Sensors and Actuators B 137 (2009) 637–643
Contents lists available at ScienceDirect
Sensors and Actuators B: Chemical
journal homepage: www.elsevier.com/locate/snb
Comparative study of nanocrystalline SnO
2
materials for gas sensor application:
Thermal stability and catalytic activity
R.G. Pavelko
a,∗
, A.A. Vasiliev
a
, E. Llobet
a
, X. Vilanova
a
, N. Barrabés
b
, F. Medina
b
, V.G. Sevastyanov
c
a
University Rovira i Virgili, Department of Electronic, Electrical and Automatic Control Engineering (DEEEA), 43007 Tarragona, Spain
b
University Rovira i Virgili, Chemical Engineering Department (DEQ), Tarragona, Spain
c
N.S. Kurnakov Institute of General and Inorganic Chemistry RAS, Moscow, Russian Federation
article info
Article history:
Received 3 October 2008

on SnO
2
. Long-term stability of the sensors made on the basis of synthesized and commercial SnO
2
was
measured as a sensor signal deviation during 590 h of operation in 0.2, 0.6 and 1.0 vol. % of propane in air
(50% RH at 20
◦
C).
© 2008 Elsevier B.V. All rights reserved.
1. Introduction
During last decades, the development of thick film materials for
tin dioxide based gas sensors enabled the fabrication of sensitive
and relatively selective gas sensors. However, long-term stability
of the sensors remains an important problem, which restricts the
application of such devices in industrial and residential gas and
fire alarm systems, electronic noses, etc. [1,2]. The attempts to
solve this problem by chemical modification of the sensing material
showed thatthis approach gives hard to predict gas-sensing proper-
ties, because each nanoparticle of SnO
2
contains only few atoms of
dopant. As a result, the number of admixture atoms is random and
can be very far from the one calculated using average concentration
values.
In this research, we apply a completely dif ferent approach
that consists of synthesizing a very pure tin dioxide material
with extremely low level of sodium, chlorine, and sulfur contam-
ination. Using different techniques (laser-spark element analysis,
thermal X-ray diffraction, temperature programmed reduction and

◦
C and
annealed at 300
◦
C for 12 h. As a reference material SnO
2
nanopow-
der fabricated by Sigma–Aldrich (p/n 549657) was used.
The impurity content in both materials was measured by
laser-spark mass-spectrometry (EMAL-2). Transmission electron
microscopy (TEM) was used to establish the morphology and to
evaluate the particle size distributions (Jeol JEM 1011, operating at
100 kV). FTIR spectra were recorded using JASCO 680 Plus spec-
trometer (absorption mode, scan times—32, resolution—2 cm
−1
)at
0925-4005/$ – see front matter © 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.snb.2008.12.025
638 R.G. Pavelko et al. / Sensors and Actuators B 137 (2009) 637–643
room temperature. The samples were mixed with KBr powder and
pressed into self-supporting disks.
The crystallite growth kinetics was studied by means of Siemens
D5000 diffractometer (Bragg-Brentano parafocusing geometry, ver-
tical  −  goniometer, Ni-filtered CuK␣ radiation) equipped with
an Anton-Paar HTK10 heating platform (TXRD). The data were
recorded during isothermal annealing at 700
◦
C after 1, 6, 18 and
30 h.Crystallite size was determined using the Fundamental Param-
eters Approach convolution algorithm implemented in the program

N
2
physisorption adsorption–desorption isotherms at 77 K was
measured using Micromeritics ASAP 2010 surface analyzer. Before
analysis, all the samples were degassed in vacuum at 393 K for 6 h.
2.2. Sensor fabrication
Both synthesized and commercial SnO
2
materials were impreg-
nated with water solution of palladium complex Pd(NH
3
)
4
(NO
3
)
2
.
Then, the powders were dried at 110
◦
C, calcined at 300
◦
C for 10 h
and annealed at 500
◦
C in order to achieve the Pd complex decom-
position. The preset catalyst to carrier ratio was equivalent for all
sensing materials: w
Pd
= 1wt. %. Then the powders were mixed with

gas
is the sensing layer resistance in target gas atmosphere and R
air
is the resistance in pure carrier gas. Long term stability was esti-
mated as a total deviation of sensor signal after certain periods
of time and was defined as ((S
n
− S
n−1
)/S
n−1
) × 100%, where S
n
is sensor signal after 330, 460 or 590 h and S
n−1
is the previous sen-
sor (e.g. (S
330
− S
0
)/S
0
, where S
0
and S
330
are sensor signals after the
stabilization and 330 h of operation respectively).
Table 1
Concentration of main impurities in SnO

148 m
2
/g and 17m
2
/g, respectively, which corresponds to parti-
cle sizes of 3 nm for synthesized and 26 nm for commercial SnO
2
(calculated assuming spherical particle shape).
TEM investigation of the oxides shows that synthesized SnO
2
consists of uniform particles with mean diameter 3–5 nm (Fig. 1).
Particles of commercial SnO
2
are polydisperse: the least diameter
is ∼10 nm, while the largest is ∼70 nm (Fig. 2a). These results are in
good agreement with BET area of the materials.
Both materials have cassiterite type crystalline structure. The
mean crystallite size of SnO
2
nanopowders before heat treatment,
calculated using TOPAS sof tware, was equal to 1.3 and 30 nm for
synthesized and commercial SnO
2
, respectively. The temperature
range for TXRD study of crystallite growth kinetics was chosen
by several reasons. The usual process of formation of metal oxide
sensing layers in semiconductor and thermocatalytic gas sensors
requires annealing at 700–800
◦
C [5]. This could be used not only

0
is the initial crystallite size, D the current crystallite size,
K the rate constant, N the grain growth exponent factor and t the
annealing time.
The difference between both materials becomes more evident,
if we compare their integral crystallite growth rate. The integral
rate, calculated as D/t (where D the change in crystallite size
for 30 h of annealing and t the annealing time in hours) was found
equal to 0.026 and 0.14 nm/h for synthesized and commercial SnO
2
,
respectively. The main contribution to high integral growth rate of
commercial SnO
2
was made during the first 2 h of annealing when
the rate was by a factor of 7 higher than for the synthesized SnO
2
.
Fig. 3. Crystallitesizeofsynthesized and commercialSnO
2
as a function of annealing
time at 700
◦
C.
The difference in crystallite growth rates could be explaine d
by several factors. First, the materials have different morphology,
which complicates the comparison of the results. According to
[6] for example, porosity of nanomaterials plays significant role
in crystallite growth kinetics. The nitrogen physisorption revealed
that commercial SnO

sized SnO
2
equals ∼7 nm and particles remain uniform in shape,
while for commercial SnO
2
the particles are inthe range 20–150 nm.
The change in the particle size affects the electrical properties of the
material and this will have negative impact on long-term stability
of the sensor [8].
In addition, we found that after 1 h of isothermal annealing, a
new phase appears on the surface of commercial SnO
2
particles
(Fig. 2b). This new phase is a result of the thermal migration of
impurities (Na and Cl ions), which presence on the surface can lead
to inhibition of catalytic processes [9].
FTIR spectroscopy of nanomaterials can provide useful informa-
tion about chemical modification of the solid. The spectra were
recorded on JASCO 680 Plus spectrometer at room temperature
for SnO
2
samples, which undergone isothermal annealing at 700
◦
C
during 1, 6, 18 and 30 h. The spectra of commercial and synthesized
powders are shown in Fig. 4.
While the spectrum of commercial SnO
2
is quite simple with
only one intense band between 400 and 800 cm

corresponds to ter-
minal vibration of Sn–O bond, whereas vibrations at higher wave
numbers can be ascribed to bridged ones [10]. Thus, a decrease
in adsorption band at around 500 cm
−1
could be a result of ter-
minal group lessening. This allows us to conclude that either
chemical modification of the surface or growth processes occur
during isothermal annealing. The following discussion will show
that the chemical modification of the surface is more signifi-
cant.
The bands in the range of 2000–3650 cm
−1
can be assigne d
to hydroxyl stretching vibrations 
OH
(Sn–OH) of both, terminal
and bridged groups, and to stretching vibration in hydrogen bond
OH···O. During the annealing, the band becomes more intense and
several shoulders appear on the right and on the lef t side of the
band maximum, located at 3416 cm
−1
. The latter, together with
a small left shoulder around 3230 cm
−1
, should be ascribed to
stretching vibrations of bridged Sn–OH bond. The right shoulder
around 3545 cm
−1
is related, probably, to stretching vibrations of

The evolution of the peak shape with annealing time indicates that,
after approximately 6 h, there are mostly dimers and polymers of
water molecules on the SnO
2
surface (1616 and 1635 cm
−1
). The
same situation is observed with commercial material. The weak
peak at 1620 cm
−1
is a superposition of two peaks at 1616 and
1635 cm
−1
. However, the quantity of adsorbed water increases with
annealing time only in the case of synthesized SnO
2
, whereas com-
mercial material loses molecular water during annealing (decrease
in peak intensity).
In a range of 900–1250 cm
−1
, a broad intense band appears after
1 hof annealing (Fig. 5b). According to numerous investigations, this
band corresponds to the deformation mode ı
OH
(Sn–OH) of termi-
nal OH groups [10]. It is remarkable that the spectrum of the SnO
2
powder dried after wet synthesis does not have any corresponding
vibration mode. The intensity of the band increases with annealing

,
O
−
and O
2−
) the synthesized SnO
2
probably will demonstrate in the
same temperature range high catalytic activity in oxidation reaction
and as a result higher sensitivity to reducing gases [12].
To investigate the oxidation activity of synthesized and com-
mercial materials, we performed TPR study of SnO
2
with and
without Pd (Fig. 6). TPR profile of the synthesized SnO
2
displays
one broad intense peak at 360
◦
C, followed by the onset of bulk
reduction (that is the reduction of SnO
2
particle bulk) starting from
550
◦
C. Commercial SnO
2
shows a different profile with small peak
around 200
◦

the reduction of surface species is observed), the profile of syn-
thesized Pd-SnO
2
solid shows a clear shift to lower temperatures.
This reduction peak is shifted from 360
◦
Cto200
◦
C. However,
the reduction peak for bulk SnO
2
species remained constant. The
H
2
oxidation already starts at 100
◦
C and goes on up to 400
◦
C,
demonstrating significant decrease in activation energy of oxi-
dation processes. This fact is only observed for the synthesized
samples.
Thus, the TPR study has proved our assumption about high cat-
alytic activity of synthesized SnO
2
and it becomes apparent that
high content of chemisorbed water could be an evidence for high
oxidation ability of material at elevated temperatures.
3.2. Sensor characterization
Long-term stability measurements were performed for 590 h.

2
results in the highest signals
(defined as R
air
/R
gas
) to all propane concentrations. For example,
the signal to 0.4% of propane in synthetic air at 50% RH after sensor
stabilization is equal to 42 for synthesized SnO
2
and 4.5 for com-
mercial SnO
2
. The drastic difference between both materials can be
explained by higher catalytic activity and high specific surface of
the synthesized material.
A distinctive feature of commercial SnO
2
is that the signals to all
propane concentrations increase with operation time. The crystal-
Fig. 7. Sensor layer resistance of synthesized (a) and commercial (b) SnO
2
with 1 wt.
% of Pd to 0.4% of propane in synthetic air at 50% RH.
lite growth cannot lead to this phenomenon. We explain this relying
on TEM results, which demonstrated the formation of a new phase
on the SnO
2
surface (Fig. 2b). After 100 h of phase stabilization at
450

concentrations. The highest signal drift is obtained for sensing
materials on the basis of commercial SnO
2
. The highest contribu-
tion to the signal deviation for both types of sensors was observed
after 330 h of sensor operation. The total signal drift after 590 h
varies (depending on the target gas concentration) from 0.7% to
1.4% for synthesized SnO
2
and from 16% to 25% for commercial
642 R.G. Pavelko et al. / Sensors and Actuators B 137 (2009) 637–643
Fig. 8. Total signal drift depending on propane concentration against operation time
of the sensors.
SnO
2
in the concentration range under investigation. The high
stability of the synthesized material can be explained only by ther-
mal stability of nanoparticles which is a result of low impurity
content.
4. Conclusion
XRD crystallite size and TEM particle size analysis indicate the
remarkable thermal stability of synthesized SnO
2
with low impu-
rity level. On the other hand commercial SnO
2
with Na and Cl
impurities has poor thermal stability resulting in higher crystallite
growth rate and poly-dispersed particle size after annealing. It was
shown that during isothermal annealing, impurities tend to form

signals to different propane concentrations. The low impurity level
of the synthesized materials results in the lowest signal drift.
We found that crystallite growth causes signal decrease by 1%
during 590 h of operation for this material. On the other hand,
high impurity concentration in commercial material leads to signal
increase by 25% after 590 h. This indicates that, apart from crys-
tallite growth (resulting in signal decrease), impurities take part
in surface oxidation processes causing inhibition of the propane
oxidation and remarkable signal drift.
Acknowledgments
This work has been funded in part by CICyT under Grant
no. TEC2006-03671. R. Pavelko gratefully acknowledges a Ph.D.
scholarship from URV. Also authors would like to thank Dr. F.
Gispert—Guirado from the Laboratory of X-Ray Diffraction (URV,
Spain) for his contribution to XRD experiments and spectra
processing.
References
[1] Y. Ozaki, S. Suzuki, M. Morimitsu, M. Matsunaga, Enhanced long-term stability
of SnO
2
-based CO gas sensors modified by sulfuric acid treatment, Sens. Actuat.
B: Chem. 62 (2000) 220–225.
[2] C. Pijolat, B. Riviere, M. Kamionka, J.P. Viricelle, P. Breuil, Tin dioxide gas sensor
as a tool for atmospheric pollution monitoring: problems and possibilities for
improvements, J. Mater. Sci. 38 (21) (2003) 4333–4346.
[3] TOPAS. General profile and structure analysis software for powder diffraction
data. V 3.1, Bruker AXS GmbH, Karlsruhe, Germany.
[4] A.A. Vasiliev, Physical and chemical principles of the design of gas sensors
based on metal oxides and structures metal/solid electrolyte/semiconductor.
Dissertation, Doctor of Science Degree, Moscow, 2004.

and Inorganic Chemistry, (Russian Academy of Science, Moscow). At present he is
PhD student in University Rovira i Virgili (Tarragona, Spain) in the Electronic Engi-
neering Department. His research interests concern synthesis of dispersed materials,
material science, experimental and theoretical study of surface processes related to
metal oxide gas sensors.
Alexey A. Vasiliev graduated from Moscow Institute of Physics and Technology in
1980, obtained his PhD in 1986 for the “Study of the kinetics of low-temperature
reactions of atomic fluorine by ESR method”. Gained his Dr. of Science degree (habil-
itation) in solid state microelectronics in 2004 for the investigation of “Physical and
chemical principles of design of gas sensors based on metal oxide semiconductors
and MIS structures with solid electrolyte layer”. Recently he is working in Sensor
group of the University Rovira i Virgili (Tarragona, Spain) and at Russian research
center Kurchatov Institute (visiting position). Research interests are related with
the study of the kinetics and mechanisms of heterogeneous processes related with
chemical sensing, kinetics and mechanisms of electrochemical processes in liquid
and solid electrolytes.
Eduard Llobet graduated in telecommunication engineering from the Universitat
Politècnica de Catalunya (UPC), (Barcelona, Spain) in 1991, and received his PhD in
1997 from the same university. He is currently an associate professor in the Elec-
tronic Engineering Department at the Universitat Rovira i Virgili (Tarragona, Spain).
His main areas of interest are in the design of semiconductor and carbon nanotube
based gas sensors and in the application of intelligent systems to complex odor
analysis.
Xavier Vilanova graduated in telecommunication engineering from the Universi-
tat Politècnica de Catalunya (UPC), (Barcelona, Spain) in 1991, received his PhD in
1998 from the same university. He is currently an associate professor in the Elec-
tronic Engineering Department at the Universitat Rovira i Virgili (Tarragona, Spain).
His research activities are related to semiconductor gas sensors development and
characterization, as well as, gas sensors systems design.
Noelia Barrabés received MS from Chemical Engineering Department in Rovira i