An investigation of the thermal stability, crystal structure and catalytic properties of bulk and alumina-supported transition metal nitrides

Article abstract of DOI:10.1016/j.jallcom.2007.10.046

A series of bulk and alumina-supported nitrides of metals (V, Mo, Fe and Co) were synthesized by NH₃-temperature-programmed reaction and characterized by X-ray diffraction (XRD) and temperature-programmed desorption techniques. The formation of bulk Co₄N and Co₄N/γ-Al₂O₃ was further confirmed using X-ray photoelectron spectroscopic analysis. The catalytic activities of these metal nitrides for NO decomposition were evaluated. Their activities for NO decomposition ranked in the order of Co₄N > Fe₃N > Mo₂N > VN and Co₄N/γ-Al₂O₃ > Fe₃N/γ-Al₂O₃ > Mo₂N/γ-Al₂O₃ > VN/γ-Al₂O₃. The relationship between crystal structures and catalytic activities was investigated. The results indicated that metal nitrides with higher vacancy concentration exhibited higher activities for NO decomposition. There was a stronger interaction between the metal nitride phases and γ-Al₂O₃ support. It was suggested that Co₄N/γ-Al₂O₃ exhibited thermal stability significantly higher than that of bulk counterpart, owing to the strong interaction between the Co₄N phase and γ-Al₂O₃ support. We applied the XRD technique to examine the structural changes of Co₄N/γ-Al₂O₃ catalysts during the reactions. The results indicated that the rapidly loss in catalytic activity was due to the bulk oxidation of Co₄N/γ-Al₂O₃. In the NO-H₂ reaction, the oxygen generated during NO dissociation was partly reduced by H₂ and partly incorporated into the nitride lattice. By the addition of H₂ in feed gas at 600 °C, one can retain the active Co₄N/γ-Al₂O₃ phase by minimizing the presence of surface oxygen.

Full text of DOI:10.1016/j.jallcom.2007.10.046

Journal of Alloys and Compounds 464 (2008) 488–496
An investigation of the thermal stability, crystal structure and catalytic
properties of bulk and alumina-supported transition metal nitrides
∗
Zhiwei Yao , Anjie Zhang, Yuan Li, Yuzhuo Zhang, Xiaoqing Cheng, Chuan Shi
Laboratory of Plasma Physical Chemistry, Dalian University of Technology, Dalian 116024, China
Received 23 September 2007; received in revised form 4 October 2007; accepted 4 October 2007
Available online 22 October 2007
Abstract
A series of bulk and alumina-supported nitrides of metals (V, Mo, Fe and Co) were synthesized by NH
3
-temperature-programmed reaction and
N and Co N/␥-Al
characterized by X-ray diffraction (XRD) and temperature-programmed desorption techniques. The formation of bulk Co
was further confirmed using X-ray photoelectron spectroscopic analysis. The catalytic activities of these metal nitrides for NO decomposition were
evaluated. Their activities for NO decomposition ranked in the order of Co N > Fe N > Mo N > VN and Co N/␥-Al > Fe N/␥-Al > Mo N/␥-
Al > VN/␥-Al . The relationship between crystal structures and catalytic activities was investigated. The results indicated that metal nitrides
with higher vacancy concentration exhibited higher activities for NO decomposition. There was a stronger interaction between the metal nitride
phases and ␥-Al support. It was suggested that Co N/␥-Al exhibited thermal stability significantly higher than that of bulk counterpart,
owing to the strong interaction between the Co N phase and ␥-Al support. We applied the XRD technique to examine the structural changes
of Co N/␥-Al catalysts during the reactions. The results indicated that the rapidly loss in catalytic activity was due to the bulk oxidation of
Co N/␥-Al
4
4
2 3
O
4
3
2
4
2
O
3
3
2
O
3
2
2
O
3
2 3
O
2
O
3
4
2 3
O
4
2 3
O
4
2 3
O
4
2
O
3
. In the NO–H
2
reaction, the oxygen generated during NO dissociation was partly reduced by H
2
and partly incorporated into the
◦
nitride lattice. By the addition of H
oxygen.
2
in feed gas at 600 C, one can retain the active Co
4
N/␥-Al
2
O
3
phase by minimizing the presence of surface
©
2007 Elsevier B.V. All rights reserved.
Keywords: Nitride materials; Catalysis; Crystal structure; Vacancy information; X-ray diffraction
1
. Introduction
observation indicated that Co4N was unstable and undergo-
ing a stepwise decomposition with N2 evolution below 700 C:
◦
An increasing interest has developed in exploring the cat-
Co4N → Co3N + Co, Co3N → Co2N + Co, Co2N → CoN + Co
[16]. Kojima and Aika reported the use of Re3N as catalysts for
ammonia synthesis at 623 K under 0.1 MPa. Although activity
was high at the initial stage, there was deactivation due to Re3N
decomposition [17]. For the utilization of nitrides of Group
VII–VIII transition metals, a means of enhancing the thermal
stability remains as a meaningful challenge in the field.
The catalytic activities and surface properties of transi-
tion metal nitride catalysts were reported to be related to the
vacancies in the metal lattice and near the surface [18]. In
general, vacancies are partly due to the inherent nature of
these interstitial compounds and partly depend on the prepa-
ration or heat treatment conditions. Zhang and co-workers
reported that non-stoichiometric FeNx were prepared at differ-
ent temperatures and showed differently active for hydrazine
decomposition, implying that there were relationships between
the catalyst stoichiometry and activity [19]. As Omi and
alytic properties of transition metal nitrides since these materials
exhibit the noble metals-like behavior. Much of the focus has
been on bulk and supported VN, Mo2N, W2N and Mo/W-
based bimetallic (oxy)nitrides due to their competitive activities
for reactions including NH3 synthesis [1–3], CO hydrogena-
tion [4], NO removal [5–9], hydrazine decomposition [10–12],
hydrogenation (HYN), hydrodesulfurization (HDS), hydroden-
itrogenation (HDN) and hydrodeoxygenation (HDO) [13–15].
Compared to the nitrides of Group V–VI metals (e.g. V, Mo, W),
the nitrides of Fe, Co, Re (Group VII–VIII) have received far less
attention due to their poor thermal stability. The results of earlier
∗
Corresponding author. Tel.: +86 411 84708548x804;
fax: +86 411 84708548x808.
E-mail address: mezhiwei@163.com (Z. Yao).
0
925-8388/$ – see front matter © 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.jallcom.2007.10.046

Z. Yao et al. / Journal of Alloys and Compounds 464 (2008) 488–496
489
co-workers reported, the structure and composition of molyb-
denum nitride catalysts changed during cooling in a stream
of ammonia or helium gas after NH3 treating. During the
HDN of carbazole, Mo2N was more active than Mo2N0.78
for C–N hydrogenolysis. On the contrary, Mo2N0.78 was
more active than Mo2N for hydrogenation of carbazole [20].
In addition, Thompson and co-workers studied the ther-
mal desorption of H2 and NH3 over molybdenum nitrides
by H2-temperature-programmed desorption (H2-TPD) and
NH3-temperature-programmed desorption (NH3-TPD), respec-
tively. The results indicated that the adsorption capacity
of molybdenum nitrides increased in the following order:
supported oxide precursors were placed in a microreactor and a flow of NH3
(
150 ml/min) was introduced into the system. Initially, the sample was linearly
◦
heated from room temperature (RT) to 300 C over a period of 30 min, followed
by a rise in temperature from 300 to 450 C at a rate of 0.67 C/min, and a further
increase from 450 to 600 C for Fe2O3/␥-Al2O3 and to 700 C for others at a rate
of 1.67 C/min. The temperature was then kept at the final temperature for 2 h
◦
◦
◦
◦
◦
before cooling to RT in a NH3 flow. The cobalt metal catalyst was prepared from
◦
their corresponding oxide precursors using H2 at 700 C for 2 h. The material
was also passivated in 1% O2/99% N2 for 12 h before it was exposed to air.
2
.2. Catalyst characterization
The BET surface areas of the passivated samples were measured on an ASAP
2010 instrument. The N2 gas was adopted for standard five-point BET surface
Mo N7 < MoN < Mo2N for H2 and MoN < Mo N7 < Mo2N
for NH3. Moreover, they reported that the pyridine HDN
reaction rate linearly increased with the adsorping of NH3
area measurements. XRD examination was performed using a X-ray diffrac-
tometer (Rigaku D-Max Rotaflex) with Cu K␣ radiation (λ = 1.5404
particle size was estimated according to the Scherrer formula. The temperature-
programmed desorption (TPD) of the as-prepared nitrides was examined on
a flow system equipped with a mass spectrometer (MS, HP G1800A). X-ray
photoelectron spectroscopic (XPS) investigation was conducted using an Al
K␣ source operated at 10 kV and 15 mA (Leybold Heraeus-Shengyang SKL-12
with VG CLAM 4 MCD analyzer); charging effects were corrected by means
of adventitious carbon (284.6 eV) referencing.
1
6
16
A˚ ). The
(
MoN < Mo N7 < Mo2N) [21].
16
In addition, catalysis involving transition metal nitrides is
strongly related to surface structure and elemental composition
22,23]. Spreading a transition metal nitride on a support could
[
alter its surface structure and elemental composition of the tran-
sitionmetalnitride, therebyinducingacatalyticactivitydifferent
from that of its bulk counterpart. It is of significant interest to
find out whether a strong interaction between metal nitrides and
support would exist and affect the structural properties of metal
nitrides. However, no investigations have, thus far, been made
on this topic. Moreover, few investigations have been particu-
larly focused on the relationship between the crystal structures
and catalytic activities of transition metal nitrides.
In the present study, a series of bulk and alumina-supported
nitrides of metals (V, Mo, Fe and Co) were synthesized and their
catalytic performances for NO decomposition were evaluated in
a continuous-flow microreactor. The support effects on textural
and structural properties of these metal nitrides were studied by
characterizations of XRD, BET and temperature-programmed
desorption. Therelationshipbetweencatalyticperformancesand
crystal structures of the metal nitride catalysts was discussed.
The XRD measurement was adopted for the characterization of
the used alumina-supported cobalt nitride catalysts, and to give
insights into the catalytic nature of alumina-supported cobalt
nitride catalyst for NO decomposition and reduction with H2.
2
.3. Catalytic activity
The catalytic activity was evaluated using a quartz microreactor (i.d. 4 mm).
The temperature was measured with a thermocouple placed adjacent to the cata-
lyst outside the reactor. A reaction gas mixture composed of 0 or 1000 ppm H2,
1000 ppm NO and He as balance was fed through the catalyst (0.4 g) which had
◦
been pretreated in pure He at 400 C for 1 h. The effluent gases were monitored
by means of online GC with a TCD detector. A molecular sieve 5 A˚ column was
used to separate H2, O2, N2 and NO. The amount of N2 produced was used to
calculate the conversion of NO to N2.
3. Results
3.1. Catalyst charaterization
The XRD patterns of bulk and alumina-supported metal
nitrides as well as ␥-Al2O3 for comparison are shown in
Figs. 1 and 2, respectively. The XRD patterns of bulk and
2
. Experimental
2
.1. Catalyst preparation
The metal nitrides were prepared from bulk metal oxide and alumina-
supported metal oxide precursors via NH3-temperature-programmed reaction,
following the preparation procedures described in detail elsewhere [7–9].
The following bulk metal oxide precursors were used in the present study:
V2O5, MoO3, Fe2O3 and Co3O4. Alumina-supported metal oxide precursors
(
MoO3/␥-Al2O3, Fe2O3/␥-Al2O3 and Co3O4/␥-Al2O3) with theoretical load-
ing of 30 wt.% M (M = Mo, Fe, Co) were prepared by the incipient wetness
method using aqueous ammonium molybdate ((NH4)6Mo7O24·4H2O), iron
nitrate (Fe(NO3)3·9H2O), cobalt nitrate (Co(NO3)2·6H2O) solutions and ␥-
2
−1
Al2O3 (SBET ≈ 250 m g ), 40–60 mesh as support, except that V2O5/␥-Al2O3
precursor was prepared by mixing V2O5 and ␥-Al2O3 with theoretical loading
of 30 wt.% V. The mixture was thoroughly ground, pressed into pellets. All
◦
the alumina-supported metal oxide precursors followed by drying at 120 C for
1
◦
2 3 4
Fig. 1. XRD patterns of VN, Mo N, Fe N and Co N.
2 h and calcination at 500 C for 3 h. Typically, 2.0 g of the bulk and alumina-

4
90
Z. Yao et al. / Journal of Alloys and Compounds 464 (2008) 488–496
Table 1
Textural properties of bulk and alumina-supported transition metal nitrides
2
−1
)
Material
Surface area (m g
Crystal size (nm)
VN
Mo2N
17.0
62.0
14
8
Fe3N
Co4N
VN/␥-Al2O3
Mo2N/␥-Al2O3
Fe3N/␥-Al2O3
Co4N/␥-Al2O3
1.3
1.3
118.1
158.2
142.6
148.4
19
25
14
19
10
14
The results of TPD studies over the as-prepared bulk and
alumina-supported nitrides of metals (V, Mo, Fe and Co) are
shown in Fig. 4. The signal of (m/e = 28), (m/e = 17) and
(
m/e = 18) are designed to detect N2, NH3 and H2O, respectively.
◦
Fig. 2. XRD patterns of VN/␥-Al2O3, Mo2N/␥-Al2O3, Fe3N/␥-Al2O3 and
Co4N/␥-Al2O3 as well as ␥-Al2O3 for comparison.
WeobservedthatNH desorbedatlowertemperatures(<400 C)
over all the samples except Fe3N and Co4N. The TPD profiles
also showed desorption peaks of H2O below 600 C over bulk
3
◦
and alumina-supported VN and Mo2N, and a very weak broad
H2O peak within the 200–600 C range over Co4N/␥-Al2O3.
The release of N2 was featured by a broad range of desorption.
The strongest N2 peak observed over all of nitrides should be
attributed to the breaking of M–N (M = V, Mo, Fe and Co) bonds
at these temperatures.
alumina-supported nitrides of metals (V, Mo, Fe and Co) were
consistent with those reported in the literatures [24–26]. Fang
et al. reported that the grain structure of the Co4N was similar
to that of Co, which might easily lead to misinterpretation [16].
We also found that it was impossible to differentiate Co4N from
◦
◦
Co because the XRD pattern of Co3O4 reduced in H2 at 700 C
The freshly nitrided sample was cooled in a flow of NH3,
it was reasonable to assume that there were NHx (x = 1–3)
adspecies on the sample surface as suggested before [27]. Thus
the NH3 peak was a result of the desorption of weakly adsorbed
for 2 h looked exactly the same as that of Co4N (Fig. 3).
Table 1 summarizes the textural properties of bulk and
alumina-supported transition metal nitrides. Compared with
unsupportednitrides, allthealumina-supportednitridesobtained
◦
2
−1
2
−1
NH . The N desorption below 600 C was attributed to the
had higher surface area, from 118.1 m g up to 158.2 m g
.
3 2
decomposition of NHx (x = 1–3) adspecies that were strongly
adsorbed on the nitrides [7]. In other words, adsorbed NHx
Especially for Co4N/␥-Al2O3 and Fe3N/␥-Al2O3, the surface
areas were above 100 times higher than those of their bulk coun-
terparts. The crystal sizes, calculated by the Scherrer formula
based on XRD peak broadening, indicated dimensions of the
order 8–25 nm.
(
x = 1–3) species desorbed either as N2 or NH3. Over the Co4N
◦
sample, we detected an intense N2 peak at 620 C and there was
no obvious desorption of NH3 and H2O in the entire tempera-
ture range adopted for the study; a clear indication that the N2
was originated from the decomposition of Co4N. In the case
of Co4N/␥-Al2O3, the N2 peak was at 420 C and there was
desorption of NH3 and H2O within the 200–600 C range, the
◦
◦
former was strong whereas the latter weak. Therefore, It was
◦
not possible to be confirmative whether the N2 peak at 420 C
was due to Co4N decomposition or NHx (x = 1–3) dissociation.
The hydrogen originated from NHx decomposition could des-
orb as H2 or react with surface oxygen to give water. We could
not monitor H2 desorption because the MS equipment was not
designed for hydrogen detection. Nevertheless, the detection of
weak H2O desorption signal suggested that there was only a
small amount of oxygen on the surface of Co4N/␥-Al2O3 and
there was NHx (x = 1–3) dissociation. In addition, our NH3-TPD
results (not shown) on Co4N/␥-Al2O3 revealed that the temper-
◦
ature (480 C) for the N2 desorption in NH3-TPD investigation
was close to that detected in TPD (Fig. 4h). If one could accept
◦
that the N2 peak at 420 C was originated from NHx (x = 1–3),
the Co4N formed on ␥-Al2O3 should be thermally stable and
have kept its structure reasonably well up to a temperature as
high as 950 C.
Fig. 3. AcomparisonofXRDpatternsforCoandCo4NsynthesizedfromCo3O4
in H2 and NH3, respectively.
◦

Z. Yao et al. / Journal of Alloys and Compounds 464 (2008) 488–496
491
Fig. 4. TPD profiles of (a) VN, (b) Mo2N, (c) Fe3N, (d) Co4N, (e) VN/␥-Al2O3, (f) Mo2N/␥-Al2O3, (g) Fe3N/␥-Al2O3 and (h) Co4N/␥-Al2O3.

4
92
Z. Yao et al. / Journal of Alloys and Compounds 464 (2008) 488–496
Fig. 4. (Continued ).
peak fittings of the Co 2p3/2 binding energies were deconvo-
0
2+
3+
luted to Co (777.6 ± 0.8 eV), Co (779.9 ± 0.4 eV) and Co
(
781.6 ± 0.3 eV) according to the XPS study of cobalt molyb-
denum nitrides by Nagai and co-workers [28]. By means of
curve-fitting, the cobalt oxidation states and the corresponding
distribution of cobalt species were estimated (Table 2). As shown
in Fig. 5, the binding energies of Co 2p1/2 and Co 2p3/2 for Co4N
were 796.3 and 781.1 eV, respectively. The value of Co 2p3/2 was
identical with that reported by Milad et al [24]. The value was
2+
higher than that of Co (779.9 ± 0.4 eV) but slightly lower than
that of Co3 (781.6 ± 0.3 eV) [28]. Accordingly, we denoted the
+
δ+
Co species of Co4N as Co , where 2 < δ < 3. For a Co sample
obtained via H2-treating a cobalt oxide sample after passivation,
the binding energies of Co 2p3/2 were determined to be 777.9 and
7
80.5 eV, ingoodagreementwiththenumberobtainedinanXPS
0
study of Co (778.0 eV) and Co3O4 (780.5 eV) species [29]. The
results of XPS investigation indicated that there was the coexis-
tence of metal cobalt and cobalt oxide on the H2-reduced cobalt
oxide sample. As for the passivated Co N/␥-Al O sample, the
Fig. 5. Co 2p spectra of Co4N, Co4N/␥-Al2O3, Co4N/␥-Al2O3 (treated in He
◦
at 500 C for 3 h) and Co in XPS study.
4
2
3
peaks at binding energy of 781.6 and 796.8 eV were closer to
those of Co4N (781.1 and 796.3 eV) in comparison to those of
Co (778.0 and 793.0 eV) [29]. The results indicated that the
The XPS Co 2p spectra of Co4N and Co4N/␥-Al2O3 are pre-
sented in Fig. 5. For comparison, that of Co metal is also shown.
The relative intensities of spin–orbit doublet peaks are given by
the ratio of their respective degeneracy, and the I(2p3/2)/I(2p1/2)
intensity ratio for the Co(2p3/2)/Co(2p1/2) doublet is 2/1. A split-
ting energy of 15.2 eV is expected for the doublet. Since the
Co 2p spectra were very broad and difficult to deconvolute for
both Co 2p3/2 and Co 2p1/2 in the region of 775–805 eV. The
0
cobalt oxide precursor on ␥-Al2O3 was converted to Co4N but
0
not Co during the ammonolysis process. In view of the N2
peak shown in Fig. 4h could be a result of Co4N decomposition,
a portion of the Co4N/␥-Al2O3 sample was pretreated in a flow
◦
of He at 500 C for 3 h to eliminate surface NHx (x = 1–3). The
Co 2p spectrum of the treated sample was similar to that of the
Table 2
XPS parameters derived from curve-fitting the Co 2p3/2 and N 1s envelopes
Sample
Co 2p3/2 (eV)
Co4N
N 1s (eV)
Co3O4
780.5
Co0
Co4N
NHx/N2
Co4N
Co4N/␥-Al2O3
Co4N/␥-Al2O3 (500 C pretreated)
Co
781.1
781.6
781.4
397.8
397.7
397.8
399.6
399.6
399.5
399.7
◦
777.9

Z. Yao et al. / Journal of Alloys and Compounds 464 (2008) 488–496
493
Fig. 6. Variation of NO conversion to N2 as a function of reaction tem-
perature over VN, Mo2N, Fe3N and Co4N catalysts. Reaction conditions:
NO = 1000 ppm, gas flow rate = 20 cm min , W/F = 1.2 g scm .
Fig. 7. Variation of NO conversion to N2 as a function of reaction temper-
ature over VN/␥-Al O , Mo N/␥-Al O , Fe N/␥-Al O and Co N/␥-Al O
3
2
3
2
2
3
3
2
3
4
2
3
−1
−3
catalysts (reaction conditions: same as those of Fig. 6).
un-treated one. The results inferred that there was Co4N on ␥-
◦
Al2O3 aftertheN2 desorptionat420 C(Fig. 4h). Inotherwords,
the rest of this paper was devoted to investigate the genuine
nature of Co4N/␥-Al2O3 for catalytic removal of NO.
the evolution of N2 was originated from NHx (x = 1–3) decom-
position rather than from thermal decomposition of Co4N. The
region of XPS N 1s contained two peaks of N 1s binding ener-
gies at ∼398 and ∼400 eV for cobalt nitride samples and one
at ∼400 eV for cobalt metal sample (Table 2). The N 1s peak
at ∼398 eV is usually known as “nitride” peak [24]. While the
nitrogen species with the peak at ∼400 eV are assigned to the N
Fig. 8 shows the time dependence of NO conversion to N2
over Co4N/␥-Al2O3 catalystsin0.1%NO/Heand0.1%H2/0.1%
NO/He gas streams. It can be observed that the Co4N/␥-Al2O3
catalyst showed a stable activity of ca. 100% in a gas stream
◦
of 0.1% NO/He at 400 C, but deactivated rapidly after 28 h of
on-stream reaction: NO conversion to N2 decreased from ca.
1
s levels of NHx or N2 chemisorbed on the surface [30,31].
1
0
00% to ca. 58% within a period of 32 h. With the addition of
.1% H2 in the feed, the Co4N/␥-Al2O3 catalyst showed a stable
3
.2. Activity measurement
activity of ca. 100% through out the test period of 60 h, no matter
◦
whether it was at 400 or 600 C.
Fig. 6 shows the catalytic activities of VN, Mo2N, Fe3N
and Co4N for NO decomposition. It can be observed that
the trend of activity increased with increasing reaction tem-
perature. Over VN, Mo2N, Fe3N and Co4N catalysts, the
NO conversion into N2 reached 40, 51, 89 and ca. 100% at
◦
3
50 C, respectively. The catalytic activities of NO decompo-
sition ranked in the order Co4N > Fe3N > Mo2N > VN. Fig. 7
shows the catalytic activities of alumina-supported transition
metal nitride catalysts for NO decomposition. It was obvious
that the activity increased accompanied with reaction temper-
ature. Over VN/␥-Al2O3, Mo2N/␥-Al2O3, Fe3N/␥-Al2O3 and
Co4N/␥-Al2O3 catalysts, the NO conversion into N2 reached
◦
4
2, 68, 96 and ca. 100% at 350 C, respectively. Compared to
unsupported nitrides, alumina-supported nitrides showed higher
activities for NO decomposition at the entire reaction tempera-
ture range. Particularly, Fe3N/␥-Al2O3 and Co4N/␥-Al2O3 were
◦
more active than bulk ones at lower temperatures (≤300 C).
It was noteworthy that the activities for NO decomposition
followed the order Co4N/␥-Al2O3 > Fe3N/␥-Al2O3 > Mo2N/␥-
Al2O3 > VN/␥-Al2O3, which was the same to that of bulk metal
nitride catalysts. Amongst this series of bulk and alumina-
supported transition metal nitrides, Co4N/␥-Al2O3 was found
to exhibit the highest activity for NO decomposition. Therefore,
Fig. 8. NO conversion to N2 as a function of reaction time over Co4N/␥-Al2O3
catalysts. Reaction conditions: H2 = 0 or 1000 ppm, NO = 1000 ppm, gas flow
3
−1
−3
rate = 20 cm min , W/F = 1.2 g scm .

4
94
Z. Yao et al. / Journal of Alloys and Compounds 464 (2008) 488–496
Fig. 9. XRD patterns of used Co4N/␥-Al2O3 catalysts as well as fresh sample
and ␥-Al2O3 for comparison: (᭹) Co4N, (ꢀ) CoO, (ꢁ) Co3O4.
Fig. 10. Crystal structure of (a) VN, (b) Mo2N, (c) Fe3N and (d) Co4N, where
the large white spheres represent metal atoms (V, Mo, Fe and Co) and the small
black ones represent interstitial atoms of N.
3
.3. Characterization of the used samples
The XRD patterns of the used Co4N/␥-Al2O3 catalysts (func-
tioned in 0.1% NO/He and 0.1% H2/0.1% NO/He gas streams)
Most transition metal nitrides are interstitial compounds with
nitrogen atoms residing in interstices in the metal lattice. They
are often characterized by the crystal structures of face-centered
cubic (fcc), hexagonal closed packed (hcp) or simple hexagonal
are presented in Fig. 9. As for the used Co4N/␥-Al2O3 sample
◦
in 0.1% NO/He at 400 C, there were only diffraction peaks of
CoO crystallite, except that the most intense Co4N phase diffrac-
tion peak was also visible, and other diffraction peaks from this
phase were too weak to detect. The results indicated that, almost
complete oxidation of Co4N/␥-Al2O3 catalyst occurred within
(hex)andmanyofthemcanshowlargevacancyconcentrationon
the nitrogen sublattices [36]. Fig. 10 shows the crystal structures
of VN, Mo2N, Fe3N and Co4N. VN has the rock salt structure
with fcc lattice of V and all of octahedral sites are occupied by
N atoms in such a way that all adjacent N atoms centred octa-
3
2 h on stream. In the case of Co4N/␥-Al2O3 catalyst functioned
◦
in 0.1% H2/0.1% NO/He at 400 C for 60 h, the diffraction lines
of Co4N changed a lot. The line intensity of Co4N decreased
significantly and there were lines of CoO occurred. As for the
ꢀ
hedral share edges [37]. Mo2N has the L3 structure with a fcc
lattice of Mo but with N atoms filling only half of the available
octahedral sites and edge-sharing arrangements appear as well
Co4N/␥-Al2O3 catalyst functioned in 0.1% H2/0.1% NO/He at
◦
6
00 C for 60 h, the diffraction lines due to Co4N crystalline
[8]. For the structure of Fe3N, the Fe atoms form a hcp pack-
werestillstronginintensity;therewerelinesduetocobaltoxides
CoOandCo3O4), buttheintensitiesweremuchlowerthanthose
ing, with the N atoms occupying one-third of corner-sharing
octahedral [37,38]. Finally, the crystal structure of Co4N can be
appropriately described by an antiperovskite structure with fcc
Co metallic lattice, where one-fourth of all octahedral sites are
occupied by N atoms in such a way that the octahedra share cor-
ners only [16]. The crystal structure of Co4N can be also thought
of as fcc cobalt lattice with N atoms at the center of the unit cell.
Most of structure information of VN, Mo2N, Fe3N and Co4N
is tabulated in Table 3. Obviously, in the case of nonmetallic
nitrides with the cubic or hexagonal closed packed structure,
the nitrogen vacancy concentration appeared to be related to
atomic ratio of N/metal. It was expected that the NO decompo-
(
of the catalyst used in 0.1% NO/He and 0.1% H2/0.1% NO/He
◦
gas streams at 400 C.
4
. Discussion
The nitrides and carbides of transition metals have attracted
considerable attention because their catalytic properties resem-
ble those of noble metals [13]. The formation of nitrides and
carbides modifies the nature of the d band of the parent met-
als. It is believed that such a d band contraction would result
in a greater density of states near the Fermi level, and the cat-
alytic properties are different from those of the parent metals
but similar to those of noble metals [32,33]. It is well known
that noble metal catalysts are catalytically active in decomposi-
tion and reduction of NOx, and are used as active components in
three-way catalyst (TWC) converters [34,35]. The similarity in
chemical properties between transition metal nitrides and noble
metals suggests that the former could be catalytically effective
for NO elimination.
Table 3
Structure information of the transition metal nitrides
Material
Lattice structure
N/metal ratio
Vacancy concentration (at%)
VN
fcc
fcc
hcp
fcc
1/1
1/2
1/3
1/4
0
50
67
75
Mo2N
Fe3N
Co4N

Z. Yao et al. / Journal of Alloys and Compounds 464 (2008) 488–496
495
◦
sition activity strongly depended on the vacancy concentration
of transition metal nitrides. In an analogous way, metal nitrides
with higher vacancy concentration exhibited higher activities
for NO decomposition (see Figs. 6 and 7). It was believed that
the octahedral sites of transition metal nitrides, can be regarded
as oxygen vacancies due to the similar atom radius between
nitrogen and oxygen [14]. This suggested that the presence of
vacancies was favorable for the adsorption and dissociation of
NO and the adsorption and dissociation amount of NO was cor-
related with the amount of vacancies in nitride lattice, which
was similar to many perovskite-type oxide catalysts for NO
decomposition [39].
Co4N/␥-Al2O3 at 400 C with no detectable production of O2,
indicating that oxygen produced during NO dissociation was
completely captured by Co4N/␥-Al2O3. Heavy accumulation of
surface oxygen resulted in gradual diffusion of oxygen into the
nitride lattice and caused the ultimate oxidation of the bulk. This
has been verified by XRD measurement as shown in Fig. 9. In
the NO reduction reaction (Fig. 8), the Co4N/␥-Al2O3 catalyst
showed a stable activity of ca. 100% through out the test period
of 60 h. The results indicated that H2 facilitated the removal
of surface oxygen and hence one avoided the heavy accumu-
lation of surface oxygen to a certain extent. According to the
XRD characterization of the Co4N/␥-Al2O3 catalyst functioned
◦
The effects of ␥-Al2O3 support on structural and textural
properties of metal nitrides were investigated by XRD, BET and
TPD studies. From Fig. 2, it was found that the peak-shift of ␥-
Al2O3 in alumina-supported metal nitrides occurred compared
with pure ␥-Al2O3. It was indicated that there was a stronger
interaction between metal nitride phases and ␥-Al2O3 support.
The interaction between ␥-Al2O3 and metal carbides was also
observed in XRD study by other researchers [40]. As revealed
by TPD studies (Fig. 4), the decomposition temperatures of
VN/␥-Al2O3 and Mo2N/␥-Al2O3 shifted to lower temperatures
compared with those of bulk VN and Mo2N. Fe3N and Fe3N/␥-
Al2O3 had similar decomposition temperatures. Nevertheless,
Co4N/␥-Al2O3 was rather more thermally stable than Co4N and
kept its structure reasonably well up to a temperature as high
in 0.1% H2/0.1% NO/He at 400 C for 60 h, it was found that the
diffraction lines due to CoO were stronger than those of Co4N.
However, as for the Co4N/␥-Al2O3 catalyst functioned in 0.1%
H2/0.1% NO/He at 600 C for 60 h, the diffraction lines due to
Co4N crystalline were still strong in intensity, indicating that
the nitride phase was well retained in the Co4N/␥-Al2O3 cata-
lyst. The results suggested that the addition of H2 in feed gas at
◦
◦
600 C was required for oxidation prevention of Co4N/␥-Al2O3
catalyst.
5. Conclusion
A series of bulk and alumina-supported nitrides of
metals (V, Mo, Fe and Co) were synthesized via NH3-
temperature-programmed reaction. The resulting materials were
characterized using XRD, BET and TPD techniques, and
the formation of bulk Co4N and Co4N/␥-Al2O3 were fur-
ther confirmed using XPS analysis. The structure of these
materials has been investigated. VN, Mo2N and Co4N
adopted fcc structure while Fe3N crystallized in hcp struc-
ture, and the vacancy concentration was ranked as follows:
Co4N > Fe3N > Mo2N > VN. The catalytic activities of these
transition metal nitrides were evaluated using NO decompo-
sition as reaction probe. The activities of NO decomposition
were ranked in the following order: Co4N > Fe3N > Mo2N > VN
and Co4N/␥-Al2O3 > Fe3N/␥-Al2O3 > Mo2N/␥-Al2O3 > VN/␥-
Al2O3, respectively. The activities observed in this series of bulk
and alumina-supported metal nitrides studied here, were found
to correlate with the vacancy concentration. The metal nitrides
with higher vacancy concentration exhibited higher activities
for NO decomposition. We observed that the support effects of
metal nitrides on thermal stability were not negligible. Com-
pared with VN, Mo2N and Fe3N, Co4N was highly stabilized
by dispersing it on ␥-Al2O3 support. The influence of the crys-
tal sizes of these nitrides on their catalytic properties was not
significant, but the surface areas seemed to contributed to the
catalytic activities of these solids. In addition, the strong inter-
action between nitride phases and ␥-Al2O3 support might play a
role in NO decomposition. As indicated by XRD investigation,
the Co4N/␥-Al2O3 catalyst suffered from deactivation due to the
bulk oxidation of nitride. By the addition of H2 in feed gas at
the appropriate reaction temperature, the active Co4N/␥-Al2O3
phase can be retained to avoid heavy accumulation of surface
oxygen and deactivation of catalyst.
◦
as 950 C. These results might be responsible for the strong
interaction between metal nitride phases and ␥-Al2O3 support.
The strong interaction between Co4N phase and ␥-Al2O3 sup-
port was further investigated by XPS studies. As shown in
Table 2. The binding energies of Co 2p3/2 for Co4N/␥-Al2O3
◦
and Co4N/␥-Al2O3 (500 C pretreated) were 781.4 and 781.6,
respectively, whichwerehigherthanthatofCo4N(781.1). Based
on the XRD results, there were no lines that could be assigned
to cobalt oxide(s) or other nitrides, indicating that the Co4N on
␥
-Al2O3 support was phase-pure (Fig. 2). Therefore, it might be
δ+
due to the interaction of Co species with ␥-Al2O3 on the sur-
face that the Co 2p3/2 levels of alumina-supported cobalt nitrides
shifted to higher binding energy value. Taking into account the
results of Table 1, the effects of ␥-Al2O3 support on the crys-
tal sizes of the metal nitrides were negligible, but the supported
nitrides showed much higher surface areas than those of unsup-
ported ones. Based on the characterizations of XRD, BET and
TPD, we deduced that the supported metal nitrides performed
much better than their bulk counterparts for NO decomposition,
owing to high surface areas and the strong interaction between
metal nitride phases and ␥-Al2O3 support.
According to earlier reports [5–7], the deactivation of tran-
sition metal nitride (VN, Mo2N and W2N) catalysts for NO
decomposition and reduction with H2 was due to the bulk
oxidation of nitrides to oxides. For the present study, The
Co4N/␥-Al2O3 showedthesuperiorityinactivityforNOdecom-
position among this series of bulk and alumina-supported
nitrides of metals (V, Mo, Fe and Co). We focused on the explo-
ration of Co4N/␥-Al2O3 as a catalyst for NO decomposition
and reduction with hydrogen. It has been shown in Fig. 8 that
NO was converted to N2 (100%) in the absence of H2 over

4
96
Z. Yao et al. / Journal of Alloys and Compounds 464 (2008) 488–496
[20] M. Nagai, Y. Goto, A. Miyata, M. Kiyoshi, K. Hada, K. Oshikawa, S. Omi,
Acknowledgements
J. Catal. 182 (1999) 292–301.
[
[
[
[
21] M. Nagai, Appl. Catal. A 322 (2007) 178–190.
22] J. Choi, J. Catal. 182 (1999) 104–116.
23] K. Hada, M. Nagai, S. Omi, J. Phys. Chem. B 104 (2000) 2090–2098.
24] I.K. Milad, K.J. Smith, P.C. Wong, K.A.R. Mitchell, Catal. Lett. 52 (1998)
The work was supported by the National Natural Science
Foundation of China (No. 20573014) and the Natural Science
Foundation of Liao-ning Province (20041072).
113–119.
[
[
[
25] R.S. Ningthoujam, N.S. Gajbhiye, Mater. Res. Bull. 43 (2008) 1079–1085.
26] N.S. Gajbhiye, R.S. Ningthoujam, Mater. Res. Bull. 41 (2006) 1612–1621.
27] M. Nagai, Y. Goto, A. Miyoshi, K. Hada, K. Oshikawa, S. Omi, J. Catal.
References
[
[
[
[
[
[
[
[
1] R. Kojima, K.-I. Aika, Appl. Catal. A 219 (2001) 141–147.
2] R. Kojima, K.-I. Aika, Appl. Catal. A 219 (2001) 157–170.
3] C.J.H. Jacobsen, Chem. Commun. (2000) 1057–1058.
4] G.S. Ranhotra, A.T. Bell, J.A. Reimer, J. Catal. 108 (1987) 40–49.
5] H. He, H.X. Dai, K.Y. Ngan, C.T. Au, Catal. Lett. 71 (2001) 147–153.
6] C. Shi, X.F. Yang, A.M. Zhu, C.T. Au, Catal. Today 93–95 (2004) 819–826.
7] C. Shi, A.M. Zhu, X.F. Yang, C.T. Au, Appl. Catal. A 276 (2004) 223–230.
8] Z.W. Yao, A.M. Zhu, J. Chen, X.K. Wang, C.T. Au, C. Shi, J. Solid State
Chem. 180 (2007) 2635–2640.
182 (1999) 292–301.
[
[
28] K. Hada, M. Nagai, S. Omi, J. Phys. Chem. B 105 (2001) 4084–4093.
29] G. Jacobs, J.A. Chaney, P.M. Patterson, T.K. Das, B.H. Davis, Appl. Catal.
A 264 (2004) 203–212.
[
30] O. Diwald, T.L. Thompson, T. Zubkov, E.G. Goralski, S.D. Walck, J.T.
Yates Jr., J. Phys. Chem. B 108 (2004) 6004–6008.
31] Z.P. Wang, W.M. Cai, X.T. Hong, X.L. Zhao, F. Xu, C.G. Cai, Appl. Catal.
B 57 (2005) 223–231.
32] C.W. Colling, L.T. Thompson, J. Catal. 146 (1994) 193–203.
33] M.K. Neylon, S. Choi, H. Kwon, K.E. Curry, L.T. Thompson, Appl. Catal.
A 183 (1999) 253–263.
[
[
[
[
9] C. Shi, A.M. Zhu, X.F. Yang, C.T. Au, Appl. Catal. A 293 (2005) 83–90.
[
[
[
10] J.A.J. Rodrigues, G.M. Cruz, G. Bugli, M. Boudart, G. Djega-Mariadassou,
Catal. Lett. 45 (1997) 1–3.
11] X.W. Chen, T. Zhang, M.Y. Zheng, Z.L. Wu, W.C. Wu, C. Li, J. Catal. 224
[34] A. Martorana, G. Deganello, A. Longo, A. Prestianni, L. Liotta, A.
Macaluso, G. Pantaleo, A. Balerna, S. Mobilio, J. Solid State Chem. 177
(
2004) 473–478.
(
2004) 1268–1275.
12] X.W. Chen, T. Zhang, C. Li, L.G. Xia, T. Li, M.Y. Zheng, Z.L. Wu, X.D.
Wang, Z.B. Wei, Q. Xin, Catal. Lett. 79 (2002) 21–25.
[
[
[
35] R.T. Yang, N. Chen, Ind. Eng. Chem. Res. 33 (1994) 825–831.
36] L. Hultman, Vacuum 57 (2000) 1–30.
37] B. Eck, R. Dronskowski, M. Takahashi, S. Kikkawa, J. Mater. Chem. 9
[
[
[
[
13] E. Furimsky, Appl. Catal. A 240 (2003) 1–28.
14] C.C. Yu, S. Ramanathan, S.T. Oyama, J. Catal. 173 (1998) 1–9.
15] S. Ramanathan, C.C. Yu, S.T. Oyama, J. Catal. 173 (1998) 10–16.
16] J.S. Fang, L.C. Yang, C.S. Hsu, G.S. Chen, Y.W. Lin, G.S. Chen, J. Vac.
Sci. Technol. A 22 (2004) 698–704.
(
1999) 1527–1537.
38] M. Sifkovits, H. Smolinski, S. Hellwing, W. Weber, J. Magn. Magn. Mater.
04 (1999) 191–198.
[
2
[
[
39] Y. Wu, Z. Zhao, Y. Liu, X. Yang, J. Mol. Catal. A 155 (2000) 89–100.
40] J.X. Wang, S.F. Ji, J. Yang, Q.L. Zhu, S.B. Li, Catal. Commun. 6 (2005)
[
[
[
17] R. Kojima, K. Aika, Appl. Catal. A 209 (2001) 317–325.
18] J. Choi, J. Ha, J. Hong, Appl. Catal. A 168 (1998) 47–56.
19] M.Y. Zheng, X.W. Chen, R.H. Cheng, N. Li, J. Sun, X.D. Wang, T. Zhang,
Catal. Commun. 7 (2006) 187–191.
389–393.