Hydrogen production via catalytic decomposition of methane

Article abstract of DOI:10.1006/jcat.2000.3142

Methane decomposition on various Ni-supported catalysts has been investigated as a method for production of CO-free hydrogen for use in fuel cells. The low levels of CO formed due to the interaction of surface carbon (formed from methane decomposition) with the support have been quantitatively analyzed (part per million levels) by methanation of the CO and subsequent analysis by flame ionization detection (FID). This study highlights the dependence of the type of carbon formed and the amount of CO evolved on the nature of the support. No filamentous carbon was observed on Ni/H-ZSM-5 at elevated methane decomposition temperatures, whereas Ni/HY and Ni/SiO₂ showed filamentous carbon formation over the entire temperature range studied (723 K to 873 K). While two forms of carbon (carbidic and graphitic) were observed on the Ni/SiO₂ after methane decomposition at 723 K, only graphitic carbon was observed at 823 K. The rate of CO formation was observed to be highest on Ni/H-ZSM-5 and lowest on Ni/SiO₂. The CO formation rates showed a common trend for all the catalysts: high initial rates followed by a lower stabilized rate. The CO formation rates were found to increase with increasing temperature. The CO content in the hydrogen stream was ca. 50 ppm and 250 ppm for Ni/SiO₂ and Ni/HY, respectively, after the CO production rates stabilized. The low levels of CO coupled with the stability of the catalysts for methane decomposition make this an interesting conceptual process for hydrogen production for fuel cell applications. Regeneration studies have shown that there is no loss of activity for methane decomposition at 723 K on Ni/H-ZSM-5 over many reaction cycles.

Full text of DOI:10.1006/jcat.2000.3142

Journal of Catalysis 199, 9–18 (2001)
doi:10.1006/jcat.2000.3142, available online at http://www.idealibrary.com on
Hydrogen Production via Catalytic Decomposition of Methane
T. V. Choudhary, C. Sivadinarayana, C. C. Chusuei, A. Klinghoffer, and D. W. Goodman1
Department of Chemistry, Texas A&M University, College Station, Texas 77842
Received June 12, 2000; revised December 6, 2000; accepted December 6, 2000; published online February 26, 2001
stream is even more stringent (ppm level) for the more re-
cently developed proton exchange membrane (PEM) fuel
cells. Conventional processes such as steam reforming of
methane, partial oxidation, and autothermal reforming (1–
4) produce significant amounts of CO along with hydro-
gen. Removal of CO from the hydrogen stream not only
increases the complexity but also is highly detrimental to
the total process economics. It is therefore of interest to
explore other alternatives for the production of CO-free
hydrogen.
Decomposition of methane has been studied quite ex-
tensively for methane homologation reactions (5–8) and
fundamental dynamical studies (9, 10); however, methane
decomposition as a method to obtain pure hydrogen has re-
ceived relatively little attention (11, 12). We have recently
proposed stepwise steam reforming of methane to produce
CO-free hydrogen (13, 14). This process, which operates at
relatively low temperatures ( 673 K), involves the catalytic
decomposition ofmethane in a first step to produce CO-free
hydrogen and surface carbon and/or hydrocarbonaceous
species followed by a separate step in which the surface
carbon/carbonaceous species is removed via reaction with
water. However, the methane conversions are small at low
temperatures; hence, it is important to investigate conver-
sions at higher temperatures.
Methane decomposition on various Ni-supported catalysts has
been investigated as a method for production of CO-free hydrogen
for use in fuel cells. The low levels of CO formed due to the in-
teraction of surface carbon (formed from methane decomposition)
with the support have been quantitatively analyzed (part per mil-
lion levels) by methanation of the CO and subsequent analysis by
flame ionization detection (FID). This study highlights the depen-
dence of the type of carbon formed and the amount of CO evolved
on the nature of the support. No filamentous carbon was observed
on Ni/H-ZSM-5 at elevated methane decomposition temperatures,
whereas Ni/HY and Ni/SiO2 showed filamentous carbon formation
over the entire temperature range studied (723 K to 873 K). While
two forms of carbon (carbidic and graphitic) were observed on the
Ni/SiO2 after methane decomposition at 723 K, only graphitic car-
bon was observed at 823 K. The rate of CO formation was observed
to be highest on Ni/H-ZSM-5 and lowest on Ni/SiO2. The CO for-
mation rates showed a common trend for all the catalysts: high
initial rates followed by a lower stabilized rate. The CO forma-
tion rates were found to increase with increasing temperature. The
CO content in the hydrogen stream was ca. 50 ppm and 250 ppm
for Ni/SiO2 and Ni/HY, respectively, after the CO production rates
stabilized. The low levels of CO coupled with the stability of the
catalysts for methane decomposition make this an interesting con-
ceptual process for hydrogen production for fuel cell applications.
Regeneration studies have shown that there is no loss of activity
for methane decomposition at 723 K on Ni/H-ZSM-5 over many
c
reaction cycles.
2001 Academic Press
It has been shown that at temperatures above 823 K,
Ni-based catalysts exhibit stable operation for a few hours,
providing 2 mol of hydrogen per mole of methane reacted
(12). Methane decomposition can result in carbon yields as
high as 384 gC/gNi on 90% Ni/SiO2 in a vibro-fluidized cat-
alyst bed at 823 K (15, 16). The absence of rapid catalytic
deactivation is explained by the diffusion of carbonaceous
species into the bulk of the metal particles. The latter pro-
cess generates active surface sites for dehydrogenation of
subsequent methane molecules resulting in the formation
of filamentous carbon. Recently, it has been noted that de-
composition of methane may lead to CO formation via re-
action of the carbonaceous residue with the oxygen of the
oxide support (17). These studies, however, did not provide
CO and hydrogen formation rates under steady state con-
ditions and hence the CO content in the hydrogen stream
could not be determined. The CO issue is currently the
most crucial aspect in fuel processing for fuel cells. Since
Key Words: step-wise reforming; methane decomposition; CO-
free hydrogen; filamentous carbon.
INTRODUCTION
Fuel cells being environmentally benign and highly effi-
cient represent an exciting form of technology for convert-
ing the chemical energy of a fuel directly into electricity.
Methane (the main constituent of natural gas), due to its
large abundance and high C/H ratio, is an ideal source of hy-
drogen (which is the best known primary fuel for fuel cells).
Phosphoric acid fuel cells, which are the most commercially
developed fuel cells, do not tolerate CO concentrations
above 1.5% . The CO-free requirement of the hydrogen
1 To whom correspondence should be addressed. Fax: (979) 845-6822.
E-mail: goodman@mail.chem.tamu.edu.
9
0021-9517/01 $35.00
c
Copyright
2001 by Academic Press
All rights of reproduction in any form reserved.

10
CHOUDHARY ET AL.
the main advantage offered by the stepwise reforming pro- all of the hydrogen flow to the FID through the metha-
cess is production of hydrogen with extremely low levels nizer. An auxillary flow of carrier gas was employed to
of CO, accurate analysis of this CO content in the product obtain the optimum carrier/hydrogen ratio for maximum
stream and study of factors influencing the CO content be- detector sensitivity. Various amounts of CO were injected
come imperative. To address this issue, in this work efforts into the methanizer to check for 100% conversion of CO
have been undertaken to monitor extremely small concen- to methane. This process was repeated regularly to ensure
tration of CO (ppm levels could be analyzed employing a complete conversion of CO. If any aberration was observed
methanizer coupled with a flame ionization detector). The the methanizer was regenerated (oxidation followed by re-
influences of the support and reaction conditions on the CO duction) until quantitative conversion of CO to methane
content have been investigated in this work. Transmission was observed. Regular regeneration/testing of the metha-
electron microscopy (TEM) and X-ray photoelectron spec- nizer assured constant reproducible complete conversion
troscopy (XPS) were employed to study the nature of the of CO to methane.
surface carbon; CO pulse experiments were carried out to
measure the specific metal surface area.
Before the reaction, the catalyst was reduced with H2
in flowing Ar at a total flow rate of 20 ml/min for 0.5 h at
523K and at 723K for 2.5h at 723K. Followingthis, the cata-
lyst was flushed with Ar for 0.5 h and heated to the desired
reaction temperature in a flow of carrier gas. High purity
gases (20% CH4 in Ar (certified mixtute), Ar (UHP), hy-
drogen (zero grade), and air (zero grade)) were employed
for this investigation. Product analysis showed that 2 mol of
hydrogen per mole of methane consumed were formed for
all the catalysts and conditions investigated. Though small
amounts of CO2 (<500 ppm) were sometimes observed, no
emphasis was placed on CO2 analysis as CO2 does not pose
serious problems in the operation of acid fuel cells. The
catalyst was weighed before and after the methane decom-
position reactions and a carbon mass balance of 5% was
obtained.
EXPERIMENTAL
a. Catalyst Synthesis
Ni catalysts supported on H-ZSM-5, HY, and SiO2 were
synthesized by conventional wet-impregnation with a nom-
inal metal loading of 10 wt% . Ni(NO3)2 6H2O (amount
corresponding to 10 wt% Ni in the catalyst was used for
impregnation) salt was used as the source of Ni metal and
in each case the impregnation procedure was performed at
room temperature. H-ZSM-5 (Code No. DAZ-P; Si/Al =
500; surface area, 400 m2/g; and pore volume, 0.18 cm3/g)
and NaY (Code No. DAY-P; Si/Al = 100; surface area,
700 m2/g; and pore volume, 0.3 cm3/g) were obtained from
Degussa and SiO2 (Cab-O-Sil; surface area, 325 m2/g; and
surface density of hydroxyl groups, 3.055 nm 2) was ob-
tained from Cabot Corp. NaY was converted to HY via
aqueous ion exchange (ammonium hydroxide solution).
The procedure was repeated 3–6 times to achieve a high
degree of ion exchange. Prior to loading Ni on silica, the
silica was calcined overnight at 823 K. After impregnation,
the catalysts were dried overnight at 373 K and then cal-
cined at 823 K for 4 h. The powder catalysts were then
pressed, crushed, and sieved to a 20–40 mesh size.
c. Catalyst Characterization
1. X-ray photoelectron spectroscopy. XPS was per-
formed in an ion-pumped (300 L/s) Perkin-Elmer PHI 560
system using a PHI 25-270AR double-pass cylindrical mir-
ror analyzer. A Mg K anode with a photon energy of
1253.6 eV was operated at 15 kV and 300 W. Survey scans
were performed using a 100 eV pass energy; high-resolution
scans were performed with a 50 eV pass energy. Core lev-
els of the C(1s), O(1s), Al(2p), Al(2s), Si(2p), and Ni(2p)
orbitals were scanned and normalized to their respective
atomic sensitivity factors (18). The system pressure during
b. Apparatus and Analytical Techniques
The continuous-flow reactor system consisted of a con- XPSanalysiswas
1
10 8 Torr. The Cu(2p3/2), (932.7eV)
ventional fixed-bed plug-flow stainless steel (SS) reactor and Au(4 f7/2) (84.0 eV) orbitals were used as standards
and a furnace controlled by a temperature controller. The from sputter-cleaned foils to calibrate the XPS BE range
SS reactor was 10 in. long with 38 in. O.D. with a small section (19). The precision of the BE measurements was 0.2 eV.
serving as the catalyst bed. The analysis system consisted of Signal from adventitious carbon at BE = 284.7 eV for the
a thermal conductivity detector (TCD), a flame ionization C(1s) level was used to correct for sample charging (20).
detector (FID), and a methanizer. The gases to be ana- Samples were mounted onto a 1.0 cm
1.0 cm
1.0 cm
lyzed first passed through the TCD and finally reached the support using double-sided tape (Scotch, 3M) attached to
FID via the methanizer (Carle Cat. No. 0117). The FID a probe and introduced into the UHV via a turbo-pumped
was calibrated over a large dynamic range for methane. Es- antechamber. The support was attached to a probe that
sentially 100% conversion efficiency (CO to methane) was was differentially pumped using sliding seals. Curve fitting
achieved by operating the methanizer at 400 C with large of the XPS data was performed using Peakfit ver. 3.1 B
amounts of hydrogen. This was accomplished by routing (Jandel Scientific) software.

CATALYTIC DECOMPOSITION OF METHANE
11
2. Specific area measurements. The Ni surface area was 5 m2/g. The average particle size ranged from 30–250 nm for
measured by CO pulse adsorption experiments carried out the catalysts under consideration and was measured from
at room temperature assuming a CO/Ni ratio of 1.0. Prior the chemisorption data using the usual approximation (21):
to the Ni surface area determination, the catalyst was pre- (a) metal particles have a spherical geometry and (b) sur-
reduced for 0.5 h at 523 K and for 2.5 h at 723 K in flowing H2 face is formed from equal proportions of the main low-
and He, and then brought to room temperature in flowing index planes.
He. The analysis of CO was carried out online with a TCD.
2. TEM and XPS. Figure 1a shows the TEM mi-
crograph of Ni/H-ZSM-5 catalyst after interaction with
methane and complete deactivation at 723 K. The figure
shows a single crystal of H-ZSM-5 with carbon filaments
projecting from it. Such outgrowths of filamentous carbon
from the sides of the H-ZSM-5 were observed throughout
the sample. Figure 1b shows the image of a single filament
of carbon. The pear-shaped Ni particle is present at the apex
of the filament. Growth of filamentous carbon involves the
adsorption and decomposition of methane on various faces
of the Ni particle followed by dissolution of some of the car-
bon species into the bulk and subsequent diffusion through
the metal particle due to a concentration gradient (22–24).
No filamentous carbon was observed after the methane
decomposition was carried out on Ni/H-ZSM-5 at 973 K in-
stead, TEM micrographs indicate the presence of graphitic
carbon, which encapsulates the Ni particles (Fig. 2). In the
case of Ni supported on HY and SiO2, filamentous carbon
formation was observed over the entire temperature range
(723K to 873K). Figure 3showsthe presence offilamentous
carbon on the Ni/HY sample after complete deactivation
due to methane decomposition at 823 K. This represen-
tative TEM image indicates that the catalyst can support
large amounts of filamentous carbon. It is noteworthy that
the amount of filamentous carbon formed on Ni/SiO2 was
greater than that on Ni/H-ZSM-5 at 723 K. This suggests
that some species other than filamentous carbon were also
formed on Ni/H-ZSM-5. Further discussion on this topic
will be continued in the XPS section.
3. TEM characterization. TEM micrographs of the
supported Ni catalysts were obtained using a high-
resolution Jeol 2010 instrument (Electron Microscopy Cen-
ter, Texas A&M University). Samples after reaction were
ultrasonically dispersed in acetone and spread over perfo-
rated molybdenum grids. Several bright-field TEM micro-
graphs of different portions of the sample were obtained at
magnifications up to 400,000.
4. Diffuse reflectance infrared spectroscopy (DRIFTS).
The DRIFTS experiments were performed with a Perkin-
Elmer Spectrum 2000 spectrometer equipped with an MCT
detector and a DRIFTS (Harrick) cell used in a flow mode.
Mixtures of silica (10 % Ni/SiO2) and KBr (ratio of 1 : 7)
were packed in a metal cup reactor whose temperature
could be varied from ambient to 873 K. The cell was purged
with dry nitrogen for at least 4 h prior to the DRIFTS ex-
periment. The DRIFTS spectrum of SiO2 (same batch that
was used for preparing the Ni/SiO2 catalyst) was obtained
after heating the sample to 373 K for 1 h under flow of He
to remove the adsorbed water. Prior to obtaining the spec-
tra on the Ni/SiO2 catalyst, it was pretreated in the same
manner as in the conventional catalytic reactor and heated
to the reaction temperature of 823 K.
RESULTS AND DISCUSSION
a. Surface Characterization
1. CO pulseexperiments. The resultsobtained from CO
pulse experiments are shown in Table 1. Metal dispersions
have been calculated from the CO uptake assuming an ad-
sorption stoichiometry of one CO molecule per Ni site.
The metal loading was the same for all the catalysts stud-
ied (10 wt% Ni). The highest catalyst dispersion was ob-
served on H-ZSM-5 support and the lowest on HY. The
measured specific area for all the catalysts was less than
Figure 4 shows XP spectra obtained for the C(1s) level
after complete deactivation of the Ni/SiO2 as a result of car-
bon deposition from methane decomposition reaction at
823 and 723 K, respectively. While a single peak (graphitic
carbon) was observed for the higher temperature condi-
tions, two features separated by 2.8 eV were observed af-
ter the lower temperature reaction. The graphitic form of
carbon (284.7 eV) was assigned to filamentous carbon (as
seen in TEM), whereas the lower binding energy peak at
281.9 eV was assigned to carbidic carbon. This assignment
is in good agreement with previous XPS literature involv-
ing carbide features (25, 26). The presence of the single
graphitic peak at 823 K indicates that the carbidic form
does not exist at higher temperatures, which is consistent
with a previous study (27).
TABLE 1
CO Chemisorption Data on Ni-Supported Catalysts
CO adsorbed
Dispersion
(% )
Particle sizea
(nm)
1
Catalyst
(
mol g )
10% Ni/H-ZSM-5
10% Ni/HY
10% Ni/SiO2
52
7
16
3.1
0.4
1.0
32
250
100
b. Interaction of Methane with Ni/ZSM-5
Figure 5 shows the total amount of hydrogen (in mono-
layer equivalents, MLE) produced until complete catalyst
a From chemisorption data.

12
CHOUDHARY ET AL.
FIG. 1. TEM image of (a) filamentous carbon formed on Ni/H-ZSM-5 and (b) a single filament of carbon at 723 K.
deactivation as a function of reaction temperature on 10% during filamentous carbon growth, were available for fur-
1
Ni/H-ZSM-5 (GHSV : 15,000 cm3
g
h
1). The hydrogen ther methane decomposition and resulted in a greater lifes-
monolayer equivalents (MLE) were estimated from the pan of the catalyst. Encapsulation of Ni particles caused
specific metal surface area assuming one hydrogen mono- rapid deactivation of catalyst and hence the total amount
layer to correspond to a C/Nisurface ratio of 0.5. The total of hydrogen produced is extremely small at 973 K. As
amount of hydrogen produced decreased drastically with the catalyst was fairly stable in the temperature range of
increase in reaction temperature. As seen earlier in TEM 723 to 773 K CO analysis experiments were performed at
(Fig. 1a) there was evidence for filamentous carbon forma- 723 and 773 K. Figure 6 shows the rate of hydrogen and
tion at 723 K whereas only the encapsulating graphitic type CO production as a function of time for the two tempera-
is observed at 973 K. Ni particles, which had been present tures. As can be seen from the figure, no induction period

CATALYTIC DECOMPOSITION OF METHANE
13
FIG. 2. TEM micrograph of carbon formed on Ni/H-ZSM-5 at 973 K.
FIG. 3. TEM image of carbon filaments on Ni/HY after methane decomposition reaction at 823 K.

14
CHOUDHARY ET AL.
FIG. 7. Hydrogen production on Ni/H-ZSM-5 as a function of regen-
1
eration cycles at 723 K (GHSV = 15,000 cm3
g
h
1; 0.1 g of catalyst).
FIG. 4. XPS spectrum of C(1s) formed on Ni/SiO2 surface after reac-
tion with methane at 723 and 823 K.
(methane conversion did not pass through a maximum) was
observed and the rate of hydrogen production at 773 K was
more than twice that at 723 K. The initial CO production
rate was greater at 773 K, but gradually became compa-
rable with time for both temperatures. Figure 7 shows the
data obtained for regeneration cycles on 10% Ni/ZSM-5 at
a reaction temperature of 723 K. The following procedure
was used for the study: reaction with 20% methane in Ar
(60 min at 723 K) followed by regeneration in air (30 min at
723 K) and reduction with 1 : 1 H2 (15 min at 523 K and
30 min at 723 K). The catalyst was flushed with argon for
30 min before and after the reduction treatment. The fig-
ure clearly shows that there was no drop in activity of the
catalyst through the 12 cycles studied. A similar study was
carried out at a reaction temperature of 773 K (the regen-
eration with air was carried out at 773 K instead of 723 K);
FIG. 5. Effect of temperature on the production of total hydrogen however, in this case a gradual decrease in activity was
1
(MLE) on Ni/H-ZSM-5 (GHSV = 150,000 cm3
g
h
1; 0.1 g of catalyst).
observed with every cycle (Fig. 8) which we attribute to
FIG. 6. Time-on-stream data of hydrogen and carbon monoxide for-
FIG. 8. Hydrogen production on Ni/H-ZSM-5 as a function of regen-
mation during methane decomposition on Ni/H-ZSM-5 at 723 (ᮀ, ᭿) and
1
eration cycles at 773 K (GHSV = 15,000 cm3
g
h
1; 0.1 g of catalyst).
1
773 K (᭺, ᭹) (GHSV = 15,000 cm3
g
h
1; 0.1 g of catalyst).

CATALYTIC DECOMPOSITION OF METHANE
15
FIG. 9. Time-on-stream data of hydrogen and carbon monoxide for-
mation during methane decomposition on Ni/HY at 723 and 823 K
1
(GHSV = 20,000 cm3
g
h
1; 0.1 g of catalyst).
sintering of the Ni particles at higher temperature (during
regeneration).
c. Methane Decomposition on Ni/HY and Ni/SiO2
1. Ni/HY. Figure 9 shows the rate of hydrogen and CO
production as a function of time during methane decom-
position on 10% Ni/HY at 723 K and 823 K (GHSV =
1
20,000 cm3
g
h
1). Similar to the Ni/ZSM-5 system no
FIG. 11. Time-on-stream data of (a) hydrogen and (b) carbon monox-
ide formation during methane decomposition on Ni/SiO2 at 723, 823, and
induction period is observed. The rate of CO production is
higher at 823 K as compared to the rate at 723 K, but due to
much larger hydrogen formation rates at 823 K the levels
of CO in the hydrogen stream are larger at 723 K. Figure 10
shows the time-on-stream plot for the methane decomposi-
tion reaction at 823 K until deactivation. The catalyst, hav-
ing an initial conversion of 31% , gradually deactivates in
1
873 K (GHSV = 20,000 cm3
g
h
1; 0.1 g of catalyst).
ca. 14 h. The amount of CO in the hydrogen stream is high
initially (0.2% ) but then decreases to ca. 250 ppm (where it
stabilizes for 8 h) and finally increases slightly toward the
end of the reaction. This increase toward the end can be at-
tributed to the decrease in the rate of hydrogen formation
as the catalyst approaches its complete deactivation.
2. Methane decomposition on Ni/SiO2. Hydrogen and
CO formation rates on 10% Ni/SiO2 at various reaction
temperatures are shown in Figs. 11a and 11b, respectively.
The rates of hydrogen production on our catalyst (ca.
1.4 mol (H2)/gNi h) were found to be comparable with pre-
vious methane decomposition studies on Ni/SiO2 catalysts
(ca. 1.1 mol (H2)/gNi h (12), ca. 1.7 mol (H2)/gNi h (16)) at
823 K. The initial conversion (20% CH4 in Ar) ranges from
11% at 723 K to 52% at 873 K. Unlike the Ni/H-ZSM-5 and
Ni/HY systems, a modest induction period is observed in
case of the Ni/SiO2 catalyst. For the 873 K reaction, rapid
deactivation is observed after ca. 4 h, whereas the catalyst
shows stable activity for periods much longer than 4 h for
the reactions carried out at 723 and 823 K. The total amount
of hydrogen produced was found to be larger at 823 than at
873 K. The rate of CO formation on Ni/SiO2 followed the
FIG. 10. Total CO content and hydrogen production rates as a func-
1
tion of time for Ni/HY at 823 K (GHSV = 20,000 cm3
g
h
1; 0.1 g of
catalyst).

16
CHOUDHARY ET AL.
FIG. 12. Total CO content and hydrogen production rates as a func-
1
tion of time for Ni/SiO2 at 823 K (GHSV = 20,000 cm3
g
h
1; 0.1 g of
catalyst).
FIG. 13. Comparison of methane conversion as a function of time
for Ni supported on HZSM-5, HY, and SiO2 at 823 K (GHSV =
1
20,000 cm3
g
h
1; 0.1 g of catalyst).
same trends as for the Ni/HY and Ni/ZSM-5 (i.e., rate of
CO formation increased with increasing reaction tempera-
tures). Interestingly, the increase in the CO formation rate
was greater from 823 to 873 K than from 723 to 823 K. This
may be explained by the fact that high temperatures ther-
modynamically favor CO formation. Figure 12 shows the
rate of hydrogen formation and CO content in the hydrogen
a gradual deactivation is observed. No induction period is
observed in the case of Ni/H-ZSM-5 and Ni/HY, whereas
Ni/SiO2 shows an induction period. Figure 14 shows the
time-on-stream CO content (in the hydrogen stream) for
the catalysts at 723 K. For all the catalysts, the amount of
CO is highest at the beginning and then decreases until it
reaches some steady state value. This may be related to the
rapid decrease in hydroxyl groups during the initial stages
of the decomposition process. The largest CO content in
the product stream was observed for ZSM-5 and the small-
est for SiO2, which may be related to the amount/stability
of the –OH groups present on the different supports.
Figure 15 shows the rate of CO production during the
stream for methane decomposition reaction on Ni/SiO2 at
1
823 K (GHSV = 20,000 cm3
g
h
1). The catalyst shows
stable activity for methane decomposition for 8 h, and
then rapid deactivation. The CO concentration is 0.2%
initially, drops below 100 ppm in 90 min, and then remains
at ca. 50 ppm for 6 h after, followed by a slight increase dur-
ing the deactivation stage of the catalyst. Approximately 1
mol of hydrogen can be produced per gram of catalyst with
CO levels close to 50 ppm. Investigations are currently be-
ing directed toward decreasing the time required to obtain
low levels of CO in the product stream.
d. Methane Decomposition: Comparison of Supports
The conversion levels for methane are shown in Fig. 13
as a function of time for the various catalysts studied at
1
823 K (GHSV = 20,000 cm3
h
g
1). The initial methane
conversions are comparable for all three catalysts. Com-
plete deactivation is observed for Ni/ZSM-5 in ca.1 h
whereas Ni/SiO2 and Ni/HY exhibit catalytic activity for
methane decomposition in excess of 12 h. As discussed
earlier, at high temperatures the graphitic carbon formed
on Ni/H-ZSM-5 system encapsulates the Ni particles
and leads to rapid deactivation. But in the cases of Ni/HY
and Ni/SiO2, there is evidence for filamentous carbon for-
mation leading to a longer lifetime of the catalyst. A no-
table difference observed between the deactivation profiles
of Ni/SiO2 and Ni/HY is that Ni/SiO2 shows stable activity
for ca. 8 h and then deactivates rapidly, whereas for Ni/HY
FIG. 14. CO content as a function of time for Ni supported on HZSM-
1
5, HY, and SiO2 at 723 K (GHSV = 20,000 cm3
g
h
1; 0.1 g of catalyst).

CATALYTIC DECOMPOSITION OF METHANE
17
nature of the surface carbon and CO formation rates on
the nature of the support. At high temperatures an encap-
sulating type of graphitic carbon was observed on Ni/H-
ZSM-5 whereas Ni/HY and Ni/SiO2 showed the presence
of filamentous carbon, which resulted in a greater lifetime
for the latter two and a very fast deactivation for the for-
mer. The carbidic form of carbon was present at low tem-
peratures ( 723 K) but absent when higher temperatures
were employed (>773 K) on all catalysts. The CO content
in the hydrogen stream was found to be highest on Ni/H-
ZSM-5 and lowest on Ni/SiO2. In general, CO formation
rates were high initially but decreased rapidly with time and
then remain steady until catalyst deactivation. The amount
of CO in the hydrogen stream stabilized at ca. 250 ppm and
50 ppm for Ni/HY and Ni/SiO2 at 823 K. Regeneration stud-
ieshave shown that Ni/H-ZSM-5retainsitscatalytic activity
for methane decomposition over a number of reaction cy-
cles at 723 K. Future studies addressing oxidative and steam
regeneration are currently under way.
FIG. 15. Comparison of CO formation rates on Ni supported on HY
1
and SiO2 at 823 K (GHSV = 20,000 cm3
g
h
1; 0.1 g of catalyst).
methane decomposition on Ni/HY and Ni/SiO2 at 823 K
1
(GHSV = 20,000 cm3
h
g
1). The CO formation rates
ACKNOWLEDGMENTS
were similar initially for both catalysts, but with time de-
We acknowledge with pleasure the support of this work by the U.S. De-
partment of Energy, Office of Basic Energy Sciences, Division of Chemical
Sciences. We are grateful to Prof. Lunsford for kindly allowing us access
to the DRIFTS setup. T.V.C. acknowledges the Link Foundation for the
Link Energy Fellowship. C.C.C. gratefully acknowledges support from
the Associated Western Universities, Inc., and Pacific Northwest National
Laboratories operated by Battelle Memorial.
1
creased and stabilized to 0.01 and 0.035 mol min for
Ni/SiO2 and Ni/HY, respectively. The total amount of CO
formed (obtained by integrating the area under the curve)
was found to be 14.5 and 29.4 mol for Ni/SiO2 and Ni/HY,
respectively. DRIFTS experiments showed the presence of
two types of hydroxyl species on silica; isolated and hy-
drogen bonded (similar to a previous study (28)). After
the pretreatment of the catalyst and subsequent heating to
823 K, although the total hydroxyls were found to decrease
a significant quantity was retained. The fraction of hydroxyl
groups on silica remaining after heat treatment at various
temperatures is well documented in literature (29–32). Pre-
vious studies showed that in vacuo heating of silica to 873 K
resulted in elimination of approximately two-thirds of the
initial hydroxyl groups (32). Considering the above and tak-
ing into account the hydroxyl density of un-pretreated silica
and surface area of silica, calculations indicate the presence
of approximately 55 mol of hydroxyl species at 823 K on
0.1 g of Ni/SiO2. Occasionally ppm levels of CO2 were ob-
served in the product stream, but these were not accorded
rigorous attention, as trace amounts of CO2 do not pose
significant problems to the PEM fuel cells. Therefore, it
is difficult to estimate the fraction of the surface hydroxyl
groups reacted in the process at this point. Further investi-
gation will be required to quantitatively relate the hydroxyl
groups to the COx formed during the catalytic decomposi-
tion of methane.
REFERENCES
1. Rostrup-Nielsen, J. R., in “Catalytic Steam Reforming, Science and
Engineering” (J. R. Anderson and M. Boudart, Eds.), Vol. 5. Springer,
Berlin, 1984.
2. Rostrup-Nielsen, J. R., Catal. Today 18, 305 (1993).
3. Choudhary, V. R., Mamman, A. S., and Sansare, S. D., Angew. Chem.,
Int. Ed. Engl. 31, 1189 (1992).
4. Armor, J. N., Appl. Catal. 176, 159 (1999).
5. Koerts, T., Deelan, M. J., and Van Santen, R. A., J. Catal. 138, 101
(1992).
6. Koranne, M. M., and Goodman, D. W., Catal. Lett. 30, 219 (1995).
7. Zadeh, J. M., and Smith, K. J., J. Catal. 183, 232 (1999).
8. Belgued, M., Pareja, P., Amariglio, A., and Amariglio, H., Nature 352,
789 (1991).
9. Beebe, T. P., Jr., Goodman, D. W., and Yates, J. T., Jr., J. Chem. Phys.
87, 2305 (1987); Wu, M.-C., and Goodman, D. W., Catal. Lett. 24, 23
(1994); Lenz-Solomun, P., Wu, M.-C., and Goodman, D. W., Catal.
Lett. 25, 75 (1994); Wu, M.-C., and Goodman, D. W., Surf. Sci. Lett.
306, L529 (1994); Wu, M.-C., Xu, Q., and Goodman, D. W., J. Phys.
Chem. 98, 5104 (1994).
10. Choudhary, T. V., and Goodman, D. W., J. Mol. Catal. A 163, 9 (2000).
11. Muradov, N. Z., Energy Fuels 12, 41 (1998).
12. Zhang, T., and Amiridis, M. D., Appl. Catal. A 167, 161 (1998).
13. Choudhary, T. V., and Goodman, D. W., Catal. Lett. 59, 93 (1999).
14. Choudhary, T. V., and Goodman, D. W., J. Catal. 192, 316 (2000).
15. Emrakova, M. A., Emrakov, D. Y., Plyasova, L. M., and Kuvshinov,
G. G., Catal. Lett. 62, 93 (1999).
CONCLUSIONS
Methane decomposition on various Ni-supported cata-
lysts revealed that there was a strong dependence of the
16. Emrakova, M. A., Emrakov, D. Y., Kuvshinov, G. G., and Plyasova,
L. M., J. Catal. 187, 77 (1999).

18
CHOUDHARY ET AL.
17. Ferreira-Aparicio, P., Rodriguez-Ramos, I., and Guerrero-Ruiz, A.,
25. Mizokawa, Y., Nakanishi, S., Komoda, O., and Miyase, S., J. Appl.
Phys. 67, 264 (1990).
Appl. Catal. A 148, 343 (1997).
18. Wagner, C. D., Riggs, W. M., Davis, L. E., Moulder, J. F., in “Hand- 26. Belton, D. N., Harris, J. S., Schmieg, S. J., Weiner, A. M., and Perry,
book of X-ray Photoelectron Spectroscopy” (G. E. Muilenbuerg, Ed.).
Perkin-Elmer Corp., Eden Prairie, MN, 1979.
19. Seah, M. P., Surf. Interface Anal. 14, 488 (1989).
20. Barr, T. L., and Seal, S. J., J. Vac. Sci. Technol. A 13, 1239 (1995).
21. Anderson, J. R., “Structure of Metallic Catalysts,” p. 296. Academic
Press, New York, 1975.
22. Rodriguez, N. M., J. Mater. Res. 8, 3233 (1993).
23. Baker, R. T. K., Feates, F. S., Barber, M. A., Harris, P. S., and Waite,
R. J., J. Catal. 26, 51 (1972).
T. A., Appl. Phys. Lett. 54, 416 (1989).
27. Goodman, D. W., Kelley, R. D., Madey, T. E., and White, J. M., J. Catal.
64, 479 (1980).
28. White, R. L., and Nair, A., Appl. Spectrosc. 44, 69 (1990).
29. Fripiat, J. J., and Uytterhoeven, J., J. Phys. Chem. 66, 800 (1962).
30. Taylor, J. A. G., and Hockey, J. A., J. Phys. Chem. 70, 2169
(1966).
31. Zhuravlev, L. T., Langmuir 3, 316 (1987).
32. Zhuravlev, L. T., Colloids Surf. A: Physiocochem. Eng. Aspects 173, 1
(2000).
24. Rostrup-Nielsen, J. R., and Trimm, D. L., J. Catal. 48, 155 (1977).