preparation of aqueous sols of tungsten oxide dihydrate

Preparation of aqueous sols of tungsten oxide dihydrate
from sodium tungstate by an ion-exchange method
Yong-Gyu Choi
a
, Go Sakai
b
, Kengo Shimanoe
b
, Norio Miura
c
, Noboru Yamazoe
b,*
a
Department of Molecular and Material Sciences, Interdisciplinary Graduate School of Engineering Sciences, Kyushu University,
Kasuga-shi, Fukuoka 816-8580, Japan
b
Department of Molecular and Material Sciences, Faculty of Engineering Sciences, Kyushu University, Kasuga-shi, Fukuoka 816-8580, Japan
c
Advanced Science and Technology Center for Cooperative Research, Kyushu University, Kasuga-shi, Fukuoka 816-8580, Japan
Received 18 February 2002; received in revised form 16 May 2002; accepted 20 May 2002
Abstract
Aqueous sols of tungsten oxide dihydrate (WO
3
Á2H
2
O) were prepared from Na
2
WO
4
by an ion-exchange method. An aqueous solution of
Na

2
WO
4
was let to flow through a glass column packed with protonated cation-exchange resin. The effluent, initially transparent, turned into
an opaque viscous fluid (pale yellow) in a few hours, before yellow precipitate deposited to completion in three days. The precipitate was a
mixture of a crystalline phase of WO
3
Á2H
2
O and an amorphous phase, and the crystalline part could be separated from another by washing
with deionized water and centrifuging. The gel of WO
3
Á2H
2
O thus obtained consisted of platelike crystallite 25 nm thick and 42 nm wide as
evaluated from the X-ray diffractometer (XRD) peaks, and could be dispersed well into deionized water to form a stable suspension of
colloidal particles with a mean diameter of about 30 nm. The mean particle size as well as the crystallite size tended to increase gradually with
the repetition of dispersion in water under ultrasonic wave agitation and gelling by centrifuging. On heating, the gel (WO
3
Á2H
2
O) changed to
the monohydrate (WO
3
ÁH
2
O) at 100 8C, which in turn changed to the anhydride (WO
3
) at 240 8C. Remarkably XRD patterns showed
conspicuous preferred orientation of WO

3
Á2H
2
O crystallites in (0 1 0) plane after the sol was centrifuged for a long time (10 h) and, upon
dehydration, it was inherited by the dehydrated phases, resulting in the conspicuous orientation of WO
3
crystallites in (0 0 1).
# 2002 Elsevier Science B.V. All rights reserved.
Keywords: Tungsten oxide; Sol; Colloid suspension; Ion exchange; Sodium tungstate; Preferred orientation
1. Introduction
Tungsten trioxide (WO
3
) is known as a multi-functional
material applicable for electrochromic display device (ECD)
[1–3], semiconductor gas sensor [4–9], catalyst [9–11],
varistor [12,13] and so on. It can combine with water
molecules to form crystalline phases of WO
3
ÁnH
2
O(n ¼
2, 1 or 1/3), or amorphous phases of metatungstic acids and
isopoly tungstic acids. These compounds take various mole-
cular and crystal structures, as reported in many literatures
[14–19]. Such a variety in compound and structure appears
to be the origin of the multifunctionality of WO
3
. On the
other hand, this suggests that the functionality would depend
much on the method to prepare WO

3
. For applications to
semiconductor gas sensors, polycrystalline materials of
WO
3
have been prepared by various methods, i.e. pyrolysis
of (NH
4
)
10
W
12
O
41
Á5H
2
O, sputtering or evaporation from a
source of WO
3
, sol–gel method using W-alkoxide, etc. It has
been experienced well that the gas sensing performances
differ significantly by the methods and conditions of WO
3
preparation used, but why this is so has hardly been clarified
well. Tamaki et al. have shown for sintered-block type gas
sensors that the electrical resistance as well as the sensitivity
to NO
2
begins to increase sharply as the grain size (mean
diameter) of WO

3
decreases to be less than a critical value
and that the critical value would correspond to twice the
surface space charge layer thickness of WO
3
grains [20].
There are still many other factors that should affect the gas
sensing properties, such as porosity, pore size distribution,
and gas sensing layer thickness. These factors are deeply
related with the material processing of WO
3
. We reported
[21] recently that preparation and control of SnO
2
sols in an
aqueous medium were effective for controlling the physi-
cochemical properties of SnO
2
-based sensors. This finding
prompted us to extend the same approach for the WO
3
based sensors. In the present case, however, the stable
colloidal species in an aqueous medium is not WO
3
itself
but its dihydrate, WO
3
Á2H
2
O (monoclinic, a ¼ 0:75 nm,

Sensors and Actuators B 87 (2002) 63–72
*
Corresponding author. Tel.: þ81-92-583-7539;
fax: þ81-92-583-7538/7539.
E-mail address: (N. Yamazoe).
0925-4005/02/$ – see front matter # 2002 Elsevier Science B.V. All rights reserved.
PII: S 0925-4005(02)00218-6
b ¼ 0:693 nm, c ¼ 0:37 nm, b ¼ 90:58), which is dehy-
drated on heating to WO
3
ÁH
2
O (orthorhombic, a ¼
0:5238 nm, b ¼ 1:070 nm, c ¼ 0:5120 nm) and WO
3
(monoclinic, a ¼ 0:7297 nm, b ¼ 0:7539 nm, c ¼ 0:7688
nm, b ¼ 90:918). It has been reported that a colloidal suspen-
sion of platelets of WO
3
Á2H
2
Oisformedbyacidification of an
Na
2
WO
4
solution with a mineral acid (HCl), but little infor-
mation has been accumulated on how to prepare and control
the suspension from a standpoint of gas sensor applications.
In the present study, we examined the possibilities of

obtaining aqueous sols of WO
3
Á2H
2
O from an aqueous
solution of Na
2
WO
4
by an ion-exchange method; the
Na
2
WO
4
solution was let to flow through a column packed
with a cation-exchange resin converted into the proton type
in advance. This method is a modification of the acidifica-
tion method, but may be more advantageous because no
sodium salt is left in the effluent. It has been reported [22]
that, in the acidification method, a trace amount of remain-
ing Na
þ
ions gives significant effects on the kinds and
properties of the acidification products. This paper aims
at reporting the ion-exchange based preparation of
WO
3
Á2H
2
O sols from Na

2
WO
4
. Formation and character-
ization of the sols as well as the effects of post treatments are
described together with the thermal behavior of the corre-
sponding gels.
2. Experimental
A commercial cation-exchange resin (Diaion SK 1B,
Mitsubishi Chemical Co.) was immersed in an acid solution
(HNO
3
) for 1 h to convert it from Na
þ
type to H
þ
type. After
washing with distilled water five times, the resin was packed
uniformly in a glass column and washed again with distilled
water repeatedly until pH of the effluent came close to 7. The
ion-exchange capacity (content of protons) of the resin was
about 2 meq./cm
3
, as evaluated from the titration with an
NaOH solution.
Sodium tungstate was purchased as its dihydrate
(Na
2
WO
4

Á2H
2
O) and used without further purification. Its
aqueous solution was let to flow down through the glass
column at a fixed rate, and the effluent was collected into a
beaker. After standing for 3 days, the effluent precipitated a
yellow gel containing WO
3
Á2H
2
O. The particle size distri-
butions of sols were analyzed on a laser particle size
analyzer (Photal Otsuka Electronics, LPA 3100). The crys-
talline compounds were identified for dried or calcined
samples by using an X-ray diffractometer (Rigaku RINT
2100), while their crystallite sizes were evaluated from the
full width of half maximum intensity (FWHM) values of X-
ray diffractometer (XRD) peaks by using Scherrer’s equa-
tion. In some cases, colloidal particles were subjected to
direct observation on a transmission electron microscope
(JEOL JEM-4000EX). The content of Na
þ
ions in the gels
was analyzed by fluorescence X-ray spectroscopy (Rigaku,
Fluorescence X-ray Spectrometer 3270).
3. Results and discussion
3.1. Hydrolysis behavior of Na
2
WO
4

The hydrolysis of Na
2
WO
4
with a mineral acid (like HCl)
is known to be rather complex, giving different products or
intermediates depending on the pH of the reaction medium.
As a preliminary test, the hydrolysis behavior of Na
2
WO
4
with the protonated cation-exchanged resin was investi-
gated. The resin was added bit by bit to the solutions of
Na
2
WO
4
(0.1, 0.15 and 0.2 M) under agitation with a stirrer
at room temperature. The resulting titration curves are
shown in Fig. 1. The curves exhibited two inflections in
the pH ranges of 7–5 and 4–2, respectively. The color of the
solution changed from none to light yellow at the first
inflection, while a deep yellow precipitate deposited at
the second. These titration curves were confirmed to be
essentially the same as those obtained in the hydrolysis with
sulfuric acid. It is known that WO
4
2À
ions, stable in an
alkaline solution, condense together with a lowering in pH;

the condensation products are paratungstate ions like
[W
12
O
41
]
10À
and [H
2
W
12
O
40
]
6À
in the pH range of 7–4,
while metatungstic acids ((WO
3
)
n
ÁxH
2
O) and related ions
prevail in the pH range of 4–1 [19]. Based on this informa-
tion, the first inflection seems to reflect the condensation to
paratungstate ions, for instance, as follows:
12WO
4
2À
þ 14H

þ
!½W
12
O
41

10À
þ 7H
2
O (1)
12WO
4
2À
þ 18H
þ
!½H
2
W
12
O
40

6À
þ 8H
2
O (2)
The resin to Na
2
WO
4

equivalent ratio at the first inflection,
1.2–1.4, coincides well with these condensation reactions.
The second inflection, on the other hand, seems to reflect the
formation of metatungstic acids (and related ionic species),
for example, as follows.
nWO
4
2À
þ 2nH
þ
!ðWO
3
Þ
n
Á xH
2
O þðn À xÞH
2
O (3)
Here, metatungstic acids are expressed by (WO
3
)
n
ÁxH
2
O,
but they are in fact a mixture of homologous compounds
different in n and x.Theresin/Na
2
WO

4
equivalent ratio
observed (1.8) is slightly lower than expected (2.0), possi-
bly due to slow equilibration of the foregoing paratungstate
ions.
In the earlier mentioned experiment, the hydrolysis of
Na
2
WO
4
was controlled by the rate of resin addition. In the
actual ion-exchange reaction, a solution of Na
2
WO
4
is let to
flow through an ion-exchange column. The hydrolysis is
now controlled by the rate of ion exchange, allowing the pH
of the solution to decrease rapidly down to the final value
( 1) observed in the titration experiment (Fig. 1). It is thus
expected that the primary product of the ion-exchange
reaction would be metatungstic acids (and related ions). It
has been reported [23] that, in the acidification of Na
2
WO
4
with HCl, WO
3
Á2H
2

O is formed as a secondary product
derived from metatungstic acids (primary product) when the
final pH of the solution is set to 1–2. Thus WO
3
Á2H
2
O would
64 Y G. Choi et al. / Sensors and Actuators B 87 (2002) 63–72
be formed when the ion-exchanged effluent is kept for a
certain period.
3.2. Products of ion exchange and ageing
A volume of 200 cm
3
of sodium tungstate solution
(0.152 M, pH ¼ 8:3) was allowed to flow through a glass
column packed with 125 cm
3
of the protonated cation-
exchange resin at a rate of 2 cm
3
/min at room temperature.
In this setting, the resin/Na
2
WO
4
equivalent ratio was 4.40
and the contact time of Na
2
WO
4

solution was 62.5 min. The
effluent solution after the reaction ðpH ¼ 0:3Þ, colored light
yellow, was stable for a short while before it was trans-
formed into an opaque soft gel like a pudding in a few hours.
Soon, the soft gel began to separate into a yellow gel and a
transparent liquid, going to completion in 3 days.
In order to check the degree of ion exchange, the whole
effluent solution was dried up. As analyzed by fluorescence
X-ray spectroscopy, the Na
þ
content of the resulting powder
was below the detection level, confirming that the ion-
exchange reaction between Na
2
WO
4
and the resin was com-
plete. The same check was carried out for the effluents
obtained at larger rates of flow of the Na
2
WO
4
solution.
The Na
þ
content was again below the detection level at a flow
rate of 20 cm
3
/min (contact time 6.25 min), while a signifi-
cant level of Na

þ
was detected at 100 cm
3
/min (contact time
1.25 min). The flow rate was fixed at 2 cm
3
/min hereafter.
For identification, the yellow gel was collected by decan-
tation and dried in vacuum at room temperature. As shown in
Fig. 2, the gel as dried hardly exhibited XRD peaks. The gel
obtained was then calcined at selected temperatures from
100–600 8C for 2 h. After calcination at 100 8C, crystalline
phases of WO
3
Á1/3H
2
O (orthorhombic, a ¼ 0:7359 nm, b ¼
1:251 nm, c ¼ 0:7704 nm) and WO
3
ÁH
2
O appeared. The
latter phase disappeared almost completely at 200 8C, while
the former phase remained up to 400 8C and was converted
into WO
3
at 430 8C. As just mentioned, the gel dried at
room temperature hardly showed XRD peaks. It is possible
that crystalline products, even if formed, might be covered
too thick by an amorphous material. Thus the yellow gel

was suspended in deionized water (320 cm
3
) under agita-
tion briefly (3 min) and centrifuged for 1 h. The resulting
gel and liquid were vacuum-dried at room temperature
separately. Fig. 3(a) shows the XRD patterns of the gel as
dried and as calcined. Remarkably, a clear XRD pattern
of WO
3
Á2H
2
O phase was observed after drying. This
phase was converted to WO
3
ÁH
2
OandWO
3
after cal-
cinationat100and3008C, respectively. The correspond-
ing XRD patterns for the liquid part are shown in Fig. 3(b).
The sample remained almost amorphous after drying at
room temperature as well as after calcination at 100 8C.
Crystalline phases of WO
3
Á1/3H
2
O and unidentified com-
pound(s) appeared after calcination at 200–400 8C, and
those were converted to WO

3
completely after calcination
at 500 8C.
It is understood that the yellow gel precipitate deposited
from the effluent of ion-exchange reaction after the ageing
was a mixture of an amorphous product (tungstic acids) and
a crystalline product of WO
3
Á2H
2
O, and that the mixture
could be separated from each other by the washing and
centrifugal treatment. Yields of WO
3
Á2H
2
O and tungstic
acids were 72.1 and 20.3 mol%, on the basis of starting
Na
2
WO
4
, respectively, as evaluated from the masses of WO
3
after calcination at 500 8C. In order to check the material
balance in more detail, the liquid part of the ion-exchange
effluent remaining after the yellow gel (mixture) was sepa-
rated off was also dried and calcined. As a result, WO
3
was

also found as the calcination product, and it accounted for
6.3 mol% of the starting Na
2
WO
4
. The sum of these found
values amounted to 98.7 mol%, confirming that the reactant
Fig. 1. Titration of Na
2
WO
4
solutions with protonated cation-exchanged resin.
Y G. Choi et al. / Sensors and Actuators B 87 (2002) 63–72 65
Fig. 2. XRD patterns of the as-precipitated gel after drying or calcining at the indicated temperatures.
Fig. 3. XRD patterns of the powder samples derived from the solid part (a) and the liquid part (b) resulting when the as-precipitated gel was briefly washed
and centrifuged. The liquid was evaporated to dryness at room temperature.
66 Y G. Choi et al. / Sensors and Actuators B 87 (2002) 63–72
or products remaining in the ion-exchange column should be
minimal.
Based on these results, the reaction paths in the ion
exchange and ageing are summarized in the scheme shown
in Fig. 4. It is assumed that the metatungstic acids as a
primary product of the ion-exchange reaction is soluble but
they combine together to form insoluble metatungstic acids
and WO
3
Á2H
2
O gel during the ageing period (3 days). The
insoluble metatungstic acids are made soluble when washed

with water, probably because of an increase in pH. A single
phase of WO
3
Á2H
2
O gel is thus obtained at more than 70%
yield in the present method.
3.3. Properties of gel and sol of WO
3
Á2H
2
O freshly
prepared
The wet yellow gel of WO
3
Á2H
2
O obtained earlier could
be easily dispersed in deionized water to form a suspension,
if the content of WO
3
Á2H
2
O was about 5 wt.% or less on the
WO
3
basis. In addition, the suspension could be stored
comfortably for more than 1 month at room temperature.
Fig. 5 shows the mean particle size as well as the range of
particle size distribution analyzed on LPA for a freshly

prepared sol (content: 5 wt.% on the WO
3
basis) as a
function of storage time. For the fresh suspension, the
particle sizes were distributed in a fairly narrow range of
25–35 nm with a mean size of about 30 nm in diameter. With
an increase in storage time up to 28 days, the mean size
tended to increase somewhat, but the increment was minimal
(only up to about 32 nm) and the range of particle size
distribution was also fairly stable during the storage.
In order to know whether the colloidal particles are
provided with free crystallites of WO
3
Á2H
2
O, the crystallite
sizes were evaluated from the XRD peak widths of the
corresponding WO
3
Á2H
2
O gel dried at room temperature
(Fig. 3(a)). It has been reported that the WO
3
Á2H
2
O crystal-
lites tend to grow into platelets. This tendency was also
detected in the present study. For the fresh gel, the crystallite
Fig. 4. Reaction paths stating from Na

2
WO
4
.
Fig. 5. Mean and range of particle sizes of the WO
3
Á2H
2
O sol as a function of time of storage (measured on LPA).
Y G. Choi et al. / Sensors and Actuators B 87 (2002) 63–72 67
size evaluated from (0 1 0) or (0 2 0) was about 25 nm, while
that from (2 0 0) or (0 0 1) was about 42 nm. This shows that
each crystallite is a platelet with its basal plane parallel to
(0 1 0) developed about two times as large as its thickness
normal to (0 1 0). Obviously the mean particle size (30 nm)
of the sol as well as the range of particle size distribution
(25–35 nm) is placed between the width (42 nm) and the
thickness (25 nm) of the crystallites. It is thus rational to
conclude that each colloidal particle is in fact provided with
a single crystallite of WO
3
Á2H
2
O. This conclusion was
supported by TEM observation as described later.
3.4. Effects of washing and/or centrifugal treatments
It has been reported [23] that the WO
3
Á2H
2

O platelets tend
to grow with a washing treatment. This suggests a possibility
of controlling the crystallite size or colloidal particle size by
adequate post treatments. The freshly prepared WO
3
Á2H
2
O
gel was subjected to a washing and centrifugal treatment, i.e.
dispersion into deionized water (320 cm
3
) under ultrasonic
wave agitation for 20 min followed by gelling back by
centrifuging for 1 h. This treatment was repeated up to four
times. After each treatment, a portion of the colleted gel was
vacuum-dried at room temperature for XRD analysis, while
another portion was subjected to the particle size distribution
analysis by LPA. The resulting XRD data are shown in Fig. 6.
Two kinds of changes can be discerned clearly. First,
WO
3
ÁH
2
O phase began to appear in the gel after the third
treatment and became dominant after the fourth. This indi-
cates that the dehydration from WO
3
Á2H
2
OtoWO

3
ÁH
2
O
proceeds gradually during the treatments. Second, the XRD
intensities of (0 1 0) and (0 2 0) peaks of WO
3
Á2H
2
O tended
to increase relative to the other peaks, e.g. (2 0 0) and
(0 0 1), as the number of treatments increased.
In order to know the origins of these changes in more
detail, the time of washing and that of centrifuging were
prolonged independently in separate experiments by using
freshly prepared gels of WO
3
Á2H
2
O. After the washing for
10 h, the gel contained WO
3
ÁH
2
O partially or totally as
judged from the XRD patterns, while the XRD peak inten-
sities were not distorted. After the centrifuging for 10 h, on
the other hand, all the XRD peaks could be ascribable to
WO
3

Á2H
2
O but the peak intensities were extremely dis-
torted, as shown at the bottom in Fig. 7. The distortion
was in the same tendency as observed previously but was
much more marked. These results indicate that the partial
conversion of WO
3
Á2H
2
OtoWO
3
ÁH
2
O during the earlier
treatments was mainly caused by the washing treatment, and
that the intensity distortion in XRD pattern was by the
centrifugal treatment. The intensity distribution will be
discussed in detail later.
The particle size distribution analyses were carried out for
the sols dispersing the gels after these treatments. The mean
particle size and the range of particle size distribution are
shown as a function of total centrifuging time in Fig. 8
(upper). Note that the freshly prepared gel had already been
centrifuged for 1 h and that each washing and centrifugal
treatment included 1 h of centrifuging. It is seen that the
mean particle size as well as the upper and lower limit of
particle sizes tend to increase gradually with an increase in
total centrifuging time. The sizes of WO
3

Á2H
2
O crystallites
evaluated from the widths of XRD peaks are also shown in
Fig. 8 (lower). Since the platelike habit was evident as
mentioned before, the dimension (thickness) normal to
(0 1 0) was obtained by averaging the values based on
(0 1 0) and (0 2 0), while that (width) parallel to (0 1 0)
was done based on (2 0 0) and (0 0 1). These dimensions are
seen to increase with an increase in total centrifuging time.
Fig. 6. Changes of XRD patterns of the gels with the repetition of the washing and centrifugal treatment.
68 Y G. Choi et al. / Sensors and Actuators B 87 (2002) 63–72
Notably the upper and lower limits of particle sizes were
fairly close to the widths and thicknesses of WO
3
Á2H
2
O
crystallites of the corresponding gels, respectively. This
indicates that the sol particles consist of free (dissociated)
crystallites. It is also obvious that the size of colloidal
particles (crystallites) can be controlled well by adjusting
these treatments. As also indicated in the same figure
(Fig. 8), the crystallites size of dehydrated phases, WO
3
ÁH
2
O
and WO
3

, increased gradually with increasing total centri-
fuging time. This indicates that the crystallite size of WO
3
included in the sensor device can also be controlled by these
treatments. It is seen that for all of the investigated samples
of WO
3
, the dimensions normal to (2 0 0) are considerably
larger than those normal to (0 0 2), the crystallites are thus
suggested to be considerably anisotropic.
Fig. 9 shows a TEM image of WO
3
Á2H
2
O particles for the
sol obtained after the third washing and centrifugal treat-
ment (total centrifuging time 4 h). Most of the crystallites
are in the range of 25–50 nm in size, in fair agreement with
the analyses based on LPA and XRD. Although most of the
crystallites overlap too heavily with each other, some not
overlapping so heavily have their contrast (brightness) kept
uniform inside their peripheries, suggesting the platelike
nature of the crystallites.
3.5. Preferred orientation
As mentioned in Section 3.4, the XRD patterns of
WO
3
Á2H
2
O gels became distorted in intensity, that is,

(0 1 0) and (0 2 0) peaks were unusually intense, after the
repetition of washing and centrifugal treatments. The distribu-
tion was extremely marked after the prolonged (10 h) cen-
trifuging (Fig. 7), as also mentioned. Such distortion in
intensity is known to reflect preferred orientation of the
crystallites involved. It is considered that in the gelling process
the WO
3
Á2H
2
O crystallites (platelets) prefer to agglomerate
together with their basal planes of (0 1 0) oriented in parallel to
each other. Notably the preferred orientation of WO
3
Á2H
2
O
was found to be inherited well by the dehydrated phases of
WO
3
ÁH
2
OandWO
3
on calcination. The XRD pattern of
WO
3
ÁH
2
OinFig. 7 is strongly distorted by unusually high

intensities of (0 2 0) and (0 4 0), indicating preferred orienta-
tion in (0 1 0) plane. Similarly that of WO
3
with unusually
strong (0 0 2) indicates preferred orientation in (0 0 1). The
preferred orientation of each phase is interrelated as follows.
ð010Þof WO
3
Á 2H
2
O !ð010Þ of WO
3
Á H
2
O
!ð001Þ of WO
3
(4)
Fig. 7. XRD patterns of the gel centrifuged for 10 h after drying or calcining at the indicated temperatures.
Fig. 8. Mean and range of particle sizes of WO
3
Á2H
2
O sols (LPA) and
crystallite sizes of the corresponding gel after dried or calcined at the
designated temperatures (XRD) as a function of total centrifuging time.
Y G. Choi et al. / Sensors and Actuators B 87 (2002) 63–72 69
This phenomenon can be understood well if topotaxy is
assumed for the respective dehydration steps.
To discuss preferred orientation more quantitatively, pre-

ferred orientation index (POI) was defined for each phase on
the basis of the intensities (heights) of selected diffraction
peaks as follows.
where I (hkl) means the intensity of (hkl) peak, and the
suffixes (observed (obsd.) and studied (std.)) indicate the
intensity ratios for actual and standard (undistorted) XRD
patterns, respectively. The intensity ratios of the standard
XRD patterns were obtained by referring to JCPDS; 3.33
for WO
3
Á2H
2
O (JCPDS, 18-1420), 0.80 for WO
3
ÁH
2
O
Fig. 9. A TEM image of WO
3
Á2H
2
O particles of the sol after the third washing and centrifugal treatment (total centrifuging time 4 h).
POI ¼
½Ið010Þ=Ið001Þ
obsd:
=½Ið010Þ=Ið001Þ
std:
for WO
3
Á 2H

2
O
½Ið020Þ=Ið111Þ
obsd:
=½Ið020Þ=Ið111Þ
std:
for WO
3
Á H
2
O
½Ið002Þ=Ið200Þ
obsd:
=½Ið002Þ=Ið200Þ
std:
for WO
3
8
>
<
>
:
9
>
=
>
;
(5)
Fig. 10. Preferred orientation index as a function of total centrifuging time for the powder samples of WO
3

Á2H
2
O, WO
3
ÁH
2
OorWO
3
after dried or calcined
at the indicated temperatures.
70 Y G. Choi et al. / Sensors and Actuators B 87 (2002) 63–72
(43-0679), 1.01 for WO
3
(43-1035). The POI values thus
evaluated are summarized as a function of total centrifuging
time in Fig. 10. For WO
3
Á2H
2
O, POI is seen to increase
progressively with prolonging centrifuging time. As shown
previously (Fig. 8), the crystallites (platelets) of WO
3
Á2H
2
O
grew gradually in both thickness and width during these
treatments. It is likely that the growth of the basal plane
coupled with the centrifugal force causes the crystallites
to agglomerate together with their basal planes oriented

in parallel. It is notable that the behavior of preferred
orientation with a change in total centrifuging time is very
similar for the three phases. This POI behavior as well as
that shown in Fig. 7 clearly indicates that the topotaxy
mentioned earlier is maintained firmly at each dehydration
step of WO
3
Á 2H
2
O ! WO
3
Á H
2
O ! WO
3
.
4. Conclusions
Nano sized crystallites of WO
3
Á2H
2
O were prepared from
Na
2
WO
4
at more than 70% yield by an ion-exchange method
coupled with the ageing of the effluent (3 days) followed by
brief washing and centrifuging. The crystallites, platelike in
shape, could be easily dispersed in deionized water to form a

stable sol. The size of the crystallites increased gradually by
washing and/or centrifugal treatments of the sol, providing a
method to control the crystallite size. Because of the plate-
like nature of the crystallites, preferred orientation became
marked for the WO
3
Á2H
2
O crystallites after prolonged
centrifuging. Notably the preferred orientation was inherited
well by the dehydrated phases of WO
3
ÁH
2
O and WO
3
,
indicating the existence of topotaxy at the dehydration steps.
Acknowledgements
The authors would like to thank Dr. M. Uehara of faculty
of Engineering, Kyushu University, for technical support in
TEM observation.
References
[1] C.G. Granqvist, A. Azens, A. Hjelm, L. Kullman, G.A. Niklasson, D.
Ronnow, M. StroMMe Mattsson, M. Veszelei, G. Vaivars, Recent
advances in electrochromics for smart windows applications, Solar
Energy 63 (1998) 199–216.
[2] S. Papaefthimiou, G. Leftheriotis, P. Yianoulis, Advanced electro-
chromic devices based on WO
3

thin films, Electrochim. Acta 46 (2001)
2145–2150.
[3] A. Bessie
`
re, J C. Badot, M C. Certiat, J. Livage, V. Lucas, N.
Baffier, Sol–gel deposition of electrochromic WO
3
thin film on
flexible ITO/PET substrate, Electrochim. Acta 46 (2001) 2251–2256.
[4] M. Akiyama, J. Tamaki, N. Miura, N. Yamazoe, Tungsten oxide-
based semiconductor sensor highly sensitive to NO and NO
2
, Chem.
Lett. (1991)1611–1614.
[5] T. Inoue, K. Ohtsuka, Y. Yoshida, Y. Matsuura, Y. Kajiyama, Metal
oxide semiconductor NO
2
sensor, Sens. Actuators B 24/25 (1995)
388–391.
[6] G. Sberveglieri, L. Depero, S. Groppelli, P. Nelli, WO
3
sputtered
thin film for NO
x
monitoring, Sens. Actuators B 26/27 (1995) 89–
92.
[7] A.A. Tomchenko, I.L. Emelianov, V.V. Khatko, Tungsten trioxide-
based thick-film NO sensor: design and investigation, Sens.
Actuators B 57 (1999) 166–170.
[8] D S. Lee, S D. Han, J S. Huh, D D. Lee, Nitrogen oxides-sensing

characteristics of WO
3
-based nanocrystalline thick film gas sensor,
Sens. Actuators B 60 (1999) 57–63.
[9] A.C.C. Tseung, K.Y. Chen, Hydrogen spill-over effect on Pt/WO
3
anode catalysts, Catal. Today 38 (1997) 439–443.
[10] F. Cora
`
, C.R.A. Catlow, D.W. Lewis, Design of microporous
transition metal oxide catalysts and investigation of their synthesis
conditions, J. Mol. Catal. A: Chem. 166 (2001) 123–134.
[11] G.R. Bamwenda, H. Arakawa, The visible light induced photo-
catalytic activity of tungsten trioxide powders, Appl. Catal. A 210
(2001) 181–191.
[12] Y. Wang, Z. Aburas, K.L. Yao, Z.L. Liu, Effects of doping and
temperature on nonlinearity of WO
3
varistor, Mater. Chem. Phys. 58
(1999) 51–54.
[13] V.O. Makarov, M. Trontelj, Effect of Al
2
O
3
on the microstructure
and electrical properties of WO
3
-based varistor ceramics, J. Euro.
Ceram. Soc. 20 (2000) 747–749.
[14] D.L. Kepert, J.H. Kyle, Transient intermediates in the polymerization

of tungstate, J. Chem. Soc., Dalton (1978) 133–137.
[15] B. Gerand, G. Nowogrocki, M. Figlarz, A new tungsten trioxide
hydrate, WO
3
Á1/3H
2
O: preparation, characterization and crystal-
lographic study, J. Solid State Chem. 38 (1981) 312–320.
[16] K.P. Reis, A. Ramanan, M.S. Whittingham, Synthesis of novel
compounds with the pyrochlore and hexagonal tungsten bronze
structures, J. Solid State Chem. 96 (1992) 31–47.
[17] B. Gerand, G. Nowogrocki, J. Guenot, M. Figlarz, Structural study of
a new hexagonal form of tungsten trioxide, J. Solid State Chem. 29
(1979) 429–434.
[18] P. Delichere, P. Falaras, M. Froment, A. Hugot-Le Goff, B. Agius,
Electrochromism in anodic WO
3
films in the virgin and colored
states, Thin Solid Films 161 (1988) 35–46.
[19] T. Nanba, Y. Nishiyama, I. Yasui, Structural study of amorphous
WO
3
thin films prepared by the ion-exchange method, J. Mater. Res.
6 (1991) 1324–1333.
[20] J. Tamaki, Z. Zhang, K. Fujimori, M. Akiyama, T. Harada, N. Miura,
N. Yamazoe, Grain-size effects in tungsten oxide-based sensor for
nitrogen oxides, J. Electrochem. Soc. 141 (1994) 2207–2210.
[21] N.S. Baik, G. Sakai, N. Miura, N. Yamazoe, Hydrothermally treated
sol solution of tin oxide for thin-film gas sensor, Sens. Actuators B 63
(2000) 74–79.

[22] Y.M. Li, M. Hibino, M. Miyayania, T. Kudo, Proton conductivity of
tungsten trioxide hydrates at intermediate temperature, Solid State
Ionics 134 (2000) 271–279.
[23] C. Bala
´
zsi, J. Pfeifer, Structure and morphology changes caused by
wash treatment of tungstic acid precipitates, Solid State Ionics 124
(1999) 73–81.
Biographies
Yong-Gyu Choi received his BE degree in materials science and
engineering in 1996 and ME degree in 1998 from the Kyungsung
University in Korea. Now, he is a doctoral course student of majoring of
molecular and materials sciences in the Kyushu University. His current
research interest is development of an NO
x
sensor by spin coating method
with WO
3
sol provided by ion-exchange method.
Go Sakai has been a research associate at the Kyushu University since
1996. He received his BE degree in applied chemistry in 1991, ME degree
in 1993 and PhD in engineering 1996 from the Kyushu University. His
Y G. Choi et al. / Sensors and Actuators B 87 (2002) 63–72 71
current research work is focused on development of chemical sensors as
well as functional inorganic materials.
Kengo Shimanoe has been an associate professor at the Kyushu University
since 1999. He received his BE degree in applied chemistry in 1983 and
ME degree in 1985 from the Kagoshima University and the Kyushu
University, respectively. He joined the advanced materials and technology
laboratory in Nippon Steel Corp. and studied the electronic characteriza-

tion on semiconductor surface and the electrochemical reaction on
materials. He received PhD in engineering in 1993 from the Kyushu
University. His current research interests include the development of
chemical sensors and the analysis of solid surface.
Norio Miura joined the Kyushu University as an associate professor in
1982 and was promoted to professor in 1999. He received his BE degree in
applied chemistry in 1973, ME degree in 1975 from the Hiroshima
University and PhD in engineering in 1980 from the Kyushu University.
His current research concentrates on development of new chemical sensors
as well as other electrochemical functional devices.
Noboru Yamazoe has been a professor at the Kyushu University since
1981. He received his BE degree in applied chemistry in 1963 and PhD in
engineering in 1969 from the Kyushu University. His current research
interests include the development and application of the functional
inorganic materials.
72 Y G. Choi et al. / Sensors and Actuators B 87 (2002) 63–72