Catalytic properties of Ni-Fe-Mg alloy nanoparticle catalysts for methanol decomposition

Article abstract of DOI:10.1016/j.cattod.2016.10.036

Ni-25Fe and Ni-15Fe-13Mg (at.%) alloy nanoparticles were fabricated by a thermal plasma method, and their catalytic properties for methanol decomposition were examined in the 513–793?K temperature range. Both nanoparticles started showing activity for methanol decomposition above 593?K, with activity rapidly increasing with temperature. Ni-15Fe-13Mg showed much higher activity and less carbon deposition than Ni-25Fe. These results indicated that the addition of Mg in the Ni-Fe alloy significantly increased the activity, but suppressed carbon deposition. Characterization of the nanoparticles revealed that the Mg was segregated and a thin layer of MgO was formed on the surface of the nanoparticles. It is likely that the formation of MgO suppressed carbon deposition, but enhanced the activity by delaying the coarsening of nanoparticles during the reaction.

Full text of DOI:10.1016/j.cattod.2016.10.036

Catalysis Today 281 (2017) 669–676
Contents lists available at ScienceDirect
Catalysis Today
journal homepage: www.elsevier.com/locate/cattod
Catalytic properties of Ni-Fe-Mg alloy nanoparticle catalysts for
methanol decomposition
Haruka Mitania,b, Ya Xu , Toshiyuki Hirano , Masahiko Demura , Ryuji Tamura
b,∗
b
b
a
a
Tokyo University of Science, 6-3-1 Nijuku, Katsusika-ku, 125-8585, Japan
National Institute for Materials Science,1-2-1 Sengen, Tsukuba, 305-0047, Japan
b
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 16 June 2016
Received in revised form 23 October 2016
Accepted 27 October 2016
Available online 11 November 2016
Ni-25Fe and Ni-15Fe-13Mg (at.%) alloy nanoparticles were fabricated by a thermal plasma method, and
their catalytic properties for methanol decomposition were examined in the 513–793 K temperature
range. Both nanoparticles started showing activity for methanol decomposition above 593 K, with activ-
ity rapidly increasing with temperature. Ni-15Fe-13Mg showed much higher activity and less carbon
deposition than Ni-25Fe. These results indicated that the addition of Mg in the Ni-Fe alloy significantly
increased the activity, but suppressed carbon deposition. Characterization of the nanoparticles revealed
that the Mg was segregated and a thin layer of MgO was formed on the surface of the nanoparticles. It
is likely that the formation of MgO suppressed carbon deposition, but enhanced the activity by delaying
the coarsening of nanoparticles during the reaction.
Keywords:
Ni-Fe-Mg alloy nanoparticles
Ni3Fe
Thermal plasma method
Methanol decomposition
Hydrogen production
© 2016 Elsevier B.V. All rights reserved.
1
. Introduction
Intermetallic compounds (IMCs) are expected to be promis-
Needless to say, it is important to develop nanosized IMC
particles in order to obtain better catalytic properties. Thermal
plasma-assisted processes have been widely used for the fabrica-
tion of nanoparticles [9,10]. In contrast to ordinary wet chemical
processes, metals are evaporated in the thermal plasma processes
above 10,000 K, followed by rapid condensation of the gas phases
[11]. Therefore, they are suitable for fabricating well-crystallized
nanoparticles of alloy and intermetallic phases. We have success-
fully fabricated Ni3Al nanoparticles using arc plasma evaporation
[12,13]. In this study, we fabricate Ni3Fe (Ni-25 at.% Fe) and Ni-
Fe-Mg alloy nanoparticles using a thermal plasma method, and
examine their catalytic properties in methanol decomposition.
Carbon deposition is a significant problem for Ni-based catalysts
in hydrogen production from hydrocarbons. In fact, a large amount
of carbon fiber was produced on the Ni3Fe foil catalysts during
methanol decomposition in our previous study [8]. It is known
that basic oxides, such as MgO, suppress carbon deposition effec-
tively. Penkova et al. carried out methanol steamreforming tests
ing candidates because many IMCs exhibit unique physical and
mechanical properties due to their ordered crystalline structure,
such as a shape memory effect and superconductivity [1,2]. The
catalytic properties of some IMCs, such as Pt3Ti and Pt3Ge, have
been reported [3,4]. Recently we found that Ni-based IMCs, such as
Ni3Al, Ni3(Sn,Ti), and Ni3Fe showed good catalytic activity in the
form of foils for methanol decomposition [5–8]. The mechanism of
the activity appeared to be related to IMC composition. The activity
of Ni3Al and Ni3(Si,Ti) was due to the formation of fine Ni particles
through the selective oxidation of Al, Si, and Ti during the reaction.
In other words, Ni3Al and Ni3(Si,Ti) serve as a precursor of for-
mation of active Ni particles. In contrast, the activity of the Ni3Fe
foil was attributed to the formation of Ni3Fe fine particles on the
foil surface during the reaction and the Ni3Fe particles themselves
might serve as the active site though the formation mechanism of
the particles remains unclear. These results suggest that Ni-Fe is
an interesting system for investigating the catalytic properties of
IMCs.
using NiSn/MgO-Al O3 catalysts with different MgO contents [14].
2
Catalystscontaining more MgO showed better resistance to carbon
deposition and higher selectivity for hydrogen production. Djaid-
jas et al. measured the amounts of carbon deposition after CO2
reforming of methane using Ni catalysts with different MgO con-
tents, reporting that samples with higher Mg contents showed less
carbon deposition [15]. Tomishige et al. compared the effects of dif-
ferent MgO loadings in Ni catalysts for CO2 reforming of methane,
also reporting that higher loadings of MgO reduced carbon deposi-
∗ Corresponding author.
E-mail addresses: 8215655@ed.tus.ac.jp (H. Mitani), XU.Ya@nims.go.jp
(
(
Y. Xu), HIRANO.Toshiyuki@nims.go.jp (T. Hirano), DEMURA.Masahiko@nims.go.jp
M. Demura), tamura@rs.noda.tus.ac.jp (R. Tamura).
http://dx.doi.org/10.1016/j.cattod.2016.10.036
920-5861/© 2016 Elsevier B.V. All rights reserved.
0

6
70
H. Mitani et al. / Catalysis Today 281 (2017) 669–676
tion [16]. However, as far as we know, there have been no reports on
the carbon deposition behavior on Ni-Fe-Mg alloy catalysts. In this
study, we analyze the effects of the addition of Mg to Ni-Fe IMCs on
carbon deposition and catalytic properties for methanol decompo-
sition based on a detailed characterization of the catalysts before
and after the reactions.
3. Results and discussion
3.1. Characterization before catalytic tests
Fig. 1a and b show SEM images of both samples before reaction,
and Fig. 1c and d show the distribution of particle size obtained
from the SEM observation. Both samples show a similar distribution
of particle size. The particle size distribution in both samples was
estimated to be between 20 and 300 nm in diameter. The average
particle sizes were calculated to be 100 and 112 nm for Ni-25Fe
and Ni-15Fe-Mg, respectively (Table 1), showing that there was
no significant difference in particle size and distribution between
them before reaction.
The XRD profiles of Ni-25Fe and Ni-15Fe-13Mg nanoparticles
before reaction are shown in Fig. 2a and b. The diffraction pat-
terns were analyzed using Rietveld method. Both samples were
confirmed to be mainly Ni3Fe phase.
2
. Experimental
2.1. Catalyst preparation
Two kinds of nanoparticles, 75Ni-25Fe and 72Ni-15Fe-13Mg
(
at.%), were fabricated by a radio-frequency induction thermal
plasma method (Nisshin Engineering, Inc.). The raw materials for
fabrication were pure powders of Ni (99.9%, 2–3 ␮m diameter),
Fe (99.9%, <53 ␮m diameter), and Mg (99.5%, <180 ␮m diame-
ter) (Kojundo Chem. Lab., Ltd.). The composition of the fabricated
nanoparticles was analyzed by fluorescence X-ray and inductively
coupled plasma (ICP) analyses.
However, the lattice constants were estimated to be 3.553 and
3
.542 Å for Ni-25Fe and Ni-15Fe-15Mg, respectively (Table 2). This
small difference in lattice constants was supposedly due to the
different compositions of the two samples.
In addition, the crystallite sizes were estimated to be 179.4 nm
for Ni-25Fe, and 89.9 nm for Ni-15Fe-15Mg (Table 2). The difference
in crystallite size was relatively larger compared with the difference
in particle sizes obtained from SEM analysis. This contradiction was
probably due to the limited particle number in SEM analysis.
The BET specific surface areas of Ni-25Fe and Ni-15Fe-13Mg
nanoparticles were measured before reaction, as shown in Table 3.
2
.2. Catalytic performance
Two types of reaction tests, isochronal and isothermal tests,
were carried out in a conventional fixed-bed flow reactor, as
described in our previous report [6]. Isochronal tests were carried
out stepwise from 513 K to 793 K at intervals of 40 K. The holding
time at each temperature was 30 min. Isothermal tests were carried
out at 673 K for 25 ∼ 27 h. The amount of catalyst used was 20 mg
for the tests. After hydrogen reduction at 723 K for 1 h, methanol
was introduced into the reactor at a gas hourly space velocity of
2
Both samples had specific surface areas around 5 m /g, with no
significant difference between them.
In order to understand the distribution of Mg in the Ni-15Fe-
1
3Mg sample before reaction, we carried out STEM observation
and EDS elemental mapping analysis. Fig. 3a shows the high-angle
annular dark-field (HAADF) STEM image, and Fig. 3b shows the cor-
responding elemental mapping result. It was seen that Ni and Fe
were uniformly in the core of the particle, while Mg was segregated
on the surface of the particles, forming a thin film on the surface.
EDS point analysis was carried out on the surface and center
of the particles, as marked in Fig. 3a. The obtained results for Mg,
Ni, Fe, and O are summarized in Table 4. As shown in Table 4, Ni
and Fe were mainly detected on the center of particles, a higher
concentration of Mg and O was detected around the particles; we
supposed that a MgO layer had formed on the surface because Mg
oxidizes more easily when exposed to air than Ni and Fe. There-
fore, Ni-15Fe-13Mg nanoparticles fabricated by the thermal plasma
method might contain nanocomposite structures of Ni3Fe with a
MgO shell.
−
1
8
0,000 h . The catalytic properties of samples were evaluated by
measuring the outlet composition of gaseous products with two
on-line gas chromatographs (GC) equipped with a thermal conduc-
tivity detector (GL Science, GC323) and the total flow rate of outlet
gases was measured with a soap bubble meter.
2.3. Catalyst characterization
The specific surface areas of the samples before and after reac-
tions were measured by N2 adsorption (Brunauer-Emmett-Teller
BET) method) using a surface area analyzer (Micromeritics, ASAP
020). The morphologies and microstructures of the samples were
analyzed using scanning electron microscopy (SEM; JEOL, JSM-
000F), and transmission electron microscopy (TEM; JEOL,2100F)
(
2
7
with an energy dispersive X-ray spectroscopy (EDS) system. Since
Ni-Fe nanoparticles are ferromagnetic, the TEM samples were pre-
pared by slicing the nanoparticles bonded on a silicon plate using
a focused ion beam system (FIB; JEM-9320FIB, JEOL).
Table 1
Average particle size of samples before reaction and after isothermal tests.
Average particle size (nm)
The crystal structures of both the fabricated samples and sam-
ples after the isochronal tests were measured with synchrotron
X-ray diffraction analysis (SXRD) at the NIMS Beam-line BL15XU
in Spring-8 [17]. The wavelength of the incident X-ray beam was
Ni-25Fe
Ni-15Fe-13Mg
Before reaction
100
319
355
112
85.2
105
6h
27h
0.06525 nm, which was near the Nb K absorption edge. Also, a
laboratory X-ray diffractometer (XRD; RIGAKU, SmartLab) using a
Cu-K␣ source was used for the analysis of samples after isothermal
tests with a parallel beam method.
Table 2
Lattice constants and crystallite sizes of the samples before and after stepwise tests.
The amount of carbon deposition was examined after catalytic
tests using a thermogravimetric (TG) measurement (SHIMADZU,
DTG-60H). The amount of the sample used was 10 mg. The mass
change was measured at a heating rate of 3 K/min from room tem-
perature to 1173 K in a flow of pure air at 50 mL/min.
Lattice constant(A) Crystallite size(nm)
Ni-25Fe
Ni-15Fe-13Mg
Ni-25Fe
Ni-15Fe-13Mg
Before stepwise tests
after stepwise tests
3.553
3.552
3.542
3.542
179.4
217.7
89.9
124.9

H. Mitani et al. / Catalysis Today 281 (2017) 669–676
671
Fig. 1. SEM images of (a) Ni-25Fe nanoparticles, and (b) Ni-15Fe-13Mg nanoparticles before reaction, the corresponding distribution of particle size for Ni-25Fe (c) and
Ni-15Fe-13Mg (d).
Table 3
The above results showed that the Ni-25Fe and Ni-15Fe-13Mg
BET specific surface area of the nanoparticles before and after reactions.
nanoparticles were successfully fabricated by the thermal plasma
2
method. Both of them were mainly composed of Ni Fe phase. The
Samples
BET specific surface area(m /g)
3
Mg in Ni-15Fe-13Mg was found to segregate to the particle surface,
probably forming a MgO layer or shell.
Before
After 6 h reaction
Ni-25Fe
Ni-15Fe-13Mg
5.48
4.56
99.3
72.4
3.2. Catalytic tests
Table 4
EDS results of the points marked in Fig. 3a for Ni-15Fe-13Mg sample before reaction.
3.2.1. Isochronal tests
At first, we carried out isochronal tests between 513 and 793 K
No.
Position
Concentration (at.%)
to examine the temperature dependence of catalytic properties of
the fabricated nanoparticles. Fig. 4 shows methanol conversion as
a function of temperature. The inner pressure (inside the reactor)
was recorded during the reaction (also shown in Fig. 4) to mon-
itor the blockage of reactor due to carbon deposition. Methanol
conversion was low below 593 K for both Ni-25Fe and Ni-15Fe-
Mg
Fe
Ni
O
1
2
3
4
5
Center
Center
Center
Surface
Surface
5.28
2.76
1.59
6.5
15.15
16.1
16.17
0.45
64.35
69.93
71.45
1.73
15.22
11.21
10.8
91.33
94.61
4.65
0.43
0.31
1
3Mg nanoparticles, and began increasing above 593 K. However,
this increase was faster in Ni-15Fe-13Mg than Ni-25Fe, reveal-
ing that Ni-15Fe-13Mg had higher activity than Ni-25Fe. At 753 K,
about 90% conversion was obtained for Ni-15Fe-13Mg, but only
about 60% for Ni-25Fe. The inner pressure was observed to increase
quickly for Ni-15Fe-13Mg above 730 K and above 753 K for Ni-25Fe,
showing that carbon deposition with Ni-15Fe-13Mg progressed
more rapidly than with Ni-25Fe at temperatures above 753 K. To
avoid massive carbon deposition, we carried out isothermal tests
at 673 K in which the inner pressure did not increase markedly
during isochronal tests for both samples.
3.2.2. Isothermal tests
Fig. 5 shows methanol conversion and inner pressure as func-
tions of time on stream at 673 K using Ni-25Fe and Ni-15Fe-13Mg
nanoparticles. At the beginning of the test, Ni-25Fe achieved a

6
72
H. Mitani et al. / Catalysis Today 281 (2017) 669–676
1
00
0.2
Ni-25Fe
Ni-15Fe-
8
6
4
2
0
0
0
0
0
0.18
0.16
0.14
0.12
0.1
13Mg
Ni-25Fe
(inner
pressure)
Ni-15Fe-
1
3Mg (inner
pressure)
5
00
600
700
800
Tempreture (K)
Fig. 4. Methanol conversion and inner pressure as a function of temperature during
isochronal tests over Ni-25Fe and Ni-15Fe-13Mg nanoparticles.
Fig. 2. SXRD results of (a) Ni-25Fe nanoparticles before reaction, (b) Ni-15Fe-13Mg
nanoparticles before reaction, (c) Ni-25Fe nanoparticles after isochronal test, and (d)
Ni-15Fe-13Mg nanoparticles after isochronal test. The wavelength of the incident
X-ray beam was 0.06525 nm.
1
00
0.22
Ni-15Fe -13Mg
Ni-25Fe
8
0
0
.18
conversion of about 20% and about 80% for Ni-15Fe-13Mg, indi-
cating that Ni-15Fe-13Mg had much higher initial activity than
Ni-25Fe, in agreement with the results of isochronal tests. As the
test progressed, the conversion gradually increased for Ni-25Fe
until becoming saturated at about 40% after 5 h. In contrast, the
conversion slightly decreased in the first 5 h, and then stabilized at
about 70% for Ni-15Fe-13Mg. These results revealed that Ni-15Fe-
60
4
2
0
0
0
0.14
0.1
1
3Mg showed much higher activity at steady state than Ni-25Fe.
Fig. 5 shows that, for Ni-25Fe, the inner pressure increased very
0
5
10
15
20
25
Time on stream (h)
slowly during the first 10 h, but quickly after that, finally reach-
ing 0.2 MPa, which was the limit value for forced stoppage of the
reactor. In contrast, for Ni-15Fe-13Mg, the inner pressure increased
slowly during the whole test, giving an inner pressure of about
Fig. 5. Methanol conversion over Ni-25Fe and Ni-15Fe-13Mg nanoparticles as a
function of time on stream for isothermal tests at 623 K.
0
.13 MPa after 27 h, which was much lower than the limit value
the production rates of H2 and CO with Ni-15Fe-13Mg were much
higher than with Ni-25Fe, showing that Ni-15Fe-13Mg showed
higher catalytic activity for hydrogen production from methanol
decomposition than Ni-25Fe.
of 0.2 MPa. Obviously, Ni-15Fe-13Mgt caused much less carbon
deposition during the isothermal test than Ni-25Fe.
The production rates of the gas products were calculated based
on GC analysis and the total flow rate of outlet gases. The produc-
tion rate of carbon was also estimated by considering the balance
of reactants and products involved in the possible reactions. The
details of this method have been described in our previous work [6].
Fig. 6 shows the production rates of all products as a function of time
on stream. The major products were H2 and CO, while minor prod-
Fig. 7 shows the selectivities for all products as a function of time
on stream for (a) Ni-25Fe and (b) Ni-15Fe-13Mg. The selectivities
for H2 and CO were stable, at around 90%, during the whole test
in both samples, whereas selectivities for the other products, CH4,
CO , H O, and C, were less than 10% during the test. These results
2
2
revealed that methanol decomposition (Eq. (1)) was the main reac-
tion during the test, while other side reactions, which produced
ucts (CO2, CH4, H O, and C) were found in both samples. However,
2
Fig. 3. (a) STEM image of Ni-15Fe-13Mg nanoparticles before reaction, (b) The corresponding elemental mapping by STEM-EDS. The Ni distribution is represented with blue,
Fe with green, and Mg with red.

H. Mitani et al. / Catalysis Today 281 (2017) 669–676
673
Fig. 6. Production rates of H2, CO, CO2, CH4, and C over Ni-25Fe (a) and Ni-15Fe-13Mg (b) as a function of time on stream for isothermal tests at 623 K.
Ni-25Fe
Ni-15Fe-13Mg
1
00
1
00
8
6
4
2
0
0
0
0
0
8
6
4
0
0
0
HH 2
CO
CH
H O
2
CH4
4
H2
CC OO 2
CO
2
O
CC OO 2
CC HH 4
C
2
2
4
HH 2 OO
C
HH2
2
2
20
0
0
5
10
15
20
0
5
10
15
20
25
Time on stream (h)
Time on stream (h)
Fig. 7. Selectivities to H2, CO, CO2, CH4, and C over Ni-25Fe (a) and Ni-15Fe-13Mg (b) as a function of time on stream for isothermal tests at 623 K.
Table 5
CH4, CO , H O, and C, such as the water-gas shift reaction (Eq. (2)),
2
2
EDS results of the points marked in Fig. 9a for Ni-15Fe-13Mg sample after 6 h of
reaction at 673 K.
methanation (Eq. (3)), and Boudouard reaction (Eq. (4)), occurred as
side reactions. These results indicated that the addition of Mg to the
Ni-Fe alloy did not significantly affect the main and side reactions
during methanol decomposition.
No.
Position
Concentration (at%)
Mg
Fe
Ni
O
CH OH(g) ꢀ 2H (g) + CO(g)
(1)
(2)
(3)
(4)
1
2
3
4
5
6
7
Surface
Surface
Surface
Interface
Interface
Center
2.39
4.47
1.16
7.12
1.03
1.01
0.10
1.71
14.02
9.06
12.37
12.48
17.00
19.42
8.03
87.86
11.63
36.99
19.04
17.64
20.65
17.15
3
2
69.87
52.79
61.47
68.86
61.34
63.33
CO(g) + H O(g) ꢀ CO (g) + H (g)
2
2
2
CO(g) + 3H (g) ꢀ CH (g) + H O(g)
2
4
2
2
3
3
2
CO(g) ꢀ C(s) + CO (g)
2
Center
.3. Characterization after catalytic tests
of Ni-25Fe became about four times larger after 6 h of reaction,
whereas those of Ni-15Fe-13Mg show no obvious change, even
after 27 h reacting. These results suggested that the addition of Mg
suppressed nanoparticle agglomeration during the reaction.
A detailed analysis of Ni-15Fe-13Mg was carried out after the
reaction, using TEM and STEM-EDS, in order to evaluate Mg distri-
bution during the reaction. Fig. 9a,b show HAADF-STEM images of
the samples after 6 h and 27 h of reaction, and Fig. 9c,d show the
corresponding EDS elemental mapping images. EDS point analysis
was also performed at the positions marked in Fig. 9a. The results
are shown in Table 5. A higher concentration of Mg was detected on
the surface of the particles, compared to the center of the particles.
In contrast, high concentrations of Ni and Fe were detected on the
.3.1. Morphology and microstructure analyses
The morphology of Ni-25Fe and Ni-15Fe-13Mg after 6 h and
5–27 h of isothermal tests at 673 K was analyzed using SEM, as
shown in Fig. 8. Many carbon nanofibers were formed with fine
particles attached at their ends after 6 h of reaction for both Ni-
2
5Fe (Fig. 8a) and Ni-15Fe-13Mg (Fig. 8c). The diameters of the
particles and the carbon nanofibers were approximately the same.
After 25–27 h of reaction, similar microstructures were observed
(Fig. 8b,d). The average particle size in sample was calculated based
on SEM images, as shown in Table 1. The size of carbon nanofibers
and particles in Ni-15Fe-13Mg were much smaller than for Ni-25Fe.
Compared with their sizes before the reaction (Table 1), particles

6
74
H. Mitani et al. / Catalysis Today 281 (2017) 669–676
Fig. 8. SEM images of Ni-25Fe nanoparticles (a) after 6 h, (d) after 25 h, and Ni-15Fe-13Mg nanoparticles (c) after 6 h, (d) after 27 h, of reaction at 673 K.
Fig. 9. STEM images of Ni-15Fe-13Mg nanoparticles after 6 h of reaction (a), 27 h of reaction (b), and the corresponding elemental mappings (c), (d). The Ni distribution is
represented with blue, Fe with green, and Mg with red.

H. Mitani et al. / Catalysis Today 281 (2017) 669–676
675
Fig. 10. XRD profiles of Ni-25Fe and Ni-15Fe-13Mg nanoparticles, (a)(d) before tests, (b) (d) after 6 h, (c)(f) after 27 h.
Table 6
Lattice constants of samples before and after isothermal tests.
Lattice constant (Å)
Ni-25Fe
Ni-15Fe-13Mg
Before reaction
3.553
3.553
3.552
3.543
3.553
3.534
6
2
h
7 h
center of particles. It was seen that the concentrations of Ni and Fe
were much higher even on the surface of the particles (positions
1
–3), compared to these before reaction (Table 4). This might be
due to the influence of the neighbor particles (Fig. 9a).
Elemental mapping (Fig. 9c,d) showed that the MgO layers
around the particles observed before reaction (Fig. 3b) were not
clearly visible after 6 h, forming aggregates neighboring the Ni3Fe
particles. Obviously, MgO re-distributed during the reaction. In
addition, the existence of Mg on the center of the particles was also
confirmed, although the concentration was lower compared to that
before reaction (Table 4), suggesting that some Mg still remained
inside the particles during the reaction.
Fig. 11. TG measurement of Ni-25Fe and Ni-15Fe-13Mg nanoparticles after 6 h of
reaction at 673 K.
The BET specific surface areas of the samples were measured
after the isothermal test at 673 K for 6 h, as shown in Table 3. Both
Ni-25Fe and Ni-15Fe-13Mg showed a significant increase in spe-
cific surface area after the reaction, with the specific surface area of
Ni-25Fe larger than Ni-15Fe-13Mg. The increase in specific surface
area was seemingly due to carbon deposition during the reaction,
with greater carbon deposition resulting in larger specific surface
areas.
3.3.2. XRD and BET specific surface area measurements
The crystal structure and lattice constants of the samples, after
isochronal and isothermal tests, were examined by SXRD and XRD
measurements. The SXRD profiles of the Ni-25Fe and Ni-15Fe-
1
3Mg nanoparticles after isochronal tests are shown in Fig. 2c and d.
3.3.3. Measurement of carbon deposition
The profiles of both samples corresponded to the Ni3Fe phase, and
showed no obvious changes compared with those before the reac-
tion. This suggested that the Ni3Fe phase was maintained during
the isochronal tests in both samples.
The XRD profiles of the samples after isothermal tests at 673 K
for 6 h and 25–27 h are shown in Fig. 10. The Ni3Fe phase was con-
firmed in all the samples. In addition, the carbon phase (graphite)
was also identified in samples after 25–27 h reaction time. The
calculated lattice constants are shown in Table 6. The lattice con-
stants for Ni-25Fe were unchanged before and after the reaction.
However, a small variation was observed during the reaction for Ni-
The amount of carbon produced during the reactions was mea-
sured by TG analysis. Fig. 11 shows the TG profiles of Ni-25Fe and
Ni-15Fe-13Mg after reacting at 673 K for 6 h. Mass losses started
around 693 K in both samples, and finished around 863 K. The mass
loss for Ni-25Fe was about 46%, while that of Ni-15Fe-13Mg was
about 18%. Since mass losses correspond to the combustion of
carbon in the samples, Ni-15Fe-13Mg thus had less carbon deposi-
tion than Ni-25Fe. These results revealed that Ni-15Fe-Mg showed
much higher resistance to carbon deposition than Ni-25Fe, indi-
cating that Mg addition to Ni-Fe catalysts significantly suppressed
carbon deposition during methanol decomposition.
1
5Fe-13Mg, i.e., the lattice constant increased after 6 h reaction, but
decreased after 27 h reaction. The cause of this change was probably
related to an alloy composition change during the reaction due to
the diffusion of Mg toward the outside of the particles, as indicated
by STEM-EDS analysis below.
3.4. Effect of Mg addition in Ni-15Fe-13Mg
As shown above, the Ni-15Fe-13Mg nanoparticles showed much
higher activity and much less carbon deposition in methanol

6
76
H. Mitani et al. / Catalysis Today 281 (2017) 669–676
decomposition than Ni-25Fe nanoparticles (Section 3.2), despite
having similar particle size distributions, similar average parti-
cle sizes and BET specific surface area, and the same Ni3Fe phase
structure before reaction (Section 3.1). This strongly suggested that
Mg addition played an important role in improving the catalytic
properties. We considered the effects of Mg on the basis of the
characterization results as discussed below.
2. Ni-15Fe-13Mg nanoparticles showed much higher activity and
much less carbon deposition in methanol decomposition than
Ni-25Fe nanoparticles.
3. The addition of Mg significantly suppressed the coarsening
process of alloy nanoparticles during the reaction, which con-
tributed to maintain the higher activity of Ni-15Fe-13Mg.
4. As the reaction progressed, the MgO shell around the particles in
the Ni-15Fe-13Mg lost its shape, and segregated near the alloy
particles. MgO appeared to play the role of a promoter and sup-
port for alloy particles, contributing to the suppression of carbon
deposition.
Firstly, Mg addition significantly suppressed the coarsening of
alloy nanoparticles during the reaction (Fig. 8 and Table 1). STEM-
EDS analysis of the samples before the reaction showed that Mg was
segregated around the Ni-Fe-Mg particles, forming a thin layer or
shell of MgO (Fig. 3b). This shell might have protected against fur-
ther oxidation of Ni and Fe in the particles, resulting in less Ni and Fe
oxides on the surface of Ni-15Fe-13Mg particles than in the Ni-25Fe
sample. Fewer Ni and Fe oxides should contribute to a much higher
initial activity than Ni-25Fe (Figs. 4 and 5). Furthermore, the high
initial activity must have resulted in the quick formation of car-
bon fibers on each nanoparticle, as indicated by the results of inner
pressure changes during isochronal tests (Fig. 4), i.e., the inner pres-
sure increased faster Ni-15Fe-13Mg than with Ni-25Fe. The carbon
fibers spread, and prevented coalescence of, the nanoparticles dur-
ing the reaction. In contrast, for Ni-25Fe, it was speculated that
more Ni and Fe oxides existed on the surface, which resulted in a
much lower initial activity than Ni-15Fe-13Mg. As the reaction pro-
gressed, these Ni and Fe oxides could be reduced by the H2 product
Acknowledgments
Synchrotron diffraction experiments were performed under the
approval of the NIMS Synchrotron X-ray Station at the Spring-
8 facility (Proposal Nos. 2013A4501,2014A4907, 2015B4901). The
authors are grateful to Dr. Masahiko Tanaka, Dr. Yoshitake Mat-
sushita, and Mr. Yoshio Katsuya in NIMS for their assistance in
synchrotron diffraction experiments. This work was supported, in
part, by a Grants-in-Aid for Scientific Research (KAKENHI) grant
from the Japan Society for the Promotion of Science (No. 15k06572).
Also, a part of this work was supported by NIMS microstructural
characterization platform (NMCP) as a program of “ Nanotechnol-
ogy Platform” of the Ministry of Education, Culture, Sports, Science
and Technology (MEXT), Japan. We are grateful to Mrs. Isaka,
Mrs.Nishimiya and Mrs. Shichiri in NMCP for preparing samples
for TEM analysis and technological supporting.
(Fig. 6), resulting in a gradual increase in activity during the first
several hours of the reaction (Fig. 5).
Secondly, the formation of MgO seemed to effectively sup-
press carbon deposition (Fig. 11). As has been shown, an MgO
shell formed on the surface of particles (Fig. 3). As the reaction
progressed, the MgO lost its shell form, and segregated near the
particles (Fig. 9c, d). It was suggested that MgO played a role as a
promoter and support for the alloy particles. It is known that the
MgO doped with cations of Ni and Fe also shows activity for the
methanol decomposition [18]. We consider that the MgO might also
contribute to the increase of activity in Ni-15Fe-13Mg, although it
was confirmed that pure MgO powder did not show obvious activity
for methanol decomposition at the present experimental condition.
Also, it is known that MgO contributed to the suppression of car-
bon deposition in a similar fashion to the mechanism reported for
Ni/MgO catalysts, in which MgO inhibited the Boudouard reaction
References
[
1] K. Otsuka, C.M. Wayman, Shape Memory Materials, Cambridge Univ Press,
Cambridge, 1998.
[
2] N.S. Stoloff, C.T. Liu, S.C. Deevi, Intermetallics 8 (2000) 1313–1320.
[3] T. Komatsu, D. Satoh, A. Onda, Chem. Commun. (2001) 1080–1081.
[4] T. Komatsu, S. Hyodo, T. Yashima, J. Phys. Chem. B 101 (1997) 5565–5572.
[
[
[
[
[
5] D.H. Chun, Y. Xu, M. Demura, K. Kishida, M.H. Oh, T. Hirano, D.M. Wee, Catal.
Lett. 106 (2006) 71–75.
6] D.H. Chun, Y. Xu, M. Demura, K. Kishida, D.M. Wee, T. Hirano, J. Catal. 243
(2006) 99–107.
7] Y. Kaneno, T. Kondo, Y. Fujimoto, H. Tsuda, Y. Xu, M. Demura, H. Iwai, T.
Hirano, T. Takasugi, Mater. Trans. 51 (2010) 1002.
8] Y. Xu, M. Demura, T. Hirano, Y. Kamedo, T. Takasugi, Mater. Sci. Forum
706–709 (2012) 1052–1057.
9] Z. Turgut, N.T. Nuhfer, H.R. Piehler, M.E. Mchenry, J. Appl. Phys. 85 (1999)
(
Eq. (4)) over Ni [19]. In addition, it was shown that some Mg might
have remained in the particles, even after 6 h of reaction at 673 K
Table 5). The Mg contained in the Ni-Fe-Mg alloy particles might
also have affected the activity and carbon deposition.
4406–4408.
[
[
[
[
[
[
10] J. Zheng, R. Yang, L. Xie, J.L. Qu, Y. Liu, X.G. Li, Adv. Mater. 22 (2010)
1451–1473.
11] Y. Xu, J. Yang, M. Demura, T. Hirano, Y. Matsushita, M. Tanaka, Y. Katsuya,
Mater. Sci. Forum 783–786 (2014) 2040–2045.
12] Y. Xu, J. Yang, M. Demura, T. Hara, T. Hirano, Y. Matsushita, M. Tanaka, Y.
Katsuya, Appl. Catal. A 478 (2014) 165–174.
13] Y. Xu, J. Yang, M. Demura, T. Hara, T. Hirano, Y. Matsushita, M. Tanaka, Y.
Katsuya, Int. J. Hydrogen Energy 39 (2014) 13156–13163.
14] A. Penkova, L. Bobadilla, M.I. Dominguez, F. Romero-Sarria, A.C. Roger, Appl.
Catal. A 392 (2011) 184–191.
(
4
. Conclusions
Ni-25Fe and Ni-15Fe-13Mg nanoparticles were fabricated using
a radio-frequency induction thermal plasma technique, and their
catalytic properties were studied in methanol decomposition. The
following results were obtained:
15] A. Djaidja, S. Libs, A. Kiennemann, A. Barama, Catal. Today 113 (2006)
194–200.
[
[
16] K. Tomishige, Y. Chen, K. Fujimoto, J. Catal. 181 (1991) 91–103.
17] M. Tanaka, Y. Katsuya, A. Yamamoto, Rev. Sci. Instrum. 79 (2008)
075106-1–075106-6.
1
. Both Ni-25Fe and Ni-15Fe-13Mg nanoparticles were mainly
composed of a Ni3Fe phase. They had a BET specific surface area
around 5 m2/g, and their particle sizes were distributed between
[
[
18] M. Morita, K. Takeda, S. Tanabe, Y. Matsuda, Chem. Soc. Jpn. 9 (1991)
1238–1240.
19] E. Ruckenstein, Y.H. Hu, Appl. Catal. A 133 (1995) 149–161.
5
0 and 300 nm in diameter. Mg in the Ni-15Fe-13Mg sample was
found to be segregated on the surface, forming a thin MgO shell
around the particles.