Temperature-programmed and X-ray diffractometry studies of hydrogen-reduction course and products of WO3 powder: Influence of reduction parameters

Article abstract of DOI:10.1016/j.tca.2011.05.004

The hydrogen-reduction course and products of synthetic tungsten(VI) oxide (WO₃) were examined by means of temperature-programmed reduction (TPR) and X-ray powder diffractometry (XRD) studies. A set of model tungsten compounds was procured and examined similarly for reference purposes. Results obtained could help resolving two subsequent reduction stages: (i) a low-temperature stage (<1050 K) through which WO₃ is reduced to the tetravalent state (WO₂) via formation and subsequent reduction of intermediate WO_2.96, WO_2.9, WO_2.72 oxides; and (ii) a high-temperature stage (>1050 K) through which WO₂ thus produced is reduced to the metallic state (W^o) via two intermediate oxide species (tentatively, WO and W₂O-W₃O). Reduction events involved in the high-temperature stage were found to be relatively more sensitive to the reduction parameters; namely, the starting oxide mass, heating temperature and rate, and gas flow rate and composition. They were also found to require lower activation energies than those required by events occurring throughout the low-temperature stage, a fact that may suspect compliance of the high-temperature reduction events to autocatalytic effects.

Full text of DOI:10.1016/j.tca.2011.05.004

Thermochimica Acta 523 (2011) 90–96
Contents lists available at ScienceDirect
Thermochimica Acta
journal homepage: www.elsevier.com/locate/tca
Temperature-programmed and X-ray diffractometry studies of
hydrogen-reduction course and products of WO powder: Influence of reduction
3
parameters
∗
M.I. Zaki , N.E. Fouad, S.A.A. Mansour, A.I. Muftah
Chemistry Department, Faculty of Science, Minia University, El-Minia 61519, Egypt
a r t i c l e i n f o
a b s t r a c t
Article history:
The hydrogen-reduction course and products of synthetic tungsten(VI) oxide (WO ) were examined by
means of temperature-programmed reduction (TPR) and X-ray powder diffractometry (XRD) studies. A
3
Received 8 October 2010
Received in revised form 30 April 2011
Accepted 4 May 2011
set of model tungsten compounds was procured and examined similarly for reference purposes. Results
obtained could help resolving two subsequent reduction stages: (i) a low-temperature stage (<1050 K)
through which WO3 is reduced to the tetravalent state (WO2) via formation and subsequent reduction of
intermediate WO2.96, WO2.9, WO2.72 oxides; and (ii) a high-temperature stage (>1050 K) through which
Available online 11 May 2011
Keywords:
Tungsten(VI) oxide
Temperature-programmed reduction
Reduction course
Reduction products
o
WO2 thus produced is reduced to the metallic state (W ) via two intermediate oxide species (tentatively,
WO and W2O–W3O). Reduction events involved in the high-temperature stage were found to be rela-
tively more sensitive to the reduction parameters; namely, the starting oxide mass, heating temperature
and rate, and gas flow rate and composition. They were also found to require lower activation energies
than those required by events occurring throughout the low-temperature stage, a fact that may suspect
compliance of the high-temperature reduction events to autocatalytic effects.
Impacts of reduction parameters
X-ray diffractometry
©
2011 Elsevier B.V. All rights reserved.
1
. Introduction
a result of dissociative chemisorption of H . Upon further reduc-
2
4+
tion, oxygen vacancies are progressively formed and, hence, W
o
Hydrogen-reduction of tungsten(VI) oxide (WO3) powder
ions are generated and paired with the metallic state (W ) to show
an apparent oxidation state of “2+”. These reductive modifications
of the surface chemistry of WO3 particles have been, a few years
later, allocated the responsibility of alkene and alkane reforming
catalytic activity shown to develop with time of thermal treatment
of the oxide particles in H2 atmosphere [14–17]. Precisely speak-
ing, it is the pertinent population of density of states in the 5d
and 6s orbitals [1] that develops the metallic character required
for the initial hydrogenation/dehydrogenation of the hydrocarbon;
whereas, the generated OH-groups provide the necessary Brönsted
acidity for the subsequent skeletal rearrangement (isomerization)
[18]. Further development of the metallic character pertaining to
has been industrially favored as a feasible means of recovery
of tungsten metal (W ) particles, which enjoy a number of
o
application-worthy properties ([1] and references therein cited).
Hence, the reduction process has been the focus of numerous
research investigations including extensive thermodynamic [1,2],
kinetic [3–5], morphological [1,6,7], and characterization [1,8–12]
studies. These studies employed a wide range of bulk and surface,
isothermal and nonisothermal, thermoanalytical (TGR, TPR, TPO
and TPD), microscopic (TEM and SEM) and spectroscopic (XRD, IR,
Raman, XPS, etc.) methods. A thorough inspection of the results
reveals a complex nature for the reduction events, pathways and
products of the oxide.
o
deepening of the reduction into the metallic (W ) state would then
Being a solid state reaction occurring at gas/solid interfaces,
trigger the undesirable hydrogenolysis of the hydrocarbon refor-
mation products [18].
the overall reduction process (WO3(s) + 3H2(g) = Wo + 3H O(g)) has
(s)
2
been suggested [13] to commence via adsorptive interactions of
Incursion of the reduction interface (i.e., reaction front) into
the bulk of the oxide particles has been found to yield a number
of thermodynamically feasible intermediate oxide compositions
H2 molecules with the WO particles. XPS examination [13] dis-
3
5
+
closed that W ions are initially formed on the surface due to
conversion of one of the oxide oxygen ligands into OH-groups as
o
underway to formation of the tungsten metal (W ) particles [1].
Reportedly [1–12], oxides having compositions like WO2.96, WO2.9
,
WO2.72, WO , WO, and W O may be formed. Temperature, kinetics,
2
3
pathway and sequence of formation of these intermediate oxides
∗ Corresponding author. Tel.: +20 102217470; fax: +20 862360833.
E-mail address: mizaki@link.net (M.I. Zaki).
have been suggested to be critically controlled by the H -reduction
2
0
040-6031/$ – see front matter © 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.tca.2011.05.004

M.I. Zaki et al. / Thermochimica Acta 523 (2011) 90–96
91
parameters (heating rate, gas flow rate, etc.) [1]. A shortcut to the
nucleation of Wo species during early stages of WO3 reduction is
likely, but under specific reduction conditions [1,4,5]. Such an early
generation of Wo nuclei has been found to evolve autocatalysis in
the reduction course [4,5]. The known hydrogen dissociative activ-
ground to an average particle size <200 ␮m. H2 and N2 gases were
99% pure products of Egyptian Company of Industrial Gases (El-
Hawamddyia/Egypt). They were used as supplied.
2.2. Temperature-programmed reduction (TPR)
o
ity of W sites [1], and consequent spillover of the hydrogen atoms
thus generated [19] onto the underlying WO3 species, catalytically
enhance their further reduction to the metallic state [4,5]. Atomic
hydrogen has, also, been found [20,21] to interact with WO3 par-
ticles near ambient conditions leading to formation of hydrogen
TPR profiles were automatically measured during heating test
samples (mass = 5–350 mg) at various temperature ramping rates
(5–20 K/min), using a ChemBET 3000 (Quantachrome/USA). Mea-
surements were carried out in nonisothermal mode up to 1373 K, or
isothermal mode at various temperatures (873–1273 K) for 60 min,
under a streaming gas mixture of 5% (or 50%) H /N of varied flow
tungsten bronze (HxWO , with x ≤ 0.5).
3
It is worth noting, that volatile tungsten oxide hydrate
2
2
(
WO (OH) ) is formed during the H -reduction course of WO by
3
2
2
2
3
rates (20–80 cm /min). Only in the isothermal mode, the hydrogen
supply was turned off during initial heating to the set temperature.
Then, it was turned on at the set temperature and maintained so
throughout the run duration (60 min). TPR profiles were deconvo-
reaction of the oxide with water that forms during the reduction
1,22,23]. WO (OH) partial pressures are built up above unstable
[
2
2
reduction products (e.g., WO ) than above stable ones (e.g., W O
2
3
o
and W ). The pressure imbalance thus established has been sug-
gested [1] to drive a chemical vapor transport (CVT) of tungsten,
which finally leads to a dissolution (oxidation) of unstable products
2
luted using standard PeakFit software, and areas (in cm ) of peaks
thus resolved were determined and used to calculate correspond-
ing hydrogen consumption (␮mol/mol of oxide) on the basis of a
(
WO → WO (OH) ) and the growth of stable ones through deposi-
2
2
2
2
predetermined calibration constant (␮mol/cm ). The calibration
tion of tungsten metal via decomposition (reduction) of WO (OH) .
2
2
constant was determined from a TPR profile encompassing the
This CVT process, which is considered a fascinating peculiarity
of the W–O–H reduction system [1], has been found to critically
control particle morphological characteristics of WO3 reduction
products [6,7].
o
complete reduction course (WO → W ) of a precisely weighted
3
amount of WO (M); see Fig. 3.
3
Activation energy (ꢀE/kJ/mol) for each of the reduction steps
resolved (in Fig. 3) was calculated from shifts conceded by TPR
Overall, it has been emphasized [1] that a meaningful compari-
son between results of reduction studies of metal oxides, performed
in various laboratories, demands obtaining the results under identi-
cal processing parameters. Therefore, the present investigation was
designed to examine influence of reduction parameters (namely,
starting oxide mass, heating temperature and rate, and gas compo-
sition and flow rate) on the hydrogen-reduction events, course and
peak temperature (Tp/K) as a result of changing of the heating rate
−1
(ˇ/K s ), implementing the following linear relationship [25]:
ꢀ
ꢁ
ꢂ
ꢃ
ˇ
ꢀE
RTp
AR
ln
= −
+ ln
(1)
2
T
ꢀE
p
where A is the frequency factor and R is the universal gas con-
stant. Thus, the activation energy was derived from slope of plots
products of finely-divided WO particles. To accomplish this objec-
3
2
of ln(ˇ/T ) versus (1/Tp).
tive, (i) reduction events occurring throughout the reduction course
p
(
up to 1373 K) have been monitored in nonisothermal and isother-
2
.3. X-ray powder diffractometry (XRD)
mal temperature-programmed reduction (TPR) profiles measured
as a function of the processing parameters concerned, (ii) TPR pro-
files have been deconvoluted and quantified in order to facilitate
reduction stoichiometry calculations, (iii) not only synthetic WO ,
but also model (commercial) WO , WO2.9, WO2 and W powders
XRD analysis of test samples was performed by means of a model
JSX-60PA Jeol diffractometer (Japan) equipped with Ni-filtered
CuK␣ radiation (0.15418 nm, 40 kV and 30 mA). The data were
acquired stepwise (2 /s) in the 2ꢁ range 5–100 with a divergence
slit of 2 . For identification purposes of crystalline phase compo-
sition, diffraction patterns obtained were matched with standard
diffraction data (JCPDS) files [26].
3
o
3
◦
◦
were subjected to TPR analysis in order to help assigning the reduc-
tion events observed to specific reduction steps and products, and
◦
(
iv) crystalline phase composition of isothermal reduction products
has been identified by X-ray powder diffractometry (XRD).
It is worth noting, that the above shown breadth of test pro-
cess variables and materials, as well as the diverse measurement
modes (isothermal and nonisothermal mode), are not encounter-
able in a single relevant investigation of those previously reported
and herein cited. This is meant, foremost, to bridge existing gaps
of knowledge regarding the present subject matter, and provide
quantitative basis for assignments to be made for reduction events
and products resolved by deconvolution of the TPR profiles.
3. Results and discussion
3.1. Reduction course and products
3.1.1. Nonisothermal reduction
Nonisothermal TPR profiles obtained in a hydrogen-poor atmo-
sphere (5% H /N ) for the model oxides WO (M), WO2.9(M) and
2
2
2
WO (M) are compared in Fig. 1 (a–c, respectively). A TPR pro-
3
file obtained only for WO (M) in a hydrogen-rich atmosphere
3
2
. Experimental
(50% H /N ) is also included in Fig. 1(d). Profile-(a) indicates that
2
2
reduction of WO (M) does not commence detectably unless the
2
2
.1. Materials
temperature exceeds 1200 K. The corresponding reduction event is
shown not to reach its peak even at 1400 K. Hence, the shoulder
and peak monitored in profile-(b) at 1066 and 1210 K, respectively,
o
Powders of model (commercial) tungsten metal (W ) and oxides
WO , WO
(
and WO ) were ≥99% pure products of BDH. They
may account for a two-step reduction of WO2.9(M) to WO , whose
3
2.9
2
2
are discerned throughout the text by suffixing (M), for model, to
the respective molecular formula. Powder of synthetic tungsten(VI)
oxide (denoted WO (S)) was obtained by calcination of ammonium
reduction is shown to commence at the immediate vicinity of its
formation. Similar two reduction events, manifested by two well
resolved peaks at 1066 and 1140 K, are shown to be preceded by an
3
paratungstate particles ((NH4)10[H W12O42]; >99% pure product of
ill-defined shoulder (at 969 K) in profile-(c) obtained for WO (M)
2
3
Fluka Chemie) at 773 K for 10 h in a still atmosphere of air accord-
ing to Fouad et al. [24]. Prior to application, test powders were
(Fig. 1). This shoulder, which is not monitored in profile-(b), may
be plausibly assigned for reduction of WO →WO2.9. Hence, the
3

9
2
M.I. Zaki et al. / Thermochimica Acta 523 (2011) 90–96
Fig. 1. TPR profiles obtained for the indicated model tungsten oxides (WO2–3(M)),
under the inset invariant conditions: oxide mass, heating rate (H.R.), gas flow rate
(
F.R.), and composition of reduction atmosphere (Rd.A.). The dashed profile (d) was
obtained for WO3(M) at similar conditions, but in the hydrogen-richer atmosphere
of 50% H2/N2.
Fig. 2. X-ray powder diffractograms obtained for WO3(S) and its isothermal reduc-
tion product under the conditions specified. Diffractograms obtained for model
o
WO3(M) and W (M) are inset for comparison purposes.
succeeding two peaks are most likely those related to reduction
ing that obtained for model Wo metal (cubic-W , Fig. 2), which is
o
of the product WO2.9→WO , whereas reduction of WO thus pro-
2
2
the same crystalline phase composition assumed by the so-called
duced might be that responsible for the following incomplete peak
commencing near 1200 K.
o
␣
-W [1,10].
Accordingly, the composite reduction peak (at 800–1273 K)
monitored in the TPR profile of WO (M) under the H -rich atmo-
sphere (profile (d), Fig. 1) is proven to encompass a complete
When reduction of WO (M) was carried out in the hydrogen-
3
rich atmosphere of 50% H , the obtained TPR profile (d, Fig. 1)
3
2
2
displays a very broad, composite peak extending over the tem-
perature range of 800–1273 K. It is shown to resolve 6 maxima:
I at 969 K, II at 1016 K, III at 1061 K, IV at 1118 K, V at 1189 K,
and VI at 1241 K. An analogous sort of experiment was carried
out on WO (S) (i.e., synthetic WO ) in pure (100%) H atmosphere.
o
reduction course, i.e., WO → W . The fine structure of the peak may
3
thus account for a complex course of reduction. As a matter of fact,
the 6 reduction steps resolved and their corresponding positions
at the temperature scale (I–VI) were determined in the light of the
deconvolution results demonstrated in Fig. 3. Based on the amount
3
3
2
The sample mass was watched thermogravimetrically, and results
obtained (Fig. S1; Supporting Information File) helped determining
a total mass loss of 21.8% at 1073 K. When the gas atmosphere was
of the starting oxide used (5-mg WO ), the integrated (total) area of
the composite peak (=126.33 cm ) and area of each of the resolved
peaks, amounts of hydrogen consumed in the various reduc-
tion steps resolved were determined and set out in Table 1. These
3
2
6
changed into pure N , the thermogram obtained (Fig. S1) on heat-
2
ing up to 1373 K revealed high thermal stability for WO (S). This
3
result proves that the mass loss affected in the H2 atmosphere is
due to thermochemical reduction rather than thermal decomposi-
tion. Fig. S1 shows, furthermore, that the total mass loss determined
(
21.8%), which would then correspond to quantitative reduction
o
of WO →W (20.7% expected), is made up in three cumulative
3
mass loss steps: 1.7% at 800–873 K, 7.3% at 873–923 K, and 21.8% at
1
020–1100 K. Molecular stoichiometry calculations help attribut-
ing the first mass loss step (1.7%) to reduction of WO →WO
3
2.72
(
1.9% expected), the second (7.3%) to further reduction to WO (6.9%
2
expected), whereas the third step (21.8%) to the reduction to the
metallic state (20.7% expected).
In order to identify nature of the final reduction product, a simi-
lar TPR experiment was performed on a sample of WO (S) but under
3
isothermal conditions (at 1273 K for 60 min). Both of the starting
oxide and the final product were then subjected to XRD analysis.
The powder diffractogram obtained for WO (S) is shown (Fig. 2) to
3
be very similar to that obtained for the corresponding model oxide
(
WO (M)) in identifying a sole crystalline phase having the com-
3
position of monoclinic-WO . On the other hand, the final reduction
3
Fig. 3. Deconvoluted TPR profile obtained for reduction of WO (M) in the hydrogen-
3
product is found to give rise to a diffraction pattern quite match-
rich gas atmosphere (50% H2/N2) (profile (d) in Fig. 1).

M.I. Zaki et al. / Thermochimica Acta 523 (2011) 90–96
93
Table 1
o
Hydrogen consumption as derived from areas of peaks resolved in the deconvoluted TPR profile (Fig. 3) for the complete reduction of WO3 to W .
Peak numbera
Tmaxa (K)
Area (cm2)
H2 consumption (␮mol)
b
H2 consumption(cum)c
Expected reductiond
Product
(
␮mol)
(%)
4.0
9.3
33.2
64.8
(%)
I
II
III
IV
969
1016
1061
1118
5.05
6.70
30.20
39.93
2.587
3.430
15.462
20.444
2.587
6.017
WO2.9
WO2.7
WO2
WO
3.7
8.3
33.3
21.479
41.923
66.7
83.3–
88.9
ꢄ
ꢅ
ꢄ
ꢅ
W2O−
e
V
1189
1241
26.62
17.83
13.631
9.134
55.559
64.693
85.9
99.9
W3O
o
VI
W
100.0
Total
126.33
64.688
a
According to Fig. 3.
b
2
Derived from each corresponding peak area implementing the calibration constant (0.512 ␮mol/cm ) as determined from the sample mass (5-mg WO3) and the integrated
2
(
total) peak area (126.33 cm ).
c
Cumulative H2 consumption as derived from areas of peaks (I); (I) + (II); (I) + (II) + (III), etc.
WO3 was the starting oxide for each step.
Metal-rich species, where W3O has been indexed as ␤-W [1].
d
e
o
o
data could help calculating the cumulative hydrogen consumption
listed (in ␮mol and %), on the basis of which expected reduction
products and percentages are also given alongside in Table 1. It
is obvious from these results that values of % cumulative hydrogen
consumption are satisfactorily close to % reduction values expected
to accompany a stepwise reduction of WO →WO (3.7%; 969 K),
events to WO2 reduction to W . It is worth mentioning, that the
results of Wilken et al. [6] were obtained by following changes in
the dew point of the gas atmosphere, whereas those of Venables
and Brown [11] were obtained by TCD tracing. Thus, the two stud-
ies employed two different experimental methods, which are yet
different from the method applied in the present study.
Therefore, the 1-h isothermal reduction product at each of the
reduction temperatures adopted (Fig. 4) was subjected to XRD
analysis and diffractograms thus obtained are exhibited in Fig. 5.
Also exhibited in Fig. 5 are diffractograms given rise by the model
3
2.9
WO
(8.3%; 1016 K), WO2 (33.3%; 1061 K), WO (66.7%; 1118 K),
2.72
o
W O–W O (83.3–88.9%; 1189 K) and W (100%; 1241 K). Accord-
2
3
ingly, the TPR deconvolution results (Fig. 3 and Table 1) do not
account for WO3 reduction to WO2.96 that has been reported [6]
to be the initial step in the oxide reduction course.
oxides WO (M), WO (M) and WO (M), for comparison purposes.
3
2.9
2
Several literature reports [1,10,11] have also suggested forma-
tion of the suboxides WO2.9 and WO2.7 during the reduction of
WO →WO . Formation of these suboxides is thermodynamically
It is obvious from Fig. 5 that a great deal of the diffraction peaks
observed for the reduction product of WO (S) at 873 K (for 60 min)
3
owe their existence to the starting oxide, whereas the rest are
shown to be assignable to monoclinic-WO2.9 [JCPDS 73-2182]. This
result may sustain an earlier suggestion [1] ascribing the fastest
3
2
sustained [2,27], and should be preceded by formation of WO2 as
.96
the earliest reduction product of WO3 [6]. In contrast, formation of
the suboxide WO in the course of reduction of WO2 to the metallic
reduction event observed (the 167-s peak, Fig. 4) to H -uptake and
2
o
state (W ) has only infrequently reported [1]. Whereas, tungsten-
rich oxides (WnO, where n ≥ 2) with obvious metallic character,
have been encountered and reported previously [1]. Namely, W O
3
has been observed repeatedly in the reduction course not only of
WO , but also of WO , and denoted ␤-Wo [1,12].
2
2.9
3.1.2. Isothermal reduction
Fig. 4 compares TPR profiles obtained isothermally for WO (S)
3
at each of the indicated temperatures (873–1273 K) for 60 min.
The test sample was firstly heated at 20 K/min up to the set tem-
perature in pure N2 atmosphere. Then, the gas atmosphere was
switched to 5% H /N2 and the measurements started. At 873 K,
2
the profile monitors an early sharp peak after the elapse of 167 s,
which is shown to be maintained occurring at the same speed in
the other high-temperature profiles (973–1273 K). This peak is fol-
lowed by two weak maxima at 833 and 1487 s, which are shown
to be speeded up and merge into a single maximum with temper-
ature increase. A third maximum is shown to emerge at 2282 s in
the profile obtained at 973 K. It is also speeded up with temper-
ature to occur after the elapse of 1730 s at 1073 K, and 1423 s at
1
273 K. At 1073 K, the profile monitors an indication of the occur-
rence of a slow reduction event (at >2500 s) that seems not to be
enhanced significantly at the higher temperature of 1273 K. Similar
isothermal reduction events have been encountered previously by
Wilken et al. [6] and Venables and Brown [11]. The initial, sharp
peak was attributed by the former authors to formation of WO2.96
,
and by the latter authors [11] to formation of WO . Hence, the sub-
2
sequent events were thought by the former authors [6] to account
for a stepwise (WO2
lic state (W ), whereas the latter authors [11] ascribed subsequent
.96
→ WO
2.72
→ WO ) reduction to the metal-
2
Fig. 4. Isothermal TPR profiles obtained for WO (S) being heated at each of the
indicated temperatures (873–1273 K) for 60 min, under the conditions inset.
3
o

9
4
M.I. Zaki et al. / Thermochimica Acta 523 (2011) 90–96
A few residual, weak peaks that do not relate to the WO2 phase are
comparable to small proportions of unreduced WO2.9 and WO2.72
Fig. 5). With reference to the standard data filed in JCPDS 04-
806 (Fig. 2), coexistence of minority crystallites of Wo cannot be
(
0
excluded with certainty. Accordingly, the further enhanced isother-
mal reduction events maximized at 576 and 1423 s in the TPR
profile measured for WO (S) at 1273 K (Fig. 4) evidently encompass
3
the formation of WO2.9, WO2.7 and WO2 as intermediate prod-
ucts in the reduction course of the trioxide to the metallic state.
The sharp peak at 167 s may, according to literature reports [1],
be considered indicative for the formation of WO2.96 as the initial
reduction product. The rapidity and expected minute consumption
of H responsible for the reduction of WO into WO may explain
2
3
2.96
whey it went unnoticed in the corresponding nonisothermal TPR
profiles (c and d, Fig. 1). The slow event appearing at ≥2000 s (Fig. 4)
o
is presumably corresponding to WO2 reduction to W via forma-
tion and subsequent reduction of WO and W O–W O. It is worth
2
3
noting, that compositions of the intermediates WO and W O–W O
2
3
are suggested solely on basis of TPR deconvolution results (Fig. 3
and Table 1).
The increase of rate of the isothermal reduction events
attributable to the reduction of WO2 to WO2.9, WO2.72, WO , WO,
.96
2
o
W O–W O and W with temperature is often ascribed to enhanced
2
3
kinetics of solid state reactions involved [1]. In previous investi-
gations performed in this laboratory [4,5], however, experimental
evidences were brought about for the involvement of autocatal-
ysis in the hydrogen reduction course of WO . Genesis of a few
3
nuclei of metallic tungsten has been considered [4,5] sufficient to
affect dissociative chemisorptions of H and consequent spillover of
2
active hydrogen atoms to catalyze further reduction of unreduced
oxides. Hence, involvement of autocatalysis in the present isother-
mal reduction experiments (Fig. 4), particularly at ≥973 K, cannot
be excluded with certainty.
Fig. 5. X-ray powder diffractograms obtained for isothermal reduction products
of WO3(S) at each of the indicated temperatures (873–1273 K) for 60 min. Diffrac-
tograms obtained for model oxides WO3(M), WO2.9(M) and WO2(M) are inset for
comparison purposes.
3.2. Influence of reduction parameters
Considering
the
conceptual
reduction
equation
WO3(s) + 3H2(g) = Wo + 3H O(g), the reduction progress is con-
(s)
2
reduction of WO3 into WO2.96, whereas the subsequent peak at
trolled by the pressure ratio P(H )/P(H O). This is in the sense
2 2
8
33 s to a further reduction of the product WO2.96 to give WO2.9
.
that the higher the ratio is, the more enhanced the reduction
Formation of WO2.96 has been considered [6] to follow formation
progress [1]. Within this context, the WO reduction enhancement
3
and subsequent decomposition of tungsten bonze (HxWO ) [20,21].
observed in the hydrogen-rich atmosphere (50% H , profile-d,
3
2
The shoulder emerging a bit later (at 1487 s) in the TPR obtained
at 873 K (Fig. 4) may be tentatively related to a much slower event
Fig. 1) as compared to the limited reduction progress accom-
plished in the hydrogen-poor atmosphere (5% H , profile-c, Fig. 1),
2
(
at this temperature) which does not lead to an XRD-detectable
product.
Diffractograms obtained for the isothermal reduction products
should be understandable. Results obtained for the influence of
other reduction parameters (viz. oxide mass, heating rate and gas
flow rate) are presented and discussed below.
of WO (S) at 973 and 1073 K were rather similar in identifying
3
WO2.9 as the major crystalline phase (Fig. 5). Some other (weak)
peaks that are not related to the WO2.9 (Fig. 5), are displayed and
found to account for monoclinic-WO2.72 (JCPDS 84-1516]. As is
shown in Fig. 4, isothermal reduction events assignable to forma-
tion at 873 K of WO2.9 (peak at 833 s) and WO2.72 (peak at 1487 s)
are expedited to occur after the elapse of 538 and 1050 s, respec-
tively, upon increasing the temperature up to 973 K. Upon further
increase of the temperature up to 1073 K the two peaks (partic-
ularly that due, suggestively, to formation of WO2.72) converge in
one broad peak maximized at 666 s. The peak emerging at 2282 s
in the profile obtained at 973 K, which is also enhanced to occur at
3.2.1. Oxide mass
Nonisothermal TPR profiles obtained as a function of WO (S)
3
mass (2–118 mg) are compared in Fig. 6. The general trend exhib-
ited by these profiles is the high-temperature shifts conceded by
the hydrogen consumption peaks monitored upon increasing the
oxide mass. Relatively speaking, however, low-temperature peaks,
i.e. those corresponding to formation of the suboxides of WO3 (viz.
WO2.9 and WO2.72) are less influenced by the mass increase than
the high-temperature peaks assignable to formation and reduction
of WO . Though these profiles were measured in the hydrogen-
2
poor atmosphere of 5% H , the reduction is shown to be advanced
2
1
730 s upon increasing the temperature up to 1073 K, is shown to
be followed by a slight consumption of hydrogen (at ≥2400 s) at
073 K. Reduction product(s) of these latter events seem to have
significantly when the starting oxide mass is decreased to less than
5 mg.
1
Increasing the oxide mass has been considered by many authors
not been developed into XRD-detectable crystalline phase(s).
Isothermal reduction product at 1273 K is shown by XRD to con-
sist of a major crystalline phase exhibiting a diffraction pattern very
[1,28] to be pertained by increase of the build up of P(H O)
2
within the sample particles. Consequently, the P(H )/P(H O) ratio
2
2
is decreased and, in turn, the reduction progress is hampered.
This becomes more obvious in the later stage of reduction (i.e.,
similar to that feild (JCPDS 48-1827) for orthorhombic-WO (Fig. 5).
2

M.I. Zaki et al. / Thermochimica Acta 523 (2011) 90–96
95
Table 2
The activation energy values (ꢀE) derived for each of the indicated reduction events
(
(
I–IV) from nonisothermal TPR profiles obtained as a function of heating rate of WO3
S).
Reduction eventa
ꢀE/(± 3 kJ/mol)
I
93
II
109
III
IV
64
73
a
According to Fig. 7.
the peak temperatures to increase with heating rate, as has often
been encountered in the literature [1,28]. The four peaks monitored
(I–IV) correspond to the reduction of WO3 to WO2.9 (I), WO2.72 (II),
WO2 (III), and WO (IV). The hydrogen consumption monitored to
follow peak-IV is indicative of commencement of the reduction of
2
WO . Implementing Eq. (1) plots of ln(ˇ/T ) versus (1/Tp) were
2
p
constructed and shown in Fig. S2. Activation energy values there
from derived are set out in Table 2. The activation energy values
obtained for peak-I (93.4 kJ/mol) and -II (109.4 kJ/mol) are higher
than those obtained for peak-III (63.7 kJ/mol) and -IV (72.7 kJ/mol).
Fouad et al. [4,5] determined similar values (90–123 kcal/mol) for
corresponding H -reduction steps of WO , adopting isothermal
Fig. 6. TPR profiles obtained for WO3(S) as a function of the oxide mass (2–118 mg),
under the inset invariant conditions.
2
3
and nonisothermal gravimetry measurements. These authors have
attributed [4,5] the low energy values to involvement of autocat-
alytic reduction due to development of nuclei of metallic tungsten.
A similar reasoning has been put forward by other authors [28].
o
WO → W ) because of the expected release of larger amounts of
2
water than in the earlier stages (WO → WO
). However, in the
3
2.72
presence of only minute amounts of the oxide (<5 mg) the build up
of effective water vapor pressure is not facilitated and, hence, ren-
Autocatalysis has been suggested [6] to be more likely involved
dered insufficient to hamper the advance of WO reduction towards
the metallic state. The TPR profile obtained for 2-mg WO3 in the
hydrogen-poor atmosphere of 5% H (Fig. 6) is almost analogous to
o
3
in the reduction course of WO –W (second stage) than in that of
2
WO –WO2 (first stage).
3
2
that obtained for 5-mg WO3 in the hydrogen-rich atmosphere of
3
.2.3. Gas flow rate
TPR profiles obtained for WO (S) at 20 K/min as a function
5
0% H2 (Fig. 1).
3
3
of the gas (5% H /N ) flow rate (20–80 cm /min) are exhibited
in Fig. 8. Results indicate that increasing the gas flow rate leads
2
2
3.2.2. Heating rate
TPR profiles obtained in 5% H /N2 for WO (S) as a function of
2
3
heating rate (5–20 K/min) are compared in Fig. 7. The results show
Fig. 7. TPR profiles obtained for WO3(S) as a function of heating rate (5–20 K/min),
under the inset invariant conditions. Peak designations (I–IV) are the same as in
profile-d of Fig. 1.
Fig. 8. TPR profiles obtained for WO3(S) as
(20–80 cm /min), under the inset invariant conditions.
a function of gas flow rate
3

9
6
M.I. Zaki et al. / Thermochimica Acta 523 (2011) 90–96
to decreasing peak temperatures. Accordingly, the TPR obtained
under 20 cm /min displays only two strongly overlapping peaks at
Appendix A. Supplementary data
3
1
183 and 1260 K. These two peaks are shifted to the lower tem-
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.tca.2011.05.004.
peratures of 1063 and 1163 K, and further to 1026 and 1070 K
upon increasing the gas flow rate up to 50 and 80 cm /min, respec-
3
tively. Moreover, a third incomplete reduction peak is monitored
at 50 cm /min. This peak, as well as a fourth one, is shown to be
References
3
[
1] E. Lassner, W.-D. Schbert, Tungsten Properties, Chemistry, Technology of the
Element, Alloys and Chemical Compounds, Kluwer Academic/Plenum Publish-
ers, New York, 1999.
completed, maximizing at 1213 and 1250 K, upon increasing the
3
gas flow rate up to 80 cm /min. The four peaks monitored in the
3
TPR profile obtained at 80 cm /min are most probably those (I–VI)
[2] L. Sha, Z.J. Qiu, Thermodynamic calculation on reduction of tungsten oxide in
H2 atmosphere, Int. J. Refract. Met. Hard Mater. 26 (2008) 362–366.
monitored in the TPR profile-c displayed in Fig. 7 and assigned,
[
3] J.A. Bustnes, D. Sichen, S. Seetharaman, Application of a nonisothermal thermo-
gravimetric method to the kinetic study of the reduction of metallic oxides. Part
II. A theoretical treatment of powder bed reduction and its application to the
reduction of tungsten oxide by hydrogen, Metall. Trans. B 24 (1993) 475–480.
4] N.E. Fouad, Impacts of hydrogen spillover on the reduction behavior of tung-
sten oxide: isothermal and non-isothermal approaches, J. Anal. Appl. Pyrol. 44
(1997) 13–28.
respectively, to the stepwise reduction of WO3 to WO2.9, WO2.72
,
WO and WO as in profile-d of Fig. 1. The communicated direct rela-
2
tionship between the gas flow rate and easiness of reduction has
[
been attributed [1,28] to faster removal of H O vapor the higher the
2
gas flow rate, thus increasing the pressure ratio P(H )/P(H O). This
2
2
is in the sense that accumulated water vapor (reduction product)
within the oxide particles may re-oxidize the oxide thus hampering
the reduction progress [1].
[5] N.E. Fouad, K.M.E. Attyia, M.I. Zaki, Thermogravimetry of WO3 reduction in
hydrogen: kinetic characterization of autocatalytic effects, Powder Technol. 74
(
1993) 31–37.
[
[
[
[
6] T.R. Wilken, W.R. Morcom, C.A. Wert, J.B. Woodhouse, Reduction of tungsten
oxide to tungsten metal, Metall. Trans. B 7 (1976) 589–597.
7] V.K. Sarin, Morphological changes occurring during reduction of WO3, J. Mater.
Sci. 10 (1975) 593–598.
8] M.G. Charlton, Hydrogen reduction of tungsten oxides, Nature 169 (1952)
109–110.
4
. Conclusions
The above presented and discussed results may help drawing
the following conclusions:
9] J.O. Hougen, R.R. Reeves, G.G. Mannlla, Reduction of tungsten oxides with
hydrogen, Ind. Eng. Chem. 48 (1956) 318–320.
[
[
10] R. Hasegawa, T. Kurosawa, T. Yagihashi, Hydrogen reduction of tungsten tri-
oxide and ammonium para-tungstate under high pressure, J. Jpn. Inst. Met. 15
1
. Two stages are discerned in the H -reduction course of syn-
2
(
1974) 75–82 (Engl. Transl.).
thetic WO3 particles: a low-temperature stage (WO → WO at
3
2
11] D.S. Venables, M.E. Brown, Reduction of tungsten oxides with hydrogen and
with hydrogen and carbon, Thermochim. Acta 285 (1996) 361–382.
<1050 K) involving formation and reduction of three intermedi-
ate suboxides of WO3 (viz., WO2.96, WO2.9 and WO2.72), and a
[12] A. Lackner, A. Filzwieser, P. Paschen, W. Köck, On the reduction of tungsten
blue oxide in a stream of hydrogen, Int. J. Refract. Met. Hard Mater. 14 (1996)
o
high-temperature stage (WO → W at >1050 K) involving for-
2
3
83–391.
13] J. Haber, J. Stoch, L. Ungier, Electron spectroscopic studies of the reduction of
WO , J. Solid State Chem. 19 (1976) 113–115.
mation and reduction of two intermediate suboxides of WO2
[
(
tentatively, WO and W O–W O).
2 3
3
2
3
. WO2.96 is the first reduction product of WO , whose formation
is detectable only in isothermal TPR profiles after the elapse of
[14] A. Katrib, F. Hemming, P. Wehrer, L. Hilaire, G. Maire, The multi-surface struc-
ture and catalytic properties of partially reduced WO3, WO2 and WC + O2 or
W + O2 as characterized by XPS, J. Electron Spectrosc. Relat. Phenom. 76 (1995)
3
1
67 s (Fig. 4). The corresponding hydrogen consumption peak is
195–200.
most prominent in the TPR profile obtained at 873 K (Fig. 4).
[15] V. Logie, P. Wehrer, A. Katrib, G. Maire, Catalytic activity of bulk WO3 for the
reforming of hexanes and hexenes, J. Catal. 189 (2000) 438–448.
o
. The high-temperature stage of WO3 reduction (WO → W ) is
2
[
16] A. Benadda, A. Katrib, J.W. Sobczak, A. Barama, Hydroisomerization of n-
heptane and dehydration of 2-propanol on bulk and supported WO2(Hx)ac on
TiO2, Appl. Catal. A 260 (2004) 175–183.
more facile (ꢀE = 63.7–72.7 kJ/mol) than the low-temperature
stage (ꢀE = 93.4–109.4 kJ/mol), most probably due to contribu-
tion of autocatalytic reduction associated with early nucleation
of metallic tungsten species (the metal or metal-rich oxides).
[
17] H. Al-Kandari, F. Al-Kharafi, N. Al-Awadi, O.M. El-Dusouqui, A. Katrib, Surface
electronic structure–catalytic activity relationship of partially reduced WO3
bulk or deposited on TiO2, J. Electron Spectrosc. Relat. Phenom. 151 (2006)
128–134.
4
. Reduction of WO is enhanced upon increasing the P(H )/P(H O)
3
2
2
[
18] J.H. Sinfelt, Catalytic reforming, in: G. Ertl, H. Knözinger, J. Weitkamp (Eds.),
Handbook of Heterogeneous Catalysis, vol. 4, Wiley-VCH, Weinheim, 1997, pp.
pressure ratio, with the hydrogen pressure being more influen-
tial in the low-temperature stage, and the water vapor pressure
is likewise but in the high-temperature stage. This can be accom-
plished by increasing amount of hydrogen in the gas atmosphere,
1939–1986.
[19] P.A. Sermon, G.C. Bond, Hydrogen spillover, Catal. Rev. 8 (1973) 211–239.
[
[
20] P.G. Dickens, J.H. Moore, D.J. Neild, Thermochemistry of hydrogen tungsten
bronze phases HxWO3, J. Solid State Chem. 7 (1973) 241–244.
21] B. Gerand, M. Figlarz, Spillover of adsorbed hydrogen, in: G.M. Pajonk, S.T.
Teichner, J.E. Germain (Eds.), Spillover of Adsorbed Species, Elsevier Science
Publishers B.V., Amsterdam, 1983, pp. 275–283.
or decreasing amount of the starting oxide (WO ) and/or increas-
3
ing the gas flow rate. The specified changes of the latter two
reduction parameters are reflected in decreasing the build up of
water vapor pressure within the oxide particles.
. A justifiable comparison of reported reduction results of WO3 is
established only on grounds of identical (or very close) reduction
parameters (heating rate, gas composition, oxide mass and gas
flow rate).
[
22] O. Glemser, H.G. Voltz, Gasformige hydroxyde. 1. Uber gasformiges WO2(OH)2,
Naturwissenschaften 43 (1956) 33–133.
5
[23] O. Glemser, H. Ackermann, Gasformige hydroxyde. 6. Erganzende untersuchun-
gen uber gasformige WO2(OH)2, Z. Anorg. Allg. Chem. 325 (1963) 281–286.
[
24] N.E. Fouad, A.K.H. Nohman, M.I. Zaki, Spectro-thermal investigation of the
decomposition intermediates developed throughout reduction of ammonium
paratungstate, Thermochim. Acta 343 (2000) 139–143.
[
25] J. Sˇ esták, V. Sˇ atava, W.W. Wendlandt, The study of heterogeneous processes by
thermal analysis, Thermochim. Acta 7 (1973) 333–556.
26] International Center for Diffraction Data, 12 Campus Boulevard, Newton
Square, PA 19073-3273.
Acknowledgement
[
The authors are indebted to the Alexander von Humboldt-
Foundation (Bonn) for an equipment donation (V-815/03029). AIM
appreciates a financial support from Minia University Research
Administration.
[27] Chinese
Academy
of
Science,
Basic
Science
Data
Center,
http//mole.ipe.ac.cn/all thermochemistry/all thermochemistry query.aspx.
28] P. Arnoldy, J.C.M. de Jonge, J.A. Moulijn, Temperature-programed reduction of
molybdenum(VI) oxide and molybdenum(IV) oxide, J. Phys. Chem. 89 (1985)
4517–4526.
[