Sensors and Actuators B 232 (2016) 402–409

Contents lists available at ScienceDirect

Sensors and Actuators B: Chemical

This document is the Accepted Manuscriptunder the CC-BY-NC-ND 4.0 that appeared in final form in Sensors and Actuators B 232 (2016)
This manuscript version is made available version of a Published Work license http://creativecommons.org/licenses/by-nc-nd/4.0/.
Link to the final version: http://dx.doi.org/10.1016/j.snb.2016.03.091.
402–409. http://dx.doi.org/10.1016/j.snb.2016.03.091.
journal homepage: www.elsevier.com/locate/snb

Site-selectively grown SnO2 NWs networks on micromembranes for
efficient ammonia sensing in humid conditions
Jordi Samà a , Sven Barth b , Guillem Domènech-Gil a , Joan-Daniel Prades a , Núria López c ,
Olga Casals a , Isabel Gràcia d , Carles Cané d , Albert Romano-Rodríguez a,∗
a
Universitat de Barcelona (UB), MIND-Departament of Electronics and Institute of Nanoscience and Nanotechnology (IN2UB), c/Martí i Franquès 1,
E-08028 Barcelona, Spain
b
Vienna University of Technology (TUW), Institut of Materials Chemistry, Getreidmarkt 9/BC/02, A-1060 Vienna, Austria
c
Institute of Chemical Research of Catalonia (ICIQ), Av. Països Catalans 16, E-43007 Tarragona, Spain
d
Consejo Superior de Investigaciones Científicas (CSIC), Institut de Microelectrònica de Barcelona (IMB-CNM), Campus UAB, E-08193 Bellaterra, Spain

a r t i c l e

i n f o

Article history:
Received 13 November 2015
Received in revised form 26 February 2016
Accepted 20 March 2016
Available online 22 March 2016
Keywords:
SnO2 nanowires network
Ammonia gas sensor
Localized growth
Low power consumption

a b s t r a c t
SnO2 NWs networks on heated micromembranes have been characterized as ammonia sensors. The
approach allows achieving reproducible growth and stable and long-lasting ammonia sensors with sitespecific grown SnO2 NWs. The devices have been tested both in dry and humid conditions showing
response time down to two minutes. Sensors have been tested up to 1 month, only presenting variation
of the base resistance with full retention of the response towards the gaseous analytes. Different concurrent sensing mechanisms have been identified relating the determined sensing kinetics with previous
theoretical calculations. Specifically, oxygen dissociation seems to play a key role in the overall ammonia
sensing sequence. In humid conditions, moisture reduces the response to ammonia but also lowers the
activation energy of the reaction process.
© 2016 Elsevier B.V. All rights reserved.

1. Introduction
Ammonia is a toxic gas with irritant properties that can injure
the respiratory tract [1]. Anthropogenic emissions mainly comes
in a 95% from agriculture, where ammonium salts are widely used
as a fertilizer [2]. On the other hand, swine farms can achieve high
concentrations of ammonia up to 100 ppm generated by decomposition of pig manure via metabolic activities of bacteria and fungi
[3].
Furthermore, ammonia, together with urea, is used in automotive applications for selective catalytic reduction (SCR) of nitrogen
oxide (NOx ) to nitrogen in order to avoid harmful gas emissions
in diesel heavy weight engines, which gives also H2 O as a byproduct [4,5]. Accurate measurements of the ammonia concentration
for SCR are needed for an efficient process and the recycling of the
non-reacted ammonia fraction.
SnO2 and other metal oxides have been explored as chemoresistive gas sensing materials since the development of Taguchi
sensors four decades ago [6]. More recently, nanostructured metal
oxide materials were developed in order to improve the sensing

∗ Corresponding author.
E-mail address: aromano@el.ub.edu (A. Romano-Rodríguez).
http://dx.doi.org/10.1016/j.snb.2016.03.091
0925-4005/© 2016 Elsevier B.V. All rights reserved.

properties by enhancing the surface to volume ratio [7]. Among
them, nanowires are one of the most studied and promising
structures because of their well-controlled chemical and physical
properties [8]. Nanowire-based gas sensors have been developed by
using individual structures [9] or in a network form where the electrical current flows through the nanowires along the axial direction
[10]. Single SnO2 nanowire gas sensors have been demonstrated
as sensitive and effective devices, but are constrained by ad hoc
contact fabrication, which is time consuming and difficult to systematize. A possibility for its industrial processing would be the
alignment of nanowires using dielectrophoretic forces [11].
Monocrystalline SnO2 structures offer better stability to gas
response than multigrained ones because the former expose more
stable, homogeneous and well defined crystal planes to gases. Single crystalline nanowires can be prepared by metal seed supported
gas phase growth techniques, such as chemical vapor deposition
(CVD) or pulsed laser deposition (PLD) [8]. The growth of nanowires
often requires high temperatures; for instance, SnO2 nanowires
form around 700 ◦ C [12]. The growth process according to the
vapor-liquid-solid mechanism (VLS) is frequently performed by
CVD, involving long times for heating up and cooling down due
to the high thermal mass of the furnace; additionally, the entire
heated substrate is coated and not only the active area. Consequently, the transfer of nanowires to the final device is required,

J. Samà et al. / Sensors and Actuators B 232 (2016) 402–409

403

Fig. 1. (a) SEM general view of the micromembrane with Pt interdigitated electrodes (IDE) on the top. Brighter area corresponds to the grown SnO2 NWs; (b) SEM image of
SnO2 NWs contacting the Pt electrode (on the left side);.

adding complexity to the approach. Precise growth and integration
of nanowires in an efficient and reproducible way is one of the most
challenging aspects for gas sensors based on such structures.
Thus, nanostructures grown directly on the final electronic
device severely reduces fabrication steps and manufacturing cost.
At the same time, a low power consumption during the growth
is achieved due to the small mass of the heated area allowing
fast heating and cooling processes. The approach presented here is
based on direct growth of nanowires onto micromembranes, small
membrane areas of 400 × 400 ␮m with an incorporated heater
that provides the temperature necessary for the growth of SnO2
nanowires and for the sensor operation [13]. The nanostructures
are available after the growth for resistor-type sensing applications without requiring any additional fabrication step due to the
bridging of the interdigitate on top of the heater.
In this work, we present an ammonia sensor based on SnO2
nanowires grown by this direct integration process of nanostructures with the electronic platform and their response in dry and
humid air.

Constant flow of 200 ml/min was kept both for dry and humid air
measurements. Water vapor was added by deviating a part of the
synthetic air flow through a bubbler. Relative humidity provided by
the bubbler was calibrated as a function of the air flow before NWsbased sensor characterization using a commercial humidity sensor
at 20 ◦ C. The relative humidity conditions indicated throughout
the paper are referenced to the RH at room temperature. Keithley
2602A dual Source Measure Unit allowed to control simultaneously the resistance of the sensor and the voltage for heating the
micromembrane. Electrical measurements and flowing gases were
controlled by a home-developed Labview software.
SnO2 NWs based gas sensors were characterized towards
ammonia by allowing 5 hours to stabilize the baseline, and then
first exposing for 1 h to ammonia. The sensors were then exposed
to air for 2 h to recover the baseline. This sequence was repeated
for the different ammonia concentrations for both dry and humid
characterization.

2. Experimental details

3.1. Temperature behavior in absence of ammonia and humidity

Bulk micromachined substrates were used as a platform for the
growth of nanowires. Chips consist of 4 isolating micromembranes
with a thickness about 1.1 ␮m with reduced thermal mass that
provides fast thermal response. Each micromembrane contains a
poly-Si doped buried heater embedded in Si3 N4 with an isolating
layer of SiO2 on the top; Pt interdigitated electrodes are on top of
the layers. The details of a very similar platform are described elsewhere [13]. The micromembranes are mounted onto a TO-8 holder,
and wire bonded to them.
The synthesis of NWs on top of micromembranes is carried out
according to the growth process reported in [14]. A non-continuous
Au layer is deposited by 5–10 seconds sputtering and used as a catalyst of the VLS mechanism growth based on the decomposition of
the Sn(Ot Bu)4 precursor. In order to carry out the growth, the chip
with 4 micromembranes is introduced in a quartz-chamber, where
the heaters are externally biased while the gaseous precursor is
flowing through the chamber. Synthesis is carried out locally only
on the heated membranes where the thermal decomposition of
the precursor occurs. Low power consumption is required for the
growth: 48 mW are needed to achieve this temperature. Growth
of NWs takes 30 min including the heating up and cooling down
ramps. SnO2 NWs are grown in low vacuum conditions at a pressure
of 0.1 mbar and at temperature around 700 ◦ C.
The response of SnO2 NWs was characterized in a home-made
stainless steel chamber of 8.6 ml volume connected to a Gometrics MGP2 gas mixer with 4 Bronkhorst Mass-Flow Controllers.

A low magnification SEM image of a micromembrane, where
SnO2 NWs have grown site-selectively, is presented in Fig. 1(a).
The brighter area in this figure denotes the growth of the nanostructures, which also follows the temperature profile provided by
the heater. A higher magnification SEM picture of NWs close to the
Pt electrode is shown in Fig. 1(b), where no NWs are found on top of
it; a Pt/Au alloy formed during annealing has not acted as a catalyst
for nanowires growth.
The crystalline structure of tin dioxide NWs has been analyzed
by High Resolution Transmission Electron Microscopy (HRTEM),
represented in Fig. 1 Fig. 2(a). As reported in a previous work
[14], the locally grown SnO2 nanowires are monocrystalline, with
tetragonal (rutile) phase, with predominant [101] growth direction. A cross-section SEM image of a device with slightly shorter
nanowires is shown in Fig. 2(b), where a high density of NWs is
found with a length between 2 and 5 ␮m, and the contact between
the network of NWs can be observed. Besides, a very thin layer
is observed between the nanowires and the SiO2 layer. The twodimensional growth occurs simultaneously during the synthesis
of the nanowires with a much lower growth rate and leads to a
nanocrystalline layer of SnO2 because it is a non-catalyzed process.
The thickness of this layer is between 30 and 80 nm and the layer
roughness observed is due to the base of nanowires that remained
after the sample preparation.
The behavior of the electrical resistance of SnO2 NWs at different temperatures in synthetic air atmosphere has been investigated

3. Results and discussion

404

J. Samà et al. / Sensors and Actuators B 232 (2016) 402–409

Fig. 2. (a) HR-TEM image of a monocrystalline SnO2 NW that crystallizes in rutile phase. Inlet FFT illustrates the [101] predominant growth direction; (b) Cross-section of
SnO2 NWs-based sensor. A very thin layer is observed below the NWs of a thickness between 30–80 nm.

Fig. 3. (a) Resistance of SnO2 locally grown NWs as a function of temperature. The minimum in resistance reflects the change in the adsorbed oxygen specie; (b) Transient
response of the sensor resistance in a change of temperature illustrating the increase in resistance for increasing temperatures above 200 ◦ C. Blue line represents the evolution
of temperature.

in order to study the oxygen species that are chemisorbed at the
surface of the metal oxide nanostructure (Fig. 3(a)). The resistance
of SnO2 NWs at room temperature is 2850 , and decreases for
increasing temperatures up to 200 ◦ C. Above this temperature and
up to 450 ◦ C, the maximum studied temperature, the resistance
increases. At temperatures above 500 ◦ C, the resistance changes
the tendency and decreases for increasing temperatures, which is
related to the lower coverage of O− oxygen species. The same temperature behavior has been obtained in other sensors based on SnO2
[15]. The SnO2 nanowires present a low resistance, which is probably due to a highly defective surface. I–V curves of the sensor show
an ohmic behavior even at room temperature, which remarks the
absence of a Pt/SnO2 Schottky barrier that would be expected for a
pure Pt/SnO2 interface. The interface between the Pt electrodes and
the SnO2 NWs is partially responsible for a lower resistance than if
a Schottky barrier would be present.
At low temperatures atmospheric oxygen is physisorbed [15].
This mechanism takes place without electron transfer and therefore, cannot change the resistivity of the semiconductor. During
physisorption, the change in resistance observed is purely due
to the excitation of charge carriers. Molecular oxygen (O2 − ) is
chemisorbed at the surface at temperatures around 150–200 ◦ C
with some degree of charge transfer from the oxide towards
the adsorbate. Above 200 ◦ C the oxygen trapped by means of
chemisorption can dissociate in two atomic oxygen species (O− )
according to:

−
−
O2(g) + e− → 2O(g)

(1)

Thus, O− leads to capture more electrons from the semiconductor and thus, increase its resistance [15,16]. The observed resistance
behavior has also been reported for other SnO2 structures at similar temperatures [16]. Therefore, the resistance is related to the
particular oxygen species adsorbed at the metal oxide surface. The
high temperature range starting at 600 ◦ C renders oxygen atoms
that share two electrons (O2− ) with the metal oxide [15]. However,
these high temperatures are beyond the scope of the present work.
Transient response of the SnO2 NW device resistance with
increasing temperature from 400 to 450 ◦ C is represented in
Fig. 3(b). The low thermal mass of the micromembranes provides
a fast thermal response of the order of tens of microseconds to
achieve 200 ◦ C [13] which leads to the initial fast decay in resistance when the temperature is changed, but is afterwards followed
by a slower increase. The gradual increase in resistance is related
to the chemisorption of atomic oxygen species, since temperature
changes induce a new dynamic equilibrium between the wire and
the adsorbates.

3.2. Response to ammonia in dry synthetic air
The change in resistance of the locally grown nanowires against
ammonia in synthetic air has been measured. The concentration
of ammonia has been varied between 10 and 40 ppm, which is in
the range of time-weighted average exposure limit recommended
by NIOSH (25 ppm) for up-to 10 h workday [17]. The evolution
of the sensor resistance at different temperatures is represented
in Fig. 4(a), where the decrease in resistance in the presence of

J. Samà et al. / Sensors and Actuators B 232 (2016) 402–409

405

Fig. 4. (a) Evolution of SnO2 NWs resistance in front of different concentration of ammonia in synthetic air; (b) Response of the test represented in a) in function of temperature;
(c) Arrhenius plot of the response time for pulses of 30 ppm of NH3 . Symbols are experimental values and lines are the fitted exponential decay.

ammonia, expected for an n-type semiconductor gas sensor like
tin dioxide, is observed.
The response of the sensor is defined as:
Response (%) =

Rair − RNH3
Rair

(2)

The response as a function of temperature is represented in
Fig. 4(b). Localized grown SnO2 NWs show a response of up to 36%
for 40 ppm of NH3 at 400 ◦ C, temperature at which the response
presents a maximum. The response time is defined as the time to
evolve from 10% to 90% of the steady state value. A response time
as low as 2 min is obtained, again, at 400 ◦ C. An Arrhenius plot of
the response time as a function of temperature is represented in
Fig. 4(c), where an exponential behavior is deduced.
Three competitive reactions are described in the literature for
ammonia oxidation on metal oxide surfaces [18–20]:
−
˛2NH3(g) + 3O˛(s) → ˛N2(g) + ˛3H2 O(g) + 3e− T < 400◦ C

(3)

−
˛2NH3(g) + 4O˛(s) → ˛N2 O(g) + ˛3H2 O(g) + 4e− T < 400◦ C

(4)

−
˛2NH3(g) + 5O˛(s) → ˛2NO(g) + ˛3H2 O(g) + 5e− T > 400◦ C

(5)

where (g) stands for gas, (s) for surface, e− is a conduction electron, ˛ is a coefficient that is equal to 1 for atomic oxygen O− or
−
2 for molecular oxygen O2 ionosorbed species and consequently,
depends on the working temperature. The three reaction mechanisms are supported by chemisorbed oxygen at the surface of the
metal oxide. Our work is focused at temperatures above 200 ◦ C in
order to achieve acceptable response time for the gas sensing and,
as a consequence, chemisorbed atomic oxygen is the specie that
oxidizes ammonia molecules at the surface of the SnO2 nanowires.
It is well known from catalysis that in a large number of metal
oxide catalysts, depending on the temperature one specific reaction
is preferred over the others [20,21]. In this direction, the sensing
Eqs. (3)–(5) are by no means easy to disentangle. Indeed, these
reactions account for oxygen adsorption in the form of the active
species and subsequent H stripping from the ammonia molecules
till a relevant intermediate containing the NN or NO bond is formed.
When this takes place, further H elimination with the concomitant
formation of H2 O is carried out. Thus, Eqs. (3)–(5) share more than
70% of their elementary steps. Consequently, in the low temperature regime, i.e., below 400 ◦ C approximately, the Reactions (3) and
(4) take place simultaneously.
Reaction (5) dominates at temperatures above 400 ◦ C and surface coverage governs the reaction since higher energy barriers
can be surpassed at this temperature, the NO byproduct becomes
dominant [20,21]. The nitrogen containing products of the three
reactions are therefore oxidized as a function of temperature: the

higher the temperature at which the reaction takes place, the more
oxidized is the nitrogen containing product. At the same time, the
more oxidized is the nitrogen product also the higher the activation
act
act
act
energy, following the relationship E(3) < E(4) < E(5) [20].
The activation energy of the sensor response towards ammonia
has been obtained from the response time, which is represented as
a function of temperature in Fig. 4(c) in an Arrhenius plot. Three
different kinetic regions are defined from the plot according to the
experimental measurements: (I) temperatures below 200 ◦ C; (II)
between 200 and 400 ◦ C and (III) above 400 ◦ C, each of them with
different activation energies.
The main reaction that takes place at lower temperatures
(region I) is ammonia oxidation involving molecular oxygen ions,
since this species is the main oxygen form at this temperature
range, which gives a different activation energy to the one deduced
from region II. Activation energy from this region cannot be derived
due to the small number of points measured, but a lower slope
than region II in the ␶(1000/kT) is traced. The activation energy
act
obtained from Arrhenius plot in region II is EII = 0.35 ± 0.04eV ,
where ammonia oxidation is supported by atomic oxygen. This
process was also found to provide higher activation energy than
molecular oxygen in [22]. Therefore, differences in sensing kinetics
are explained by the different oxygen species promoting the NH3
oxidation.
On the other hand, this region comprehends the temperature
range where Reactions (3) and (4) can take place on the tin dioxide surface. The exponential behavior of the response time in the
Arrhenius plot suggests that the same mechanism takes place over
the whole temperature range. As pointed out before, both mentioned reactions can occur simultaneously at these temperatures;
both share many elementary steps. The rate determining step O2
activation is the same and, thus, a unique dependence is found. Furthermore, the ratio between oxygen and ammonia partial pressures
(pO2 /pNH3 ) also influences the selectivity towards one of the three
reactions. As shown in [20], for pO2 /pNH3 > 10, the selectivity is lost
and N2 and N2 O are produced in equivalent percentage at temperatures below 400 ◦ C. On the other hand, for low pO2 /pNH3 ratios
(<0.1), N2 production is enhanced [18,20], and N2 O production is
almost negligible. Consequently, as our experiments are carried out
in high pO2 /pNH3 , sensing mechanism in region II can be considered
as the addition of Reactions (3) and (4) taking place in parallel.
The activation energy detailed in this work is lower than in [20]
and in a previous work [23], which obtained an activation enthalpy
of 0.74 eV (range between 210–260 ◦ C, reaction mixture of 10% of
ammonia and 90% of O2 ) and 0.5 eV (150–300 ◦ C), respectively. The
study of SnO2 as a catalyst for NH3 oxidation concluded that the
formation of Reaction (3) started at temperatures below 200 ◦ C,

406

J. Samà et al. / Sensors and Actuators B 232 (2016) 402–409

and Reaction (4) takes place for temperatures higher than 200 ◦ C
[20]. Furthermore, in previous density functional theory (DFT) studies we concluded that the rate-determining step of NH3 sensing
mechanism in a single SnO2 monocrystalline nanowire is oxygen
adsorption, with an associated activation energy of 0.5 eV, similar to
the experimental value obtained in that work [23]. Notice, though
that the errors intrinsic to DFT in particular when dealing with the
description of O2 make the DFT value a qualitative estimate. Still,
if oxygen adsorption is energetically favored, i.e. oxygen binding
energy to the surface increases, then the activation energy of the
whole process could be lowered.
A change in tendency of the response time as a function of temperature is observed again at 450 ◦ C in Fig. 4(c), the labelled region
III. The variation in transient behavior at 450 ◦ C suggests that a different reaction takes place, which could correspond to Reaction (5),
since this latter starts to dominate at about 400 ◦ C. The temperature intervals for NO reaction are in agreement with other works,
and are also supported by the transient decay of the sensor resistance at 400 and 450 ◦ C. The initial interaction of ammonia with the
metal oxide surface leads to a fast decay of the NWs resistance at
these temperatures, followed by a slow increase during the exposure to NH3 . The same behavior has also been observed in metal
oxide sensors like WO3 [19] and In2 O3 /MgO bilayer structure [24]
in the characterization of their responses towards ammonia.
Nitric oxide (NO) is easily and readily oxidized to NO2 at 400 ◦ C
[24,25]. SnO2 NWs are known to respond to less than 100 ppb of
NO2 [26], a value easy to reach during the ammonia and nitric
oxide oxidation. Other products generated in the reactions cannot
explain the observed increasing resistance. On the one hand, H2 O
acts as a reducing agent to tin dioxide, i.e., diminishes the resistance of tin oxide NWs and, consequently, as a product of Reactions
(3)–(5), could not provide the increase in resistance. N2 is inert at
the analyzed temperature range and does not affect the conductivity. Furthermore, SnO2 without additives on its surface requires
tens of ppm of N2 O [27] to achieve an observable response. As a
result, the only reaction that will give rise to the observed behavior is the oxidation of NO to NO2 , which is the only species able to
promote the increasing resistance. Thus, NO2 attaches to an oxygen
vacancy at the SnO2 surface [28], trapping an electron and therefore the resistance of SnO2 is increased as a consequence of the
reactions:
2NO(g) + O2 → 2NO2(g)
−

NO2(g) + e →

−
NO2(s)

(6)
(7)

This fact is confirmed by the reduction of the response for
temperatures above 400 ◦ C. Actually, the initial resistance decay
towards ammonia at 450 ◦ C is larger than the decay when the sensor is heated at 400 ◦ C. However, the further resistance increase due
to NO2 adsorption reduces the final response (see Fig. 5(d)). Therefore, at this temperature the response of the sensor is explained by
Reactions (5)–(7) that take place simultaneously, being (6) and (7)
secondary reactions.
The observed low resistance of the NWs network at room temperature is an indicator of the presence of surface defects, i.e., that
oxygen vacancies are widely present at the surface of the NWs.
Consequently, the gas response is probably lowered partially since
chemisorbed oxygen is necessary for the ammonia sensing mechanism, as Reactions (3)–(5) describe.
3.3. SnO2 NWs sensing mechanisms in humid conditions
3.3.1. SnO2 NWs interaction towards H2 O
SnO2 NWs exhibit a reversible response towards different concentration of water vapor at different examined temperatures.
Water vapor is expressed as relative humidity in %, i.e., the ratio

of partial water vapor pressure and saturation pressure obtained at
room temperature (20 ◦ C). The transient in electrical signal of the
sensor towards the presence of water vapor is shown in Fig. 5(a) at
an operation temperature of 400 ◦ C. A clear decrease in resistance
due to the presence of water vapor is observed, which is reversible
as can be seen in the same figure since the baseline value is fully
recovered when water vapor is removed from the chamber.
The behavior of the NWs resistance as a function of the relative
humidity denotes that water vapor acts as a reducing gas for SnO2
nanostructures. SnO2 NWs have been reported as high sensitivity
humidity sensor [29]. There are several mechanisms reported in
literature for the interaction of H2 O with the tin dioxide surface.
Dissociation of water involves the reaction of hydroxyl groups with
oxygen atoms from the lattice, providing terminal OH group, and
thus, releasing two electrons per molecule to the conduction band
according to the following Eqs. [30–32]:
..
H2 O + 2Snlat + Olat ↔ 2 (OH − Snlat ) + VO + 2e−

(8)

where Snlat and Olat stands for a tin and oxygen atom in lattice
..
position, respectively, and VO is a doubly ionized oxygen vacancy.
Another mechanism related to chemisorbed oxygen at the SnO2
surface is the competitive reaction of water with pre-adsorbed oxygen ions, which is similar to Eq. (8), but involved oxygen is an atomic
chemisorbed specie that releases an electron to the metal oxide
[30]:
−
H2 O + 2Snlat + O(s) ↔ 2 (OH − Snlat ) + e−

(9)

Eq. (9) is supported by diffuse reflectance infrared Fourier
transform (DRIFT) spectroscopy measurements that conclude that
only OH terminal groups are formed at the surface of SnO2 [30].
The authors observed that the concentration of these surface
hydroxyl groups increased with oxygen partial pressure, reaching
a saturation, and showed that this effect was reversible. Therefore, the concentration of hydroxyl groups which provides the
increase in conductivity to the metal oxide is strongly dependent
on chemisorbed oxygen. Consequently, when another reducing gas
is present, together with H2 O, both will compete for reacting with
chemisorbed oxygen atoms.
The resistance as a function of temperature in both dry and
humid conditions (30% and 60% of relative humidity) is represented in Fig. 5(b). As can be seen, the sensor resistance decreases at
higher concentrations of water vapor. On the other hand, resistance
of SnO2 NWs grows with increasing temperature above 200 ◦ C,
showing a U-shaped resistance-temperature curve. The increasing
resistance ensures that atomic oxygen is available to react with
other species; there is an incomplete coverage of chemisorbed oxygen with hydroxyl groups since otherwise, a decreasing resistance
for increasing temperature would be experimentally observed.
3.3.2. Ammonia sensing in humid conditions
Sensing capabilities of tin dioxide nanowires towards ammonia
added to water vapor have been investigated at a temperatures of
200, 300 and 400 ◦ C, where mainly atomic oxygen is chemisorbed at
the semiconductor’s surface. Two different relative humidity levels
were employed for each temperature: 30 and 60% for the same
sequence of ammonia pulses performed in dry air experiments.
The normalized sensor resistance to the baseline value in synthetic air at 400 ◦ C is represented in Fig. 5(c) for dry and humid
conditions towards different concentrations of ammonia. Clearly,
the response is reduced in the presence of increasing amounts of
water as compared to dry conditions. The sensor resistance does not
show the overshoot in resistance observed in the response towards
ammonia in dry conditions at 450 ◦ C (see Fig. 5(d) for comparison),
suggesting that Reactions (3) and (4) are the main ones that take
place in the studied range, and the influence of (5) is here negligible.

J. Samà et al. / Sensors and Actuators B 232 (2016) 402–409

407

Fig. 5. (a) Electrical response of SnO2 NWs to different concentrations of water vapor in synthetic air at 400 ◦ C; (b) Resistance of SnO2 NWs for different humidity levels at
different temperature. U-shaped R-T is also obtained in humid conditions due to dissociation of molecular to atomic oxygen at a temperature of 200 ◦ C; (c) Electrical response
of SnO2 NWs in dry, 30% and 60% of relative humidity (RH at room temperature) towards different ammonia pulses in synthetic air. Three tests have been performed by
keeping 400 ◦ C constant temperature; (d) Sensor response towards 10 ppm of NH3 at different conditions. The overshoot in resistance is visible for 450 ◦ C and RH = 0% curve.

The reduced response of tin dioxide nanowires against ammonia in the presence of water vapor can be explained by several
concurring mechanisms. On the one hand, competitive adsorption between water vapor and ammonia with chemisorbed oxygen
takes place. Atomic oxygen and water vapor compete for the same
adsorption sites, diminishing the probability of Reactions (3)–(5)
in the steady sensing state. Furthermore, 60% of relative humidity
at 20 ◦ C corresponds to approximately 14300 ppm of water, a concentration 3 orders of magnitude higher than ammonia in these
experiments. Thus, the partial pressure of water is considerably
higher than that of ammonia, which leads to much higher coverage
of these adsorbed species on the metal oxide surface.
The Arrhenius plot of the response time of the sensor in dry
and humid conditions for 30 ppm of ammonia (Fig. 6(a)) shows
that the response is slower in humid than in dry conditions, and
that the higher the temperature, the higher the difference. Larger
response times are a direct consequence of the competitive adsorption between atomic oxygen and water vapor that reduces the
probability of oxidizing ammonia and, thus, slows down the reaction. The effect is more pronounced at higher temperatures because
under dry conditions the reaction is promoted without any competitive adsorption phenomenon.
At the same time, the activation energies shown in Fig. 6(b),
deduced from Fig. 6(a) for different ammonia concentrations and
in the different studied relative humidity values, clearly decrease
in the presence of water vapor and are essentially independent on
the ammonia concentration, within the experimental uncertainty.

As mentioned before, water vapor can dissociate in two hydroxyl
species and chemisorb at the surface of tin oxide assisted by a Snlat
and a chemisorbed O− (s) as described in (8). A tentative explanation for the behavior could be derived from the interaction of
hydroxyl groups with O2 , as proposed by Epling et al. [33,34] in
monocrystalline rutile TiO2 according to:
−
OH(s) + O2(g) → O(s) + (Oa − OH)(s)

(9)

where Oa stands for an oxygen adatom. This mechanism creates a
new path for oxygen dissociation, which can occur even at lower
temperatures than oxygen dissociation described in (1), diminishing the activation barrier of oxygen dissociation. According to our
previous study of ammonia interaction mechanism, oxygen dissociation is precisely the energy limiting step [23]. Thus, a mechanism
similar to (9) could explain the lowering in the activation energy
of ammonia sensing in presence of water. There is an extra term
of adsorbed oxygen (Oa ) which grows with increasing number of
OH groups at the surface that can act as H scavenging centers, thus
improving the probability of adsorbed ammonia to react with active
bases in the intermediate ammonia decomposition steps.
As for endurance, the sensors were operated for 1 month in dry
and humid conditions, after which a drift in resistance of 7% from
the initial values was observed, but where the resistance change
due to the presence of the gases was almost invariant in its value,
only about 1%. Fig. 6(d) shows the resistance change at 300 ◦ C
and 30% of RH obtained at the 10th and 25th day of operation,
and illustrates the repeatability of measurements made with the

408

J. Samà et al. / Sensors and Actuators B 232 (2016) 402–409

Fig. 6. (a) Response time of sensor of NH3 pulses of 30 ppm in dry and humid condition, represented in an Arrhenius plot. Symbols are experimental values and lines are the
fitted exponential decay; (b) Activation energy obtained from response time for ammonia response in function of relative humidity, for all the concentrations studied; (c)
Response in humid condition of SnO2 NWs against ammonia at 400 ◦ C. The sensor distinguishes the concentration of ammonia in a precision of 30 ppm in realistic operational
conditions, where humidity is present; (d) Comparison between the response of the sensor at 300 ◦ C and 30% of RH from 10th and 25th day of measurements. A change of
5% in resistance baseline is obtained.

nanowire-based device. The sensor has shown a good durability,
and offers relatively fast response times of 6 minutes in presence
of water vapor, and 2 minutes for dry conditions towards 30 ppm
of NH3 .

4. Conclusions
The sensing mechanisms of locally grown SnO2 NWs towards
ammonia gas diluted in air, in both dry and humid conditions, have
been studied. Different temperature regimes have been identified
in the sensor kinetic response. In dry conditions, the promotion
of NO as byproduct at high operating temperatures reduces the
response of the sensor and provides a maximum ammonia response
at 400 ◦ C. When operated in humid conditions, the response of
tin dioxide nanostructures is reduced and slowed down by the
presence of water vapor. Simultaneously, the activation energy is
lowered by moisture, which could be explained through the reaction between O2 and the OH adsorbed groups, consequence of
water decomposition. Finally, the integrated growth of SnO2 on
micromembranes has been demonstrated as a fast, reproducible
and low power consuming approach, which gives rise to ammonia sensors with good repeatability, fast response and long term
stability.

Acknowledgements
The research leading to these results has received funding from
the Spanish Ministerio de Economía y Competitividad, through
project TEC2013-48147-C6 (TEMIN-AIR) and from the European
Research Council under the European Union’s Seventh Framework
Programme (FP/2007-2013)/ERC Grant Agreement n. 336917. J.D.
Prades acknowledges the support of the Serra Húnter Program.
Transmission electron microscopy investigations were carried out
using facilities at the University Service Centre for Transmission
Electron Microscopy, Vienna University of Technology, Austria.

References
[1] T.M. Banhazi, J. Seedorf, D.L. Rutley, W.S. Pitchford, Identification of risk
factors for sub-optimal housing conditions in Australian piggeries: Part 1.
Study justification and design, J. Agric. Saf. Health 14 (2008) 5–20.
[2] J.N. Galloway, F.J. Dentener, D.G. Capone, E.W. Boyer, R.W. Howarth, S.P.
Seitzinger, et al., Nitrogen cycles: Past, present, and future, Biogeochemistry
(2004), http://dx.doi.org/10.1007/s10533-004-0370-0.
[3] F.-X. Philippe, J.-F. Cabaraux, B. Nicks, Ammonia emissions from pig houses:
Influencing factors and mitigation techniques, Agric. Ecosyst. Environ. 141
(2011) 245–260, http://dx.doi.org/10.1016/j.agee.2011.03.012.
[4] S. Brandenberger, O. Kröcher, A. Tissler, R. Althoff, The State of the Art in
Selective Catalytic Reduction of NOx by Ammonia Using Metal-Exchanged
Zeolite Catalysts, Catal. Rev. 50 (2008) 492–531, http://dx.doi.org/10.1080/
01614940802480122.

J. Samà et al. / Sensors and Actuators B 232 (2016) 402–409
[5] M. Mehring, M. Elsener, O. Kröcher, Diesel soot catalyzes the selective
catalytic reduction of NOx with NH3, Top. Catal. 56 (2013) 440–445, http://dx.
doi.org/10.1007/s11244-013-9993-5.
[6] N. Taguchi,Patent, 45-38200, 1962.
[7] N. Yamazoe, New approaches for improving semiconductor gas sensors, Sens.
Actuators B Chem. 5 (1991) 7–19, http://dx.doi.org/10.1016/09254005(91)80213-4.
[8] S. Barth, F. Hernandez-Ramirez, J.D. Holmes, A. Romano-Rodriguez, Synthesis
and applications of one-dimensional semiconductors, Prog. Mater. Sci. (2010)
563–627, http://dx.doi.org/10.1016/j.pmatsci.2010.02.001.
[9] F. Hernandez-Ramirez, J.D. Prades, a Tarancon, S. Barth, O. Casals, R.
Jiménez–Diaz, et al., Portable microsensors based on individual SnO2
nanowires, Nanotechnology 18 (2007), http://dx.doi.org/10.1088/0957-4484/
18/49/495501, 495501.
[10] D.T.T. Le, N. Van Duy, H.M. Tan, D.D. Trung, N.N. Trung, P.T.H. Van, et al.,
Density-controllable growth of SnO2 nanowire junction-bridging across
electrode for low-temperature NO2 gas detection, J. Mater. Sci. 48 (2013)
7253–7259, http://dx.doi.org/10.1007/s10853-013-7545-9.
[11] J. Guilera, C. Fàbrega, O. Casals, F. Hernández-Ramírez, S. Wang, S. Mathur,
et al., Facile integration of ordered nanowires in functional devices, Sens.
Actuators B Chem. 221 (2015) 104–112, http://dx.doi.org/10.1016/j.snb.2015.
06.069.
[12] S. Mathur, S. Barth, H. Shen, J.-C. Pyun, U. Werner, Size-Dependent
Photoconductance in SnO2 Nanowires, Small 1 (2005) 713–717, http://dx.doi.
org/10.1002/smll.200400168.
[13] J. Puigcorbé, D. Vogel, B. Michel, A. Vilà, I. Gràcia, C. Cané, et al., Thermal and
mechanical analysis of micromachined gas sensors, J. Micromechanics
Microengineering. 13 (2003) 548–556, http://dx.doi.org/10.1088/0960-1317/
13/5/304.
[14] S. Barth, R. Jimenez-Diaz, J. Samà, J. Daniel Prades, I. Gracia, J. Santander, et al.,
Localized growth and in situ integration of nanowires for device applications,
Chem. Commun. 48 (2012) 4734–4736, http://dx.doi.org/10.1039/
c2cc30920c.
[15] S. Ahlers, T. Becker, W. Hellmich, C. Braunmühl, G. Müller, Temperature- and
Field-Effect-Modulation Techniques for Thin-Film Metal Oxide Gas Sensors,
in: T. Doll (Ed.), Adv. Gas Sens. SE − 6, Springer, US, 2003, pp. 123–159, http://
dx.doi.org/10.1007/978-1-4419-8612-2 6.
[16] S. Chang, Oxygen chemisorption on tin oxide: Correlation between electrical
conductivity and EPR measurements, J. Vac. Sci. Technol. 17 (1980) 366,
http://dx.doi.org/10.1116/1.570389.
[17] National Institute for Occupational Safety and Health (NIOSH), NIOSH Pocket
Guide to Chemical Hazards—Human Services, Saf. Heal. (2007).
[18] M. de Boer, H.M. Huisman, R.J.M. Mos, R.G. Leliveld, a. J. van Dillen, J.W. Geus,
Selective oxidation of ammonia to nitrogen over SiO2 -supported MoO3
catalysts, Catal. Today 17 (1993) 189–200, http://dx.doi.org/10.1016/09205861(93)80023-T.
[19] I. Jiménez, M.A. Centeno, R. Scotti, F. Morazzoni, J. Arbiol, A. Cornet, et al., NH3
interaction with chromium-doped WO3 nanocrystalline powders for gas
sensing applications, J. Mater. Chem. 14 (2004) 2412, http://dx.doi.org/10.
1039/b400872c.
[20] N.I. Il’chenko, Catalytic Oxidation of Ammonia, Russ. Chem. Rev. 45 (1976)
1119–1134, http://dx.doi.org/10.1070/RC1976v045n12ABEH002765.
[21] V.A. Sadykov, L.A. Isupova, I.A. Zolotarskii, L.N. Bobrova, A.S. Noskov, V.N.
Parmon, et al., Oxide catalysts for ammonia oxidation in nitric acid
production: properties and perspectives, Appl. Catal. A Gen. 204 (2000)
59–87, http://dx.doi.org/10.1016/S0926-860X(00)00506-8.
[22] C.-M. Hung, Decomposition kinetics of ammonia in gaseous stream by a
nanoscale copper-cerium bimetallic catalyst, J. Hazard. Mater. 150 (2008)
53–61, http://dx.doi.org/10.1016/j.jhazmat.2007.04.044.
[23] F. Shao, M.W.G. Hoffmann, J.D. Prades, J.R. Morante, N. López, F.
Hernández-Ramírez, Interaction mechanisms of ammonia and Tin oxide: A
combined analysis using single nanowire devices and DFT calculations, J.
Phys. Chem. C. 117 (2013) 3520–3526, http://dx.doi.org/10.1021/jp3085342.
[24] Y. Takao, High Ammonia Sensitive Semiconductor Gas Sensors with
Double-Layer Structure and Interface Electrodes, J. Electrochem. Soc. 141
(1994) 1028, http://dx.doi.org/10.1149/1.2054836.
[25] K. Skalska, J.S. Miller, S. Ledakowicz, Kinetics of nitric oxide oxidation, Chem.
Pap. 64 (2010) 269–272, http://dx.doi.org/10.2478/s11696-009-0105-8.
[26] J.D. Prades, R. Jimenez-Diaz, F. Hernandez-Ramirez, S. Barth, A. Cirera, A.
Romano-Rodriguez, et al., Ultralow power consumption gas sensors based on
self-heated individual nanowires, Appl Phys. Lett. 93 (2008) 123110, http://
dx.doi.org/10.1063/1.2988265.
[27] E. Kanazawa, G. Sakai, K. Shimanoe, Y. Kanmura, Y. Teraoka, N. Miura, et al.,
Metal oxide semiconductor N2O sensor for medical use, Sens. Actuators B
Chem. 77 (2001) 72–77, http://dx.doi.org/10.1016/S0925-4005(01)00675-X.
[28] P.T. Moseley, Solid state gas sensors, Meas. Sci. Technol. 8 (1997) 223 http://
stacks.iop.org/0957-0233/8/i=3/a=003.
[29] F. Hernandez-Ramirez, S. Barth, A. Tarancon, O. Casals, E. Pellicer, J. Rodriguez,
et al., Water vapor detection with individual tin oxide nanowires,
Nanotechnology 18 (2007), http://dx.doi.org/10.1088/0957-4484/18/42/
424016, 424016.
[30] D. Koziej, N. Bârsan, U. Weimar, J. Szuber, K. Shimanoe, N. Yamazoe,
Water-oxygen interplay on tin dioxide surface: implication on gas sensing,
Chem. Phys. Lett. 410 (2005) 321–323, http://dx.doi.org/10.1016/j.cplett.
2005.05.107.

409

[31] S.H. Hahn, N. Bârsan, U. Weimar, S.G. Ejakov, J.H. Visser, R.E. Soltis, CO sensing
with SnO2 thick film sensors: role of oxygen and water vapour, Thin Solid
Films 436 (2003) 17–24, http://dx.doi.org/10.1016/S0040-6090(03)00520-0.
[32] G. Heiland, D. Khol, Physical and Chemical Aspects of Oxidic Semiconductor
Gas Sensors, Kodansha Ltd., 1988, http://dx.doi.org/10.1016/B978-0-44498901-7.50007-5.
[33] W.S. Epling, C.H.F. Peden, M.A. Henderson, U. Diebold, Evidence for oxygen
adatoms on TiO2(110) resulting from O2 dissociation at vacancy sites, Surf.
Sci. 412–413 (1998) 333–343, http://dx.doi.org/10.1016/S00396028(98)00446-4.
[34] M.A. Henderson, W.S. Epling, C.H.F. Peden, C.L. Perkins, Insights into
Photoexcited Electron Scavenging Processes on TiO2 Obtained from Studies of
the Reaction of O2 with OH Groups Adsorbed at Electronic Defects on TiO2
(110), J. Phys. Chem. B. 107 (2003) 534–545, http://dx.doi.org/10.1021/
jp0262113.

Biographies
Jordi Samà was born in Barcelona in 1985. He received the degree in Physics at
the University of Barcelona (UB) in 2010. Nowadays he is a predoctoral researcher
in MIND group in Electronics Departament at UB. His current research is focused
on the development and fabrication of nanostructured metal oxide gas sensors. His
works also includes the structural and electrical characterization of the metal oxide
nanostructures and the interaction mechanisms with toxic gases.
Sven Barth is a group leader at Vienna University of Technology, where he received
the venia docendi for inorganic chemistry in 2015. He graduated in chemistry (2003)
and received his PhD (2008) from Saarland University. His core experience is related
to the molecule-based synthesis and characterization of nanoscaled metal oxide and
group 14 semiconductors. The nanomaterials are prepared in gas and liquid phase
processes. He has published over 50 scientific papers, reviews and book chapters.
Guillem Domènech-Gil was born in Barcelona in 1986. He graduated in physics at
University of Barcelona in 2009 and obtained a master degree in nanoscience and
nanotechnology at the same faculty in 2014. His research activities started with the
growth of metal oxide nanowires for their employment as gas sensors main element towards toxic gases and fabricating nanodevices for monitoring environment.
Coauthor at a MRS 2014 and in Eurosensors 2015 with poster and oral presentations.
His doctoral studies are focused in enhancing the gas sensing technology through
the use of nanostructures with a deeper comprehension of the sensing mechanisms
ocurring.
Joan Daniel Prades was born in Barcelona in 1982. He graduated in Physics and Electronic Engineering at the University of Barcelona and obtained his PhD at the same
institution in 2009. He has experience in modelling of the electronic and vibrational
properties of nanostructured metal oxides and in their experimental validation. He
is actively involved in the development of innovative device prototypes based on
nanomaterials. He has published more than 50 papers in peer-reviewed journals and
contributed to more than 150 international conferences. He has also contributed to
five industrial patents.
Núria López got her PhD at the University of Barcelona, 1999. After a post-doctoral
stage in the group of professor Noskov she moved to the Institute of Chemical
Research of Catalonia. She received the ERC Starting grant in 2010.
Olga Casals was born in Barcelona in 1973. She received her diploma in Optics
and Optometry at the Polytechnic University of Catalonia in 1994 and her degree in
Physics at the University of Barcelona in 2001, and her PhD in 2012. Her professional
experience is set in development of new technologies on solid-state gas sensors.
Isabel Gràcia received her Ph.D. degree in physics in 1993 from the Autonomous
University of Barcelona, Spain, working on chemical sensors. She joined the National
Microelectronics Center (CNM) working on photolithography, currently she is full
time senior researcher in the Micro-Nano Systems department of the CNM and her
work is focused on gas sensing technologies and MEMS reliability.
Carles Cané is Telecommunications Engineer and he received his Ph.D. in 1989.
Since 1990 he is full time senior researcher at CNM and has been working in the
development of CMOS technologies and also on mechanical and chemical sensors
and microsystems. Since 2012, he is the director of the IMB-CNM in Barcelona. He is
member of the technical committee of EURIMUS-EUREKA programme since 1999.
Over the years he has been coordinator of several R&D projects, both at national and
international level in the MST field.
Albert Romano-Rodríguez is a Professor in electronics at Universitat de Barcelona.
He is active in the fields of characterisation of semiconducting materials and fabrication processes by using different structural, physical and chemical characterisation
techniques, in the development of electronic materials for solar cells and in the
design, fabrication and testing of different kinds of physical and chemical sensors, with special emphasize in the fabrication of nanosensors based on nanowires
towards the development of ultra-low power gas sensing systems based on these
materials. He has co-authored over 160 peer-reviewed scientific and technical
papers in these fields.