Synthesis of transition-metal phosphides from oxidic precursors by reduction in hydrogen plasma

Article abstract of DOI:10.1016/j.jssc.2009.03.026

A series of transition metal phosphides, including MoP, WP, CoP, Co₂P, and Ni₂P, were synthesized from their oxidic precursors by means of hydrogen plasma reduction under mild conditions. The effects of reduction conditions, such as

Full text of DOI:10.1016/j.jssc.2009.03.026

ARTICLE IN PRESS
Journal of Solid State Chemistry 182 (2009) 1550–1555
Contents lists available at ScienceDirect
Journal of Solid State Chemistry
journal homepage: www.elsevier.com/locate/jssc
Synthesis of transition-metal phosphides from oxidic precursors by reduction
in hydrogen plasma
Ã
Jie Guan a, Yao Wang b, Minglei Qin a, Ying Yang a, Xiang Li a,b, Anjie Wang a,b,
a Department of Catalytic Chemistry and Engineering, State Key Laboratory of Fine Chemicals, Dalian University of Technology, Dalian 116012, PR China
b Liaoning Key Laboratory of Petrochemical Technology and Equipments, Dalian University of Technology, Dalian 116012, PR China
a r t i c l e i n f o
a b s t r a c t
Article history:
A series of transition metal phosphides, including MoP, WP, CoP, Co2P, and Ni2P, were synthesized from
their oxidic precursors by means of hydrogen plasma reduction under mild conditions. The effects of
reduction conditions, such as metal to phosphorus molar ratio, power input, and reduction time, on the
synthesis of metal phosphides were investigated. The products were identified by means of XRD
characterization. It is indicated that metal phosphides were readily synthesized stoichiometrically from
their oxides in hydrogen plasma under mild conditions.
Received 16 December 2008
Received in revised form
9 March 2009
Accepted 28 March 2009
Available online 8 April 2009
Keywords:
& 2009 Elsevier Inc. All rights reserved.
Metal phosphide
Synthesis
Hydrogen plasma
Reduction
Non-thermal
1. Introduction
dure, metal phosphides can only be obtained at a low heating rate
and a high H2 flow velocity so as to keep very low vapor pressure
on the surface. Another disadvantage of the conventional method
is that excess phosphorus is needed because phosphorus loss
occurs due to the migration of volatile phosphorus species in the
long time and high temperature reduction.
Recently, we reported a new approach to synthesizing metal
phosphides, i.e. reduction from the oxidic precursors in H2 plasma
as a communication [9]. In this approach, metal phosphides can
be obtained in H2 plasma under mild conditions in short time. In
the present work, crystalline WP, MoP, CoP, Co2P, Ni2P were
synthesized in hydrogen plasma, and the effects of reduction
conditions were investigated.
Transition-metal phosphides possess unique physical and
chemical properties which make them attractive and promising
in the fields of catalysis, electronics, optoelectronics, and magnetic
applications [1,2]. There are a variety of methods for synthesizing
metal phosphides, including direct reaction of metals and
phosphorus for prolonged periods at high temperature, reaction
of phosphine with metals or metal oxides, thermal decomposition
of single-source precursors, reduction of metal phosphates by
hydrogen, electrolysis of molten metal phosphate salts, solid-state
metathesis, and self-propagation high-temperature synthesis
[3–5]. Nevertheless, these methods usually require extremely
harsh conditions (high temperature), toxic reagents or very long
synthesis time. In addition, most of these methods are not suitable
for preparing the supported metal phosphides as catalysts.
Metal phosphides show high catalytic activities in hydrogena-
tion [6], hydrodesulfurization (HDS) [3], and hydrodenitrogena-
tion (HDN) [7]. The conventional synthesis procedure for metal
phosphide catalysts is temperature-programmed reduction in
hydrogen. However, it is neither thermodynamically nor kineti-
cally favorable to transform oxide precursors to phosphides by
hydrogen reduction. It is therefore essential to perform the
reduction at high temperature and low water vapor pressure [8].
In the conventional temperature-programmed reduction proce-
2. Experimental
All the reagents, including Ni(NO3)2 ꢀ 6H2O, Co(NO3)2 ꢀ 6H2O,
(NH4)2HPO4, (NH4)6Mo7O24 ꢀ 4H2O, (NH4)6W12O39, were of AR
grade. Hydrogen, with a purity of 99.99%, was supplied by
Guangming Gas Co. (Dalian, China).
The oxidic precursors were prepared by co-precipitation from
metal salts and (NH4)2HPO4 by adjusting the pH of the aqueous
solution, followed by drying at 120 1C for 12 h, and temperature-
programmed calcination in air at 500 1C for 3 h (heating rate
10 1C minꢁ1 ). The oxidic precursors were pelleted, crushed, and
sieved to 20–40 meshes.
Ã
Corresponding author. School of Chemical Engineering, Dalian University of
The synthesis of metal phosphides by hydrogen plasma reduction
was conducted in a dielectric barrier discharge (DBD) reactor
Technology, Dalian 116012, PR China. Fax: +86 41188993693.
E-mail address: ajwang@dlut.edu.cn (A. Wang).
0022-4596/$ - see front matter & 2009 Elsevier Inc. All rights reserved.
doi:10.1016/j.jssc.2009.03.026

ARTICLE IN PRESS
J. Guan et al. / Journal of Solid State Chemistry 182 (2009) 1550–1555
1551
consisting of a quartz tube and two electrodes. Configuration of the
DBD discharge reactor was shown in Fig. 1. The high-voltage
electrode was a stainless-steel rod with a diameter of 2.5 mm,
which was installed along the axis of the quartz tube and connected
to an alternating current supply. The grounding electrode was an
aluminum foil, which was wrapped around the quartz tube and
linked to ground by a wire. For each run, 0.8 g precursor was charged
into the quartz tube. Hydrogen from a gas cylinder was allowed
to pass through the bed at 150 ml minꢁ1 to flush air out of the
reactor. At a constant H2 flow, a high voltage of 9–11 kV, varying with
the generator input voltage (40ꢁ90 V), was supplied by a plasma
generator. The discharge frequency was tuned to be around 10 kHz.
The total input was measured by the product of current and voltage
of the generator input. Due to the high voltage, it is impossible to
measure the bed temperature during plasma discharge. Our pre-
vious measurement by means of an infrared imager (ThermoVision
A40M) [9] showed that the temperature rise in the bed was not
significant. In a typical reduction, the input voltage of the generator
was kept at a specific value (45ꢁ90 V) for desired time (15–240 min).
After reduction, the plasma generator was turned off, and the reactor
was allowed to cool down to room temperature in H2 flow. Then, the
obtained material was passivated in a 10 ml minꢁ1 flow of 0.5 mol%
O2 in He for 2 h at room temperature [9].
plasma reduction [9]. Because Ni2P is more important in
hydrotreating catalysis, the effects of reduction conditions on
the formation of Ni2P were investigated systematically. Fig. 2
shows the XRD patterns of the materials obtained from Ni and P
oxides by H2 plasma reduction at various power input. The power
was tuned by adjusting the input voltage of the plasma generator.
At 45 V for 60 min, only Ni12P5 was obtained. Increasing the
voltage or power input led to the transformation of Ni12P5 to Ni2P.
At 60 V ꢂ 0.40 A, a pure Ni2P phase was obtained. HRTEM images
in Fig. 3 indicate that crystalline Ni2P particles of 100–150 nm
were synthesized from the oxidic precursor by H2 plasma
reduction.
Reduction time is another important factor determining
the formation of crystal phases in the reduction. As seen in
Fig. 2, a mixture of Ni2P and Ni12P5 were obtained at 50 V for
120 min. When the reduction time increased to over 120 min,
high crystalline Ni2P particles were synthesized at 50 V
(Fig. 4). It is interesting to note that Ni12P5 phase formed in
15 min.
The atomic ratios of Ni2P and Ni3P as well as their oxidic
precursors (Table 1) indicate that there was little difference
between metal phosphides and their oxidic precursors, suggesting
that no loss of phosphorus occurred in the plasma reduction
procedure.
The crystal phases of the synthesized materials were character-
ized by XRD patterns, which were measured on a Rigaku D/Max
2400 diffractometer with nickel-filtered CuKa radiation at 40 kV and
3.2. Cobalt phosphides
100 mA. HRTEM images of Ni2P were taken on a Philips CM200 FEG
instrument. The atomic ratios of metal to phosphorus in the oxidic
precursor and the synthesized metal phosphide were determined by
means of ICP-OES system (Optima 2000DV, Perkin Elmer).
Recently, Autumn et al. reported cobalt phosphides also
showed fairly high HDS activity [10]. In addition, CoP can be used
as a potential contact material to InP for application in electronic
devices, owing to its metallic nature [11]. It is also a potential
candidate in thermoelectric applications [12].
3. Results
Fig. 5 shows the XRD patterns of the synthesized materials
from Co and P oxides with atomic ratio of 1 by H2 plasma
reduction at various power input for 60 min. A mixture of cobalt
phosphate and CoP formed at low power input. With increasing
power input, the portion of CoP phase was increased. Pure CoP
crystals were obtained at 90 V for 60 min. Alternatively, increasing
reduction time at lower power input, such as 80 V ꢂ 0.30 A , also
led to the formation of pure CoP crystal phase (Fig. 6). Compared
3.1. Nickel phoshpide
Our previous study showed that Ni2P and Ni3P could be
synthesized stoichiometrically from their oxide precursors by H2
Ni2P (PDF 74-1385)
60V × 0.40A
55V × 0.38A
50V × 0.44A
45V × 0.42A
Ni12P5 (PDF 22-1190)
10
20
30
40
50
60
70
80
2 theta / o
Fig. 1. The configuration of the DBD reactor: (1) high-voltage discharge electrode;
(2) grounding electrode; (3) ground; (4) quartz tube; (5) insulator; (6) solid
precursor bed; (7) gas inlet; (8) gas outlet.
Fig. 2. XRD patterns of the synthesized materials by H2 plasma reduction at
various power input for 120 min from Ni and P oxides with Ni/P atomic ratio of 2.

ARTICLE IN PRESS
1552
J. Guan et al. / Journal of Solid State Chemistry 182 (2009) 1550–1555
Fig. 3. HRTEM images of Ni2P particles synthesized by H2 plasma reduction.
CoP (PDF 29-0497)
90V × 0.28A
Ni2P (PDF 74-1385)
240 min
120 min
80V × 0.30A
70V × 0.30A
60 min
15 min
0 min
Precursor
Ni12P5 (PDF 22-1190)
a-Co2P2O7 (PDF 39-0709)
10
20
30
40
50
60
70
80
2 theta/o
10
20
30
40
50
60
70
80
Fig. 4. XRD patterns of the materials obtained at different time from Ni and P
oxides with Ni/P atomic ratio of 2 by H2 plasma reduction at 50 V ꢂ 0.44 A input.
2 theta/o
Fig. 5. XRD patterns of the materials synthesized from Co and P oxides with
atomic Co/P ratio of 1 by H2 plasma reduction for 60 min.
Table 1
Atomic ratio of metal to phosphorus in precursors and metal phosphides.
3.3. Molybdenum phosphide and tungsten phosphide
Oxidic precursor
Metal phosphide
In our previous study, we found that the H2 plasma reduction
method was also applicable in the synthesis of MoP and WP. In the
present work, the effects of the reduction conditions were
investigated in more detail.
Ni2P
Ni3P
CoP
2.06
3.08
0.98
1.90
2.07
3.05
0.98
1.92
Co2P
Fig. 8a shows the XRD patterns of the materials synthesized
from Mo and P oxides with atomic Mo/P ratio of 1 at different
power input for 60 min. It is indicated that the synthesis of MoP is
easier than that of CoP, but more difficult than that of Ni2P. In the
synthesis of MoP, no intermediate phase appeared at lower power
input. The transformation of W and P oxides to WP is similar to
that of Mo–P oxides to MoP (Fig. 8b) in that no intermediate phase
was observed.
From Fig. 9, it can be seen that pure crystalline MoP and WP
were obtained when the reduction time was prolonged. Fig. 10
indicates that highly crystalline MoP or WP could be obtained
only stoichiometrically from the oxidic precursor. An excess of
with the synthesis of Ni2P, more severe reduction conditions are
necessary to obtain pure CoP crystals.
Fig. 7 illustrates the XRD patterns of the synthesized materials
from different atomic Co/P ratio. Similar to the synthesis of
nickel phosphides [9], pure CoP and Co2P crystals were obtained
stoichiometrically from their oxidic precursors with Co/P ratio
of 1 and 2, respectively. The atomic ratios of CoP and Co2P
as well as their oxidic precursors (Table 1) indicate that there was
little difference between metal phosphides and their oxidic
precursors.

ARTICLE IN PRESS
J. Guan et al. / Journal of Solid State Chemistry 182 (2009) 1550–1555
1553
metal or phosphorus in the precursor led to lower crystallinity of
MoP or WP.
CoP (PDF 29-0497)
4. Discussion
180 min
120 min
The synthesis of metal phosphides from their oxidic precursors
by H2 reduction could be expressed as the following equation:
zMOxðsÞ þ POyðsÞ þ ðnx þ yÞH ðgÞ sÐlow MzPðsÞ þ ðnx þ yÞH OðgÞ
2
2
very fast
(1)
60 min
Transition-metal phosphides, such as MoP, WP, Ni2P, CoP, and
Co2P, react with H2O readily, leading to severe structure destruc-
tion of the formed phosphide crystal [13,14]. It is therefore
essential to remove water, which is the by-product of the
reduction, from the surface of the metal phosphide particles in
the course of reduction. In conventional temperature-pro-
grammed reduction procedure, a very low heating rate (i.e.
1 1C minꢁ1) and a high H2 flow rate are used to keep low partial
pressure of moisture.
In the conventional temperature-programmed reduction
method, the reaction (1) usually takes place at elevated tempera-
tures because there exist threshold temperatures for the synthesis
of metal phosphides, such as Ni2P, WP and MoP [15–17]. This is
probably due to the low reduction ability of molecular hydrogen.
Because of the high temperature and low heating rate in the
temperature-programmed reduction, the synthesis of metal
phosphides generally takes more than 6 h. In H2 plasma reduction,
Ni2P could be obtained in 60 min at power input of 60 V ꢂ 0.40 A
[9]. Although more harsh conditions were needed to synthesize
CoP, Co2P, MoP, and WP, the reduction times were dramatically
shortened, compared with the conventional TPR method (Figs. 6
and 9).
a-Co2P2 O7 (PDF 39-0709)
10
20
30
40
2 theta
50
60
70
80
o
/
Fig. 6. XRD patterns of the synthesized materials from Co and P oxides with
atomic Co/P ratio of 1 by H2 plasma reduction at 80 V ꢂ 0.30 A for different
reduction time.
Co/P=1
CoP (PDF 29-0497)
In H2 plasma, high-energy electrons generated by plasma
collide inelastically with hydrogen molecules, which lead to
the production of excited hydrogen species and ions with a
dramatically enhanced reduction ability [18]. These active
species can promptly reduce the oxidic precursors whereas the
plasma helps to remove the by-product water from the surface
of metal phosphides. Consequently, the forward reaction in
Eq. (1) will be significantly accelerated under plasma irradiation,
leading to a quick formation of metal phosphides under mild
conditions.
Co/P=2
Co2P PDF 32-0306
10
20
30
40
50
60
70
80
2 theta/ o
Fig. 7. XRD patterns of the synthesized materials from Co and P oxides with
atomic Co/P ratio of 1 and 2 by H2 plasma reduction 90 V ꢂ 0.28 A for 60 min.
WP (PDF 29-1364)
90V × 0.30A
MoP (PDF 24-0771)
90V ×0.30A
80V × 0.30A
80V × 0.30A
70V × 0.30A
60V × 0.40A
70V × 0.30A
Precursor
10
20
30
40
50
60
70
80
10
20
30
40
50
60
70
80
o
2 theta / o
2 theta /
Fig. 8. XRD patterns of the synthesized materials from Mo and P oxides (Mo/P ¼ 1) (a) and W and P oxides (W/P ¼ 1) (b) by H2 plasma reduction at different power input.

ARTICLE IN PRESS
1554
J. Guan et al. / Journal of Solid State Chemistry 182 (2009) 1550–1555
WP(PDF29-1364)
120 min
MoP(PDF24-0771)
240 min
120 min
60 min
30 min
0 min
60 min
30 min
10
20
30
40
2 theta / o
50
60
70
80
10
20
30
40
2 theta / o
50
60
70
80
Fig. 9. XRD patterns of the synthesized materials from Mo and P oxides (Mo/P ¼ 1) at 70 V ꢂ 0.30 A (a) and from W and P oxides (W/P ¼ 1) at 80 V ꢂ 0.30 A (b) by H2 plasma
reduction for different reduction time.
WP(PDF29-1364)
MoP(PDF24-0771)
W/P=1/2
Mo/P=1/2
Mo/P=1
W/P=1
Mo/P=2
W/P=2
10
20
30
40
50
60
70
80
10
20
30
40
50
60
70
80
o
/
2 theta/o
2 theta
Fig. 10. XRD patterns of the synthesized materials from Mo–P oxides and W-P oxides with various Mo/P and W/P atomic ratios by H2 plasma reduction at 70 V ꢂ 0.30 A for
120 min.
Two solid reactants and a gaseous hydrogen molecule are
involved in the forward reaction in Eq. (1). One of the reactant or
its derivatives must migrate to meet the other reactant before the
reaction can proceed. It is generally believed that the metal oxide
is reduced to metal and then phosphorus species are reduced to
volatile P or PH3 [3]. Carenco et al recently reported the synthesis
of Ni2P nanoparticles from Ni(0) complexes and P [19]. Yang et al.
synthesized Ni2P from phosphidation of Ni by PH3 [4]. The volatile
species migrate to meet metal particles so as to form metal
phosphide. At elevated temperature and high velocity of hydrogen
flow, part of the volatile species may be flushed off the solid
surface, leading to phosphorus loss in the course of high
temperature reduction. It is therefore essential that excess
phosphorus is present in the precursors to compensate the loss
in the conventional TPR synthesis.
Our previous study showed that Ni2P and Ni3P could be
synthesized stoichiometrically from their oxidic precursors [9]. In
the present study, crystalline CoP and Co2P were obtained
stoichiometrically from their oxidic precursors (Fig. 6). Fig. 10
illustrates that highly crystalline MoP and WP were synthesized
only at stoichiometric atomic Mo/P and W/P ratios. In the excess
of metal or phosphorus, poor yield of MoP or WP was obtained. All
these results indicate that excess phosphorus is not necessary in
H2 plasma reduction. This is probably due to the low temperature,
which decreases the volatility of phosphorus species, suppressing
the loss of phosphorus species.
According to the TPR profiles in the literatures [16,20], the
main H2 consumption peak temperatures of the phosphide
precursors increased in the following order: Ni2PoMoPo
CoPoWP, in accordance with the threshold conditions in the

ARTICLE IN PRESS
J. Guan et al. / Journal of Solid State Chemistry 182 (2009) 1550–1555
1555
synthesis of the metal phosphides. It is therefore suggested that
References
hydrogen plasma reduction does not change the transformation
mechanism but enhance the forward reaction in Eq. (1), due to the
generation of highly reactive hydrogen species and to the surface
cleaning effects of plasma irradiation.
[1] C.M. Lukehart, S.B. Milne, S.R. Stock, Chem. Mater. 10 (1998) 903–908.
[2] S.L. Brock, K. Senevirathne, J. Solid State Chem. 18 (2008) 1552–1559.
[3] S.T. Oyama, J. Catal. 216 (2003) 343–352.
[4] S. Yang, C. Liang, R. Prins, J. Catal. 237 (2006) 118–130.
[5] S.L. Brock, K. Senevirathne, J. Solid State Chem. 181 (2008) 1552–1559.
[6] N.P. Sweeny, C.S. Rohrer, O.W. Brown, J. Am. Chem. Soc. 80 (1958) 799–800.
[7] M. Lu, A. Wang, X. Li, X. Duan, Y. Teng, Y. Wang, C. Song, Y. Hu, Energy Fuels 21
(2007) 554–560.
5. Conclusions
[8] V. Zuzaniuk, R. Prins, J. Catal. 219 (2003) 85–96.
[9] A. Wang, M. Qin, J. Guan, L. Wang, H. Guo, X. Li, Y. Wang, R. Prins, Y. Hu,
Angew. Chem. Int. Ed. 47 (2008) 6052–6054.
[10] A.W. Burns, K.A. Layman, D.H. Bale, M.E. Bussell, Appl. Catal. A. 343 (2008)
68–76.
[11] D. Swenson, Y.A. Chang, Mater. Sci. Eng. B 8 (1991) 225–230.
[12] R.C. DeMattei, A. Watcharapasorn, R.S. Feigelson, J. Electrochem. Soc. 148
(2001) D109–D111.
[13] D.C. Phillips, S.J. Sawhill, R. Self, M.E. Bussell, J. Catal. 207 (2002) 266–273.
[14] Z. Wu, F. Sun, W. Wu, Z. Feng, C. Liang, Z. Wei, C. Li, J. Catal. 222 (2004) 41–52.
[15] A. Wang, L. Ruan, Y. Teng, X. Li, M. Lu, J. Ren, Y. Wang, Y. Hu, J. Catal. 229
(2005) 314–321.
Our results demonstrated that H2 plasma reduction is a more
efficient approach to synthesizing metal phosphides than the
conventional TPR method. The high efficiency is probably related
with the highly reactive hydrogen species and the surface cleaning
effects of plasma irradiation. Moreover, all the metal phosphides,
including MoP, WP, CoP, Co2P, and Ni2P, were synthesized
stoichiometrically from their oxidic precursors.
[16] P. Clark, X. Wang, S.T. Oyama, J. Catal. 207 (2002) 256–265.
[17] P.A. Clark, S.T. Oyama, J. Catal. 218 (2003) 78–87.
Acknowledgments
[18] Y. Zhang, W. Ding, S. Guo, K. Xu, Chin. J. Nonferrous Met. 14 (2004) 315–321.
The authors acknowledge the financial supports from the NSFC
(20333030, 20503003, and 20773020), the NCET, and the 111
Project. We thank Prof. R. Prins for helpful discussions.
´
[19] S. Carenco, I. Resa, X. Le Goff, P. Le Floch, N. Mezailles, Chem. Commun. (2008)
2568–2570.
[20] X. Wang, P. Clark, S.T. Oyama, J. Catal. 208 (2002) 321–331.