TAP investigations of the CO2 reforming of CH4 over Pt/ZrO2

Article abstract of DOI:10.1006/jcat.1997.1539

The adsorption and reaction characteristics of methane, carbon dioxide, carbon monoxide, and hydrogen have been investigated over a ZrO₂ support and a Pt/ZrO₂ catalyst by using a temporal analysis of products reactor system. It was observed that on Pt/ZrO₂ both methane and carbon dioxide dissociate independently of one another. The dissociation of carbon dioxide acts as an oxygen supplier, while the decomposition products of methane scavenge the oxygen from the catalyst. When an abundance of oxygen is present, pulsing of methane leads to the production of carbon dioxide. It is concluded that both the selectivity with which methane produces carbon monoxide or carbon dioxide and the carbon dioxide conversion is determined by the same reaction: CO_ads + O_ads ? CO_2,ads?

Full text of DOI:10.1006/jcat.1997.1539

JOURNAL OF CATALYSIS 166, 306–314 (1997)
ARTICLE NO. CA971539
TAP Investigations of the CO2 Reforming of CH4 over Pt/ZrO2
,
,1
ꢀ
ꢀ
A. N. J. van Keulen, † K. Seshan,† J. H. B. J. Hoebink,‡ and J. R. H. Ross
ꢀ
College of Science, University of Limerick, Limerick, Ireland; †Catalytic Processes and Materials Group, Faculty of Chemical Technology, University
of Twente, P.O. Box 217, 7500 AE Enschede, The Netherlands; and ‡Schuit Institute of Catalysis, Laboratorium voor Chemische Technologie,
Eindhoven University of Technology, P.O. Box 513, 5600 MB Eindhoven, The Netherlands
Received June 27, 1996; revised November 18, 1996; accepted November 19, 1996
they are also more stable in comparison to Pt/Al2O3 and
Pt/TiO2 catalysts (11). The aim of the present study was
to investigate the mechanism of the CO2 reforming of CH4
using a temporal analysis of products (TAP) reactor system
(12).
The adsorption and reaction characteristics of methane, carbon
dioxide, carbon monoxide, and hydrogen have been investigated
over a ZrO2 support and a Pt/ZrO2 catalyst by using a temporal
analysis of products reactor system. It was observed that on Pt/ZrO2
both methane and carbon dioxide dissociate independently of one
another. The dissociation of carbon dioxide acts as an oxygen sup-
plier, while the decomposition products of methane scavenge the
oxygen from the catalyst. When an abundance of oxygen is present,
pulsing of methane leads to the production of carbon dioxide. It is
concluded that both the selectivity with which methane produces
carbon monoxide or carbon dioxide and the carbon dioxide conver-
EXPERIMENTAL
Catalyst Preparation
The ZrO2 support, used for the catalyst investigated here
(1 wt% Pt/ZrO2), was prepared by heating Zr(OH)4 extru-
dates (MEL Chemicals, XZ0706/3) at a rate of 5ꢂC minꢃ1 to
800ꢂC and maintaining that temperature for 15 h in flowing
air. The sample was then crushed and sieved to obtain the
fraction with particle sizes in the range 0.21–0.50 mm. The
1 wt% Pt/ZrO2 catalyst was prepared by wet impregnation
of the zirconia support with a 0.5 wt% H2PtCl6 solution
(PGP Industries). After allowing overnight equilibration,
the resultant sluꢃr1ry was dried and subsequently heated at a
rate of 5ꢂC min to 600ꢂC in flowing air, maintaining that
temperature for 6 h. A more detailed description is given
elsewhere (11). The BET surface areas of the fresh sam-
ples were 41 m2/g for the Pt/ZrO2 and 42 m2/g for the ZrO2
support.
sion is determined by the same reaction: COads + Oads ⇔ CO2,ads
.
c
ꢁ 1997 Academic Press
INTRODUCTION
Over the past few years, the carbon dioxide reforming
of CH4 (dry reforming) has attracted interest from both an
environmental and an industrial perspective. The environ-
mental viewpoint stems from the fact that both CO2 and
CH4 are viewed as “harmful” greenhouse gases (1–4) and
hence the reaction provides a method of disposing of these
gases. However, the use of the carbon dioxide reforming
reaction as a CO2 scavenger is debatable. This is due to
the endothermic nature of the reaction, which requires a
heat input and hence fuel combustion and CO2 generation.
From the industrial view the reaction is attractive as it pro-
duces synthesis gas with a higher purity (5) and a lower H2
to CO ratio than does either partial oxidation or steam re-
forming. However, compared with the latter two processes,
the dry reforming has a higher risk of catalyst deactivation
due to carbon deposition; this is thermodynamically more
favoured when the carbon content in the feed mixture in-
creases.
The TAP experiments with the ZrO2 were conducted
with 0.90 g of sample. For the experiments with Pt/ZrO2 a
total of 0.30 g was used, corresponding to a total amount of
9 ꢄ 1018 atoms of Pt and 1.3 ꢄ 1021 atoms of Zr. To minimise
the void space in the microreactor, nonporous ꢀ-Al2O3
(particle size 0.25–0.30 mm) was used as packing material.
TAP Reactor System
A detailed description of the TAP system has been given
by Gleaves et al. (12). In essence, it consists of a microreac-
tor, two high-speed pulse valves which can inject between
1014 and 1018 molecules per pulse and a mass spectrometer
which can continuously monitor selected m/e values. Three
different types of transient experiments were performed:
single pulse, pump-probe, and multipulse. The single pulse
experiments were conducted by pulsing a single gas through
one of the pulse valves and observing the interaction of that
Several patents (6–8) and our own research (9–11) have
shown that ZrO2-based catalysts are very effective for the
CO2 reforming of CH4; not only are they more active but
1 Current address: Johnson Matthey Technology Centre, Blount’s
Court, Sonning Common, Reading RG4 9NH, UK. Fax: 01734-24 23 60.
306
0021-9517/97 $25.00
c
Copyright ꢁ 1997 by Academic Press
All rights of reproduction in any form reserved.

CO2 REFORMING OF CH4 USING TAP
307
gas with the sample. The pump-probe experiments were
In experiments with the ZrO2 support, a total of 0.90 g
performed by operating the two pulse valves, each pulsing was used. The pretreatment consisted of evacuation at a
a different gas, independently of each other, at various time pressure of 10ꢃ8 bar and a temperature of 250ꢂC. Single
intervals. Both the single pulse and the pump-probe experi- pulse experiments were then conducted. The interaction
ments consisted of repeated pulse cycles, with signal averag- of each gas was first measured at 50ꢂC and subsequently
ing to improve the signal to noise ratio. This procedure can at higher temperatures, up to 700ꢂC. After completing the
be applied if the responses obtained do not vary during the experiments with one gas, the sample was maintained under
measurement. This can be satisfied when (i) enough time is vacuum at 700ꢂC to allow for the desorption of the gas used
allowed between each pulse for the gases to exit the reac- before it was cooled down under vacuum to 50ꢂC for the
tor and (ii) the pulse intensity is sufficiently small as not to experiments with the next gas.
create a significant change on the sample surface. In general,
A total of 0.30 g of the 1 wt% Pt/ZrO2 sample was used;
a pulse intensity smaller than 1% of the number of atoms this was pretreated by heating at a rate of 2ꢂC ꢅ minꢃ1 to
exposed at the sample surface is considered acceptable. The 700ꢂC in a hydrogen stream with continuous evacuation,
multipulse experiments were a series of single experiments maintaining that temperature for 1 h. The sample was then
for which no signal averaging was applied and all the indi- cooled down overnight to 100ꢂC under high vaccum. The
vidual responses were recorded as a function of time.
single pulse experiments were performed in the same man-
ner as with the ZrO2 support. However, before starting
the experiments with 13CH4, the catalyst was treated ex-
tensively with CO2 pulses.
When possible, the conversions were calculated based on
the quantity of reactant and product measured. If that was
not possible, the conversionswere calculated usingthe reac-
tant and Ar peak areas. The CO and CO2 selectivities (dur-
ing CH4 pulsing) were based on the quantity of products
detected, assuming a carbon balance of 100% . The maxi-
mum absolute error in the conversions and selectivities was
determined to be 5% .
Experimental Conditions
The gases used were methane (purity: 99.9995% ), hydro-
gen (99.999% ), carbon monoxide (99.997% ), carbon diox-
ide (99.999% ), and argon (99.99995% ), all supplied by Air
Products. The 13CH4 had a purity of 99% and was supplied
by Messer Griesheim (Lot F-5007). As the responses were
measured with mass spectrometry, the following m/e values
were selected as being characteristic of each of the individ-
ual gases: H2, 2; CH4, 15 (pump-probe experiments) and
16 (single pulse experiments); 13CH4, 17; H2O, 18; CO, 28;
13CO, 29; Ar, 40; CO2, 44; and 13CO2, 45. Since CO2 also
gave a response at m/e = 28 and 13CO2 one at 29, the frag-
mentation patterns of CO2 and 13CO2 were obtained and
the appropriate corrections were made to give the CO or
RESULTS
Before the results are presented, there are some points
13CO responses, whenever either of these gases was present which need to be discussed. First, the residence time of the
in the reactor effluent. It should be noted that H2 is diffi- gases in the reactor is determined by the degree of inter-
cult to detect in a quadrupole mass spectrometer and so action (adsorption/desorption) with the sample and by gas
quantification of the H2 signal was not attempted.
phase diffusion. Gleaves et al. (12) has discussed that when
All gas volumes dosed contained 10 vol% of Ar, which the pulse intensity is less than 1016 molecules, the gas phase
was added as a tracer to allow for the determination of the diffusion is governed by Knudsen diffusion. This has the
pulse sizes and the conversions of (some of) the reactants. consequence that the residence time of each molecule is
Only with 13CH4 was no Ar added due to experimental directly proportional to the square root of the molecular
limitations. The pulse intensity used in the single pulse ex- weight. If no interaction takes place between the gases and
periments varied between 4 ꢄ 1015 and 7 ꢄ 1016 molecules the sample in the reactor, the molecules will exit the reac-
per pulse, whereas the pulse intensities with the pump- tor at different times, with the lightest first and the heaviest
probe experiments were of the order of 5 ꢄ 1015 to 10 ꢄ 1015 last. For the pump-probe experiments and the experiments
molecules. With the Pt/ZrO2 sample, these pulse sizes cor- with the 13CH4, the pulse intensities were 1016 or smaller,
responded to coverages of 0.1–1.4% of the total available and so gas phase diffusion was mainly Knudsen diffusion
number of Pt atoms (1 wt% Pt/ZrO2, 0.30 g, and 45% dis- in those experiments. This indicates that the order in which
persion). For the multipulse experiments with 13CH4 the the various molecules exited the reactor was H2 followed
pulse intensity was estimated to be 5 ꢄ 1015 molecules. Note by CH4, H2O, CO, Ar, and finally CO2.
that most of these pulse intensities were too large for the
A second important point is that no carbonaceous species
gas transport mechanism to be determined by Knudsen dif- remained on the catalyst during any of the experiments.
fusion only. For mechanistic investigations (as opposed to This was observed in preliminary experiments and the ev-
kinetic studies) the gas transport mechanism is less rele- idence for this conclusion is presented towards the end of
vant. These pulse intensities were chosen as they resulted this section, in relation to the results of the long term 13CH4
in better signal to noise ratios.
experiment.

308
VAN KEULEN ET AL.
Single Pulses over the ZrO2 Support
average residence time of 0.79 s, while the corresponding
Ar average residence time was 0.10 s. Since no interaction
was observed between CO and the zirconia, the CO most
likely interacts with the Pt.
CH4 was the only molecule which appeared before Ar at
all temperatures, suggesting that it has little or no interac-
tion with the Pt/ZrO2. However, the peak area obtained at
550ꢂC was only between 20 and 25% of the total peak area
obtained at 50ꢂC, indicating that 75 to 80% of the CH4 was
consumed. It appeared that the majority of the CH4 had
(irreversibly) reacted over the catalyst.
When the individual gases were pulsed over the sample of
0.90 g of the zirconia support at temperatures of 50, 550, and
700ꢂC, the following results were obtained. When H2 was
pulsed, it appeared rapidly at all temperatures examined
(50, 550, and 700ꢂC). At 550 and 700ꢂC the ratio of the H2
to Ar peak areas was an order of magnitude smaller than at
50ꢂC, suggesting that most of the hydrogen was converted
at the higher temperatures. As small quantities of H2O were
detected it seems probable that the H2 reduced the support.
When CO2 was pulsed, a strong interaction was observed
at all temperatures. For example, at 50ꢂC no CO2 was de-
tected in the gas phase. Subsequent heating at 20ꢂC ꢅ minꢃ1
revealed that the CO2 which had been adsorbed at 50ꢂC
desorbed from the zirconia above 200ꢂC. At 550ꢂC the CO2
pulses took more than 10 s to exit the reactor, while at 700ꢂC
the delay was more than 2 s.
Combined Pulses
Two sets of experiments were conducted in which gases
were pulsed simultaneously over the Pt/ZrO2 catalyst: (i)
H2 and CO2 and (ii) CH4 and CO2.
As Pt/ZrO2 is an effective (reverse) water-gas shift cata-
lyst (13, 14), the influence of simultaneous pulsing of H2
and CO2 was investigated. The pulse experiments were con-
ducted by pulsing H2, CO2, and Ar in the ratio 1 :1 :1. At no
temperature examined were H2O and H2 peaks recorded;
only a continuous low partial pressure of each of these
molecules was observed, indicating that slow desorption of
both took place. Evidence that the reverse water-gas shift
reaction had occurred was obtained by comparison of the
peak areas of CO and CO2. At a catalyst bed temperature of
550ꢂC, the CO area was 46% of the CO2 area; after correc-
tion for the fragmentation pattern of CO2 this corresponds
to a CO2 conversion of about 20% . At a temperature of
700ꢂC, the CO peak area was 63% of that of the CO2 peak
area, corresponding to a CO2 conversion about 30% .
In the second set of experiments, the influence of simulta-
neous pulsing of CH4 and CO2 was investigated at catalyst
bed temperatures of 550 and 700ꢂC. Each pulse consisted
of an equimolar mixture of CH4, CO2, and Ar (added as
reference) with a total pulse intensity of 9 ꢄ 1015 molecules.
The results obtained at 700ꢂC are presented in Fig. 1, in
which the normalised responses of CO2, CO, H2, CH4, and
Ar are plotted as a function of time. The response of H2O is
not presented as it had a significant level of noise. The CH4
peak appeared first, with a peak area corresponding to a
conversion of around 95% . Both CO and CO2 appeared
after Ar, while H2 appeared last. (It should be noted that
the apparent increase of the H2 signal at around 0.01 s is
a contribution from the fragmentation of CH4.) Using the
CO2 and Ar peak areas, a CO2 conversion of 86% was cal-
culated, which is lower than the CH4 conversion. The CO
produced corresponded (within 10% ) to the value expected
from the CH4 and CO2 conversions, indicating a reasonable
carbon balance.
Both CH4 and CO eluted rapidly, each appearing before
Ar at each of the temperatures examined. If only Knudsen
diffusion occurs, these gases would appear before Ar, so it
is concluded that they did not have a significant interaction
with the zirconia support. Note that the ratio of the CH4 to
the Ar peak areas at 50 and 700ꢂC were within 2% , showing
that no CH4 had reacted over the support.
Single Pulses over the Pt/ZrO2 Catalyst
The individual gases were pulsed over 0.30 g of a 1 wt%
Pt/ZrO2 sample at temperatures of 50, 550, and 700ꢂC. H2
exhibited a strong interaction with the Pt/ZrO2. Multipulse
experiments at 50ꢂC revealed that the H2 was completely
adsorbed in the first 55 pulses. Assuming that one H atom is
adsorbed per one Pt atom, this corresponds to a Pt disper-
sion of 55% . This figure compares well with a value of 45%
–
obtained using H2 O2 titrations (13). At the other tempera-
tures, more pulses of H2 also needed to be admitted before a
H2 response became visible. This indicates that the Pt/ZrO2
adsorbs H2 well.
CO2 again exhibited a strong interaction. No CO2 was
detected in the gas phase, even after 300 pulses had been
admitted to the sample at 50ꢂC. This corresponds to ad-
sorption of more than 1.5 ꢄ 1019 molecules of CO2, signif-
icantly more than the total number of Pt atoms present
(9 ꢄ 1018 atoms). It is therefore probable that the CO2 ad-
sorbs strongly on zirconia support, disguising any interac-
tion of CO2 with the Pt.
CO also exhibited a stronginteraction with Pt/ZrO2. Mul-
tipulse experiments with the sample at 50ꢂC revealed that
all the CO present in the first 90 pulses was adsorbed; this
amount corresponds to 3.6 ꢄ 1018 CO molecules. If one as-
sumes a CO to Pt ratio of 1 : 1, then this number indicates
a dispersion of 40% , agreeing well with the Pt dispersion
of 45% measured using H2–O2 titrations (13). At a sample
Pump-Probe Experiments
To gain more insight as to how the CO is produced,
temperature of 550ꢂC, the CO appeared after Ar, with an from the CH4 or the CO2, pump-probe experiments were

CO2 REFORMING OF CH4 USING TAP
309
FIG. 1. The normalised responses of CH4, Ar, CO2, CO, and H2 as a function of time when pulsing CH4, CO2, and Ar at t = 0 s and 700ꢂC. Note
that all signals have been normalised to the largest response, which was that of CO2.
conducted at 700ꢂC. The first series of experiments was car-
Variation of the time interval between the CO2 and the
ried out by first pulsing (at t = 0 s) a CO2/Ar mixture, and CH4 pulses in the other pump-probe experiments gave no
this was followed by pulsing a CH4/Ar mixture after vari- changes in the peak shapes, all being identical in form to
ous time intervals. The total time span of each pump-probe those presented in Figs. 2a and 2b. Only when the time
cycle was 3 s. The CO2 mixture consisted of 90.0% CO2 interval between both pulses was short (<0.5 s) were the
and 10.0% Ar, with a pulse intensity of 10 ꢄ 1015 molecules, signal responses apparently different. This, however, was
while the CH4/Ar mixture contained 90.0% CH4 and 10.0% the result of an overlap of the signal responses of the two
Ar, with a pulse intensity of 8 ꢄ 1015 molecules. The results pulses. Though the peak shapes were identical, it was ob-
obtained for the pump-probe with CH4/Ar pulsed at t = 2 s served that the peak areas were significantly different. For
are presented in Fig. 2a and 2b, in which the normalised example, it was found that when CO2 was pulsed before
responses of CO2, CO, CH4, and Ar are plotted as a func- CH4, the CO2 conversion increased when the time interval
tion of time. Note that time scale ranged from 0 to 0.25 s between the CO2 and CH4 pulses was increased. Since the
in Fig. 2a and from 2.0 to 2.25 s in Fig. 2b, while the signal total time of each pulse sequence was 3 s, a larger time in-
normalisation is the same in both figures.
terval between the first CO2 and following CH4 will also
Figure 2a, which presents the situation after a CO2/Ar result a shorter time interval between the CH4 and the next
pulse, shows that the CO and CO2 peak maximums ap- CO2 pulse.
peared after the Ar signal. The (nonnormalised) CO peak
A second series of pump-probe experiments was there-
area was 106% of the (nonnormalised) CO2 peak area, this fore conducted, in which either CO2 or CH4 was pulsed
corresponding to a CO2 conversion of 45% . (The observa- first and the second gas was pulsed after a time interval of
tion that the CO appeared before CO2, even though it is a 1 s. Since the peak shapes were identical to those shown in
product from CO2, may be explained (as seen earlier) by Figs. 2a and 2b, these are not presented. Instead, Table 1
the fact that the CO2 has a stronger interaction with the lists the CO2 and CH4 conversions based on the peak areas
Pt/ZrO2.)
as well as with the selectivities in which the CH4 is con-
In Fig. 2b, where the situation after the CH4/Ar pulse is verted into CO and CO2. The results from the first series
presented, CH4 appeared first. The (nonnormalised) peak are also presented in Table 1. Note that between the two
area corresponded to a conversion of 95% , twice the CO2 series several experiments were conducted; these resulted
conversion obtained from the results of Fig. 2a. The Ar, CO, in the catalyst being more reduced in the second series. As
and CO2 signals appeared together. That CO2 was observed will be shown in the long-term 13CH4 addition section the
suggests that it is a product of the CH4. The H2 response, degree of reduction of the catalyst affects the conversions
which for clarity is not presented in the figure, was similar and selectivities. Therefore the values should be compared
to that shown in Fig. 1.
within the series, not between the series.

310
VAN KEULEN ET AL.
FIG. 2. (a) The normalised responses of CH4, Ar, CO, and CO2 as a function of time when pulsing CO2 and Ar at t = 0 s and 700ꢂC. Note that all
signals have been normalised to the largest response, which was that of CO2. (b) The normalised responses of CH4, Ar, CO, and CO2 as a function of
time when pulsing CH4 and Ar at t = 2.0 s and 700ꢂC. Note that all signals have been normalised to the largest response, which was that of CO2 in (a).
For both series of experiments there are two trends: be concluded that oxygen was removed from the Pt/ZrO2
(i) the sooner the CO2 is pulsed after the CH4, the higher is material.
the CO2 conversions; and (ii) the sooner the CH4 is pulsed
after the CO2, the lower is the selectivity for the conver-
Pump-Probe with 13CH4
sion of the CH4 to CO. In addition to these trends, it is
The CO2 which evolved upon pulsing of CH4 could have
important to note that the CH4 conversion remained con- been produced by oxidation of the CH4 or by desorp-
stant at almost 100% for all experiments, but that the CO2 tion of previously adsorbed CO2 species. Therefore, a se-
conversion varied. The CO2 conversion was lower than ries of experiments with methane containing labelled car-
that of CH4 for each experiment and this resulted in the bon (13CH4) was carried out. Furthermore, in order to
oxygen balance being more that 100% . It must therefore gain more information on the incomplete oxygen balance

CO2 REFORMING OF CH4 USING TAP
311
Long-Term 13CH4 Addition
TABLE 1
The CO2 and CH4 Conversions and the Selectivities by Which
CH4 is Converted into CO and CO2 for Several CO2/CH4 Pump-
Probe Experiments
Extensive CO2 pretreatment before the pump-probe ex-
periments resulted in lower CO2 conversions and higher
13CO2 selectivities than when no CO2 pretreatment was
carried out. Therefore, it was decided to examine the effect
of a long-term sequence of 13CH4 pulses on the Pt/ZrO2
catalyst. A total of 2500 13CH4 pulses were admitted to the
sample, these corresponding to a total of 1.3 (ꢆ0.2) ꢄ 1019
molecules, or 1.4 (ꢆ0.2) methane molecules per Pt atom
present in the sample. During these pulses 13CH4, 13CO,
and 13CO2 were monitored continuously. The 13CH4 con-
version remained stable, while the selectivity to 13CO2
dropped, from 91% at the start of the sequence to 13% after
1200 pulses and to 2% after 2500 pulses, while the 13CO se-
lectivity increased accordingly. Unfortunately, accurate de-
termination of how much oxygen had been removed from
the Pt/ZrO2 sample was not possible as none of the H2O
produced could be detected. However, if one assumes that
Time of pulsing (s)
CO2
CH4
XCO (% )
XCH (% )
SCO a (% )
SCO a (% )
2
4
2
First series
0
0
0
0.5
1
2
33
38
46
98
97
97
63
59
51
37
41
49
Second series
0
1
1
0
46
53
97
96
49
45
51
55
a These are products of CH4.
Note. The time between each pulse time was 3 s and therefore pulsing
CH4 1 s after CO2 can also be viewed as pulsing CH4 2 s before CO2.
reported above, the Pt/ZrO2 sample was pretreated by an complete methane conversion occurs, with only CO2, CO,
extended series of CO2 pulses. Pump-probe experiments and H2 as products, the amount of oxygen extracted dur-
with sequences of CO2/Ar and 13CH4 were then carried ing the sequence corresponded to 1.7 oxygen atoms per Pt
out (without Ar present in the 13CH4); the CO2 mixture atom or 0.01 oxygen atom per Zr atom. However, the true
was the same as that used with the previous pump-probe value will be higher, since H2O will also have been formed.
experiments, while the 13CH4 contained 99% 13CH4 and
After 1200 pulses of the labelled methane had been ad-
1% 12CH4, with a pulse intensity of 5 ꢄ 1015 molecules. mitted, 12CO2 was pulsed and the formation of 12CO and
The experiments were carried out by first pulsing the 13CO was monitored to see if any carbonaceous species had
CO2/Ar mixture (at t = 0 s), followed by pulsing the 13CH4 been formed on the catalyst surface. The 12CO2 conver-
at t = 1 s.
sion was 86% and the products were 12CO together with a
The results obtained are presented in Figs. 3a and 3b, small amount of 13CO. However, the 13CO peak area cor-
in which the normalised responses of 12CO2, 13CO2, 12CO, responded to what would be expected on the basis of the
13CO, 13CH4, and Ar are plotted as a function of time. The naturally occurring of 13CO2 in 12CO2. It was therefore con-
time-scale ranges from 0 to 0.25 s in Fig. 3a and from 1.0 cluded that no carbonaceous species were formed on the
to 1.25 s in Fig. 3b, while the signal normalisation is the sample surface during the pulsing of the 13CH4.
same in both figures. In Fig. 3a, where the effect of puls-
After 2500 pulses of labelled methane, a 13CH4 and
ing 12CO2/Ar is examined, only 12CO2, 12CO, and Ar are CO2/Ar pump-probe experiment was again conducted,
detected. A slight change in the signals of 13CO2 and 13CO both gases being pulsed at the same time. A conversion
was also detected, but this could be ascribed to the natural of the 13CH4 of 98% was observed, this being identical to
occurrence of 13CO2 in 12CO2 (due to this the fragmenta- what has been found previously. However, the selectivity to
tion pattern of 12CO2 will contain a trace of the 13CO2). In 13CO2 was now only 1% , while 13CO was the predominant
all other ways the figure is identical to Fig. 2a. Figure 3b product. Furthermore, the CO2 conversion was 90% , the
shows that mainly 13CO2 and 13CO are produced after puls- highest observed in all the pulse experiments carried out.
ing of 13CH4. Small quantities of 12CO2 and 12CO were also It can therefore be concluded that removal of oxygen from
formed, but these could be accounted for by an impurity the catalyst not only enhances the CO2 conversion but also
level of 12CH4, present in the 13CH4. It can be concluded diminishes the production of CO2 from CH4.
that the CO2 detected after pulsing CH4 is a reaction prod-
uct of the CH4. In these experiments the 13CH4 conversion
DISCUSSION
was nearly complete at 95% , identical to what had been
observed with the nonlabelled species. However, selectiv-
Table 1 shows that when the time interval between the
ity to 13CO2 was 90% , much higher than was found in the CH4 and the CO2 pulses was varied, the CH4 conversion
previous pump-probe experiments, while the CO2 conver- remained constant. However, the conversion of CO2 did
sion was 3% much lower than observed before. Since the vary, increasing when the CO2 was pulsed sooner after the
CO2 did not supply much oxygen, these results indicate that CH4 pulse. Also, when the CH4 was pulsed sooner after the
13CH4 must have extracted oxygen from the Pt/ZrO2 sam- CO2, the selectivity to CO2 also increased. Both changes
ple, producing 13CO2 and 13CO.
could have been caused by some adsorbed species still being

312
VAN KEULEN ET AL.
FIG. 3. (a) The normalised responses of 13CH4, Ar, 12CO, 13CO, 12CO2, and 13CO2 as a function of time when pulsing 12CO2 and Ar at t = 0 s and
T = 700ꢂC. Note that all signals have been normalised to the largest response, which was that of 12CO2. (b) The normalised responses of 13CH4, Ar,
12CO, 13CO, 12CO2, and 13CO2 as a function of time when pulsing 13CH4 at t = 1.0 s and T = 700ꢂC. Note that all signals have been normalised to the
largest response, which was that of 12CO2 in (a).
present on the surface and/or by the surface gradually being and CO2 pulses. Solymosi et al. (15) have suggested that an
modified by the pulsing of CH4 or CO2.
enhanced dissociation of CO2 occurs in the presence of H2,
From the single pulse experiments, it can be concluded and Basini et al. (16) reported that CO2 chemisorption on
that the only species which remain sufficiently long on the Rh clusters was enhanced by hydrogen due to the forma-
catalyst to be able to react with subsequent pulses are ei- tion of a hydridocarbonyl cluster. It could thus be argued
ther adsorbed hydrogen or hydroxyl groups. It is possible that pulsing the CO2 soon after the CH4 will result in an en-
that the adsorption of CO2 is enhanced by the presence of hanced CO2 reaction due to the presence of more hydrogen
hydroxyl groups or by the formation of HCOꢃ3 species. On on the surface.
the other hand, adsorbed hydrogen does react with CO2
However, the changes in the selectivity with which CO2
as was demonstrated in the experiments with combined H2 is produced from CH4, brought about by pulsing CH4 at

CO2 REFORMING OF CH4 USING TAP
313
FIG. 4. A scheme of the reaction pathway of the over Pt/ZrO2.
various intervals after the CO2, cannot be explained by the the oxygen could be present as adsorbed oxygen species,
presence of hydrogen or hydroxyl groups on the surface. hydroxyl species, or perhaps even as an carbonate species.)
Furthermore, the low CO2 conversions occurring after pre-
treatment with CO2 cannot be explained in this manner over a Pt/ZrO2 catalyst. It is probable that the CH4 decom-
either. poses on the Pt, producing Cads and Hads, since no reaction
Figure 4 shows a possible mechanism for the reaction
The second possibility, a modification of the surface, pro- of CH4 was observed on the ZrO2. What happens to the
vides a good explanation for the observations above. As all Hads cannot be ascertained from the present experiments,
the pump-probe and labelled experiments exhibited a con- but it may be assumed that it will desorb, mainly as H2 but
stant methane conversion while the CO2 conversion varied also partly as H2O. The CO2 also dissociates on the catalyst,
considerably, it can be concluded that these two molecules producing an oxygen species and CO. Though the CO2 dis-
react independently of each other. As CO2 is converted sociation on supported noble metals has been previously
into CO and the CH4 into CO, CO2, H2, and H2O, this im- observed (15, 16), the literature is not clear on how it disso-
plies that the intermediate species between CO2 and CH4 ciates. Tol et al. (19) reported that CO2 does not dissociate
is an oxygen-containing species. We therefore suggest that on Pt, even though dissociation did occur on Rh (20). On
there is an oxygen pool on the surface, with CO2 contribut- the other hand, Huinink (21) found that oxygen exchange
ing and CH4 extracting oxygen. As the long-term pulsing of CO2 occurred when it was pulsed over a Pt surface and
of methane resulted in the removal of at least 1.7 oxygen Bitter et al. (22) observed, using infrared spectroscopy, that
–
atoms per Pt atom, this gives an indication of the size of a Pt CO bond was formed.
this pool. What exactly is the nature of the pool is not yet
The Cads formed subsequently reacts with the Oꢀ of the
known. However, as the Pt/ZrO2 catalyst is more active and pool to produce CO; as there was no evidence for the exis-
stable than Pt supported on other materials (11, 17), it may tence of Cads, it must be assumed that the reaction of Cads
be assumed that the zirconia plays a role. There is evidence is fast. The CO formed can either be oxidised to CO2 or
that ZrO2 can be reduced partially in the presence of Pt remain as CO, depending on the concentration of Oꢀ. If
(13, 18); Tournayan et al. (18) explained this reduction of there is an abundance of Oꢀ, the CO will be oxidised pref-
the ZrO2 by the formation of a Pt1ꢃxZrx alloy. This suggests erentially and CO2 will not dissociate. On the other hand,
that the oxygen pool used for the reactions described here if there is a shortage of Oꢀ, CO2 will dissociate to produce
is present in the vicinity of the Pt crystallites.
CO and CO will not be oxidised. The quantity of Oꢀ present
The existence of an oxygen pool explains the observation determines the direction of the CO2 dissociation reaction:
that the oxygen balance was never complete. Furthermore, CO2, ads, ⇔ COads + Oaꢀ.
the low CO2 conversion which occurred after the CO2 pre-
treatment can be explained by assuming that the oxygen
pool was full so that no oxygen was needed. Moreover, it
may also explain that the CO2 pretreatment resulted in the
CONCLUSIONS
It was found that CO2 and CH4 react independently of
CH4 being converted preferentially into CO2 instead of CO. each other on Pt/ZrO2. An oxygen pool is present on the
In other words, the quantity of oxygen (Oꢀ) present in the material; CO2 actsasan oxygen supplier, while CH4 extracts
pool determines whether or not the CO2 dissociation reac- the oxygen. When there is an abundance of oxygen on the
tion occurs:
catalyst surface, the CH4 can be oxidised to CO2. The quan-
tity of oxygen present in the pool determines the selectivity
with which CH4 produces CO and CO2. The direction of
CO2,ads ⇔ COads + Oꢀ.
(Note that the nature of the Oꢀ has not been identified; the CO2 dissociation reaction CO2,ads ⇔ COads + Oads (and

314
VAN KEULEN ET AL.
its reverse) is affected by the quantity of oxygen in the pool 10. Seshan, K., ten Barge, H. W., Hally, W., van Keulen, A. N. J., and Ross,
J. R. H., Stud. Surf. Sci. Catal. 81, 285 (1994).
11. van Keulen, A. N. J., Hegarty, M. E. S., Ross, J. R. H., and van den
and the presence of adsorbed hydrogen.
Oosterkamp, P. F., Stud. Surf. Sci. Catal., in press.
12. Gleaves, J. T., Ebner, J. R., and Kuechler, T. C., Catal. Rev. Sci. Eng.
ACKNOWLEDGMENTS
30, 49 (1988).
The authors thank the European Union for the financial support
13. van Keulen, A. N. J., Doctorate Thesis, University of Twente, The
through the JOULE II programme (Energy from fossil fuels: hydrocar-
Netherlands, 1996.
bons, Contract JOU2-CT92-0073) and T. Sommen and E. P. J. Mallens for
their assistance during the experimental work.
14. Xue, E., O’Keeffe, M., and Ross, J. R. H., Catal. Today 30, 107 (1996).
15. Solymosi, F., Kutsan, Gy., and Erdohelyi, A., Catal. Lett. 11, 149
(1991).
REFERENCES
16. Basini, L., Marchionna, M., and Aragno, A., J. Phys. Chem. 96, 9431
(1992).
1. Ashcroft, A. T., Cheetham, A. K., Green, M. L., and Vernon, P. D. F., 17. Tsipouriari, V. A., Efstathiou, A. M., Zhang, Z. L., and Verykios, X. E.,
Nature 352, 225 (1991).
Catal. Today 21, 579 (1994).
2. Richardson, J. T., and Papipatyadar, S. A., Appl. Catal. 61, 293 (1990).
3. Edwards, J. H., and Maitra, A. M., Fuel Proc. Tech. 42, 269 (1995).
4. Rostrup-Nielsen, J. R., and Bakhansen, J.-H., J. Catal. 144, 38 (1993).
5. Teuner, S., Hydrocarbon Processing 64, 107 (1985).
6. Czimczik, A., German Patent DE 39 33 285, 1991.
7. Basini, L., Marchionna, M., Rossini, S., and Sanfilippo, D., German
Patent DE 41 02 185, 1991.
8. Seshan, K., Ross, J. R. H., Mercera, P. D., and Xue, E., German Patent
DE 94 00 513, 1994.
9. Bitter, J. H., Hally, W., Seshan, K., van Ommen, J. G., and Lercher,
J. A., Catal. Today 29, 732 (1996).
18. Tournayan, L., Auroux, A., Charcosset, H., and Szymanski, R., Ads.
Sci. Tech. 2, 55 (1985).
19. van Tol, M. F. H., Gielbert, A., Wolf, R. M., Lie, A. B. K., and
Nieuwenhuys, B. E., Surf. Sci. 287/288, 201 (1993).
20. van Tol, M. F. H., Gielbert, A., and Nieuwenhuys, B. E., Appl. Surf.
Sci. 67, 166 (1993).
21. Huinink, J. P., Doctorate Thesis, Eindhoven University of Technology,
The Netherlands, 1995.
22. Lercher, J. A., Bitter, J. H., Hally, W., Niessen, W., and Seshan, K.,
in “Proceedings of the 11th International Congress on Catalysis,”
Baltimore, MD, Stud. Surf. Sci. Catal. 101, 463 (1996).