Effect of hydrogen annealing on the composition and particle size of molybdenum nanopowders

Article abstract of DOI:10.1134/S002016850912005X

We have studied the behavior of molybdenum nanopowder consisting of particles with an average size of 23.7 nm. The powder was prepared by hydrogen plasma reduction of molybdenum trioxide and contained considerable amounts of oxygen in the form of amorphou

Full text of DOI:10.1134/S002016850912005X

ISSN 0020-1685, Inorganic Materials, 2009, Vol. 45, No. 12, pp. 1337–1341. © Pleiades Publishing, Ltd., 2009.
Original Russian Text © Yu.V. Levinsky, Yu.V. Blagoveshchenskii, G.M. Vol’dman, N.V. Kobzarev, 2009, published in Neorganicheskie Materialy, 2009, Vol. 45, No. 12,
pp. 1431−1435.
Effect of Hydrogen Annealing on the Composition
and Particle Size of Molybdenum Nanopowders
Yu. V. Levinskya, Yu. V. Blagoveshchenskiib, G. M. Vol’dmana, and N. V. Kobzareva
a Lomonosov State Academy of Fine Chemical Technology, pr. Vernadskogo 86, Moscow, 117571 Russia
b Baikov Institute of Metallurgy and Materials Science, Russian Academy of Sciences,
Leninskii pr. 49, Moscow, 119991 Russia
e-mail: levinsky@hotbox.ru
Received November 20, 2007
Abstract—We have studied the behavior of molybdenum nanopowder consisting of particles with an average
size of 23.7 nm. The powder was prepared by hydrogen plasma reduction of molybdenum trioxide and con-
tained considerable amounts of oxygen in the form of amorphous molybdenum oxides and hydroxides. Anneal-
ing in hydrogen for 1 h at temperatures from 300 to 1000°C is shown to cause crystallization of oxides, which
then vaporize to form the volatile hydroxide MoO2(OH)2. Most of the oxygen is removed by annealing below
700°ë. Starting at this temperature, particle growth is observed. The main mechanism behind the coagulation
of molybdenum particles is vapor transport. Annealing at 1000°C for 30–50 min in hydrogen with a dew point
of –5°C increases the particle size by about one order of magnitude.
DOI: 10.1134/S002016850912005X
INTRODUCTION
The nondimensional thermodynamic particle-size
criterion DL was calculated by the formula
A widespread approach to producing nanopowders
of refractory metals is the reduction of their oxides in a
low-temperature hydrogen plasma [1]. The powders
thus obtained contain considerable levels of oxygen.
The oxygen can be eliminated—or at least its content
can be reduced to an acceptable level—by annealing in
a reducing atmosphere. This process, however, leads to
coagulation of the metal nanoparticles.
DL = 1.2 × 10–11σ/(Gb2d)
(0.1)
where σ is the surface energy, G is the shear modulus,
and b is the Burgers vector [2]. The particle-size crite-
rion DL for the as-prepared powder was found to be
11.8. Anneals were carried out for 5–60 min at a fixed
temperature from 300 to 1000°C.
The purpose of this work was to study the behavior of
molybdenum nanopowders during hydrogen annealing.
RESULTS AND DISCUSSION
The annealing conditions and analysis results for the
annealed samples are summarized in Table 1. XRD
examination showed that the only crystalline phase in
these samples was metallic molybdenum. This implies
EXPERIMENTAL PROCEDURE
AND CALCULATIONAL APPROACH
The starting powder, prepared by plasma reduction that oxygen was present in the powder either as
of molybdenum trioxide [1], consisted of equiaxed par- adsorbed species or in X-ray amorphous phases.
ticles and contained 15.0 wt % oxygen. The specific
surface area S of the powder before and after storage
was 24.8 3 m2/g, which corresponded to an average
particle size d= 23.7 nm.
In addition to about fifteen stable and metastable
oxides [3], the system in question contains the
åÓé2(éç)2 hydroxide, which has the highest vapor
pressure among the phases of this system [4].
Loose powder samples were annealed in flowing
hydrogen with a dew point of –5°ë. The specific sur-
face area of the powder was determined by low-temper-
ature nitrogen adsorption measurements. The average
particle size was evaluated as d = 6/(γS) (where γ is the
density of molybdenum. Oxygen content was deter-
mined on a LECO TC-436 gas analyzer. X-ray diffrac-
tion (XRD) characterization was performed with CuKα
Analysis of the data in Table 1 leads us to propose
the following sequence of processes that take place dur-
ing annealing of the as-prepared powder in flowing
hydrogen: Heating from room temperature to 600–
700°C leads to simultaneous vaporization of amor-
phous oxides and hydroxides, their crystallization, and
reduction with hydrogen. In this process, the particle
size remains essentially unchanged. At higher tempera-
radiation. Weight changes were monitored with an ana- tures, particles of almost pure molybdenum coagulate.
lytical balance to an accuracy of 1 mg. This sequence is supported by the observed changes in
1337

1338
LEVINSKY et al.
Table 1. Effect of hydrogen annealing on the properties of molybdenum nanoparticles
Annealing conditions
Phase com-
position
Weight loss, % S, m2/g dmeas, nm DL
wt % oxygen
d
, nm
calc
temperature, °C
time, min
As-prepared
After storage
300
–
–
24.8
24.8
30.4
30.2
3
3
3
3
23.7 11.8
23.7 11.8
19.3 14.6
19.5 14.4
15.0
17.7
–
–
–
–
Mo, Mo2O3
Mo, Mo2O3
Mo, Mo2O3
Mo, Mo2O3
Mo, Mo2O3
Mo
60
60
60
60
60
15
30
60
5
5.51
5.93
7.17
11.25
–
400
–
500
15.8 1.8
18.6 2.6
28.2 2.6
23.0 2.3
11.8 1.1
10.0 0.9
–
37.2
31.6
7.58
8.92
–
600
–
700
20.9 13.5
25.6 11.0
4.87
800
–
–
60
74
93 Mo
–
800
–
49.9
58.8
–
5.65
–
–
800
–
4.79
–
2.58
900
29.24
28.15
–
–
–
–
900
10
15
30
30
60
60
30
5
–
–
–
–
900
9.9 0.9
59.4
4.74
2.68
4.50
3.54
–
137
–
–
–
900
–
5.6 0.7 105.0
174
900
24.80
–
9.4 1.1
62.6
79.5
–
174
900
7.4 0.8
1.24
218 Mo
900*
900**
1000
1000
1000
1000
1000
1000
18.74
5.22
22.58
21.33
–
–
–
–
–
–
–
Mo
–
–
–
–
–
–
–
–
–
–
–
–
–
–
10
15
15
30
60
–
–
–
–
2.5 0.2 235.3
4.3 0.5 136.8
3.5 0.3 168.1
3.2 0.5 183.8
1.19
2.06
1.67
1.53
–
282
282
355
–
–
–
–
–
0.76
447 Mo
Note:
d
and d
are the average particle sizes, and D is the particle-size criterion.
calc L
meas
*Powder compact with an initial relative density of 20%.
**Annealing in helium.
sample weight, phase composition, specific surface
area, and oxygen content.
After heating at 200–600°C for 1 h, the samples con-
sisted of two crystalline phases: Mo and åÓ2é3. The
contents of these phases, evaluated from the intensity of
XRD peaks from molybdenum (d = 2.22 Å) and åÓ2é3
(d = 3.44 Å), are listed in Table 2. At annealing temper-
atures of 700°C and above, the samples consisted
entirely of Mo. The observed weight loss was due to the
removal of volatile species from the powder.
Only two crystalline phases were identified in our
samples by XRD: molybdenum and åÓ2é3. Note that
åÓ2é3 was not detected in earlier structural studies of
molybdenum oxides prepared by a method other than
plasma reduction [4]. It is, however, completely identi-
cal to the åÓ2é3 obtained by a plasma process and
described by Tsvetkov et al. [1]. It seems likely that this
oxide is a metastable form of åÓ4é11 [4].
Control experiments with loose powder samples and
with the powder compacted to a relative density of 20%
(marked by an asterisk in Table 1) showed approxi-
mately the same weight loss. Raising the temperature
from 300 to 600°C (1-h anneals) leads to a systematic
increase in weight loss, which nevertheless remains
below the initial oxygen content of the powder. Given
that it is in this temperature range that the amorphous
oxides and hydroxides in åÓ2é3 crystallize, we think
that the weight loss is due to the removal of the water of
Even though the as-prepared powder contained
17.7 wt % oxygen and consisted of fine particles, its
XRD pattern showed sharp, narrow peaks from only
one phase, molybdenum, with very weak, broad reflec-
tions from åÓ2é3. Therefore, nearly all of the bound
oxygen in the powder was present in the form of amor-
phous oxides and hydroxides.
INORGANIC MATERIALS Vol. 45 No. 12 2009

EFFECT OF HYDROGEN ANNEALING ON THE COMPOSITION
1339
Table 2. XRD results for the molybdenum nanopowders an-
hydration and hydrogen reduction of åÓ2é3. At tem-
peratures of 700–800°C and above, the weight loss cor-
responding to the removal of all the oxygen is 25–30%.
This considerably exceeds the initial oxygen content of
the powder. The reason for this is that the removal of the
water of hydration and the reduction of oxides are
accompanied by the vaporization of the oxides and the
volatile hydroxide åÓé2(éç)2.
Heating at 700°C for 1 h and even at 800°C for
15 min has little or no effect on the specific surface area
of the powder. Annealing at 800°C for a longer time or
at 900 and 1000°C for any of the annealing times tested
reduces the specific surface area as a result of particle
coagulation. The oxygen content of the powder drops
rapidly at annealing temperatures above 700°C, but
remains appreciable even after annealing at 1000°C,
which may be caused by partial oxidation of the com-
pletely reduced molybdenum during cooling in flowing
hydrogen containing trace levels of moisture.
nealed in hydrogen for 1 h
Annealing
temperature,
°C
Weight percent
IMo/I
Mo2O3
Mo
Mo2O3
As-prepared
200
50
100
76
Traces
24
26
25
35
23
0
1.5
300
1.4
1.5
0.92
1.75
∞
74
400
75
500
65
600
77
700
100
100
100
800
∞
∞
0
900
0
Note: IMo/IMo O is the intensity ratio of the Mo line with d = 2.22 Å
2
3
to the Mo O line with d = 3.44 Å.
2
3
Coagulation of molybdenum particles during hydro-
gen annealing at temperatures from 800 to 1000°C may
proceed through the growth of the contact necks
between the particles as a result of volume and surface
diffusion and vapor transport. The time needed for par-
ticle coalescence through volume and surface diffusion
was evaluated as [5]
It is convenient to consider coagulation kinetics
using the Livshits–Slezov equation in differential form:
σV C D
4
27
dr
0
∞
vol
-------------------------------
----- =
dt
,
(3)
r2RT
5
--
2
2RTr3
5σV0Dvol
where ë∞ is the equilibrium concentration of particles
with r ∞ in the host solution.
ΔL
⎛
⎝
⎞ ⎛
⎞
⎠
------ ----------------------
t =
,
(1)
(2)
⎠ ⎝
L0
In this equation, the term 2σV0/(rRT), which repre-
sents the solute concentration gradient, must be
replaced in the case of vapor transport by the product of
the gradient of the number density of particles and the
volume per particle, Va. Given that grad nV‡ =
gradpV‡/(kT), grad p = Δp/z, and Δp = 2σV‡p∞/(rkT), we
have
t = (r4RT)/(DsurfσV0δ).
Here, L is the center-to-center distance between neigh-
boring particles, L0 = 2r0 is the initial distance, V0 is the
molar volume, Dvol and Dsurf and the volume and sur-
face diffusion coefficients, R is the gas constant, í is the
absolute temperature, r is the particle radius, and δ is
the thickness of the surface diffusion layer.
2σV20P∞
2σV2a P∞
-------------------
grad(nVa) = ------------------- =
.
(4)
r2(RT)2
r2(kT)2
ΔL
2
------
Substituting
= 0.5, r = 12 nm, σ = 1.915 J/m ,
L0
It follows from Eqs. (3) and (4) that
V0 = 9.42 × 10–6 m3/mol, δ = 10–9 m, Dvol = 1.448 ×
σV20P∞
4
27
10−21 m2/s, and Dsurf = 100Dvol into Eqs. (1) and (2), we
find that at 100°C the coalescence time must be 13.8 h
for volume diffusion and 23.3 h for surface diffusion.
Both values far exceed those determined experimen-
tally. Therefore, the contribution of solid-state diffusion
to coagulation is negligible.
dr
------------------------
----- =
dt
2 Dt,
(5)
(6)
2
r (RT)
2
V0
RT
4
9
3
3
⎛
⎞
⎠
-- -------
r = r + σ
D„P∞t.
⎝
Here, Dg is the diffusion coefficient of atoms (mole-
cules) in the gas phase.
The main mechanism behind the coagulation of
molybdenum particles is vapor transport. At atmo-
spheric hydrogen pressure and a bulk density of 0.05,
the mean free path of molecules of volatile compounds
is much smaller than the interparticle distance in the
range of particle sizes studied. Therefore, the coagula-
tion process is mediated by the diffusion flow caused by
the difference between the pressure of the volatile
molybdenum compound over a flat surface and that
over particles of radius r.
The main point in evaluating the coagulation rate by
Eq. (6) is to determine the nature and p∞ of the atoms
(molecules) responsible for vapor transport. In the sys-
tem under consideration, mass transport in the vapor
phase may be due to molybdenum atoms and molecules
of various oxides and hydrates.
In the range 800–1000°C, the equilibrium molybde-
num vapor pressure is so low that substituting it in
INORGANIC MATERIALS Vol. 45 No. 12 2009

1340
LEVINSKY et al.
Eq. (6) gives values many orders of magnitude below
those observed in experiment. Therefore, in what fol-
lows the effect of transport via molybdenum vapor can
be neglected.
As shown by Jehn [4], the presence of solid molyb-
denum in the Mo–O–H system leads to the formation of
solid MoO2 (the lower molybdenum oxide) and
åÓé2(éç)2 vapor even at very small amounts of oxy-
At 800, 900, and 1000°C, the Dg of åÓO2(OH)2
molecules is 2.17 × 10–5, 2.48 × 10–5, and 2.80 × 10–5
m2/s, respectively.
Under the conditions of our experiments, pH and
2
pH O, needed to find pMoO (OH) by Eq. (8), may
2
2
2
depend on two factors. First, the starting hydrogen has
a certain humidity. In flowing hydrogen, to this humid-
ity corresponds a constant water vapor pressure over
the molybdenum powder. Second, the starting powder
contains oxygen, and its traces are present in the final
product. This suggests that, throughout a given experi-
ment, the water vapor pressure may be determined by
equilibrium in the hydrogen reduction of the lower
molybdenum oxide.
gen and, accordingly, a low
partial pressure and
pH O
2
small pH O/pH ratio. Vapor transport and growth of
2
2
powder particles proceed with the participation of
åÓé2(éç)2 molecules, which form by the reaction
åÓé2(s) + 2ç2é(g) = åÓé2(éç)2 (g) + ç2, (7)
p
p
MoO2(OH)2 H2
In the former case, the water vapor pressure must be
constant, independent of the annealing temperature and
duration. For hydrogen with a dew point of –5°C, the
water pressure is 4.162 × 10–3 Pa. Substituting this water
vapor pressure and a hydrogen pressure of 105 Pa into
Eq. (8) and using the temperature dependence of the
equilibrium constant of reaction (7), we find that
--------------------------------
Kp(7)
=
.
(8)
2
p
H2O
The temperature dependence of this constant was deter-
mined by Jehn [4].
In addition, the vapor phase contains a variety of
oxide species. åÓ3é9, one the most stable oxides, with
the highest partial pressure, may form by the reaction
pMoO (OH) p∞ at 800, 900, and 1000°C is 8.7 × 10–8,
2
2
7.8 × 10–7, and 4.9 × 10–6 Pa, respectively. Substituting
these p∞ values and the corresponding Dg values into
Eq. (6) gives an increase in the radius of molybdenum
particles below the experimentally determined value by
more than one order of magnitude.
3åÓ3é2(éç)2(g) = åÓ3é9(g) + 3ç2é.
(9)
The equilibrium constant of this reaction is such
that, even at low water partial pressures (103 Pa), the
åÓ3é9(g) partial pressure is three orders of magnitude
below the åÓ2(éç)2(g) partial pressure.
In the latter case, the hydrogen pressure is deter-
mined by equilibrium in the reaction
Volatile molybdenum oxides may also form through
direct reaction between condensed molybdenum and
oxygen impurities, which are always present in gases
nonreactive with molybdenum. As shown in our exper-
iments, the weight loss during annealing in helium
(marked by two asterisks in Table 1) is substantially
lower than that in hydrogen, and therefore the mass
transport is predominantly due to åÓé2(éç)2 mole-
cules.
0.5åÓé2(s) + ç2(g) = 0.5åÓ + ç2é(g).
(10)
The equilibrium constant of this reaction, K =
pH /pH O , as a function of temperature has the form [5]
2
2
1383.2
---------------
T
logKp = 0.8799 –
.
(11)
Combining this equation with (5) and taking into
account that the total pressure is 105 Pa, we find that
pMoO (OH) at 800, 900, and 1000°C is 5.5 × 10–4, 7.5 ×
10–3, and 6.8 × 10–2 Pa, respectively, which is about four
orders of magnitude higher in comparison with the
former case.
The gas-phase diffusion coefficient of åÓé2(éç)2
was evaluated using the well-known formula Dg =
uλ/3, where u is the average speed of diffusant mole-
cules in the gas phase (m/s):
2
2
8kT
u = --------- = ----------,
πm πM
8RT
The pMoO (OH) and Dg MoO (OH) values thus
2
2
2
2
kT
obtained were used to evaluate the increase in particle
------------------------------------------------------------
λ =
.
size by Eq. (6) with V0 = 2.02 × 10–5 m3/mol and σ =
2π(rMoO (OH) + rH )2 pΣ
2
2
2
1.0 J/m2. The calculated particle sizes of the annealed
samples dcalc are listed in Table 1. The experimentally
determined average particle sizes are of the same order
as the values calculated by Eq. (6). This provides solid
evidence that the main mechanism behind the coagula-
tion of molybdenum particles during hydrogen anneal-
ing at temperatures from 800 to 1000°C is mass trans-
port by åÓé2(éç)2 vapor.
Here m is the molecular mass of the diffusant (kg), å is
its molar mass (kg/mol), λ is the mean free path of dif-
fusant molecules (m), rMoO (OH) is the molecular
2
2
radius of åÓO2(OH)2 (3 × 10–10 m), rH is the molecu-
lar radius of hydrogen (1.35 × 10–10 m), and pΣ is the
total pressure in the system (1.013 × 105 Pa).
2
INORGANIC MATERIALS Vol. 45 No. 12 2009

EFFECT OF HYDROGEN ANNEALING ON THE COMPOSITION
1341
CONCLUSIONS
no. RNP.3.1.1.11044) and the Russian Foundation for
Basic Research (grant no. 08-03-00807a).
Heating of molybdenum powder in hydrogen below
700°C leads to åÓ2é3 crystallization, a gradual
decrease in oxygen content through removal of the
water of hydration, and vaporization of molybdenum
oxides and hydroxides. After heating to above 700°C,
the powder consists almost entirely of molybdenum,
with trace amounts of oxygen.
REFERENCES
1. Tsvetkov, Yu.V., Nikolaev, A.V., Panfilov, S.A., et al.,
Plazmennaya metallurgiya (Plasma Metallurgy), vol. 8:
Nizkotemperaturnaya
plasma
(Low-Temperature
Plasma), Novosibirsk: Nauka, 1992.
Considerable coagulation of molybdenum nanopar-
ticles begins above 700°C and proceeds through vapor-
ization.
2. Levinsky,Yu.V., Nondimensional Criterion for the Parti-
cle Size of Crystalline Solids, Pis’ma Zh. Tekh. Fiz.,
2003, vol. 29, no. 5, pp. 49–52.
3. Levinsky, Y., Pressure Dependent Phase Diagrams of
Binary Alloys, Materials Park: ASM International, 1997,
p. 1830.
Molybdenum is transported through the gas phase
primarily by åÓé2(éç)2 molecules.
4. Jehn, H., Molybdan und Sauerstoff, in Gmelin Hand-
buch der anorganische Chemie, Berlin: Springer, 1975,
pp. 21–142.
5. Blagoveshchenskii, Yu.V., Levinsky, Yu.V., and
Vol’dman, G.M., Vacuum-Annealing-Induced Coagula-
tion of Ultrafine Niobium Powders, Izv. Akad. Nauk,
Met., 2002, no. 2, pp. 32–36.
ACKNOWLEDGMENTS
This work was supported by the Federal Agency for
Education (analytical departmental targeted program
Development of the Scientific Potential of Higher Edu-
cation
Institutions
2006–2008,
project
INORGANIC MATERIALS Vol. 45 No. 12 2009