Pulsed Laser Deposited Nanostructured Vanadium Oxide Thin Films Characterized as Ammonia Sensors

G Model

ARTICLE IN PRESS

SNB-18153; No. of Pages 8

Sensors and Actuators B xxx (2015) xxx–xxx

Contents lists available at ScienceDirect

Sensors and Actuators B: Chemical journal homepage: www.elsevier.com/locate/snb

General

Pulsed Laser Deposited Nanostructured Vanadium Oxide Thin Films Characterized as Ammonia Sensors J. Huotari a,∗ , R. Bjorklund b , J. Lappalainen a , A. Lloyd Spetz a,b a b

Microelectronics and Materials Physics Laboratories, University of Oulu, P.O. Box 4500, FI-90014 Oulu, Finland Department of Physics, Chemistry and Biology, Linköping University, SE-581 83, Linköping, Sweden

a r t i c l e

i n f o

Article history: Received 19 June 2014 Received in revised form 16 February 2015 Accepted 17 February 2015 Available online xxx Keywords: Vanadium oxides Mixed phase V7 O16 V2 O5 Thin film NH3 Gas sensor

a b s t r a c t Vanadium oxide thin films were fabricated by pulsed laser deposition. The microstructure and crystal symmetry of the deposited films were studied with X-ray diffraction, scanning electron microscopy (SEM), and Raman spectroscopy, respectively. The films surface morphology was examined by atomic force microscopy. Raman spectroscopy and XRD results showed that the thin film phase-structure was composed of pure orthorhombic V2 O5 phase, or they had a mixed phase structure of orthorhombic V2 O5 and triclinic V7 O16 . Surface morphology of the films consisted of nanosized particles, although in pure V2 O5 films some bigger agglomerates and flakes were also seen. The conductivity based gas sensing measurements showed a clear response already at ppb-levels of NH3 and strong selectivity to ammonia was found when compared to NO and CO gases. Also, the films showed promising gas sensing behavior in cross-sensitivity measurements between NO and NH3 , being able to sense ammonia even in the presence of NO. This is an important property when considering possible sensing applications to control Selective Catalytic Reduction processes, e.g. in diesel engine exhausts, where introduced ammonia, or urea, transforms nitrogen oxide gases in a catalytic converter to nitrogen and water. © 2015 Elsevier B.V. All rights reserved.

1. Introduction Nitrogen oxide gases like nitrogen dioxide, NO2 , and nitric oxide, NO, are formed in combustion processes where oxygen and nitrogen are present. These gases are toxic and dangerous for both human beings and the environment, and therefore the allowed emissions of these gases are almost down to zero today. The most efficient way to reduce NO and NO2 (NOx ) from exhausts gases is to use Selective Catalytic Reduction (SCR). In the SCR process, ammonia NH3 , or urea (NH2 CONH2 ), which together with water forms NH3 and carbon dioxide CO2 at temperatures above 150 ◦ C, is injected in the exhausts of, e.g. diesel vehicles and stationary engines, and in the catalytic converter NH3 reacts with NOx and form harmless nitrogen and water. This process has to be controlled in order to be enough efficient to meet legislations, and this can be performed by either nitrogen oxide sensitive sensors or by an ammonia sensor [1,2]. The requirements on an ammonia sensor for the SCR process depend on the application, but for diesel exhausts, the demands set by the car industry for sensitivity to NH3 are in the range 0 - 100 ppm with an accuracy of ± 5 ppm for new and aged

∗ Corresponding author. Tel.: +358503502928, fax: +35885532728. E-mail address: [email protected].fi (J. Huotari).

sensors, and for response time 3 seconds from 10 - 90% of the full response, low cross sensitivity to other gases, temperature resistance between 20 ◦ C–600 ◦ C, high long-term stability, and resistivity to contaminants such as soot, sulphates, and phosphates [1–4]. Vanadium oxides form an interesting material group because of its varying oxidation states between V2+ and V5+ , the most famous compounds being V2 O5 and VO2 . These different compounds have been studied also as a possible gas sensing materials, e.g. for NH3 . VO2 has an interesting metal-insulator-transition (MIT) property where it changes its phase structure from low-temperature insulating state to high-temperature (T > 68 ◦ C) metallic state. This property has been studied in the context of, e.g. hydrogen sensor [5]. Nanofibers of V2 O5 have been studied as a ppb-level NH3 sensor [6,7], and as a sensing material to amines [8]. Also, the remarkable catalytic properties of V2 O5 have been reported in the case of sensing materials based on mixtures of V2 O5 , WO3 , and TiO2 [9,10]. The sensitivity to ammonia increased with higher content of V2 O5 . Lately, a growing interest has been shown for chemically grown vanadium oxide nanotubes (VOx -NT) whose wall are composed of triclinic V7 O16 layers [11,12] and they have been studied also as a possible ethanol sensor [13]. Some studies have also revealed the complex behavior of mixed-valence vanadium oxide thin films, when tested as conductometric gas sensors for different gases in various conditions [14]. It was concluded, that

http://dx.doi.org/10.1016/j.snb.2015.02.089 0925-4005/© 2015 Elsevier B.V. All rights reserved.

Please cite this article in press as: J. Huotari, et al., Pulsed Laser Deposited Nanostructured Vanadium Oxide Thin Films Characterized as Ammonia Sensors, Sens. Actuators B: Chem. (2015), http://dx.doi.org/10.1016/j.snb.2015.02.089

G Model SNB-18153; No. of Pages 8

ARTICLE IN PRESS J. Huotari et al. / Sensors and Actuators B xxx (2015) xxx–xxx

2

the gas sensing behaviour of vanadium oxide thin films is highly dependent of the measurement conditions, and that the presence of both V5+ in the n-type V2 O5 material and V4+ ions in the p-type V7 O16 material, respectively, causes the sensor devices to even switch between n-type and p-type semiconducting behavior. This behaviour depends on measurement temperature and whether the surrounding gas atmosphere is oxidizing or reducing. Pulsed laser deposition (PLD) has been used to manufacture vanadium oxide thin films for several applications [15], since the method has several advantages including easily controllable film composition by deposition parameters, and a good repetition of stoichiometry of the target material in the films deposited on the substrate. In our earlier studies, pure V2 O5 thin films and films with mixed phases of V2 O5 and V7 O16 were deposited by PLD and then tested as sensors for NO and H2 , for example [14]. Here we characterize these sensors for NH3 sensing for the possible application to control the SCR process. 2. Experimental Pulsed laser deposition (PLD) technique utilizing Lambda Physik Compex 201 excimer laser operating at a wavelength of 308 nm was used to deposit vanadium oxide thin films on oxidized silicon substrate with a pulse repetition rate of 5 Hz. A pure ceramic V2 O5 disc was used as rotating target and the laser pulse energy density was I = 2.6 J/cm2 . Substrate temperatures of in-situ PLD processes were T = 400 ◦ C or 500 ◦ C. The deposition chamber was first pumped down to a base pressure of below 5 ×10−5 mbar and then an atmosphere of oxygen partial pressure p(O2 ) = 6 ×10−2 mbar or 1.5 ×10−2 mbar was used for PLD. The crystal structure of the films was studied using grazing incidence diffraction (GID) of X-ray diffraction (XRD) by utilizing Bruker D8 Discover diffraction device. Raman spectroscopy using argon-ion laser with a wavelength of 488 nm was used to study the crystal symmetry of the samples by utilizing HORIBA Jobin Yvon LabRAM HR800 device. Scanning electron microscopy (SEM) studies were carried out by using Helios Dual-Beam FIB/FESEM device. Thermo Fisher Scientific ESCALAB 250Xi X-ray photoelectron spectroscopy (XPS) device was used to study the valence states of the thin-film surfaces. The surface morphology was studied using Veeco Dimension 3100 atomic force microscope (AFM). Platinum (Pt) electrodes with thickness of 400 nm together with 10 nm-thick titanium (Ti) adhesion layers were sputtered on the surfaces of the thin films. Then the sensors were glued on alumina platforms with platinum thick-film heaters and Pt-100 temperature sensors. The resistance measurements were performed with a Keithley sourcemeter and Bronckhorst flow meters were used to control the gas pulses injected into a 1 cm3 -sized gas measurement chamber. The carrier gases used were 20% of O2 in N2 , and 8% of O2 in N2 to simulate the conditions in diesel engine exhausts, and the measurement temperature was 350 ◦ C, known to be a good measurement temperature for V2 O5 sensors. In all gas measurements, the temperature was raised rapidly to 350 ◦ C and then the sensors were kept at that temperature for several hours in order to get more stable resistance baseline. Before the ppb-level ammonia measurements, the plastic gas tubes were replaced with new ones in order to prevent the buildup of NH3 in the gas line, and thus affecting the concentration of ammonia in the measurement chamber. 3. Results and discussion 3.1. Structural characterization of the thin films Raman spectroscopy and XRD experiments confirmed pure vanadium oxide phases of the films. In Fig. 1 a), Raman spectra of

Fig. 1. a) Raman spectroscopy and b) the grazing incidence X-ray diffraction results of the vanadium oxide thin films. The films with only V2 O5 phase were deposited at 400 ◦ C and in p(O2 ) = 6 ×10−2 mbar (red curves), and the films deposited at 500 ◦ C and in p(O2 ) = 1.5 ×10−2 mbar contained also V7 O16 phase (black curves). The gray circles in (a) present Raman modes originating from V7 O16 phase.

PLD deposited films are presented. According to the data, both films had pure V2 O5 phase present [16] whereas the other film, deposited at 500 ◦ C with O2 partial pressure of 1.5 ×10−2 mbar, contained also another minority phase present, indicated by small peaks with wavenumbers 847, 882, 940, and 1034 cm−1 (grey circles). All the other Raman modes shown in Fig. 1 a) originate from orthorhombic V2 O5 structure, or from silicon substrate. This other phase is identified as triclinic V7 O16 [14,17], the crystal structure typically found in the walls of vanadium oxide nanotubes (VOx -NT) [18,19]. Grazing incidence diffraction (GID) X-Ray diffraction results support the findings of Raman spectroscopy, as shown in Fig. 1 b). It should be noted here that due to high noise and background level of GID data of nanostructures, the data has been smoothed and the high background was removed from the results. While the film deposited at 400 ◦ C with O2 partial pressure of 6 × 10−2 mbar, presented by red curves in Fig. 1, shows a clear diffraction pattern of pure orthorhombic V2 O5 film, the other films had also the peaks corresponding to V7 O16 present at 2␪ ≈ 24.95◦ and 14.45◦ . This particular phase structure is now shown to exist in a solid-state thin-film form and also tested as ammonia sensing material for the first time. From now on in this study, these films containing both V2 O5 phase and V7 O16 phase will be denoted as mixed-phase films. The oxidation state of the mixed-phase thin film was studied using X-ray photoelectron spectroscopy, and in Fig. 2 a), the Voigt

Please cite this article in press as: J. Huotari, et al., Pulsed Laser Deposited Nanostructured Vanadium Oxide Thin Films Characterized as Ammonia Sensors, Sens. Actuators B: Chem. (2015), http://dx.doi.org/10.1016/j.snb.2015.02.089

G Model SNB-18153; No. of Pages 8

ARTICLE IN PRESS J. Huotari et al. / Sensors and Actuators B xxx (2015) xxx–xxx

3

Fig. 2. a) X-ray photoelectron spectra of a mixed-phase vanadium oxide thin film from V 2p3/2 orbital with Voigt function fitting, b) scanning electron microscopy crosssection micrograph from the pure V2 O5 film, c) scanning electron microscopy cross-section micrograph from the mixed-phase film, d) scanning electron microscopy surface micrograph from the pure V2 O5 film, and e) scanning electron microscopy surface micrograph from the mixed-phase film.

function fitted data of the peak of energy level V 2p3/2 used for determination of the valence states of vanadium oxide, is shown. One can see the strong presence of V5+ ions at ≈ 517.2 eV, but also a clear influence of V4+ ions at ≈ 516.2 eV. This is clear indication of the presence of lower-valence vanadium oxide phase than pure V2 O5 in the film structure. The total valence of the film was calculated to be ∼2.43 which also supports the existence of V7 O16 in the film phase structure. Scanning electron microscopy studies of the thin film crosssections and surfaces are shown in Figs. 2 b) - e). In Fig. 2 b), the cross-section SEM micrograph of the pure V2 O5 film is shown, and Fig. 2 c) presents the same data for the mixed-phase film. From the SEM cross-section micrographs of Fig. 2 b) and c), it is seen that both types of films are uniformly composed as continuous thin-film structures with thicknesses below 100 nm. From the surface micrograph of pure V2 O5 film in Fig. 2 d), it is seen that the surface is composed mostly of nanosized grains, but some bigger flake-like structures are also present. The morphology of the surface of the mixed phase film in Fig. 2 e) shows a highly porous surface formed of nanosized grains. As a conclusion, it is clear that both films exhibit a porous, nanostructured surface, but are clearly forming continuous thin films. Surface morphology micrographs measured using AFM are shown in Fig. 3. Micrographs in Figs. 3 a),b) and d),e) are of the in-situ deposited sample surfaces, whereas Fig. 3 c) and f) are micrographs from sample surfaces which have been heat-treated during gas response measurements. Micrographs of the pure V2 O5 film are shown in Fig. 3 a), b), and c), and it can be seen that the film surface consist mostly of nanosized grains, but also bigger agglomerates and flakes are present. The root mean square roughness values Rq determined from 2 ␮m x 2 ␮m micrographs, describing the standard deviation of the surface profile height values, were Rq = 27.07 nm for the in-situ deposited samples, and Rq = 8.8 nm for

the samples of pure V2 O5 films after measurements at high temperature. The surface micrographs of mixed-phase films with both V2 O5 and V7 O16 phases are shown in Fig. 3 d), e), and f). The sample surface consists only of single nanoparticles. Also, the sample surface is much more flat in the mixed-phase film, when compared to pure V2 O5 film. Values of Rq for the mixed-phase film were Rq = 5.01 nm for the in-situ deposited samples, and Rq = 2.82 nm for the samples after gas response measurements at high temperature. It is also seen, that some changes took place in the film surface morphology during the heat treatments (Fig. 3 c), f)), when compared to in-situ deposited samples. Both films show a large effective surface area suitable for gas sensing. For the comparison, the micrographs taken by AFM from the as-deposited sample surfaces are obviously consistent with the SEM surface studies shown in Fig. 2. Structural characterization of the films is based on (i) Raman spectroscopy, (ii) XRD, and (iii) XPS methods. These result show without any doubt, that there are two kind of phases included in the samples. Especially, the observed results of Raman spectroscopy reveal in a very clear way that there are two separate phases constituting two different entities, i.e. grains in these polycrystalline samples. This is due to the fact that, for example, Raman modes assigned (gray circles) only for V7 O16 in Fig. 1 a) cannot originate from V2 O5 phase due to the crystal symmetry restrictions. Nevertheless, Raman modes of V2 O5 are still evident in the same graph leading to the conclusions that the sample with data presented by the black curve in Fig. 1 a) actually consists of contributions of two different entities with different symmetries, i.e. two different crystalline structures, which are called grains in this context. Since the films are shown to be continuous, which is, on the other hand, supported by the fact that they show resistive response for the applied electric field in the form of finite current density, and that films are also polycrystalline, which is clearly supported by SEM and AFM micrographs in Figs. 2 and 3, respectively, it is inevitable that the

Please cite this article in press as: J. Huotari, et al., Pulsed Laser Deposited Nanostructured Vanadium Oxide Thin Films Characterized as Ammonia Sensors, Sens. Actuators B: Chem. (2015), http://dx.doi.org/10.1016/j.snb.2015.02.089

G Model SNB-18153; No. of Pages 8

ARTICLE IN PRESS J. Huotari et al. / Sensors and Actuators B xxx (2015) xxx–xxx

4

Fig. 3. Atomic force microscopy micrographs of the surface morphologies of (a-c) films with pure V2 O5 phase, and (d-f) the mixed-phase films. Figures a), b), d) and e) show samples in-situ deposited, whereas c) and d) samples heat-treated during the measurements.

two observed phases V7 O16 and V2 O5 form interfaces, such as grain boundaries. The existence of this structure is further supported by XRD and XPS results. Furthermore, observed film microstructure is very consistent with the observed values of resistivities of mixedphase films presented later in this manuscript. As a consequence of the different crystalline structures, the surface morphologies of the films differ from each other, most likely affecting also the gas response characteristics.

two n-type or two p-type grains. Under the constant applied electrical field during the gas response measurements, a certain number of these asymmetric grain boundary interfaces are reverse biased, the current density through them is very low, and these grain boundary

3.2. Gas sensing results 3.2.1. Detection of NH3 down to ppb-level Sensor responses for 10 minute pulses of NH3 gas in concentrations below 1 ppm, with carrier gas of 20% of O2 in N2 , are shown in Fig. 4a) for films of V2 O5 and the mixed phase (V2 O5 + V7 O16 ), respectively. Both sensors show a clear decrease of resistance during gas exposure, indicating dominant n-type behavior in both of the thin films. A clear response is seen for as low concentrations as 160 ppb of ammonia for the pure V2 O5 film, and the mixed-phase film has a response already for 80 ppb. Some change of resistance is already seen at lower concentrations, which is probably due to two different reactions mechanisms affecting the response. It is very likely that NO2 is created when adsorbed NH3 molecules react with oxygen on the sensor surface and the produced NO2 will thus screen the response from the actual ammonia exposure, and lead to a rise in the resistance at very low concentrations of NH3 [7]. This reaction is seen for 40 ppb and 80 ppb concentrations of NH3 in the case of V2 O5 samples, as shown in Fig. 4 a). The mixed-phase film shows a higher response to NH3 than the pure V2 O5 film. One reason for this might be the higher resistance level of mixed-phase material being composed of V7 O16 grains with p-type conductivity and of V2 O5 grains with n-type conductivity of [13,14]. Therefore, the mixed-phase film has evidently grain boundaries between nand p-type grains in its structure. When a resistive Taguchi-type gas sensor is fabricated of such a mixed-phase film, the current paths for electrons and holes are formed through the necks in grain boundaries, in which potential barriers are formed due to discontinuity of the crystal structure. When n-type (V2 O5 ) and p-type (V7 O16 ) grains have electrically connected grain-boundary interfaces, the potential barriers are consequently determined by the differences of Fermi levels of the grains, and they are essentially asymmetric in comparison to potential barriers formed between

Fig. 4. a) Sensor response for ppb-level NH3 at 350 ◦ C of vanadium oxide thin films in 20% of O2 in the carrier gas, and b) normalized response R/R0 curves of the same films in both 20% of O2 and 8% of O2 gas in the carrier gas. The pulse length was 10 min.

Please cite this article in press as: J. Huotari, et al., Pulsed Laser Deposited Nanostructured Vanadium Oxide Thin Films Characterized as Ammonia Sensors, Sens. Actuators B: Chem. (2015), http://dx.doi.org/10.1016/j.snb.2015.02.089

G Model SNB-18153; No. of Pages 8

ARTICLE IN PRESS J. Huotari et al. / Sensors and Actuators B xxx (2015) xxx–xxx

interfaces actually limit the total current density of the sensor. This is observed as increased resistance level. Also, it is seen from Fig. 4, that a pulse time of 10 minutes is not enough for the sensors, especially the ones containing the mixed phase, to reach saturation. The baseline drift of the mixed-phase film was also found to be almost negligible in the case of the pure V2 O5 sensor. The responses plotted as R/R0 , where R0 is the sensor resistance value in the carrier gas, for both types of sensors as a function of ammonia concentrations are shown in Fig. 4 b). The curves emphasize the fact, that the mixed-phase films show higher sensitivity to NH3 with lower detection limit, below 100 ppb. Also, no saturation in the response curves is noticed, and the mixed-phase films have clearly higher responses still at ppm-level concentrations. As can be seen from the presented results, the mixed-phase films seem to have a higher response towards ammonia than the pure V2 O5 films. The explanation to this might be the higher resistance level in the mixed phase films. However, it is also noticed that the resistance baseline is drifting more and that the recovery time from the gas pulses is longer for the mixed-phase films than for the pure

5

V2 O5 films. The different microstructures and stoichiometry of the films are probably the reasons for this behaviour. The mixed-phase structure is more oxygen deficient than the pure V2 O5 phase and hence more unstable at higher temperatures. The grain boundaries between n- and p-type materials, which necessarily exist in the mixed-phase films, make the behaviour of the electrical properties more complex, especially under gas exposure. It should be noted that both barrier height and depletion layer widths in grain boundaries between n-type V2 O5 and p-type V7 O16 phases also contribute to the mixed-phase film gas response, which is a different situation from the pure n-type V2 O5 phase films, for example reported in Refs. [13,14,20]. The p-type conductivity of V7 O16 and n-type conductivity of V2 O5 are confirmed also in our previous studies [14]. Furthermore, if there were no p-n boundaries in the samples, the total resistance should not be higher than that of pure V2 O5 , since more V4+ ions containing vanadium oxides has been proven to show smaller resistivity than pure V2 O5 [21]. The SEM micrographs in Fig. 2e) and f), also indicate a polycrystalline structure with grains of different kinds. It should also be pointed out that the V2 O5 and

Fig. 5. Gas responses for NH3 of the pure V2 O5 phase film at 350 ◦ C and a) in synthetic air at ppb levels, b) in synthetic air at ppm levels, and c) at presence of 1% of H2 O. The pulse length was 2 hours with 2 hours in between pulses.

Please cite this article in press as: J. Huotari, et al., Pulsed Laser Deposited Nanostructured Vanadium Oxide Thin Films Characterized as Ammonia Sensors, Sens. Actuators B: Chem. (2015), http://dx.doi.org/10.1016/j.snb.2015.02.089

G Model SNB-18153; No. of Pages 8 6

ARTICLE IN PRESS J. Huotari et al. / Sensors and Actuators B xxx (2015) xxx–xxx

the V7 O16 both show lower resistance as compared to the mixed phase, which can be expected when mixing p- and n-type materials [14,22]. The pure V2 O5 thin film, on the other hand, showed a very stable performance as a gas sensor, as shown in Fig. 5. The measurements were now performed in 20% O2 in N2 at 350 ◦ C with 2 h pulses during a total of 100 hours. In Fig. 5 a), the NH3 response at ppblevels is shown again, whereas Fig. 5 b) presents NH3 response at higher concentrations from 4 ppm to 32 ppm. The sensor response is still strong, even though some tendency towards saturation at higher concentrations is seen. In Fig. 5 c), NH3 response at ppblevel now also with 1% of H2 O present is shown. In dry synthetic air, the resistance baseline shows a reasonably stable behaviour during the measurements and the detection limit of NH3 is below 80 ppb. The measurement in the presence of H2 O, on the other hand, shows some drift in the resistance baseline, however, the sensor still clearly detects ammonia already at 40 ppb. From Figs. 4 and 5, it can be concluded that the vanadium oxide sensing material fulfils the requirement of accuracy as stated in the introduction, at least for the time periods investigated here. The dynamic range is probably smaller than 100 ppm but the detection limit should be no problem, not even with real exhaust gases, however, the speed of response needs improvement. 3.2.2. Cross-sensitivity testing with NH3 and NO The responses to 35 ppm of NH3 together with 35 ppm of NO in the carrier gas of 20% O2 in N2 , presented as relative change R/Ro , of pure V2 O5 films and of the mixed-phase films, are shown in Figs. 6 a) and b), respectively. Both films had a quite similar behavior during the gas exposures. It was found, that when a 2-hour pulse of 35 ppm of NO is introduced to the measurement chamber (point 1. in Fig. 6 a and b), the resistance value rises rapidly. When 35 ppm of NH3 is introduced with the NO still on and the oxidizing reaction of NO takes place simultaneously with the reducing reaction of NH3 during the 2-hour period, the resistance drops down (point 2. in Fig. 6 a and b). Next, NO gas is turned off and NH3 kept on for 2 hours (point 3), and the response level now to the pure NH3 stays about the same. When both gases are turned off for 4-hours period (point 4), the resistance value returns back to the original baseline. After this, only 35 ppm of NH3 is injected into chamber for 2 hours (point 5) and the sensors give the same response as before, compare to point 3. Then again, also 35 ppm of NO is added to the chamber with NH3 still on, and during this 2-hour period only small changes are noticed in the resistance signal (point 6). In the next phase of measurement, only NO gas is left on for two hours (point 7) and now the resistance value rises over the baseline due to the response to pure NO as compared to point 1. Finally, both gases are turned off (points 8. in Fig. 6) and the resistance value returns to the baseline. It is clearly seen from the results, that the response to NH3 is dominating the gas response during exposure to both NH3 and NO, and that the responses to NH3 and NO are in different directions during these measurement conditions. In the SCR process, NH3 and NO reacts on a catalyst and in the presence of O2 , they form nitrogen and water. This does not seem to take place here. A gas-phase reaction between the two components also takes place, but only at temperatures above 500 ◦ C. Instead, it is assumed here that the NH3 response occupies the sites for the NO response. It should be pointed out that background resistance drift is negligible for the presented 48-hours measurement shown in Fig. 6 a) for the pure V2 O5 film. The mixed-phase film in Fig. 6 b) had somewhat more drift in the resistance baseline during the 24-hours measurement. Both types of thin films were also tested with a carrier gas of 8% O2 in N2 , since this is close to the oxygen level in diesel engine exhausts. The sensors were tested for NH3 and NO. In Fig. 7 a), the responses for 1 ppm, 2 ppm, and 3 ppm of NH3 with pulse length of 10 min are shown. Both sensors had a clear response to NH3 at

Fig. 6. The cross-sensitivity measurements of the vanadium oxide thin films for 35ppm level of NH3 with 35-ppm level of NO at 350 ◦ C for a) pure V2 O5 phase film, and b) the mixed-phase film.

1 ppm level also with this type of carrier gas. The response to NO in this carrier gas composition is shown in Fig. 7 b). The detection limit for NO was found to be between 15 ppm and 20 ppm in these measurement conditions. In Fig. 7 c), the cross-sensitivity measurement results with 20 ppm of NH3 , and 20 ppm of NO are shown. The results are consistent with the measurements made with 20% of O2 , previously shown in Fig. 6. The sensors first react rapidly with NO gas, seen as a rise in the resistance level, but when NH3 is also introduced, the resistance drops down, and when NO is finally turned off and only NH3 is left on, the response level of the sensors stays the same. It is seen again, that the mixed-phase sample is more sensitive to pure NH3 , and also that a small drift is present in the background signal, whereas V2 O5 thin film shows a more stable behavior. That there is no cross sensitivity to NO and the stability of the NH3 response, especially of the V2 O5 films, indicates that vanadium oxide thin films as a sensing material is a good candidate for the application of controlling the Selective Catalytic Reduction (SCR) process. 3.2.3. Pure V2 O5 gas response for CO and NO Sensor responses for NO and CO gases of the pure V2 O5 thin films are shown in Fig. 8 a) and b), respectively. From the curves, it is seen that the detection limit for NO is between 30 and 35 ppm and for CO approximately 50 ppm. Also, the stable behaviour of the

Please cite this article in press as: J. Huotari, et al., Pulsed Laser Deposited Nanostructured Vanadium Oxide Thin Films Characterized as Ammonia Sensors, Sens. Actuators B: Chem. (2015), http://dx.doi.org/10.1016/j.snb.2015.02.089

G Model SNB-18153; No. of Pages 8

ARTICLE IN PRESS J. Huotari et al. / Sensors and Actuators B xxx (2015) xxx–xxx

7

Fig. 8. The gas responses for the pure V2 O5 phase film at 350 ◦ C and with a) NO at ppm levels, and b) CO at ppm levels. The pulse length was 2 hours with 2 hour pulses in between.

Fig. 7. Gas responses of both pure V2 O5 film and mixed-phase film in a carrier gas of 8% of O2 in N2 at 350 ◦ C and a) NH3 at 1, 2, and 3 ppm levels, b) NO at 15, 20, and 25 ppm levels, and c) at 20 ppm level of NH3 with 20 ppm of NO.

pure V2 O5 film during the measurements is again evident. Based on these results, it is obvious that the pure V2 O5 thin film has much lower detection limit towards NH3 than to NO and CO. This suggests reasonable selectivity to NH3 in the applications where a sensor is positioned in the exhaust pipe after the catalytic converter, where the NO concentration normally is very low, below 20 ppm, and the CO concentration also stays below 100 ppm [1,2]. Other interfering

compounds like hydrocarbons and soot will have to be tested as well of course. An interesting behavior is noticed for the NO gas sensing response in both Fig. 7 b) with 8% O2 in N2 as a carrier gas, and in Fig. 8 a) also with 20% O2 in N2 . The sensors have a thresholdtype behavior in the responses to NO in both cases. In Fig. 7 b), the threshold is around 20 ppm and in Fig. 8 a), around 35 ppm of NO concentration. A tentative explanation can be that there is a competition between oxygen and NO adsorption on the sensor surface, and that NO does not decompose on the surface until it has replaced a certain amount of oxygen adsorbents. Once this level is reached, which is lower for 8% than to 20% of O2 in N2 background, NO starts to decompose and add oxygen to the surface [23], which leads to resistivity increase in the sensor signal. Further studies, for example, by mass spectrometer and diffuse reflectance infrared Fourier transform (DRIFT) spectroscopy is needed to clarify the details of this gas response, which to our knowledge is not yet reported. On the other hand, the CO response shown in Fig. 8 b), is very well characterized for many sensor surfaces. The steplike behavior of the response is well understood and documented [24,25]. Here CO is competing with oxygen on the surface and at a certain concentration, which depends on the oxygen concentration and the temperature, the surface switches from predominantly being oxygen covered to being covered by CO instead. 4. Conclusions Pulsed laser deposited vanadium oxide thin films were studied as resistive gas sensors for NH3 . The film composition was determined to be either pure V2 O5 phase structure, or a mixed-phase

Please cite this article in press as: J. Huotari, et al., Pulsed Laser Deposited Nanostructured Vanadium Oxide Thin Films Characterized as Ammonia Sensors, Sens. Actuators B: Chem. (2015), http://dx.doi.org/10.1016/j.snb.2015.02.089

G Model SNB-18153; No. of Pages 8

ARTICLE IN PRESS J. Huotari et al. / Sensors and Actuators B xxx (2015) xxx–xxx

8

structure with V2 O5 phase and triclinic V7 O16 phase. The film surface morphology was determined to be mostly nanocrystalline, however in the pure V2 O5 -film surfaces larger agglomerates of nanosized grains and flakes could also be found. Film surfaces showed large specific surface area suitable for gas sensing. Both types of the films showed NH3 sensitivity already below 100 ppb concentrations. Also, both types of the films showed strong selectivity between NH3 and NO. The cross-sensitivity tests at 35 ppm of NH3 in 35 ppm of NO background showed that the ammonia response is dominating the gas response even in the presence of NO. The mixed-phase films showed higher sensitivity to NH3 than the pure V2 O5 films. However, the mixed-phase films had a slight drift in the baseline, probably rising from the complex crystal and defect structure. The more stable pure V2 O5 thin films were tested also for longer time periods up to 100 h as ammonia sensors, and very stable behavior was found with detection limit of 80 ppb for NH3 both in dry and humid air. The detection limits for NO and CO were much higher, 20 ppm and 50 ppm, respectively. The detection limits for NO and CO together with the results from cross-sensitivity measurements suggest that the vanadium oxide thin-film sensing layers are good candidates for a Selective Catalytic Reduction (SCR) process control application. However, especially the speed of response to ammonia needs further development and the stability of the sensors in humid atmosphere should be further investigated. Acknowledgements Financial support of Finnish Funding Agency for Innovations TEKES project CHEMPACK (no. 1427/31/2010) is acknowledged. The assistance of Center of Microscopy and Nanotechnology of University of Oulu is also acknowledged. J.H acknowledges the Riitta and Jorma J. Takanen Foundation and Walter Ahlström Foundation for financial support. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.snb.2015.02.089. References [1] D.Y. Wang, S. Yao, M. Shost, J. Yoo, D. Cabush, D. Racine, Ammonia sensor for close-loop SCR control, Proc. SAE 2008 World Cong. SAE Paper 2008-01-0919, 2008. [2] A. Herman, M.C. Wu, D. Cabush, M. Shost, Model based control of SCR dosing and OBD strategies with feedback from NH3 sensors. Proc. SAE 2009 World Cong. SAE Paper 2009-01-0911, 2009. [3] H. Wingbrant, H. Svenningstorp, P. Salomonsson, D. Kubinski, J.H. Visser, M. Löfdahl, A. Lloyd Spetz, Using a misic-FET sensor for detecting NH3 in SCR systems, IEEE Sens. J. 5 (2005) 1099–1105. [4] B. Timmer, W. Olthuis, A. van den Berg, Ammonia sensors and their applications–a review, Sens. Actuators B 107 (2005) 666–677. [5] J.W. Byon, M.-B. Kim, M.H. Kim, S.Y. Kim, S.H. Lee, B.C. Lee, J.M. Baik, Electrochemically induced highly responsive and highly selective vanadium oxide hydrogen sensor based on metal-insulator transition, J. Phys. Chem. C 116 (2012) 226–230. [6] V. Modafferi, S. Trocino, A. Donato, G. Panzera, G. Neri, Electrospun V2 O5 composite fibers: synthesis, characterization and ammonia sensing properties, Thin Solid Films 548 (2013) 689–694. [7] V. Modafferi, G. Panzera, A. Donato, P.L. Antonucci, C. Cannilla, N. Donato, D. Spadaro, G. Neri, Highly sensitive ammonia resistive sensor based on electrospun V2 O5 fibers, Sens. Actuators B 163 (2012) 61–68. [8] I. Raible, M. Burghard, U. Schlecht, A. Yasuda, T. Vossmeyer, V2 O5 nanofibers: novel gas sensors with extremely high sensitivity and selectivity to amines, Sens. Actuators B 106 (2005) 730–735. [9] D. Schönauer-Kamin, M. Fleischer, R. Moos, Half-cell potential analysis of an ammonia sensor with the electrochemical cell Au | YSZ | Au, V2 O5 -WO3 -TiO2 , Sensors 13 (2013) 4760–4780. [10] D. Schönauer-Kamin, M. Fleischer, R. Moos, Influence of the V2 O5 content of the catalyst layer of a non-Nernstian NH3 sensor, Solid State Ionics 262 (2014) 270–273.

[11] F. Krumeich, H.-J. Muhr, M. Niederberger, F. Bieri, B. Schnyder, R. Nesper, Morphology and topochemical reactions of novel vanadium oxide nanotubes, J. Am. Chem. Soc. 121 (1999) 8324–8331. [12] H.-J. Muhr, F. Krumeich, U.P. Schönholzer, F. Bieri, M. Niederberger, L.J. Gauckler, R. Nesper, Vanadium oxide nanotubes - a new flexible vanadate nanophase, Adv. Mater. 12 (2000) 231–234. [13] M. Yu, X. Liu, Y. Wan, Y. Zheng, J. Zhang, M. Li, W. Lan, Q. Su, Gas sensing properties of p-type semiconducting vanadium oxide nanotubes, Appl. Surf. Sci. 258 (2012) 9554–9558. [14] J. Huotari, J. Lappalainen, J. Puustinen, A. Lloyd Spetz, Gas sensing properties of pulsed laser deposited vanadium oxide thin films with various crystal structures, Sens. Actuators B 187 (2013) 386–394. [15] D.H. Kim, H.S. Kwok, Pulsed laser deposition of VO2 thin films, Appl. Phys. Lett. 65 (1994) 3188–3190. [16] X.J. Wang, H.D. Li, Y.J. Fei, X. Wang, Y.Y. Xiong, Y.X. Nie, K.A. Feng, XRD and Raman study of vanadium oxide thin films deposited on fused silica substrates by RF magnetron sputtering, Appl. Surf. Sci. 177 (2001) 8–14. [17] X. Liu, C. Huang, J. Qiu, Y. Wang, The effect of thermal annealing and laser irradiation on the microstructure of vanadium oxide nanotubes, Appl. Surf. Sci. 253 (2006) 2747–2751. [18] M. Wörle, F. Krumeich, F. Bieri, H.-J. Muhr, R. Nesper, Flexible V7 O16 layers as the common structural element of vanadium oxide nanotubes and a new crystalline vanadate, Z. Anorg. Allg. Chem. 628 (2002) 2778–2784. [19] V. Petkov, P.Y. Zavalij, S. Lutta, M.S. Whittingham, V. Parvanov, S. Shastri, Structure beyond Bragg: Study of V2 O5 nanotubes, Phys. Rev. B 69 (2004) 085410. ´ A. Kis, J.W. Seo, F. Bieri, [20] B. Sipos, M. Duchamp, A. Magrez, L. Forró, N. Bariˇsic, F. Krumeich, R. Nesper, G.R. Patzke, Mechanical and electronic properties of vanadium oxide nanotubes, J. Appl. Phys. 105 (2009) 074317. [21] O. Schilling, K. Colbow, A mechanism for sensing reducing gases with vanadium pentoxide films, Sens. Actuators B 21 (1994) 151–157. [22] S.-T. Jun, G.M. Choi, Composition dependence of the electrical conductivity of ZnO(n)–CuO(p) ceramic composite, J. Am. Ceram. Soc. 81 (1998) 695–699. [23] C.M. Gonzalez, X. Du, J.L. Dunford, M.L. Post, Copper tungstate thin-films for nitric oxide sensing, Sens. Actuators B 173 (2012) 169–176. [24] E. Becker, M. Andersson, M. Eriksson, A. Lloyd Spetz, M. Skoglundh, Study of the sensing mechanism towards carbon monoxide of platinum-based field effect sensors, IEEE Sens. J. 11 (2011) 1527–1534. [25] M. Andersson, A. Lloyd Spetz, Tailoring of field effect gas sensors for sensing of non-hydrogen containing substances from mechanistic studies on model systems, Proc. IEEE Sensors (2009) 2031–2036.

Biographies Joni Huotari received his M.Sc. degree in electrical engineering from the Microelectronics and Material Physics Laboratories of University of Oulu in 2010. He started as a Ph.D. student at the same laboratory during spring 2011. His primary research interest is pulsed laser deposition fabrication and structural and electrical characterization of nanostructured functional materials for chemical sensors. Dr. Robert Bjorklund received a PhD degree in physical chemistry from Northwestern University, USA, in 1976. After working a few years with catalyst development in the oil industry at Gulf Research and Development Company, USA, he moved to Linköping University in 1980 as a researcher in the Applied Physics scientific area. Since 2000 he has been involved with laboratory and field development of liquid and gas phase sensors. Prof. Jyrki Lappalainen received his M.Sc. and Ph.D. degrees in microelectronics and materials physics from the University of Oulu, Finland in 1993 and 2000, respectively. He is currently professor of electronics manufacturing technologies, especially in packaging technologies for electronics and optoelectronics, and micromodules in Microelectronics and Materials Physics Laboratories at University of Oulu. He also has a docentship in microelectronics, especially in microelectronics materials and fabrication techniques at University of Oulu. His research interests are in the field of functional electroceramics in the form of thin films and nanostructures. He has published over 80 papers on fabrication and characterization of structural, electrical, optical, and mechanical properties of functional electroceramics thin films and nanostructures, and their implementation in various applications including photonics and microwave components, sensors, actuators, and integrated electronic circuits. Prof. Lappalainen has been a member of the European Physical Society (EPS) since 1996. Prof. Anita Lloyd Spetz professor in Applied Sensor Science, at Linköping University and 2011–2015 FiDiPro Professor at University of Oulu, Finland. She is the Director of the VINN Excellence centre, FunMat, Functional nanostructured Materials at Linköping University and the deputy Chair of the COST network EuNetAir TD1105. Her research involves SiC-FET high temperature gas sensors with MAX material ohmic contacts, wide band gap material transducers for biosensors, resonator sensors, soot sensors, and graphene sensors. The project at University of Oulu regards development of a portable nanoparticle detector and a method to, by electrical means, measure influence of nanoparticles on cells. She runs application projects with industry, and is member of the board of SenSiC AB for commercialization of SiC-FET sensors. She has published more than 120 papers in scientific journals.

Please cite this article in press as: J. Huotari, et al., Pulsed Laser Deposited Nanostructured Vanadium Oxide Thin Films Characterized as Ammonia Sensors, Sens. Actuators B: Chem. (2015), http://dx.doi.org/10.1016/j.snb.2015.02.089