Novel Ni catalysts for methane decomposition to hydrogen and carbon nanofibers

Article abstract of DOI:10.1016/j.jcat.2005.12.027

Novel nickel oxides with controlled crystalline size and fibrous shape were successfully prepared by precipitation of nickel acetate and sodium carbonate aqueous solution with the mediation of ethylene glycol. The precipitation media and temperature were found to play crucial roles in determining the crystalline size and shape of the nickel oxides. The NiO prepared in the aqueous solutions exhibited mainly spherical particles with few nanorods, whereas the NiO samples prepared in the solution of ethylene glycol could be of spherical, fibrous, or nanorod shape, depending on the precipitation temperature. In particular, the nickel hydroxide precipitate obtained in the solution of ethylene glycol at 393 K was α-Ni(OH) ₂ with hydrotalcite-like structures; further calcination of this precipitate at 573-1073 K resulted in the formation of nickel oxides with similar fibrous shapes and crystalline size of 3.0-8.1 nm. Similar metallic Ni crystalline size of about 11 nm and fibrous structure were obtained by reduction of these NiO samples with hydrogen at 773 K. When further exposed to the reaction of methane decomposition, the metallic Ni crystallines still exhibited similar evolution behavior and rapidly stabilized at the critical size of 17-18 nm. Consequently, very close catalytic activities for methane decomposition over these novel Ni catalysts were observed with the carbon yields of 354-398 g_C/g_Ni. The Ni crystallines departed from the fibrous structures and gradually aggregated into particles by the dissociation of the deposited carbons. These nickel particles were then highly dispersed on the in situ produced carbon, which can inhibit the further growth of nickel particle. As a result, the initial fibrous structure morphology of the metallic nickel was gradually turned into pear-like shapes.

Full text of DOI:10.1016/j.jcat.2005.12.027

Journal of Catalysis 238 (2006) 412–424
www.elsevier.com/locate/jcat
Novel Ni catalysts for methane decomposition to hydrogen
and carbon nanofibers
Yong Li, Baocai Zhang, Xiaowei Xie, Junlong Liu, Yide Xu, Wenjie Shen ∗
State Key Laboratory of Catalysis, Dalian Institute of Chemical Physics, Chinese Academy of Sciences, Dalian 116023, China
Received 11 November 2005; revised 20 December 2005; accepted 24 December 2005
Available online 27 January 2006
Abstract
Novel nickel oxides with controlled crystalline size and fibrous shape were successfully prepared by precipitation of nickel acetate and sodium
carbonate aqueous solution with the mediation of ethylene glycol. The precipitation media and temperature were found to play crucial roles in
determining the crystalline size and shape of the nickel oxides. The NiO prepared in the aqueous solutions exhibited mainly spherical particles with
few nanorods, whereas the NiO samples prepared in the solution of ethylene glycol could be of spherical, fibrous, or nanorod shape, depending on
the precipitation temperature. In particular, the nickel hydroxide precipitate obtained in the solution of ethylene glycol at 393 K was α-Ni(OH)
2
with hydrotalcite-like structures; further calcination of this precipitate at 573–1073 K resulted in the formation of nickel oxides with similar fibrous
shapes and crystalline size of 3.0–8.1 nm. Similar metallic Ni crystalline size of about 11 nm and fibrous structure were obtained by reduction
of these NiO samples with hydrogen at 773 K. When further exposed to the reaction of methane decomposition, the metallic Ni crystallines
still exhibited similar evolution behavior and rapidly stabilized at the critical size of 17–18 nm. Consequently, very close catalytic activities for
methane decomposition over these novel Ni catalysts were observed with the carbon yields of 354–398 g /g . The Ni crystallines departed
C
Ni
from the fibrous structures and gradually aggregated into particles by the dissociation of the deposited carbons. These nickel particles were then
highly dispersed on the in situ produced carbon, which can inhibit the further growth of nickel particle. As a result, the initial fibrous structure
morphology of the metallic nickel was gradually turned into pear-like shapes.
© 2006 Elsevier Inc. All rights reserved.
Keywords: Ni Catalyst; Ethylene glycol; Fibrous shape; Methane decomposition; Carbon nanofiber
1. Introduction
Because the most important factor influencing carbon yield
during methane decomposition is the particle size of the metal-
lic nickel, much attention has been given to disperse and stabi-
lize nickel particles by selecting the appropriate support [7–12].
Metallic nickel supported on oxides has been the most com-
monly used catalysts for methane decomposition. It was re-
ported that a typical 40%Ni/SiO2 catalyst could give a carbon
yield as high as 491 gC/gNi during methane decomposition at
773 K [11]. Using niobium oxide (Nb2O5) to support 50.8%Ni
and 21.2% of copper as a promoter, the yield of deposited
carbon from methane reached 743 gC/gNi at 873 K [12]. It
was also found that 75%Ni–15%Cu–Al2O3 catalysts prepared
by coprecipitation provided a carbon yield of 700 gC/gNi at
898 K [10]. When taking into account the application of carbon
fibers, however, the catalyst components should be removed or
separated from the deposited carbons [13,14]. Nickel can be
practically and easily removed by leaching with acidic aque-
Catalytic decomposition of methane has recently received
much attention as a potentially economical route for produc-
ing CO-free hydrogen that can be used directly as a fuel for
H2–O2 cells [1–3]. Ni-based catalysts were extensively studied
for methane decomposition due to their relatively higher ac-
tivities compared with other transition metals [3–6]. Because
the catalytic activity of the Ni catalysts decreased with time on
stream due to the continuous deposition of carbons, the effec-
tive catalyst should exhibit sufficient carbon storage capacity on
a per-gram basis of nickel, in addition to reasonable activity to-
ward methane decomposition at relatively lower temperatures.
*
Corresponding author. Fax: +86 411 84694447.
E-mail address: shen98@dicp.ac.cn (W. Shen).
0021-9517/$ – see front matter © 2006 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2005.12.027

Y. Li et al. / Journal of Catalysis 238 (2006) 412–424
413
ous, but the separation of the support materials, like Al2O3
2. Experimental
and SiO2, may be quite burdensome [15]. Thus, various car-
bons were used to support the nickel particles, and a secondary
carbon yield of 268.5 gC/gNi was obtained over a Ni/CFC (cat-
alytic filamentous carbon) catalyst [16]. Obviously, the carbon-
supported nickel catalyst usually gave a much lower carbon
yield compared with that of the Ni particles supported by metal
oxides [15,16].
To improve the overall efficiency of the process, attempts
were made to prepare pure nickel catalysts without a textural
promoter. Unfortunately, the pure Ni catalyst showed rapid de-
activation due to the severe aggregation of nickel particles un-
der reaction conditions and gave a much lower carbon yield
(∼10 gC/gNi). But the addition of 10% Al2O3 or SiO2 caused
a nickel particle size decrease from 65 to 11 nm and carbon
yields of 161–385 gC/gNi [7,9]. Clearly, the textural promoter
is the crucial component in formulating the effective nickel cat-
alysts, and their role is to stabilize the easily aggregated metallic
nickel particles. Perhaps for this reason, there still has been no
successful report on using pure Ni catalyst for methane decom-
position with reasonable activity and acceptable carbon storage
capacity.
Meanwhile, it is also interesting to note that the synthesis
of nickel oxides with controlled particle sizes and morpholo-
gies is becoming one of the hot topics in recent materials
sciences. Thermal decomposition of solid nickel precursors,
like nickel sol–gel complex and nickel hydroxides, were used
to produce nanocrystalline NiO materials [17,18]. These pre-
pared NiO materials are nearly spherical nanoparticles and thus
tend to aggregate into rather larger particles when calcined at
higher temperatures. The reported NiO materials gave small
particle sizes (<10 nm) only after calcination at low temper-
ature (ꢀ673 K), whereas the NiO particles rapidly aggregated
into very large nanoparticles (>20 nm) after high-temperature
calcination (ꢁ973 K). Meanwhile, there have been only few
reports on the morphology-controlled synthesis of NiO materi-
als. Nanorings and nanosheets of nickel oxides have been pre-
pared by thermal calcinations of the corresponding β-Ni(OH)2
[19,20]. But these studies were aimed at preparing nanosized
NiO materials with specific morphological shapes, and little at-
tention was paid to their catalytic applications.
In the present study, pure nickel oxides with controlled crys-
talline sizes and morphologies were successfully prepared by
precipitation of nickel acetate and sodium carbonate aqueous
solution with the mediation of ethylene glycol. The influence
of preparation parameters on the physical properties of the
nickel oxides and the catalytic behaviors for methane decom-
position of the obtained Ni catalysts was extensively inves-
tigated by X-ray diffraction (XRD), temperature-programmed
reduction (TPR), and transmission electron microscopy (TEM)
techniques. The crystalline size and the morphological features
of Ni were correlated with the catalytic activity of methane
decomposition by following the evolution of nickel oxides to
metallic nickel particles during hydrogen reduction and subse-
quent methane decomposition.
2.1. Preparation of NiO catalysts
The nickel oxides were prepared by precipitation of nickel
acetate dissolved in ethylene glycol with sodium carbonate
aqueous solution at 353–433 K. A solution containing 0.05 mol
of nickel acetate (Ni(OAc)2 · 4H2O) and 150 mL of ethylene
glycol (EG) was gradually heated to the desired temperatures
(353–433 K) under stirring and maintained at the same tem-
perature for 30 min. Then 500 mL of 0.2 M Na2CO3 aqueous
solution was slowly added to the Ni-EG solution, with a final
pH value of about 10. The precipitate was aged in the mother
liquid for another 1 h. After being filtered and washed thor-
oughly with distilled water, the nickel hydroxide precipitate was
dried at 373 K overnight and finally calcined in air at 673 K for
4 h. Depending on the precipitation temperature, the obtained
solid was designed as NiO-EG-T , where T refers to the tem-
perature of precipitation (in K).
For comparison, a NiO sample was also prepared by precip-
itation of nickel acetate dissolved in water with sodium carbon-
ate aqueous solution at 353 K in the same procedure described
above. The sample thus obtained was designed as NiO-H2O-
353.
To investigate the effect of calcination temperature on the
physiochemical properties and catalytic features of the NiO ma-
terials, the nickel hydroxide precipitate prepared in the solution
of ethylene glycol at 393 K was further calcined in air at 773–
1073 K for 4 h, resulting in NO samples of varying crystalline
sizes and morphologies.
2.2. Catalyst characterization
The N2 adsorption–desorption isotherms were performed
using a Nova 4200e (Quantachrome) system operated at 77 K.
Before these measurements, the sample was degassed by vac-
uum at 573 K for 5 h. The BET surface area was calculated from
a multipoint BET analysis of the nitrogen adsorption isotherms.
The X-ray powder diffraction (XRD) patterns were recorded
on a D/MAX 2500 X-ray diffractometer (Rigaku), using Cu-
Kα radiation operated at 40 kV and 100 mA. In situ XRD
measurements for the reduction of the NiO samples were per-
formed in a high-temperature chamber installed in the diffrac-
tometer. The NiO was pressed into flakes and mounted in the
chamber. After heating to 773 K in the flow of Ar, pure hy-
drogen was introduced into the chamber and kept at 773 K for
1 h, after which the XRD patterns were recorded. The mean
crystallite sizes of nickel and NiO were calculated from the
Scherrer equation, where the particle shape factor was taken
as 0.9 [21].
The TEM images were obtained with a JEOL JEM-2000EX
operated at 100 kV. The HRTEM images were obtained with
a Philips Tecnai G220 operated at 200 kV. Specimens were
prepared by ultrasonically suspending the sample in ethanol.
A drop of the suspension was then applied onto clean holy
copper grids and dried in air. For TEM observation of the Ni
catalyst, the NiO sample was first reduced with hydrogen in

414
Y. Li et al. / Journal of Catalysis 238 (2006) 412–424
a continuous-flow fixed-bed reactor at 773 K for 1 h. After
being cooled to room temperature under Ar flow, anhydrous
ethanol was introduced through a syringe pump with Ar as a
carrier gas until the reduced sample was completely immersed.
Then it was quickly transferred and stored in the bottle of
ethanol.
Temperature-programmed reduction (TPR) was performed
in a conventional setup equipped with a thermal conductiv-
ity detector. In typical runs, 50 mg NiO samples were used
and heated to 673 K (10 K/min) under N2 flow (40 mL/min)
and kept at this temperature for 0.5 h to remove the adsorbed
carbonates and hydrates. After cooling down to room temper-
ature and introducing the reduction agent of 20 vol% H2/N2
(40 mL/min), the temperature was then programmed to rise
from 303 to 1073 K at a ramp rate of 10 K/min.
Raman spectroscopes of the deposited carbon nanofibers
were obtained over a homemade UV Raman spectrometer.
A 325-nm laser line from Kimmon was chosen as the excita-
tion source. The power of the UV laser line was 4.0 mW. The
wavenumber of Raman spectra was corrected, and the spec-
tra error was 2 cm−1. The value of ID/IG was estimated by
curve-fitting of the Raman spectrum using Gauss function.
Fig. 1. XRD profiles of the nickel oxides.
the smallest crystalline size (4.0 nm) and the largest surface
area (123 m2/g). Fig. 2 gives the corresponding TEM images
of the nickel oxides. Clearly, the morphology was determined
by the preparation conditions as well. Precipitation in water
led mainly to the formation of spherical particles (10–20 nm)
and a few nanorods of NiO. The morphology of the nickel ox-
ides prepared in ethylene glycol solutions was significantly re-
lated to the precipitation temperature. The NiO-EG-353 sample
exhibited aggregated spherical particles, and the NiO-EG-393
sample showed fibrous shapes and interlayered structures with
no spherical particles. The NiO-EG-433 sample consisted of
nanorods (mainly NiO) and spherical particles.
Fig. 3 illustrates the H2-TPR profiles of the nickel oxides.
The figure shows only one hydrogen consumption peak at 500–
800 K, representing the characteristic reduction of stoichiomet-
ric nickel oxide [22]. Although the NiO-H2O-353, NiO-EG-
353, and NiO-EG-433 samples have different crystalline sizes
and particle shapes, they exhibited similar reduction features
with a strong and sharp hydrogen consumption peak at around
600 K. This may indicate that there is little effect of crystalline
size (6–8 nm) and morphology on the reduction behaviors of
these nickel oxides. However, the NiO-EG-393 sample exhib-
ited a relatively broad hydrogen consumption peak at higher
temperature (700 K), suggesting that reduction of the NiO
species is relatively difficult. It was previously stated that the
reduction temperature of NiO increased with increasing crys-
talline size in the range of 6–17 nm [18], but there have been
only a few reports on the effect of morphology on the reduction
property of NiO. For the current NiO-EG-393, the calculated
crystalline size of NiO was only 4.0 nm, and it should be practi-
cally and easily reduced into metallic Ni at lower temperatures.
But its reduction was completed at a higher temperature than
the other samples with larger NiO crystallines. Therefore, it
is reasonable to speculate that the interlayered structure of the
NiO-EG-393 sample hindered the reduction of NiO. In fact, H2-
TPD studies of a Ni/SiO2 catalyst have already revealed that the
morphology can indeed influence the desorption characteristics
of metallic nickel, and the nickel in whisker shapes is more
reactive toward hydrogen than spherical nickel [23,24]. Accord-
2.3. Catalytic activity test
Methane decomposition reaction was carried out in a conti-
nuous-flow fixed-bed quartz reactor at 773 K under atmospheric
pressure. A 40-mg sample of NiO (40–60 mesh) was loaded and
prereduced with pure hydrogen (20 mL/min) at 773 K for 1 h
before the reaction. The Ni catalyst was then flushed with Ar
for 0.5 h. CH4 (99.995% purity, 60 mL/min) was then intro-
duced with a gas hourly space velocity of 90,000 mL/(gcat h)
through a mass flow controller. To eliminate the space lim-
itation for carbon growth, the reactor had an internal diam-
eter of 28 mm and length of 250 mm. The effluents from
the reactor were analyzed by on-line gas chromatography.
Over all of the catalysts, hydrogen was detected as the only
gaseous product, indicating that methane decomposition oc-
curs selectively as CH4 → 2H2 + C. Accordingly, methane
conversion was calculated from the amounts of hydrogen pro-
duced.
3. Results and discussion
3.1. Effect of preparation parameters
Fig. 1 shows the XRD patterns of the nickel oxides obtained
by calcination of the nickel hydroxide precipitates at 673 K,
which were prepared in water and ethylene glycol. Clearly,
all of the samples showed a typical face-center cubic struc-
ture of NiO (JCPDS#4-835). The NiO-H2O-353 had the largest
crystalline size (8.4 nm) and a surface area of 78 m2/g. The
crystalline sizes and the surface areas of the samples prepared
in ethylene glycol were strongly dependent on the precipita-
tion temperature. Precipitation at 353 and 433 K resulted in
similar NiO crystalline sizes of 6.8–7.1 nm and close surface
areas of 78–93 m2/g, whereas the NiO-EG-393 sample gave

Y. Li et al. / Journal of Catalysis 238 (2006) 412–424
415
Fig. 2. TEM images of the nickel oxides: (A) NiO-H O-353, (B) NiO-EG-353, (C) NiO-EG-393, (D) NiO-EG-433.
2
position. Fig. 4 depicts the methane conversions as a function
of time-on-stream, and the accumulated carbon decomposition
and hydrogen yield were accordingly calculated. It is surpris-
ing that the NiO-H2O-353 did not show any catalytic activ-
ity toward methane decomposition, whereas the NiO samples
prepared in the solution of ethylene glycol showed differing
catalytic activities, depending on the precipitation temperature.
The NiO-EG-353 sample exhibited pronounced catalytic activ-
ity for methane decomposition with 60 h of time on stream, with
accumulated carbon and hydrogen yields of 271 gC/gNi and
2651 molH /molNi, respectively. The NiO-EG-393 sample gave
2
the highest methane conversion with rather prolonged catalytic
activity for 75 h, and the accumulated carbon and hydrogen
yields approached 398 gC/gNi and 3894 molH /molNi. How-
2
ever, the NiO-EG-433 sample gave only marginal methane con-
versions for about 5 h, and the accumulated carbon and hydro-
Fig. 3. TPR profiles of the nickel oxides.
gen yields were calculated as 3.4 gC/gNi and 33 molH /molNi.
2
ingly, the interlayered structure of fibrous NiO may be rather
stable under hydrogen atmosphere compared with the spheri-
cal particles and thus may lead to significant differences in the
reduction features compared with other samples with spherical
NiO particles.
Thus, it seems true that both the precipitation media and the
precipitation temperature play crucial roles in determining the
catalytic properties of the Ni catalysts, probably through ef-
fective control of the crystalline and/or particle size as well
as the morphology of the nickel hydroxide precipitates and
the nickel oxides. In particular, precipitation in ethylene gly-
col at 393 K led to the formation of stable α-Ni(OH)2 with
hydrotalcite-like structure, and further calcination of this nickel
hydroxide resulted in the formation of NiO-EG-393 with fi-
3.2. Methane decomposition
The NiO samples were first reduced into metallic nickel
with hydrogen at 773 K, and then tested for methane decom-

416
Y. Li et al. / Journal of Catalysis 238 (2006) 412–424
Fig. 5. XRD patterns of the Ni catalysts obtained by hydrogen reduction of the
NiO precursors at 773 K.
Fig. 4. Methane conversion as a function of time-on-stream over the Ni cata-
lysts originated by hydrogen reduction of the nickel oxides at 773 K. (Reac-
tion temperature, 773 K; catalyst, 0.040 g; methane, 101 kPa; and flow rate,
60 mL/min.)
atively lower carbon yield. Increasing the Ni crystalline size
to about 24 nm led to extremely low catalytic activity, and
further increasing the size up to 26 nm resulted in total in-
activity toward methane decomposition. Thus, it appears that
the metallic nickel with crystalline size of 10–20 nm is suit-
able for methane decomposition. The methane decomposition
reaction depends critically on the crystalline size of metallic
nickel, with dramatically decreased catalytic activity with in-
creasing crystalline size. This observation is consistent with
previous reports concerning the effect of Ni crystalline size
on methane decomposition over various supported nickel cat-
alysts. Investigation of methane decomposition over a series
of 90%Ni/SiO2 catalysts revealed that nickel crystalline sizes
of 10–40 nm gave similar carbon yields of 300–384 gC/gNi,
whereas further increasing the nickel crystalline size to 60 nm
drastically decreased the carbon yield to 2 gC/gNi [9]. Sim-
ilarly, 96%Ni/SiO2 catalysts with nickel crystalline sizes of
20 nm showed the highest carbon yield of 300 gC/gNi, and
the carbon yields greatly decreased (to 222 and 107 gC/gNi)
when Ni crystalline sizes were slightly increased (to 25 and
38 nm, respectively). For lower nickel loading systems, studies
over a 15%Ni/CFC catalyst also concluded that Ni crystalline
sizes of 30–70 nm were suitable for the growth of carbon
filaments during methane decomposition [8,25]. The most ef-
fective crystalline size of Ni for the growth of filamentous
carbon was found to be 10–60 nm over a 23.3%Ni/SiO2 cat-
alyst [26]. It was further observed that the optimal growth
rate of carbon nanofibers could be achieved over Ni crystals
around 34 nm in size if supported by hydrotalcite materi-
als [27].
brous shapes. Unsupported Ni catalysts have been prepared
by conventional precipitation method in aqueous solution and
also tested for methane decomposition in previous studies
[7,9]. But these pure Ni catalysts only gave very low carbon
yields of about 10 gC/gNi, similar to case of the NiO-EG-433.
When small amounts of Al2O3 or SiO2 were used to sup-
port the nickel oxides, considerable increases in carbon yields
(161–384 gC/gNi) were observed. Notably, the carbon yield
of 398 gC/gNi achieved over the NiO-EG-393 was slightly
higher than the most promising value of 384 gC/gNi over a
90%Ni/SiO2 catalyst reported to date [9]. Thus, the Ni cata-
lyst prepared by precipitation of nickel acetate at 393 K with
the mediation of ethylene glycol could be an excellent candi-
date for methane decomposition.
3.3. Particle size and morphology of the reduced metallic
nickel
From the reaction results of methane decomposition, it
seems that the initial particle size and morphology of nickel
oxide precursors had important effects on the catalytic per-
formance of the reduced Ni catalysts. Because the metallic
nickel generated from hydrogen reduction of NiO is the ac-
tive component for methane decomposition, it is necessary
to establish the relationship between the particle or crys-
talline size of the metallic nickel and the methane conver-
sion. Fig. 5 shows the XRD patterns of the Ni catalysts ob-
tained by reduction of NiO samples with hydrogen at 773 K.
It can be seen that NiO was completely reduced to metallic
nickel (JCPDS#4-850). Clearly, the crystalline sizes of metal-
lic nickel were greatly increased (by about threefold) com-
pared with their corresponding NiO precursors (Fig. 1). If the
nickel crystalline size is correlated to the catalytic behavior
of methane decomposition, then it can be readily seen that
the metallic Ni with crystalline size of about 10.8 nm had
the highest carbon and hydrogen yields and that the metal-
lic Ni with crystalline size of about 20 nm produced a rel-
Although it is obvious that smaller Ni crystalline sizes show
higher catalytic activity for methane decomposition, there is
a significant discrepancy among the reported optimal metal-
lic nickel crystalline sizes for methane decomposition, which
strongly depends on the support and the Ni loading. For the cur-
rent unsupported Ni catalysts, metallic nickel with crystalline
size of 10–20 nm was found to be suitable for methane decom-
position, but increasing the crystalline size to 26 nm resulted
in total inactivity. This extreme sensitivity of metallic nickel

Y. Li et al. / Journal of Catalysis 238 (2006) 412–424
417
Fig. 6. TEM images of the Ni catalysts obtained by hydrogen reduction of (A) NiO-H O-353, (B) NiO-EG-353, (C) NiO-EG-393, and (D) NiO-EG-433 at 773 K.
2
crystalline size up to 26 nm implies that other factors may be
governing the reaction of methane decomposition, such as the
morphology of the reduced Ni and the structural evolution un-
der reaction conditions.
hydrazine and showed that this new morphology of metallic
nickel was more active for benzene hydrogenation and more
selective in acetylene hydrogenation to benzene [23,24]. How-
ever, this nonclassical nickel phase can be obtained only for
very low nickel loading (<2%), and the silica support must be
Fig. 6 provides TEM images of the reduced Ni catalysts. It
can be seen that the NiO samples with some spherical particles
changed to regular spherical shapes after reduction with hydro-
gen at 773 K. The NiO-H2O-353 with spherical NiO particles
(10–20 nm) and few nanorods exhibited 100–200 nm metallic
Ni particles. The NiO-EG-353 with aggregated spherical NiO
particles showed a change to large Ni particles (30–100 nm).
The NiO-EG-433 containing nanorods and spherical particles
of NiO were all reduced to spherical Ni particles (50–100 nm).
This may suggest that the NiO species with nonspherical shapes
in these samples were not stable under hydrogen reduction con-
ditions and finally aggregated into large Ni spherical particles.
However, it is quite interesting to note that even though hydro-
gen reduction at 773 K of the fibrous shape of the NiO-EG-
393 sample resulted in the formation of fibrous metallic Ni, no
spherical particles were observed. The finding of rather stable
fibrous shapes in the NiO-EG-393 sample is very promising for
NiO and metallic nickel nanomaterials, because the previously
reported nickel oxide and metallic nickel over supporting ma-
terials for methane decomposition were all of spherical shape
[9,17]. Boudjahem et al. successfully prepared silica-supported
whisker-like metallic nickel by reduction of nickel acetate with
of very low surface area (15 m2
/g). To date, there have been
no reports on preparing pure NiO and metallic Ni with the fi-
brous structures as observed in the NiO-EG-393 sample, and
the metallic nickel with such a unique morphology exhibited
much higher catalytic activity for methane decomposition than
the conventional spherical metallic nickel.
3.4. Evolution of nickel particles during CH4 decomposition
It was recently proposed that the nickel catalysts might be
considered as self-organizing systems during the reaction of
methane decomposition [15,28]. Nickel particles underwent ac-
tive evolution and gradually stabilized at thermodynamically
favorable sizes. Therefore, the NiO samples were reduced into
metallic Ni catalysts with hydrogen at 773 K and reacted with
methane for desired periods. XRD measurements and TEM ob-
servations were then conducted to follow the evolution of the
nickel particles. Fig. 7 shows the XRD patterns of the freshly
reduced Ni and the completely deactivated samples. Because
the typical diffraction peak of Ni (111) was overlapped with the
graphite (101) plane, the crystalline sizes were calculated ac-

418
Y. Li et al. / Journal of Catalysis 238 (2006) 412–424
cording to the half-width of the diffraction peak of the Ni (200)
plane. It can be seen that the crystalline size of nickel increased
noticeably during the first 30 min, and no further pronounced
increase was observed in the following reaction course. The
nickel crystalline size increased slightly (from 20 to 23 nm)
over the NiO-EG-353 sample, whereas the nickel crystalline
size increased significantly (from 10.8 to 17–18 nm) over the
most effective NiO-EG-393 sample. The change in metallic
nickel crystalline size at the initial stage of methane decom-
position implied that the deposited carbon would induce aggre-
gation of the fine Ni crystallines into larger sizes. But the nickel
crystalline size increased only marginally (from 24 to 26 nm)
in the NiO-EG-433 sample, which showed the lowest catalytic
activity. This critical Ni crystalline size of 26 nm is in good
agreement with the findings for the NiO-H2O-353 sample in
which the reduced metallic nickel crystalline size was 26 nm
and showed no activity for methane decomposition. Thus, it
can be speculated that nickel crystals larger than 26 nm are not
active for methane decomposition. This can be understood by
considering the dissociation of methane on Ni crystals. Gen-
erally, methane is adsorbed and decomposed on a certain face
of Ni, followed by the diffusion and deposition of carbon atoms
on the surface of nickel crystals. The catalytic life is deter-
mined by the ratio of the formation rate of carbon atoms on
the Ni metal surfaces to the diffusion rate approaching the de-
posit site. Thus, there should be an optimal size of active Ni
crystals that could achieve this dynamic equilibrium for longer
periods [7,9,11]. Avdeeva et al. studied the evolution of nickel
particles over a 75%Ni/Al2O3 catalyst during methane decom-
position and found that the nickel crystalline size increased
from the initial 9 to 16.5, 22.5, and 40 nm after reaction with
methane for 20, 60, and 180 min, respectively [7]. Accordingly,
these authors concluded that the formation of large Ni crys-
tals during the reaction is likely caused by the dissolution of
the carbon into the metal. Based on the XRD measurements
of a series of 90%Ni/SiO2 catalysts, Ermakova et al. stated
that the nickel crystalline sizes merged from 15 to 30 nm or
dispersed from 60 to 30 nm during the reaction of methane
decomposition and that the active nickel crystalline sizes sta-
bilized at 30–40 nm [28]. Reshetenko et al. reported that the
Ni crystalline size increases from the initial 13 to 27 nm over a
75%Ni-15Cu%-Al2O3 catalyst after reaction with methane for
10–20 min at 898 K [10]. Studies on the relationship between
nickel crystalline size and the catalytic activity of methane de-
composition have suggested that the deposited carbons over Ni
catalysts supported by carbon fibers modified the size of the
Ni crystallites during the reaction, and that nickel crystallites
about 20 nm in size were the most active sites for the growth
of carbon nanofibers [15]. From this standpoint, the observed
optimal Ni crystalline size of 17–23 nm for the current NiO
catalysts is in good accordance with previous conclusions con-
cerning Ni crystalline evolution under methane decomposition
conditions.
Fig. 8 further compares the TEM images of the deposited
carbons on the Ni catalysts after complete deactivation dur-
ing methane decomposition. Significant difference in the mor-
phologies of the deposited carbon can be clearly seen. For the
NiO-EG-353 and NiO-EG-393 samples, which were active for
methane decomposition, the carbons were all in nanofiber form.
The nickel particles (darker spots) were all located at the tips of
carbon nanofibers, and the diameters of the Ni particles were
almost the same as those of the growing nanofibers. The NiO-
EG-433 K sample, which was rapidly deactivated within 5 h
Fig. 7. XRD profiles of the Ni catalysts after exposure to methane decomposi-
tion: (A) NiO-EG-353, (B) NiO-EG-393, (C) NiO-EG-433.

Y. Li et al. / Journal of Catalysis 238 (2006) 412–424
419
Fig. 8. TEM images of the completely deactivated Ni catalysts: (A) NiO-EG-353, (B) NiO-EG-393, (C) NiO-EG-433, (D1)–(D2) enlargement part of NiO-EG-433.
time on stream, exhibited three types of Ni particles after com-
plete deactivation. The smaller Ni particles (ca. 30 nm) with
elongated shapes tended to form carbon nanofibers, the rela-
tively large Ni particles (ca. 100 nm) were encapsulated by
graphitic carbon that greatly inhibited the contact of Ni par-
ticles with methane, and the largest Ni particles (>100 nm)
produced no carbon around them. These findings confirm that
the particle size of metallic nickel also plays an important role
in the process of methane decomposition. It also provides con-
vincing evidence that Ni particles >100 nm cannot perform
methane decomposition, thus explaining the total inactivity of
the NiO-H2O-353 catalyst (because the Ni particles after hy-
drogen reduction were already >100 nm). This TEM obser-
vation is also similar to the previous results. Takenaka et al.
reported that 60–100 nm Ni particles estimated from SEM im-
ages exhibited the longest catalytic life for methane decompo-
sition over a 40%Ni/SiO2 catalyst [11]. Shaikhutdinov et al.
demonstrated, using TEM, that 30–70 nm Ni particles were
more active for the growth of carbon filaments than other parti-
cles [8].
The particle sizes measured by TEM or SEM observations
and the crystalline sizes obtained from XRD profiles should
be clearly distinguished. The former usually gives individual
particle sizes that are the aggregation of various Ni crystals,
whereas the latter can give only the average crystalline size of
nickel. Comparing the two approaches in terms of Ni crystalline
size, previous reports have focused on supported Ni catalysts
in which the Ni particles were all of spherical shape and the
proposed optimal Ni crystalline sizes obtained from XRD mea-
surements were mainly in the range of 20–40 nm. In the present
work, pure Ni crystalline of particle size <23 nm showed quite
pronounced catalytic activity for methane decomposition, and
a slight increase in the nickel crystalline size to 26 nm resulted
in total deactivation for methane decomposition. This relatively
narrower Ni crystalline size range may be due to the lack of
support. Besides stabilizing the Ni particles, the support may
also provide deposition sites for the growth of the carbon dif-
fused from Ni [29,30]. The absence of support in the pure Ni
catalysts requires that the nickel crystalline have high activity
toward methane cracking and also sufficient ability to facilitate
the immediate growth of carbon filaments. Such a balance be-
tween the formation rate of carbon atoms on the Ni surfaces and
the diffusion rate approaching the deposit site make the reaction
more sensitive over unsupported Ni catalysts.
3.5. Optimal Ni particle size
There are significant discrepancies among the reported
nickel crystalline and/or particle sizes suitable for methane de-
composition, usually based on XRD and TEM observations.
Considering the particle sizes observed by TEM images,
spherical Ni particles >100 nm could not produce any carbon;

420
Y. Li et al. / Journal of Catalysis 238 (2006) 412–424
Fig. 9. HRTEM images of the carbon nanofibers formed by methane decomposition over the Ni catalyst obtained by hydrogen reduction of NiO-EG-393 at 773 K:
(A) 30 min, (B) 60 min, (C) 90 min, and (D) carbon nanofiber.
only the small Ni particles produced significant amounts of car-
bon nanofibers. As a result, the Ni-H2O-353 and Ni-EG-433
samples containing mainly Ni particles >100 nm did not show
promising catalytic activities. When the reduced Ni is presented
as nonspherical shapes, the comparison in terms of particle size
becomes controversial. In the fibrous NiO-EG-393 sample, fi-
brous Ni was observed with a crystalline size of 10.8 nm, and
this highly active nickel further grew into larger crystals of
17–18 nm during the reaction. The formation of Ni particles
during the reaction should be related to this fibrous structure.
As shown in Fig. 9, the deposited carbon grew with a fila-
mentous structure. The nickel particles (darker spots) were all
located at the tips of carbon nanofibers, and the diameters of
the Ni particles were almost the same as those of the growing
nanofibers. Initially, the 17–18 nm Ni crystalline effectively ig-
nited the decomposition of methane and induced the growth of
carbon nanofibers around it. As the reaction proceeded, the Ni
Fig. 10. Raman spectra of the carbons deposited by the methane deposition over
crystallines departed from the fibrous structures and gradually
aggregated into particles over the deposited carbons, which can
be considered the in situ produced support. The nickel parti-
cles were then highly dispersed on the carbon nanofibers, and
simultaneously the initial fibrous structure morphology grad-
ually changed to pear-like shapes. Thus, it can be concluded
that the deposited carbon changed not only the sizes, but also
the shapes of the nickel particles during methane decomposi-
the Ni catalyst originated from NiO-EG-393.
tion. Together with the evolution of Ni particles, the structure
of the deposited carbon nanofibers also changed with time on
stream.
Fig. 10 shows the Raman spectra of carbons during methane
decomposition over the Ni catalyst obtained by hydrogen
reduction of the NiO-EG-393 sample. Two distinct bands

Y. Li et al. / Journal of Catalysis 238 (2006) 412–424
421
Fig. 11. HRTEM images of the deposited carbon over the Ni catalyst originated from NiO-EG-393. The sample was taken after 30 min reaction of methane
decomposition.
were detected at about 1400 cm−1 (D band) and 1580 cm−1
(G band). The former is usually ascribed to the structural im-
perfection of graphite, and the latter is attributed to the in-plane
carbon–carbon stretching vibrations of the graphite layers [11,
31,32]. The shoulder peak at 1620 cm−1 (Dꢀ band) was assigned
to the imperfect graphite or disordered carbons [33,34]. The ra-
tio of the area of the D band to that of the G band (ID/IG) was
considered an index for the crystalline order of graphite [11,
34–37]; that is, the graphitization degree of carbons is higher
at lower ID/IG values. The ID/IG value calculated by curve-
fitting of the Raman spectra in Fig. 10 increased from 0.44
to 0.82, indicating that the crystalline order of the deposited
carbon was decreasing with the continuous growth of carbon
nanofibers. Fig. 11 shows high-resolution TEM (HRTEM) im-
ages of deposited carbon nanofibers on the nickel catalysts at
the early stage of methane decomposition. Obviously, the de-
posited carbon was pure nanofibers, and no other forms of
carbon could be observed. The graphene layers were canted
with respect to the longitudinal axis of the fibers, and Ni metal
particles of pear-like shape presented at the tips of the carbon
nanofibers.
interesting structural and catalytic features of the Ni catalysts.
The crystalline size of NiO after calcination at 773 K was cal-
culated as about 4.5 nm, and it increased only slightly to 8.1 nm
even after calcination at 1073 K. Fig. 12 shows the correspond-
ing HRTEM images of the NiO samples. It can be seen that the
interlayer structure with fibrous shapes of NiO remained in all
of the samples, even though the bundles were becoming shorter
with increasing calcination temperature. Fig. 13 demonstrates
the effect of calcination temperature on the catalytic activities
of the Ni catalysts for methane decomposition. Interestingly,
no significant changes in the methane conversions over these
Ni catalysts were observed, even though the crystalline sizes of
their NiO precursors were found to increase with calcination
temperature. When the crystalline size of NiO increased from
4.5 to 8.1 nm, the carbon yield varied only slightly (between
381 gC/gNi and 394 gC/gNi), and the corresponding hydrogen
yield also varied only slightly (between 3727 molH /molNi and
2
3854 molH /molNi).
2
The NiO samples obtained by calcining the nickel hydrox-
ide precursor at 673 and 1073 K were selected to investigate the
evolution of NiO and metallic Ni particles during the hydrogen
reduction and the subsequent reaction of methane decompo-
sition. Fig. 14 shows the XRD patterns measured during the
reduction of nickel oxides with hydrogen. Clearly, the two cat-
alysts showed similar reduction behaviors with increasing re-
duction temperature. The metallic nickel phase appeared after
reduction at 573 K, and the diffractions of NiO almost van-
ished at 673 K. With further increases in reduction temperature,
3.6. Effect of calcination temperature on the NiO structure
Because the NiO-EG-393 sample with fibrous shape and
rather small crystalline size exhibited excellent catalytic activ-
ity for methane decomposition, calcination of the nickel hy-
droxyl precursor was performed at 773–1073 K to reveal the

422
Y. Li et al. / Journal of Catalysis 238 (2006) 412–424
Fig. 12. HRTEM images of the NiO samples obtained by calcining the nickel hydroxide precipitate at 673–1073 K: (A) 673 K, (B) 773 K, (C) 973 K, (D) 1073 K.
the diffractions of metallic Ni gradually became extensive and
sharper, indicating the enlarging metallic Ni crystallines. Al-
though the initial crystalline sizes of the NiO samples were 4.0
and 8.1 nm, the resulting metallic Ni crystalline sizes were al-
most the same (about 9–11 nm) after reduction with hydrogen
at 773 K. Fig. 15 shows the XRD profiles of the Ni crystalline
during methane decomposition over the Ni catalyst generated
from NiO by calcination of the precipitate at 1073 K. Simi-
lar to the case of the Ni crystalline size over the NiO-EG-393
sample (Fig. 7B), the fresh Ni metallic crystalline sizes in both
catalysts were estimated to be ca. 10 nm, increasing signifi-
cantly (to 17 nm) during the first 30 min under methane de-
composition. Thereafter, no further changes could be observed,
and the metallic Ni crystallines were stabilized at 17–18 nm.
Clearly, the Ni catalysts prepared in the ethylene glycol solu-
tion at 393 K had different crystalline sizes after calcination
at 673–1073 K, but similar metallic Ni crystalline sizes were
obtained after reduction at 773 K with hydrogen. When fur-
ther exposed to the reaction media, the metallic Ni crystallines
still exhibited similar evolution behavior and rapidly stabilized
at the critical size of 17–18 nm within 30 min. This phenom-
enon can explain the similar catalytic activities observed for
methane decomposition over the NiO catalysts prepared in the
solution of ethylene glycol at 393 K and calcined between 673
and 1073 K.
Fig. 13. Methane conversion as a function of time-on-stream over the Ni cata-
lysts. NiO precursors were prepared by calcining the nickel hydroxide precip-
itate at 773–1073 K. (Reaction temperature, 773 K; NiO, 0.040 g; methane,
101 kPa; flow rate, 60 mL/min.)
4. Conclusion
Nickel oxides with controlled crystalline sizes and fibrous
shapes were successfully prepared by the precipitation of nickel
acetate and sodium carbonate aqueous solution with the me-

Y. Li et al. / Journal of Catalysis 238 (2006) 412–424
423
Fig. 15. XRD patterns of the NiO catalysts obtained by calcining the nickel
hydroxide precipitate at 1073 K after exposure to methane decomposition.
decomposition reaction suggested that the deposited carbon
changed not only the crystalline size, but also the morphology
of the metallic nickel. The Ni crystallines departed from the fi-
brous structures and gradually changed to particles through the
dissociation of the deposited carbons. The nickel particles were
then highly dispersed on the in situ produced carbon, inhibit-
ing the further growth of nickel particle. As a result, the initial
fibrous structure morphology of the metallic nickel gradually
changed to a pear-like shape.
References
[1] T. Zhang, M.D. Amiridis, Appl. Catal. A 167 (1998) 161.
[2] T.V. Choudhary, C. Sivadinarayana, C.C. Chusuei, A. Klinghoffer, D.W.
Goodman, J. Catal. 199 (2001) 9.
[3] R.A. Couttenye, M.H. De Vila, S.L. Suib, J. Catal. 233 (2005) 317.
[4] M.A. Ermakova, D.Yu. Ermakov, G.G. Kuvshinov, Appl. Catal. A 201
(2000) 61.
Fig. 14. XRD patterns of the nickel oxides during hydrogen reduction process.
NiO was obtained by calcination of the nickel hydroxide at: (A) 673 K and
(B) 1073 K.
diation of ethylene glycol. The precipitation temperature was
found to play crucial role in determining the crystalline size and
shape of the NiO samples. Precipitation at 393 K resulted in the
formation of α-Ni(OH)2 with a hydrotalcite-like structure, and
further calcination of this precipitate precursor at 673–1073 K
led to the formation of nickel oxides with layered structures.
The crystalline size of the NiO after calcination at 673 K was
4.0 nm and increased only slightly to 8.1 nm even after cal-
cination at 1073 K. On reduction with hydrogen at 773 K,
the crystalline sizes of the obtained Ni catalysts were almost
the same (9.0–11.0 nm). More importantly, the reduced nickel
maintained the fibrous shapes of the initial NiO precursors.
The catalytic activity of the pure Ni catalysts for methane
decomposition was strongly related to the crystalline size of
the reduced Ni. The Ni catalyst with crystalline size of about
10.8 nm had the highest carbon and hydrogen yields, and the
nickel crystalline of about 20 nm gave relatively lower carbon
yields. Further increasing the nickel crystalline size to about
24 nm led to extremely low catalytic activity, and an increase up
to 26 nm resulted in total deactivation toward methane decom-
position. Further evolution of the Ni particles during methane
[5] S. Takenaka, H. Ogihara, I. Yamanaka, K. Otsuka, Appl. Catal. A 217
(2001) 101.
[6] M.A. Ermakova, D.Yu. Ermakov, Catal. Today 77 (2002) 225.
[7] L.B. Avdeeva, O.V. Goncharova, D.I. Kochubey, V.I. Zaikovskii, L.M.
Plyasova, B.N. Novgorodov, Sh.K. Shaikhutdinov, Appl. Catal. A 141
(1996) 117.
[8] Sh.K. Shaikhutdinov, L.B. Avdeeva, B.N. Novgorodov, V.I. Zaikovskii,
D.I. Kochubey, Catal. Lett. 47 (1997) 35.
[9] M.A. Ermakova, D.Yu. Ermakov, G.C. Kuvshinov, L.M. Plyasova, J. Ca-
tal. 187 (1999) 77.
[10] T.V. Reshetenko, L.B. Avdeeva, Z.R. Ismagilov, A.L. Chuvilin, V.A.
Ushakov, Appl. Catal. A 247 (2003) 51.
[11] S. Takenaka, S. Kobayashi, H. Ogihara, K. Otsuka, J. Catal. 217 (2003)
79.
[12] J. Li, G. Lu, K. Li, W. Wang, J. Mol. Catal. A 221 (2004) 105.
[13] A. Chambers, T. Nemes, N.M. Rodriguez, R.T.K. Baker, J. Phys. Chem.
B 102 (1998) 2251.
[14] K.P. de Jong, J.W. Geus, Catal. Rev.-Sci. Eng. 42 (2000) 481.
[15] K. Otsuka , H. Ogihara, S. Takenaka, Carbon 41 (2003) 223.
[16] T.V. Reshetenko, L.B. Avdeeva, Z.R. Ismagilov, A.L. Chuvilin, V.B. Fene-
lonov, Catal. Today 102–103 (2005) 115.
[17] L.C. Corrie, S. Jennifer, J.K. Kenneth, Langmuir 18 (2002) 1352.
[18] C. Wang, G. Gau, S. Gau, C. Tang, J. Bi, Catal. Lett. 101 (2005).
[19] J. Liang, Y. Li, Chem. Lett. 32 (2003) 1126.
[20] Z. Liang, Y. Zhu, X. Hu, J. Phys. Chem. B 108 (2004) 3488.

424
Y. Li et al. / Journal of Catalysis 238 (2006) 412–424
[21] B.D. Cullity, Elements of X-Ray Diffraction, second ed., Addison–
Wesley, Menlo Park, CA, 1978.
[28] M.A. Ermakova, D.Yu. Ermakov, L.M. Plyasova, G.C. Kuvshinov, Catal.
Lett. 62 (1999) 3.
[22] S.D. Robertson, B.D. Mcnicol, J.H. de Bass, S.C. Kloet, J.W. Jenkins,
J. Catal. 37 (1975) 424.
[23] A.G. Boudjahem, S. Monteverdi, M. Mercy, M.M. Bettahar, J. Catal. 221
(2004) 325.
[24] A.G. Boudjahem, S. Monteverdi, M. Mercy, M.M. Bettahar, Catal.
Lett. 97 (2004) 177.
[29] R.T.K. Baker, Carbon 27 (1989) 315.
[30] J.W. Snoeck, G.F. Froment, M. Fowles, J. Catal. 169 (1997) 240.
[31] K. Sinha, J. Menendez, Phys. Rev. B 41 (1990) 10845.
[32] R.J. Nemanich, S.A. Solin, Phys. Rev. B 20 (1979) 392.
[33] H. Darmstadt, L. Sümmchen, J.M. Ting, U. Roland, S. Kaliaguine, C. Roy,
Carbon 35 (1997) 1581.
[25] T.V. Reshetenko, L.B. Avdeeva, Z.R. Ismagilov, A.L. Chuvilin, Carbon 42
(2004) 143.
[26] P. Wang, E. Tanabe, K. Ito, J. Jia, H. Morioka, T. Shishido, K. Takehira,
Appl. Catal. A 231 (2002) 35.
[27] D. Chen, K.O. Christensen, E. Ochoa-Fernández, Z. Yu, B. Tøtdal, N. La-
torre, A. Monzón, A. Holmen, J. Catal. 229 (2005) 82.
[34] T. Jawhari, A. Roid, J. Casado, Carbon 33 (1995) 1561.
[35] A. Cuesta, P. Dhamelincourt, J. Laureyns, A. Martinez-Alonso, J.M.D.
Tascon, Carbon 32 (1994) 1523.
[36] J.O. Herrera, D.E. Resasco, Chem. Phys. Lett. 376 (2003) 302.
[37] W.E. Alvarez, F. Pompeo, J.E. Herrera, L. Balzano, D.E. Resasco, Chem.
Mater. 14 (2002) 1853.