This document is the Accepted Manuscript version of a Published Work that appeared in final
form in ACS SENSORS, copyright © American Chemical Society after peer review and technical
editing by the publisher. To access the final edited and published work see
https://pubs.acs.org/doi/abs/10.1021/acssensors.7b00504.

Temperature-Dependent NO2 Sensing Mechanisms over Indium Oxide
Sergio Rosoa, b, David Deglerc, Eduard Llobeta, Nicolae Barsanc, Atsushi Urakawab,*
a

b

Minos-Emas, Universitat Rovira i Virgili, Av. Països Catalans 26, 43007 Tarragona, Spain

Institute of Chemical Research of Catalonia (ICIQ), The Barcelona Institute of Science and Technology,
Av. Països Catalans 16, 43007 Tarragona, Spain

c

Institute of Physical and Theoretical Chemistry (IPTC), University of Tuebingen, Auf der Morgenstelle
15, D-72076 Tuebingen, Germany
* Corresponding Author: aurakawa@iciq.es

Keywords: Gas sensor, In2O3, NO2, DRIFTS, multivariate analysis

Abstract
The surface species responsible for NO2 gas sensing over indium oxide was studied by
operando DRIFTS coupled to a multivariate spectral analysis. It revealed the important
roles of surface nitrites on the temperature-dependent gas sensing mechanism and
the interaction of such nitrites with surface hydroxyls. A highly hydroxylated surface
with high concentration of surface adsorbed H2O is beneficial to enhance the
concentration of adsorbed NO2, present as nitrites, thus explaining superior sensing
response at lower operating temperatures.

1

Indium oxide (In2O3) has been recognized as an excellent material for the detection of
oxidising gases such as NO2 and O3 through electrical resistance measurements with
excellent sensitivity at the concentration in a ppb range at low temperature and even in
the presence of humidity.1 Some studies on NO2 sensing by In2O3 proposed a surface
vacancy based model in which oxygen vacancies act as the active adsorption sites of
NO2, further reducing it NO and increasing adsorbed oxygen species.2 As temperature
increases, the amount of adsorbed oxygen species also increases as verified by XPS.3 At
the typical operating temperatures of 150-400 oC, the predominant species on the
surface of In2O3 gas sensors are reported to be oxygen species, either O2- or O- arising
from NO2. An increase in the amount of ionosorbed oxygen species further increases the
number of electrons trapped via the conduction band of the n-type indium oxide
semiconductor and this translates in the observed increase in the sensor resistance.
However, these results are not based on direct observation and cause controversies in
the sensing mechanism. An ideal approach for the mechanistic study is a direct
measurement of surface species under working, operando, conditions where electrical
resistance is measured simultaneously to clarify the interaction of NO2 with In2O3
surface and establish firmly the relationship between surface structure/species and gas
sensing response.
This work aims at shedding light on the sensing mechanism of In2O3 towards NO2 and its
temperature dependencies by surface-sensitive diffuse reflectance infrared Fourier
transform spectroscopy (DRIFTS) under operando conditions. The demanding and
difficult spectral analyses of complexly overlapping IR bands were facilitated by
multivariate analysis, specifically multivariate curve resolution (MCR),4, 5 to allow
spectral separation, extracting subtle spectral changes and obtaining respective
concentration profiles without a priori knowledge and reference spectra of surface
species as required in the evaluation using linear spectral combination or principal
component analysis as demonstrated for the spectral analyses of IR and X-ray
spectroscopy.6, 7, 8

Operando DRIFTS and sensing mechanism at 350 °C
First, we performed an operando DRIFTS study of the In2O3 sensor exposed to 1 ppm of
NO2 at 350 °C (Figure 1). The experiment consisted of three cycles of 3 h of NO2
exposure followed by 6 h recovery in dry air. Figure 2f shows the electrical resistance
of the In2O3 sensor under the periodic NO2 exposure. Despite the relatively low
response (ca. 1.28) as gas sensor, the electrical response was excellent with an
outstanding signal-to-noise ratio. The time-resolved DRIFT spectra (2D plot) of the
entire experiment (Figure 1) show the presence of the bands obviously responding to
the NO2 in the gas atmosphere (at ca. 1220, 1520 and 3630 cm-1) and of those
continuously growing (at ca. 1300 and 3630 cm-1). These spectral changes clearly
2

indicate that there are nitrites and/or nitrates and also hydroxyl groups on/near the
In2O3 surface which are involved in the reception of NO2. However, the common
difficulty encountered in the analysis and interpretation is that the spectral features of
some chemical species are overlapping, rendering them difficult or often impossible to
correctly identify them by the naked eye. Therefore, we have employed a multivariate
analysis, particularly MCR, to disentangle the overlapping bands and to determine
their concentration profiles based on the kinetic resolution of comprising bands.9

Figure 1: Time-resolved DRIFT spectra of the In2O3 gas sensor under alternating exposure to 1 ppm of NO2 (3 h)
and air (6 h) at 350 oC. The scale is in the unit of absorbance and the spectra are shown taking the state of
the sensor in air before the NO2 exposure as the background.

The MCR analysis identified that the system comprises two surface chemical species
(or two chemical states with multiple surface species) with varying distinct
concentrations under the NO2 exposure and subsequent recovery phases. These
chemical species are now called “components” and simply denoted as “C” with a label
1, 2, etc. (i.e. C1, C2 for 1000-2000 cm-1 and C1’ and C2’ for 2900-3800 cm-1; the
numbering has no meaning and there is no intended relation between CN and CN’
(N=1,2)).
Figures 2a and 2b present the operando DRIFT spectra of C1 and C2 and corresponding
concentration profiles which are shown on the same time-scale as that of the gas
sensing response (Figure 2f). It should be noted that MCR yielded the component
3

spectra and concentration profiles directly from the time-resolved DRIFT spectra
shown in Figure 1 and the spectra can be reconstructed by multiplying the component
spectra and concentration and summing up all component contributions. Two
prominent bands at 1221 and 1524 cm-1 were detected for C1, whereas for C2 clear
bands were observed at ca. 1050, 1260 and 1313 cm-1. Generally, asymmetric N-O
stretching vibrations of nitrites and nitrates are reported to appear in the range of
1200-1500 cm-1. Nitrates and nitrites adsorbed over various surfaces have been
extensively studied and there is profound knowledge on their vibrational modes and
frequencies. However, unfortunately, the assignments and observed frequencies are
diverse; one can find almost any assignment one wishes to have in the literature. It has
been reported that depending on the location of the nitrate ions, i.e. surface, subsurface or bulk, the vibrational frequencies vary10 and this factor besides potential
intermolecular interactions among adsorbates or with surface chemical groups (e.g.
hydroxyl) may further complicate firm band assignments. For these reasons, herein we
have attempted to assign the vibrational modes based mainly on the experimental
observation and indications by quantum chemical calculations with minimal reference
to available literature.

Figure 2: a) and c) Component spectra and b) and d) concentration profiles obtained after the MCR analysis of the
time-resolved DRIFT spectra shown in Figure 1 for the In2O3 gas sensor exposed at 350 °C under periodic
exposure to 1 ppm NO2. e) Hydroxyl bands identified in the H2O/D2O exchange experiment (alternatingly
passing 10% H2O/D2O vapor over the sensor). The spectrum is shown taking the sensor state in D2O as the
background. f) Response of the sensor towards 1 ppm of NO2 gas at 350 oC.

The main characteristic IR-active vibrations of isolated nitrate and nitrite ions are
asymmetric stretching modes (Figure S2, Supporting Information). Due to the
symmetry of the ions, nitrate ion has the degenerate states for the vibrational mode at
1476 cm-1, while nitrite ion has one state at 1374 cm-1 according to the quantum
chemical calculation. In one literature, the frequencies of 1380 and 1260 cm-1 are
reported for free nitrate and nitrite ions, respectively.11 Despite the large difference in
4

the numbers, it is consistent that the vibrational frequency of the stretching mode is
about 100 cm-1 higher for nitrate. Based on this, it is reasonable to conclude that the
band at 1221 cm-1 of C1 is assigned to nitrite species and the two bands observed at
1260 and 1313 cm-1 of C2 are due to nitrate species with non-degenerated vibrational
states because of a specific spatial orientation of the adsorbed ion on the surface. The
prominent band at 1524 cm-1 could be assigned to nitrates or nitrites according to the
literature11 , but here we assign that the band is due to the surface hydroxyls directly
interacting with nitrite species as discussed later.
The concentration profiles of C1 and C2 clarify important insights into the surface
species involved during gas sensing. The concentration of C1, assigned to surface
nitrite species, responds clearly and reversibly to the gaseous NO2 concentration,
matching well with the sensor response (Figure 2f). In contrast, the concentration of
C2 responds irreversibly with a gradual, step-wise increase during the NO2 pulse. These
observations imply that the reversibly adsorbed surface nitrite species is responsible
for the change in the resistance of the sensor, whereas nitrate species is formed
gradually during the NO2 pulse, likely due to slow oxidation of surface nitrites because
of the low NO2 concentration at 350 °C to form surface, sub-surface and bulk nitrates,
which cannot be decomposed in air and, therefore, irreversibly accumulate over time.
Figures 2c and 2d show MCR-processed operando DRIFT spectra in the OH stretching
region with two distinguishable hydroxyl species with distinct chemical nature (Figure
2c) with characteristic concentration profiles (Figure 2d). Generally, a presence of
isolated hydroxyl groups is indicated by the sharp bands appearing in the 3500-3700
cm-1 while the presence of the bridging hydroxyls, via hydrogen-bonds, is clarified by
the broad bands appearing in 2900-3500 cm-1. The component spectra and
concentration profiles of C1’ and C2’ show that, when the In2O3 gas sensor is exposed
to 1 ppm NO2, at first there is a sharp rise in the population of C1’, i.e. the terminal OH
band located at higher frequency at 3632 cm-1 with a shoulder band at 3648 cm-1.
Subsequently the concentration of the species sharply drops and reaches a steady
concentration (Figure 2d, C1’). The other hydroxyl species (C2’) possesses a prominent
band at 3621 cm-1 with a broad band feature (appearing as a tail) down to 3000 cm -1,
which was absent for C1’. This indicates that in C2’ an interaction of OH group with
another molecular/structural entity in various configurations like hydrogen-bonds is
present. Also, the concentration profile of C2’ is very different from that of C1’,
exhibiting increase in the concentration upon the NO2 pulse with an irreversible
profile, i.e. accumulative increase over the multiple NO2 pulses. It should be noted that
the hydroxyl bands presented in Figure 2c is relative (difference spectra) compared to
the initial state of In2O3 before the exposure to NO2 pulses. Hence, the positive values
in the concentration spectra indicate that the hydroxyl bands were newly formed
and/or that their absorbance has been enhanced due to NO2 interaction with the In2O3
surface. Since it is not trivial to quantitatively discuss the amount of hydroxyl groups
5

present initially, formed and interacting at the surface with the diffuse reflection
sampling configuration, in this study the type of hydroxyl groups present at 350 °C in
air was identified by exchanging the hydroxyl group from OH to OD. Figure 2e shows
the IR spectrum of OH groups which could be exchanged to OD groups (and vice versa)
by alternatingly passing 10% H2O and D2O vapor over the In2O3. The DRIFT spectrum
shows that the hydroxyl groups responding to NO2 pulses (C1’ and C2’) are already
present without NO2 pulses, mostly as isolated hydroxyls. This suggests that the
changes are due to an increase in the absorption cross section of the hydroxyl groups,
rather than to the formation of new hydroxyl groups upon NO2 surface adsorption.
Assuming the constant concentration of surface hydroxyl groups during NO2 sensing,
we can summarize the roles of hydroxyls and surface chemical processes as follows.
Firstly, when NO2 adsorbs over In2O3, terminal hydroxyl (C1’) promptly interacts with
adsorbed NO2 forming nitrite species (C1) and senses its presence (based on this the
prominent band at 1524 cm-1 has been assigned to nitrites interacting with hydroxyls,
vide supra). Subsequently a transformation of surface configuration of adsorbed NO2
takes place as indicated by the decrease of C1’, most likely by the formation of
(subsurface or bulk) nitrate species by the oxidation of adsorbed nitrites as strongly
suggested by the increase of C2 (nitrates) in the middle of NO2 pulses (Figure 2b, C2).
The presence of hydroxyls in multi-configurations with possible hydrogen-bonds (C2’)
seem influenced by that of the nitrate species as indicated by irreversible, step-wise
increase of the concentration of C2’ over NO2 pulses at the expense of lowered initial
peak height of C1’ concentration (Figure 2d).
We can describe the sensing mechanism at 350 °C as follows. First, the formation of
the nitrite takes place:
−
𝑁𝑂2,𝑔𝑎𝑠 + 𝑒 − ⇄ 𝑁𝑂2,𝑎𝑑𝑠

This reaction is responsible for the resistance change of the sensor. Following this
reaction, oxidation of nitrite with lattice oxidation (Olat) can take place, irreversibly
forming nitrate species.
−
−
𝑁𝑂2,𝑎𝑑𝑠 + 𝑂 𝑙𝑎𝑡 → 𝑁𝑂3,𝑎𝑑𝑠

Operando DRIFTS and sensing mechanism at 130 °C
Importantly, the response of the In2O3 sensor against 1 ppm NO2 is about 450 times
higher at 130 °C than at 350 °C, but to date the origin of the superior sensitivity at
lower temperature is unknown. Aiming at elucidating the origin of the high sensitivity,
we have investigated the roles of surface species during NO2 sensing at 130 °C by
operando DRIFTS. The MCR analysis of the obtained spectra identified three
components in both nitrite/nitrate and hydroxyl regions (Figures 3a-d). For this
6

experiment, the first cycle was a 7 h exposure to 1 ppm NO2 followed by recovery in air
for 12 h. The following two cycles were with 1 h NO2 pulse and 2 h recovery in dry air.
Similarly to Figure 2, the gas sensing response during the measurements and the DRIFT
spectrum of exchangeable hydroxyl groups obtained by the H2O/D2O experiment at
130 °C are presented in Figures 3f and 3e, respectively. The broad band feature below
3500 cm-1 observed in Figure 3e clearly shows the presence of a significantly higher
number of hydroxyls on the sample at 130 °C than 350 °C, interacting with other
surface species, likely via hydrogen-bonding.

Figure 3: a) and c) Component spectra and b) and d) concentration profiles obtained after the MCR analysis of the
time-resolved DRIFT spectra shown in Figure S3 (Supporting Information) for the In2O3 gas sensor exposed
at 130 °C under periodic exposure to 1 ppm NO2. e) Hydroxyl bands identified in the H2O/D2O exchange
experiment (alternatingly passing 10% H2O/D2O vapor over the sensor). The spectrum is shown taking the
sensor state in D2O as the background. f) Response of the sensor towards 1 ppm of NO2 gas at 130 oC.

Notably, the spectral features of the obtained components in both regions at 130 °C
were very different from those observed at 350 °C. Compared to the component
spectra at 350 °C (Figure 2a), there were two new prominent bands at ca. 1430 and
1560 cm-1, which were present for all three components. These bands arise from the
interaction of surface hydroxyls and particularly adsorbed water with adsorbed NO2 in
a specific configuration as discussed further below. A careful look into the
nitrite/nitrate spectral region allows categorizing the three component spectra (all
nitrite species) into two families. The first family is C1 (Figure 3a) characterized by a
band at 1223 cm-1 as well as a shoulder band at ca. 1530 cm-1. These band positions
are obviously identical to that formerly assigned to nitrites interacting with surface
isolated hydroxyls (Figure 2a, C1; Figure S4, Supporting Information). The species
responsible for C1 increases gradually with time, more pronouncedly during the NO2
pulse. The second family includes C2 and C3 for which a very broad band at ca. 1200
cm-1 was observed. A noticeable difference between C2 and C3 is the higher
absorbance of the band at ca. 1200 cm-1 and the broader feature of the band at ca.
1560 cm-1 for C3. The profile of C3 concentration is most closely matching with the
7

response of the gas sensor. Interestingly, the concentration profiles of C2 and C3 are
behaving oppositely and counteracting (Figure 3b), which implies that there is a
transition of the chemical states between C2 and C3 during the NO2 pulse and recovery
phase.
The characteristic frequency region of hydroxyls provides also unique insights into the
chemistry during NO2 sensing process at 130 °C (Figures 3c and 3d). In this region the
identified three components can also be categorized into two families; ones with the
strong hydrogen-bonding features (C1’ and C3’) and one without (C2’). These
components show distinct stretching frequencies of isolated OH in 3600-3700 cm-1.
The first family (C1’ and C3’) shows similar, largely negative bands in 3100-3550 cm-1
with a characteristic isolated hydroxyl at 3546 cm-1 with a broad band centred at ca.
3250 cm-1 characteristic of hydrogen-bonds. C1’ and C3’ mainly differs in the terminal
hydroxyl stretching frequency at 3640 cm-1 for C1’ and 3629 cm-1 for C3’. The
concentration profiles of the components (Figure 3d) show that the concentration of
C3’ is dominating, but first the concentration of C1’ then that of C3’ increase. In
practice, this means that the terminal hydroxyl band characteristic of C1’ appears and
then red-shifts to the frequency of C3’ by the transformation to a different nature of
hydroxyl induced by increased adsorbed NO2. By MCR this band shift is described by
two components with distinct, varying concentrations.
On the other hand, the spectrum of C2’ is markedly different without the feature of
hydrogen-bonds but with the band of terminal hydroxyls with the high stretching
frequency at 3658 cm-1. According to the concentration profile of C2’ (Figure 3d), C2’
increases gradually during the recovery phase in dry air. This can be reasonably
understood that the terminal hydroxyls of the In2O3 surface become more freely
available during the recovery and they are highly isolated without strongly interacting
species nearby, as implied by the high vibrational frequency (i.e. less dragging
interaction of the OH stretching mode) and the absence of hydrogen-bonds. In
contrast, as discussed above, the water-like spectral features with isolated and
hydrogen-bonded features are present for C1’ and C3’ and they appear as negative
bands. This is a proof that the hydroxyl groups, most likely due to surface adsorbed
water, are “consumed” when NO2 adsorbs on the surface (thus appear negatively with
respect to the initial state before the exposure to NO2) in a way that the hydrogenbonds are broken or rather that the IR absorption by the hydrogen-bonds is weakened.
This also accompanies freeing the hydroxyls on the In2O3 surface as evident from the
positive bands at ca. 3600-3700 cm-1. This process takes a while to reach a stabilized
state as suggested by the continuously changing sensor response for 7 h (Figure 3f).
Based on these observations and interpretations, it can be concluded that C2 in the
nitrite/nitrate region (Figure 3a) interacts with adsorbed NO2 and reversibly forms C3,
enhancing the bands at ca. 1200 and at ca. 1550 cm-1 (C2 vs. C3 in Figure 3a) ascribed
to nitrites interacting with surface water and hydroxyls. The broad features of the two
8

bands also indicate the highly flexible configuration of the surface nitrite and support
the view of dissolved-in-water-like interaction of nitrites on the In2O3 surface. The
surface hydroxyls of In2O3 seems also affected by the presence of adsorbed NO2 as
indicated by the red/blue shifts of the bands (C1’ vs. C3’).
Regarding the irreversible change and thus the deactivation of the gas sensor, 130 °C is
too low for the oxidation of nitrites to nitrates to take place as in the case at 350 °C.
However, there is another cause of deactivation at such low operating temperature.
The gradual increase in C1 (Figures 3a and 3b) with two prominent features at 1223
and 1530 cm-1 was confirmed, indicating the formation of nitrites irreversibly
increasing in concentration and interacting with surface hydroxyls of In2O3. Although a
similar chemical process was reversibly taking place at 350 °C (C1, Figures 2a and 2b),
the desorption process is likely too slow at 130 °C and the nitrite cannot be desorbed
during the recovery phase. This means that this NO2 adsorption mode is mainly
responsible for the changes in the electrical response at 350 °C, but at 130 °C the same
mode is responsible for deactivation. This is well plausible because the magnitude of
the electrical response is markedly different at the two temperatures and the factor
responsible for resistance change at high temperature may cause poor response at low
temperature. Further investigation at an intermediate temperature (250 °C) confirms
that the behaviour of gas sensors and surface species lies between the two
temperatures, supporting the temperature dependent sensing mechanisms suggested
above (Figures S4 and S5, Supporting Information).
The facts that the absorbance of the bands in the nitrites/nitrates region is one order
of magnitude higher at 130 °C (Figure S3, Supporting Information) than at 350 °C
(Figure 1) and that the hydrogen-bonded hydroxyl groups are consumed when the
sensor is exposed to NO2 at 130 °C suggest the important roles of hydroxyl groups and
water on the surface of In2O3 in the sensing mechanism via facilitated NO2 adsorption,
thus withdrawing more electrons from the In2O3 surface and consequently increasing
the resistance. This view is also supported by the experimental observation where the
presence of water vapor indeed improved the response of an In2O3 sensor at different
concentrations of NO2 at 130 °C.1c

Conclusions
By means of operando DRIFTS, we have elucidated the surface species responsible for
NO2 sensing using In2O3 material at different temperatures where gas-sensing
characteristics remarkably alter. It has been shown that the actual sensing mechanism
is more complicated than previously reported and, importantly, that distinct sensing
mechanisms are active depending on the temperature of sensor.

9

At all temperatures, the electrical response originates from NO2 sorption processes,
although the NO2 adsorption and thus sensing mechanisms differ due to the
characteristic chemical nature of the In2O3 surface, especially of hydroxyls and
adsorbed water, at different temperatures. At high temperature (350 °C), isolated
hydroxyl groups defined by the material and synthesis method are affected by the NO2
adsorption and directly interact with adsorbed NO2, inducing changes in electrical
response. This NO2 sorption process is reversible at 350 °C. At the same time, oxidation
of surface nitrites to nitrates takes place at 350 °C and this can lead to deactivation of
the surface for NO2 sorption.
In contrast, at low temperature (130 °C ), the gas sensing mechanism involves
condition-dependent surface hydroxyls with and without hydrogen-bonds present over
the In2O3 surface. This “wet” surface is obviously beneficial to adsorb larger amount of
NO2 and thus responsible for higher sensitivity in NO2 sensing at the temperature.
Interestingly, at 130 °C deactivation of the sensor surface is caused by the mechanism
of the active gas sensing at 350 °C, i.e. surface nitrite formation and interaction with
the isolated OH group of In2O3. This surface nitrate irreversibly increases in
concentration, but the electrical response due to this NO2 adsorption is too low to
impact on the overall response which is dominated by the “wet” NO2 adsorption
mechanism.

Associated Content
Acknowledgments
SR and AU thank the Generalitat de Catalunya for financial support through the CERCA
Programme and recognition (2014 SGR 893) and MINECO (CTQ2012-34153 and
CTQ2016-75499-R (FEDER-UE)) for financial support and support through Severo
Ochoa Excellence Accreditation 2014–2018 (SEV-2013-0319). EL thanks ICREA (Icrea
Academia Award 2012), MINECO (71663-R (FEDER-UE)) and AGAUR (2014 SGR 1267).

References
1.
(a) Ivanovskaya, M.; Gurlo, A.; Bogdanov, P., Mechanism of O3 and NO2 detection and
selectivity of In2O3 sensors. Sensors and Actuators B: Chemical 2001, 77 (1–2), 264-267; (b)
Sowti khiabani, P.; Marzbanrad, E.; Hassani, H.; Raissi, B., Fast Response NO2 Gas Sensor Based
on In2O3 Nanoparticles. Journal of the American Ceramic Society 2013, 96 (8), 2493-2498; (c)
Roso, S.; Bittencourt, C.; Umek, P.; Gonzalez, O.; Guell, F.; Urakawa, A.; Llobet, E., Synthesis of
single crystalline In2O3 octahedra for the selective detection of NO2 and H2 at trace levels.
Journal of Materials Chemistry C 2016, 4 (40), 9418-9427.

10

2.
Rout, C. S.; Ganesh, K.; Govindaraj, A.; Rao, C. N. R., Sensors for the nitrogen oxides,
NO2, NO and N2O, based on In2O3 and WO3 nanowires. Applied Physics A 2006, 85 (3), 241246.
3.
Gurlo, A.; Bârsan, N.; Ivanovskaya, M.; Weimar, U.; Göpel, W., In2O3 and MoO3–In2O3
thin film semiconductor sensors: interaction with NO2 and O3. Sensors and Actuators B:
Chemical 1998, 47 (1–3), 92-99.
4.
Jaumot, J.; de Juan, A.; Tauler, R., MCR-ALS GUI 2.0: New features and applications.
Chemometrics and Intelligent Laboratory Systems 2015, 140, 1-12.
5.
Jaumot, J.; Tauler, R., MCR-BANDS: A user friendly MATLAB program for the evaluation
of rotation ambiguities in Multivariate Curve Resolution. Chemometrics and Intelligent
Laboratory Systems 2010, 103 (2), 96-107.
6.
Voronov, A.; Urakawa, A.; Beek, W. v.; Tsakoumis, N. E.; Emerich, H.; Rønning, M.,
Multivariate curve resolution applied to in situ X-ray absorption spectroscopy data: An efficient
tool for data processing and analysis. Analytica Chimica Acta 2014, 840, 20-27.
7.
Fakhrnasova, D.; Chimentão, R. J.; Medina, F.; Urakawa, A., Rational and Statistical
Approaches in Enhancing the Yield of Ethylene Carbonate in Urea Transesterification with
Ethylene Glycol over Metal Oxides. ACS Catalysis 2015, 5 (11), 6284-6295.
8.
Hyakutake, T.; van Beek, W.; Urakawa, A., Unravelling the nature, evolution and spatial
gradients of active species and active sites in the catalyst bed of unpromoted and K/Bapromoted Cu/Al2O3 during CO2 capture-reduction. Journal of Materials Chemistry A 2016, 4
(18), 6878-6885.
9.
Jaumot, J.; Gargallo, R.; de Juan, A.; Tauler, R., A graphical user-friendly interface for
MCR-ALS: a new tool for multivariate curve resolution in MATLAB. Chemometrics and
Intelligent Laboratory Systems 2005, 76 (1), 101-110.
10.
Hesske, H.; Urakawa, A.; VandeVondele, J.; Baiker, A., Insight into Fundamental,
Overtone, and Combination IR Bands of Surface and Bulk Ba(NO3)2 by Ab Initio Molecular
Dynamics. The Journal of Physical Chemistry C 2010, 114 (35), 15042-15048.
11.
Hadjiivanov, K. I., Identification of Neutral and Charged N x O y Surface Species by IR
Spectroscopy. Catalysis Reviews 2000, 42 (1-2), 71-144.

11

Supporting Information

Temperature-Dependent NO2 Sensing Mechanisms
over Indium Oxide
Sergio Rosoa, b, David Deglerc, Eduard Llobeta, Nicolae Barsanc, Atsushi Urakawab*
a

b

c

Minos-Emas, Universitat Rovira i Virgili, Av. Països Catalans 26, 43007 Tarragona, Spain

Institute of Chemical Research of Catalonia (ICIQ), The Barcelona Institute of Technology, Av. Països
Catalans 16, 43007 Tarragona, Spain

Institute of Physical and Theoretical Chemistry (IPTC), University of Tuebingen, Auf der Morgenstelle
15, D-72076 Tuebingen, Germany

Experimental details
The sensors were produced as follows. First, commercial In2O3 (MaTeck 99.99%)
powder was mixed with 1-2 propanediol and a printable ink was made. After that, the
paste was printed onto an alumina substrate that contained a Pt electrode on one side
and a Pt heater on the other side. The printed sensors were dried at 70 oC and
annealed at 500 oC. A SEM image of the In2O3 material can be observed in Figure S1 of
the Supporting Information.
Operando measurements (DRIFTS and electrical resistance) have been recorded at the
same time over the In2O3 sensors. The reference spectrum was recorded in dry air
before exposure to NO2 gas. For the calculation of the absorbance spectra we have
used:
𝑎𝑏𝑠𝑜𝑟𝑏𝑎𝑛𝑐𝑒 = −𝑙𝑜𝑔10 (

𝑠𝑖𝑛𝑔𝑙𝑒 𝑐ℎ𝑎𝑛𝑛𝑒𝑙 𝑡𝑒𝑠𝑡 𝑔𝑎𝑠
)
𝑠𝑖𝑛𝑔𝑙𝑒 𝑐ℎ𝑎𝑛𝑛𝑒𝑙 𝑟𝑒𝑓𝑒𝑟𝑒𝑛𝑐𝑒

For the DRIFT spectroscopy, a Bruker Vertex80v FT-IR spectrometer (narrow-band MTC
detector, internal glowbar; 1024 scans per spectrum, 4 cm-1 spectral resolution) was
used. The sensor was placed in a homemade operando cell with a KBr window, which
was mounted in a diffuse reflectance spectroscopy cell (Harrick “Praying Mantis”), and
it was heated using the backside heater of the substrate to the desired temperature of
130, 250 or 350 oC. Additionally, the resistance of the sensing layer was recorded by a
12

digital multimeter. Gases were dosed using a homemade mixing station. Finally, all the
experiments were conducted using a flow of 200 ml min-1. During all the experiments,
the resistance of the active layer was recorded using a digital multimeter (Keithley
2000). In order to avoid any kind of contamination, new and fresh sensors were used
for every new DRIFTS measurement at different experimental conditions (i.e. different
temperatures).

SEM micrograph of In2O3 gas sensor

Figure S1: SEM image of the material deposited on the sensor substrate

Figure S1 shows In2O3 material deposited on top of the alumina substrate. A high
density of structures that range around 100 nm with no specific morphology can be
observed.

Characteristic asymmetric vibrational modes of nitrite and nitrate ions

O

N

Unrestricted CCSD/aug-cc-pVTZ (without scaling)

1476 cm-1 (degenerate)

1374 cm-1

Figure S2: Asymmetric stretching modes and vibrational frequencies of nitrate (NO3-) and nitrite (NO2-) ions

13

Time-resolved DRIFT spectra at 130 oC

1600
1400
1200
1000
8000
400
200
0

Time-Domain DRIFT
Spectra

3
5
0
0

3
2
0 Wavenum
5
-1
0 ber / 0
cm
0
0

2
0
0
0

1
5
0
0

0.03
0.03
5
0.02
0.02
5
0.01
0.01
5
0.00
0
5
0.00
5

Figure S3: Time-resolved DRIFT spectra of exposure to several cycles of 1 ppm of NO2 at 130 oC (time scale
identical to that of Figure 3).

Temperature-dependency of the components DRIFT spectra and
concentration profiles

Figure S4: a) component DRIFT spectra and b) concentration profiles obtained by MCR in the nitrite/nitrate region
at the three temperatures.

14

Figure S5: a) component DRIFT spectra and b) concentration profiles obtained by MCR in the hydroxyl region at
the three temperatures.

15