociated oxygen atoms react periment is crucial, and it is desirable to conduct the TPO
with the coke or if CO oxidation is the rate-limiting step. experiment directly after the coking in the same apparatus.

COKE IN ACETYLENE HYDROGENATION ON Pd/ꢀ-Al2O3
57
The design of the TPO reactor had some important fea-
tures that proved to work very well in the experiments. First,
the extra reactor with the heated Pt foil was necessary to be
able to detect the amount of coke in the desorption peak.
Another approach would be to try to determine the com-
position directly, either in the gas phase during the TPO
(33) or by extracting the soluble parts of the coke (11). In
both cases a GC-MS analysis is desirable because of the
very complex composition. In this work we have tried to
determine only the quantity and the H/C ratio, which can
be difficult by the methods discussed above. A further fea-
ture of the experimental equipment was the possibility to
add oxygen after the catalyst bed and burn everything that
desorbs during the TPD experiment. The experiment was
fast to perform and evaluate, and gave accurate results.
6. McGown, W. T., Kemball, C., and Whan, D. A., J. Catal. 51, 173
(1978).
7. Cider, L., and Scho¨o¨n, N.-H., Ind. Eng. Chem. Res. 30, 1437 (1991).
8. God´ınez, C., Canes, A. L., and Villora, G., Chem. Eng. Proc. 34, 459
(1995).
9. Duca, D., Frustreri, F., Parmaliana, A., and Deganello, G., Appl. Catal.
146, 269 (1996).
10. Yayun, L., Jing, Z., and Xueru, M., in “Proceedings, Joint Meeting of
Chemical Engineering, Chemical Industry and Engineering Society
of China and American Institute of Chemical Engineers, Beijing,”
Vol. 2, p. 688. 1982.
11. Boitiaux, J. P., Cosyns, J., Derrien, M., and Le´ger, G., Hydrocarbon
Process. 64, 51 (1985).
12. LeViness, S., Nair, V., Weiss, A. H., Schay, Z., and Guczi, L., J. Mol.
Catal. 25, 131 (1984).
13. Weiss, A. H., Gamphir, B. S., LaPierre, R. B., and Bell, W. K., Ind.
Eng. Chem. Proc. Des. Dev. 16, 352 (1977).
14. Asplund, S., J. Catal. 158, 267 (1996).
15. Asplund, S., Fornell, C., Holmgren, A., and Irandoust, S., Catal. Today
24, 181 (1995).
16. Barbier, J., Churin, E., Parera, J. M., and Riviere, J., React. Kinet. Catal.
Lett. 29, 323 (1985).
17. Fung, S. C., and Querini, C. A., J. Catal. 138, 240 (1992).
18. Querini, C. A., and Fung, S. C., Appl. Catal. A 117, 53 (1994).
CONCLUSIONS
We found that the total amount of coke was of minor im-
portance in explaining the undesired increased selectivity
to ethane formation. Neither did the TPO analyses reveal 19. Larsson, M., Hulte´n, M., Blekkan, E. A., and Andersson, B., J. Catal.
164, 44 (1996).
any evidence that the characteristics of the coke are impor-
tant for ethane selectivity. Instead, the surface coverage of
hydrogen has a crucial role in the selectivity phenomenon.
20. Gandman, Z. E., Aerov, M. E., Men’shchikov, V. A., and Getmantsev,
V. S., Int. Chem. Eng. 15, 183 (1975).
21. Larsson, M., Jansson, J., and Asplund, S., J. Catal. 162, 365 (1996).
A low surface coverage leads to a larger increase in ethane
selectivity than does a higher surface coverage of hydro-
gen. More “harmful” coke is formed, but this coke cannot
be identified in the TPO analyses. The coke formation rate
is also increased when more hydrogen is present on the sur-
face, at least at relatively low hydrogen pressures. The role
of carbon monoxide is to reduce the surface coverage of hy-
drogen, leading to higher ethane selectivity and diminished
coke formation.
22. Guczi, L., LaPierre, R. B., Weiss, A. H., and Biron, E., J. Catal. 60, 83
(1979).
23. Margitfalvi, J., Guczi, L., and Weiss, A. H., React. Kinet. Catal. Lett.
15, 475 (1980).
24. Sermon, P. A., Vong, M. S. W., and Martin-Luengo, M. A., Stud. Surf.
Sci. Catal. 88, 319 (1994).
25. Augustine, S. M., Alameddin, G. N., and Sachtler, W., J. Catal. 115,
217 (1989).
26. Butt, J. B., and Petersen, E. E., “Activation, Deactivation, and Poison-
ing of Catalysts,” Academic Press, San Diego, 1988.
27. Cider, L., and Scho¨o¨n, N.-H., Appl. Catal. 68, 207 (1991).
28. Liu, Y., Yang, J., and Yang, G., in “Proceedings, Int. Conf. Pet. Refin.
Petrochem. Process.” (X. Hou, Ed.), p. 362. Int. Acad. Publ., Beijing,
1991.
ACKNOWLEDGMENT
Support from the Swedish Research Council for Engineering Sciences
(TFR) is gratefully acknowledged.
29. Pieck, C. L., Verderone, R. J., Jablonski, E. L., and Parera, J. M., Appl.
Catal. 55, 1 (1989).
30. Sa´rka´ny, A., Lieske, H., Szila´gyi, T., and To´th, L., in “Proceedings, 8th
International Congress on Catalysis, Berlin, 1984,” Vol. 2, p. II613.
Dechema, Frankfurt-am-Main, 1984.
REFERENCES
1. Derrien, M. L., Stud. Surf. Sci. Catal. 27, 613 (1986).
2. Battiston, G. C., Dalloro, L., and Tauszik, G. R., Appl. Catal. 2, 1
(1982).
3. Sa´rka´ny, A., Guczi, L., and Weiss, A. H., Appl. Catal. 10, 369 (1984).
4. Bos, A. N. R., and Westerterp, K. R., Chem. Eng. Proc. 32, 1 (1993).
5. Bos, A. N. R., Bootsma, E. S., Foeth, F., Sleyster, W. J., and Westerterp,
K. R., Chem. Eng. Proc. 32, 53 (1993).
31. Duprez, D., Hadj-Aissa, M., and Barbier, J., Appl. Catal. 49, 67 (1989).
32. Basso, T. C., Zhang, Z., and Sachtler, W. M. H., Appl. Catal. 79, 227
(1991).
˚
33. Wrammerfors, A., and Andersson, B., J. Catal. 147, 82 (1994).
34. Augustine, S. M., Alameddin, G. N., and Sachtler, W. M. H., J. Catal.
115, 217 (1989).
35. Blyholder, G., and Eyring, H., J. Phys. Chem. 61, 682 (1957).