Improved H2-sensing performance of nanocluster-based highly porous tungsten oxide films operating at moderate temperature

Sensors and Actuators B 174 (2012) 65–73

Contents lists available at SciVerse ScienceDirect

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

Improved H2 -sensing performance of nanocluster-based highly porous tungsten oxide films operating at moderate temperature Meng Zhao a,b , Chung Wo Ong a,∗ a b

Department of Applied Physics and Materials Research Center, The Hong Kong Polytechnic University, Hong Kong, China Department of Electronic Science and Technology, Changzhou University, Changzhou 213614, China

a r t i c l e

i n f o

Article history: Received 18 May 2012 Received in revised form 6 August 2012 Accepted 7 August 2012 Available online 17 August 2012 Keywords: H2 sensor WO3 nanoclusters High porosity Fast response High cyclic stability

a b s t r a c t Many nano-metal oxides are claimed to be responsive to H2 at room-temperature, but their response rates may be too low for the signal to reach equilibrium quickly. We investigated the H2 -induced resistive response of palladium-coated supersonic cluster beam deposited tungsten oxide (Pd/SCBD WO3 ) films at temperatures Tsen between 20 and 140 ◦ C. They are constructed of genuine nano WO3 clusters (3–5 nm) loosely packed together, and have a high porosity. Slight increase of Tsen from room-temperature greatly speeds up the response rate and gives a strong apparent sensor response. Other advantages include excellent cyclic stability and selectivity against vapors of various organic compounds; negligible influence from moisture and mild ambient pressure dependence. © 2012 Elsevier B.V. All rights reserved.

1. Introduction Hydrogen (H2 ) is important in industrial processing and renewable energy research [1]. Therefore, sensing technology of the gas is highly envisaged for safety consideration [2,3]. Conventional n-type metal oxide (MOx ) semiconductors show H2 sensing properties [4,5]. At low operation temperature, oxygen from the surrounding is mainly physisorbed on the MOx surface in molecular form (O2 ). When operation temperature increases, say up to 400 ◦ C, some O2 molecules trap conduction electrons from the MOx to become chemisorbed ions like O2 − , O− and O2− [4,6], and the resistance of the MOx is high. When H2 molecules appear to react with oxygen ions, electrons are released and return to the oxide. The depletion layer becomes thinner and the resistance drops. In both processes, the oxide must have dense enough charge carriers. This requires a large-bandgap MOx sensor to operate at elevated temperature Tsen , usually above 300 ◦ C, for maintaining certain electron–hole pair population [3]. However, at a high Tsen , the material could be annealed, such that the structure of the material, as manifested by composition; oxygen vacancy concentration and location; [7–9] and grain size etc., could be altered gradually to result in instability of gas sensing performance [7,10]. This fact leads

∗ Corresponding author. Tel.: +852 2766 5689; fax: +852 2333 7629. E-mail addresses: [email protected] (M. Zhao), [email protected] (C.W. Ong). 0925-4005/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.snb.2012.08.018

to a strong intention of attaining H2 sensors workable at ambient temperature. Associated with the emergence of nano-metal oxides (nanoMOx ), many of them are found to have resistive response to H2 in ambient temperature environment [11–20]. The achievement is ascribed to the large area-to-volume ratio in a substance containing a great number of tiny particles, on which gas–solid phase reaction is greatly facilitated. However, a low Tsen may also result in low reaction rate, such that the sensor’s response cannot be able to reach equilibrium in a reasonably short period [12,13]. Apparently, the output keeps drifting throughout the measurement process, such that the true concentration of target gas is not readily determined. One is immediately motivated to see whether an optimum Tsen of a nano-MOx H2 sensor is achieved to compromise sensor response and response rate, and how other aspects of its performance are changed. In this study, we investigated the resistive sensing properties to H2 of palladium coated supersonic cluster beam deposited tungsten oxide films (Pd/SCBD WO3 ). The Pd coating is added to assist catalytic dissociation of H2 . Different from other nano-MOx based H2 sensors, an SCBD WO3 film is made of loosely connected 3–5 nm nanoclusters and has a very high porosity. It provides a large active area to interact with the gas, and allows the gas to diffuse agilely in the structure. These features are expected to result in a larger sensor response and a faster response rate concomitantly. In this paper, we report how the H2 detection signal; response rate; stability; selectivity; relative humidity and ambient air pressure dependences are affected by slight increase in Tsen from 20 to 140 ◦ C (mainly

66

M. Zhao, C.W. Ong / Sensors and Actuators B 174 (2012) 65–73

concentrated on the case of 80 ◦ C). Results lead to many generic concepts in using this sensor. Comparison of sensor response and response time with other reported nano-MOx based H2 sensors was made.

2. Experimental methods An SCBD system [21] (Tethis, Italy) was used to fabricate Pd/SCBD WO3 films, Fig. 1. A tungsten metal rod of 1/4 in. in diameter and 99.95% purity was mounted in a Pulsed Micro-plasma Cluster Source (PMCS). The target was uniformly rotated with a motor. Argon (Ar) gas pulses were admitted at a frequency of 5 Hz. Electric pulses of −900 V at the same frequency were applied to the target, delayed by 220 ␮s from the gas pulses. Micro-plasma pulses were generated to remove tungsten atoms from the target. The atoms aggregated into nanoclusters inside the chamber, which were carried by the Ar gas flow and enter an expansion chamber. In the expansion chamber, the Ar gas experienced a supersonic expansion and passed through a set of aerodynamic lens, which was then reshaped to become a fine jet. Nanoclusters in the gas, with diameters falling in a certain range, were selected and collinated to travel along the direction of the axis of the lens set. They finally entered a deposition chamber, in which they were deposited on a substrate. The average kinetic energy of the nanoclusters was low, such that they were not deformed when bomdarding on the substrate surface, and were rather immobile after landing. Hence, they mainly stacked loosely on each other and generate a very porous film structure. The film was then post-annealed at 250 ◦ C in air for 12 h in order to fully oxidize the tungsten atoms. At last, a 5-nm Pd coating was sputtered on top as a catalyst to assist dissociation of H2 . For structural characterization, scanning electron microscopy (SEM, model: JEOL JSM-6335F) and transmission electron microscopy (TEM, model: JEOL JEM-2011) was used to observe nanoclusters in selected area. Porosity of the film was estimated by first measuring the mass of an as-deposited tungsten metal film with a quartz monitor. The data was used to calculate the mass-equivalent thickness of the post-annealed WO3 film according to the density of bulk WO3 . It is then compared with the real thickness of the layer as seen in a cross-sectional TEM image. Resistive H2 sensing properties of film samples were investigated by using a homemade system [22], Fig. 2. It has a stainless steel measurement chamber used to house a sample. The flanges had electrical feedthroughs, gas inlets and pumping ports for performing the test. Cyclic testing mode was employed. A cycle started with filling up the chamber with synthetic air to one atmospheric pressure. The chamber was then evacuated with a rotary pump. A gas mixture of 20,000 ppm (=2%) H2 in air from a cylinder was used to hydrogenate the film. The H2 concentration can be diluted by premixing it with synthetic air in another chamber. The gas can also be directed to pass through a water tank to introduce moisture. The gas was then admitted into the chamber to one atmospheric pressure to react with the film. After waiting for a preset period, the chamber was evacuated again and filled up with synthetic air. After waiting for another preset period, a new cycle of test can be started. The resistance of the film material between two silver paste electrodes on its surface, separated by a distance of 5 mm, was monitored by using a Keithley 617 electrometer (input impedance > 2 × 1014 ). Note that cyclic test mode facilitates quick exchange of gaseous environment, such that response time can be evaluated more accurately. The substrate holder had a resistive heater, which was used to adjust the temperature of operation, Tsen . Tsen was automatically stabilized through a feedback mechanism. Three values of Tsen were used. They are 20, 80 and 140 ◦ C. Sensor response, S, is defined as (Ro − RH2 )/RH2 , where Ro was the base resistance of the film measured in dry synthetic air,

and RH2 was that observed at the end of the H-loading process. Response time, tR is defined as the period for RH2 to go through 90% of total resistance change in a H-loading process, i.e. 0.9(Ro − RH2 ). It is extracted from plot of RH2 versus time. The number of loading–unloading cycles showing stable amplitude of variation of resistance represents the cyclic stability of the film sensor. The experiment may also be referred to as an accelerated test of durability. Selectivity against the influence of the vapors of several organic compounds was examined. The influence of moisture was observed by setting the relative humidity of the H2 -containing gas at 0 and 90% respectively. Finally, the ambient air pressure dependence of the resistive response of the film was investigated. 3. Results and discussion 3.1. Film structure Fig. 3(a) is an SEM image of an annealed SCBD WO3 film without a Pd coating. It shows that the film surface is very rough and has many nanopores. Fig. 3(b) is a cross-sectional TEM image showing the layered structure of the film, containing a 5-nm Pd surface coating and a 140-nm WO3 layer. Fig. 3(c) is another cross-sectional TEM image magnifying a region near the film surface, where Pd gains are seen. The inset shows the electron diffraction patterns of Pd and WO3 nanocrystals. Fig. 3(d) is a cross-sectional TEM image captured from a region located deeper under the film surface. It indicates that the oxide film contains genuine WO3 nanoclusters of 3–5 nm in diameter, which is referred to as the characteristic length, L, of the gas sensing elements in our samples. The mass-equivalent thickness of the asdeposited tungsten metal film was measured to be 14.1 nm, and that of the WO3 film obtained after oxidation was calculated to be 48 nm by making use of the densities of bulk metal tungsten and WO3 (19,300 and 7160 kg m−3 ). On the other hand, the real thickness of the WO3 layer observed in a TEM image is 140 nm. This value is reconfirmed by the height of the edge of a film from the substrate surface, obtained from the surface profile collected by using an alpha-step surface profiler. The porosity of the SCBD WO3 layer is estimated to be 66%. 3.2. Sensor response and response time Fig. 4(a) and (b) shows the dependence of RH2 of a Pd/SCBD WO3 film on H2 concentration measured at Tsen = 20 ◦ C. Regarding the low response rate at this temperature, the H-loading period in a switching cycle was set to be 60 min long. Even with such a long loading time, RH2 was not entirely stabilized throughout the process. Differently, at Tsen = 80 ◦ C, RH2 responded much faster as seen in Fig. 4(c) and (d). The duration of a H-loading process was thereby reduced to 5 min. It was found long enough for RH2 to be stabilized. We first inspected the result of sensor response derived straightly in accordance with the definition. The values based on the data of 20 and 80 ◦ C are denoted as S20 and S80 . They are plotted in Fig. 5(a) as functions of H2 concentration. In this range of H2 concentration, S80 is apparently larger than S20 . However, one noticed that at 20 ◦ C the film resistance varied slowly and may not have reached the genuine equilibrium in the hydrogenation processes. Fig. 5(b) further reflects that the response time determined at 20 ◦ C, tR20 , is much longer than that of 80 ◦ C, tR80 (<1 s). Based on these observations, we suggest that S20 obtained in this test is underestimated, and the real tR20 may be even longer, and not indicated by the data acquired in that test. This conjecture was further examined by observing the change of film resistance in three prolonged hydrogenation processes for more than 2 h performed at Tsen = 20, 80 and 140 ◦ C, respectively.

M. Zhao, C.W. Ong / Sensors and Actuators B 174 (2012) 65–73

67

Fig. 1. Schematic diagram of SCBD system.

Fig. 2. Schematic diagram of gas sensing measurement system. V: valve; MFC: mass flow controller; P: pressure sensor; RH: relative humidity sensor; A/D: analog digital converter; FET: field effect transistor; SSR: solid state relay.

20,000 ppm H2 in air was used in this test. Inter-digital electrodes were made on the film surface to lower the base resistance to facilitate the measurements. Fig. 6 shows that the film resistance, RH2 , has long enough time to approach equilibrium. With this experimental condition, the sensor response observed at 20 ◦ C, S20 = 1.8 × 105 is and higher than that of 80 ◦ C, S80 = 1.6 × 105 . It

further drops to a lower value of S140 = 2.3 × 104 when Tsen is increased to 140 ◦ C (Table 1). This trend is contradictory to that obtained in the previous experiment as indicated in Fig. 5(a), but should be more reliable. We believe that S80 and tR80 represent the sensor response and response time at 80 ◦ C, and are mainly used for discussion in this paper.

Table 1 Summary of nano-MOx H2 sensor operating at different Tsen . Tsen (◦ C)

S

Measurement condition

tR (s)

3–5

80

1.9 × 105

20,000 ppm in air

<1

Pt/WO3 Pd–WO3 Pt/InSnO2 Pt/InSnO2 Pt/InSnO2 Pt/ZnO ZnO ZnO SnO2 TiO2

35–65 20 6–7 3 – 50–150 30–150 100 80 300

20

0.89 2.5 × 104 1.1 × 103 2.0 × 105 1.0 × 105 0.10 0.08 0.52 0.65 1

Pure H2 1300 ppm in air 40,000 ppm in air 900 ppm in N2 900 ppm in N2 500 ppm H2 in N2 500 ppm H2 100 ppm H2 in air 20,000 H2 in N2 500 ppm H2

3000 1300 10,000 500 750 500 1–2 64 220 300

[11] [12] [13] [14] [15] [16] [17] [18] [19] [20]

Nanoclusters

Pt/TiO2 SnO2 SnO2 –SWCNT Au/WO3 Pt–WO3 –TiO2 Pd/WO3 Pt–WO3 /ZrO2 WO3 –MWCNT Ag–TiO2

∼20 4 15 – 55 33 0.5–0.9 42.3 22

500 350 250 262 200 300 300 300 360

90 140 1.5 370 20 391 48 1.2 3.6

500 ppm in N2 1000 ppm in air 1500 ppm 10,000 ppm in air 1000 ppm in air 2300 ppm 16,000 ppm in air 12,000 ppm in air 200 ppm in air

5–10 0.15 2–3 30–60 60 102 10–20 3 23

[25] [26] [27] [28] [29] [30] [31] [32] [33]

Nanobelt Nanotubes

Pd/SnO2 MgZnFe2 O4

50–80 20–25

200 350

3.4 10

13 ppm in vacuum 1660 ppm in air

120 60

[34] [35]

Geometry

Material

Nanoclusters

Pd/SCBD WO3

Nanoclusters

Nanorods Nanorods Nanowires Nanobelts Nanofibers

L (nm)

Ref.

68

M. Zhao, C.W. Ong / Sensors and Actuators B 174 (2012) 65–73

Fig. 3. (a) SEM and (b–d) TEM images of a Pd/SCBD WO3 film. Inset in (c) is a selected area electron diffraction pattern.

The main features observed in Fig. 6 are primarily interpreted as follows. When Tsen increases, more free charge carriers are generated to make the oxide more conductive. Consequently, both the base resistance Ro and hydrogenated resistance RH2 of the film drop. Furthermore, the populations of the surface sorbed oxygen species, mainly O2 molecules and O2 − ions below 150 ◦ C [4,6], would drop successively when operation temperature rises from 20 to 80 and then to 140 ◦ C. The depletion layer on the oxide surface becomes thinner and Ro drops. Correspondingly, the value of sensor response, relying on the fractional change of film resistance, is lowered. The drop of tR with Tsen is attributed to the faster dissociation rate, diffusion rate and reaction rate of the reactant gas molecules, and larger mobility of charge carriers in the oxide. We further derived the lowest detection limit of H2 concentration of a Pd/SCBD WO3 at this temperature. This limit should be even lower than the lowest H2 concentration used in all our tests, but must be derived semi-theoretically. Fig. 7 shows that S80 depends almost linearly on H2 concentration below 55 ppm, and can be described by a linear fit of S = 0.2 × H2 concentration. The estimated lowest detectable H2 concentration was suggested to be the one corresponding to an S80 which is at least three times above the noise level, Snoise [23,24]. Snoise was determined experimentally as Rnoise /Ro , where Rnoise was the standard deviation of 600 data points of base Ro measured in an H2 -free environment, and Ro  the mean. Results show that Snoise at 80 ◦ C was 0.002, so that the lowest detectable limit of H2 concentration of the film at this temperature was 3Snoise /0.2 = 0.03 ppm. This is lower than those of most reported nano-MOx based H2 sensor prototypes.

S80 and tR80 of a Pd/SCBD WO3 film are compared with those of other nano H2 sensors operating at room temperature [11–20] in Fig. 8(a) and (b). Data are also tabulated in Table 1. We notice that most nano-MOx based H2 sensors reported in literature can be classified into two groups. The first group include those containing small nanoclusters embedded in a rather dense matrix [11–15]. The second group includes those made of loosely packed onedimensional (1-D) elements [16–20], like wires, fibers, and belts. The thicknesses of these 1-D elements are referred to as the characteristic lengths L of the geometries, which are actually rather thick and fall in the range of 30–300 nm. The data of these two groups of sensors are presented by solid and hollow symbols in the figures. One sees that S80 of our Pd/SCBD WO3 film in the H2 concentration range concerned is much higher than those of Group 2 sensors composed of thick 1-D elements [16–20]. In addition, tR80 of the Pd/SCBD WO3 film is much shorter than those of Group 1 sensors having dense matrixes containing nanoclusters [11–15]. Only one sensor composed of loosely packed ZnO nanorods [17] can be able to respond a bit faster than the Pd/SCBD WO3 film. We further compare S80 and tR80 of a Pd/SCBD WO3 film with those of other nano H2 sensors operating at Tsen ≥ 200 ◦ C [25–35]. Data are plotted in Fig. 9(a) and (b), and are tabulated in Table 1. One sees that even operating at a lower Tsen , S80 of a Pd/SCBD WO3 film is considerably larger than those of others in the range of H2 concentration concerned. In addition, tR80 of the Pd/SCBD WO3 film is mostly shorter or comparable to those of others operating at much higher Tsen .

M. Zhao, C.W. Ong / Sensors and Actuators B 174 (2012) 65–73

69

Fig. 4. (a–d) Resistive response of a Pd/SCBD WO3 film with increasing H2 concentration measured at Tsen = 20 and 80 ◦ C, respectively.

Combining the insights gained from these comparisons, we assert that the genuine nanocluster-based and highly porous structure of an SCBD oxide film greatly facilitate the achievements of concomitant large sensor response and fast response rate in H2 detection. We further note that from application point of view, manufacturers of H2 fuel cell driven vehicles ask for H2 sensors having a tR < 1 s [3]. However, most existing H2 sensor products fail to meet this criterion. The only one H2 sensor capable of having a tR close to 1 s we know so far is an MOx -based sensor [26], but it needs to work at a much higher Tsen (250–350 ◦ C) and hence is readily interfered by many flammable reducing gases via combustion [2,3,26]. Results shows that tR80 of our Pd/SCBD WO3 film can be reduced from 218 to 3 s when H2 concentration in air increases from 2.5 and 1000 ppm. It further drops to below the shortest acquisition time of the measurement system (1 s) when H2 concentration exceeds 1700 ppm.

Fig. 5. (a and b) Plots of S20 , S80 , tR20 and tR80 of a Pd/SCBD WO3 film versus H2 concentration. Fig. 6. RH2 of a Pd/SCBD WO3 film versus time of measurement observed at 20, 80 and 140 ◦ C respectively. 20,000 ppm H2 in air was used for hydrogenation.

70

M. Zhao, C.W. Ong / Sensors and Actuators B 174 (2012) 65–73

Fig. 7. A plot of S80 of a Pd/SCBD WO3 film against H2 concentration in the range of 2.5–55 ppm. The red line is a linear fit to the data point. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Therefore, a Pd/SCBD WO3 film can satisfy the requirement for that application. 3.3. Cyclic stability Instability of the output of a chemical sensor is constantly a critical problem in use [7–10]. It can be caused by many reasons, especially when operating in higher Tsen , where oxygen vacancies could be relocated, or the concentration of vacancies is changed, or the composition is altered, or even the size of the grains grows. Instability of the performance of a gas sensor is manifested by

Fig. 9. (a and b) Comparison of S80 and tR80 of a Pd/SCBD WO3 film with those of other nano-MOx H2 sensors operating at Tsen ≥ 200 ◦ C. Solid and hollow symbols are data of nano-MOx H2 sensors made of nanoclusters and 1-D sensing materials respectively. Pt/TiO2 [25], SnO2 [26], SnO2 –CNT[27], Au/WO3 [28], Pt–WO3 –TiO2 [29], Pd/WO3 [30], Pt–WO3 /ZrO2 [31], WO3 [32], Ag–TiO2 [33], Pd/SnO2 [34] and MgZnFe2 O4 [35].

progressive drift of base resistance and/or hydrogenated resistance even when the concentration of the target gas in the environment is fixed, whereas the sensor response is not stabilized as a consequence. Different from S and tR , studies on stability of gas sensors are much more seldom reported [14,36–38], because durability tests are very time consuming and have no exiting standard of assessment to follow. We designed an experiment to investigate this aspect of our Pd/SCBD WO3 films in detecting H2 . The film was repeatedly exposed to 20,000 ppm H2 –air mixture for 1 min and synthetic air for 14 min over a large number of cycles. In this test, the film was shunted with a 20-G resistor in order to cap the maximum overall resistance. Fig. 10 shows the resistive responses of selected switching cycles observed at different stages of the cyclic tests performed at Tsen = 20 and 80 ◦ C respectively. Result shows that for both the tests, the overall resistance change is highly repeatable over 2400 switching cycles. One notices that the resistance change observed at 20 ◦ C is much smaller than that observed at 80 ◦ C, because H-loading time used in the former process is too short for the film to fully react with H2 . The excellent cyclic durability of the film sensor is attributed to the use of moderate Tsen in the tests, where post-annealing effect usually encountered by those operating at high Tsen does not occur. In particular, at 80 ◦ C the overall resistance dropped from 20 G to 4.75 M within 1 min during hydrogenation. This amplitude of drop does not deviate from the initial one by more than ±5% after 2400 switching cycles. Fig. 8. (a and b) Comparison of S80 and tR80 of a Pd/SCBD WO3 film with those of other nano-MOx H2 sensors at room temperature. Solid and hollow symbols are data of nano-MOx H2 sensors made of nanoclusters and 1-D sensing materials respectively. × Pd/SCBD WO3 ,  Pt/WO3 [11],  Pd-WO3 [12],  Pt/InSnO2 [13], 䊉 Pt/InSnO2 [14],  Pt/InSnO2 [15],  Pt/ZnO [16], ZnO [17],  ZnO [18], ♦ SnO2 [19] and 夽 TiO2 [20].

3.4. Selectivity Fig. 11 shows the resistive response of a Pd/SCBD WO3 at 20 and 80 ◦ C to 100,000 ppm (=10%) concentration of vapor of

M. Zhao, C.W. Ong / Sensors and Actuators B 174 (2012) 65–73

Fig. 10. Resistive response of the sample observed in cyclic tests performed at 20 and 80 ◦ C. 20,000 ppm H2 in air was used for hydrogenation.

various organic compounds (VOCs), including methanol, ethanol, iso-propanol (IPA), formaldehyde (FD) and acetone. The change of resistance is negligibly small compared with that associated with 20,000 ppm H2 in air at both 20 and 80 ◦ C. Conservatively speaking, the interference caused by these VOCs would not be more than 1.1 × 10−5 of the signal due to H2 . This is explained by considering that combustion of these VOCs did not occur in this moderate temperature range, while the Pd layer can still be able to dissociate H2 molecules for hydrogenating the oxide layer. Excellent selectivity of the Pd/SCBD WO3 film against these VOCs is thereby confirmed. 3.5. Dependence on relative humidity Fig. 12(a) and (b) shows the influence of relative humidity (RH) on the resistive response of a Pd/SCBD WO3 film at Tsen = 20 and 80 ◦ C, respectively. 20,000 ppm H2 in air was used in the hydrogenation processes. All the gases used in the test are dry and assumed to have an RH of 0%. If the gas passed through a water-containing chamber, RH was measured to be around 90%. In this test, the film was shunted with a 20-G resistor for the convenience of measurements. For the case of Tsen = 20 ◦ C, the resistance of the sample in humid air was 10 times lower than that in dry air. This is because a layer of moisture was present on the film surface to result in a drop of overall resistance. Evacuation of the measurement chamber also

Fig. 11. Selectivity of the sample at 20 and 80 ◦ C against vapors of several VOCs, including methanol, ethanol, iso-propanol (IPA), formaldehyde (FD), acetone, with each set at a concentration of 100,000 ppm in air. The resistive response to 20,000 ppm H2 in air at 20 and 80 ◦ C are shown for comparison.

71

removed the moisture inside, such that the measured resistance returned to 20 G promptly. Admission of humid H2 –air admixture followed, which caused the resistance to drop rapid first due to the influence of moisture; and the measured resistance dropped further gradually due to ensuing hydrogenation of the SCBD WO3 layer. At this stage, the measured resistance was still higher than that observed in a test using dry H2 –air gas. This is because many water molecules were still adsorbed on the film to block the film from direct contact with H2 molecules. From these results, one knows that a Pd/SCBD WO3 film is hardly used to determine H2 concentration accurately in a very humid environment. For the case of Tsen = 80 ◦ C, the trend of the film resistance observed in a hydrogenation process performed at RH = 90% was found to be the same as that observed at RH = 0%. Both the cases showed the same sensor response, and a very short tR below 1 s. We explain this finding by assuming that the moderate Tsen of 80 ◦ C is able to remove most physisorbed water molecules from the sensor’s surface, but this Tsen has not been high enough to initiate chemisorption of water molecules which may cause the resistance of a MOx to drop substantially [6]. It is also noticed that the recovery time of the resistance in the hydrogenation process is shortened markedly from 226 to 60 s with RH increased from 0 to 90%. The effect is explained by the phenomenon observed by other research groups [36,39,40], who claimed that the presence of moisture in the detected area can be adsorbed on the surface of WO3 to form a “water bridge”, which facilitates hydrogen species to hop through and hence their mobility on the oxide surface is prominently enhanced [39,40].

3.6. Pressure dependence One possible application of H2 sensors is for detecting leakage of the gas in H2 fuel cell driven vehicles. A vehicle may be required to travel in a vertical range between sea level and 4000 m. As such, a H2 sensor installed inside must be less sensitive to the corresponding span of ambient pressure, namely, ranging from one atmospheric pressure to 62 kPa [3]. Most existing H2 sensors do not meet this requirement [3]. For research type nano-MOx based H2 sensors, Shukla et al. reported a typical example indicating that the sensor response of nanocrystalline InSnO2 measured at 101 kPa is double of that measured at 26 kPa [41]. Referring to this background, we performed an experiment to investigate how S20 and S80 of a Pd/SCBD WO3 film depend on ambient pressure over a range of 30.7–101.2 kPa for simulating the change of condition that the sensor would encounter if it is used over a vertical range from sea level to an elevation of 9000 m. 20,000 ppm H2 in air was used in hydrogenation process. Each hydrogenation process was set to last for 5 min. A 20 G resistor was shunted to the film sample to facilitate resistance measurements. Fig. 13 shows how the resistance of the film measured at 20 and 80 ◦ C depends on ambient pressure. For both Tsen , the variation of the amplitude of resistive change is mild and below 9%. In particular, the resistive response measured at 20 ◦ C has not been stabilized within the pre-set time of the hydrogenation process, this condition still allows the sensor to give leakage alert if accurate determination of H2 concentration is not necessary. We further suggest that due to the highly porous structure, both hydrogen and oxygen molecules can diffuse in it agilely with similar rates, such that a pressure change in the environment would not cause significant variation in their molar ratio everywhere inside. This helps to diminish the ambient pressure dependence of the response. If this inference is correct, the ambient pressure dependence of a more porous nano-MOx based H2 sensor would be less significant.

72

M. Zhao, C.W. Ong / Sensors and Actuators B 174 (2012) 65–73

Fig. 12. (a and b) Resistive response of the sample in dry (RH = 0%) and humid (RH = 90%) environment measured at 20 and 80 ◦ C, respectively.

WO3 film is constructed of genuine 3–5-nm nanoclusters and has a very high porosity around 66%, such that it can give both larger sensor response and faster response rate simultaneously. Its overall performance is superior to most reported nano-MOx based H2 sensors operating at room or elevated temperatures ≥200 ◦ C. A Pd/SCBD WO3 film sensor operating at 80 ◦ C was also found to have high cyclic stability over 2400 cycles without any sign of degradation. It also showed strong interference resistance against vapors of a number of organic compounds. Importantly, the sensor operating at 80 ◦ C eliminates most influence from moisture which severely affects the sensing properties of the sensor at 20 ◦ C. Moreover, the recovery rate is much faster at 80 ◦ C. Ambient pressure dependence of the sensing performance is mild. More other advantages of the sensor include miniaturizibility; compatible with silicon-based microfabrication techniques, scalability in mass production; high batch-to-batch reproducibility based on the physical nature of an SCBD process.

Acknowledgments The work described in this paper is substantially supported by Research Grants Council of the Hong Kong Administrative Region (Project No. PolyU 5016/08P, account code: B-Q10N), a PolyU internally granted project (account code: 1-ZV94) and an Innovative Technology Fund project (Project No. ITS/558/09, PolyU Project No. ZP2U, account code: K.11.27.ZP2U). The technical support from Mr. M.H. Wong is greatly appreciated. Fig. 13. Ambient air pressure dependence of resistive response of the sample measured at 20 and 80 ◦ C, respectively.

4. Conclusions Tsen dependence of the H2 sensing properties of Pd/SCBD WO3 films was investigated. At low Tsen , sensitivity of detection is higher, but the response time is too long for the resistive response to reach equilibrium within a normal hydrogenation period. At a slightly higher Tsen of 80 ◦ C, sensitivity of detection only drops mildly, but the response is substantially reduced to be around 5 min. A Pd/SCBD WO3 hydrogen sensor operating at 80 ◦ C shows more other advantages and is recognized to be very useful in practice. Sensor response and response time measured at 80 ◦ C are compared with other nano-MOx based H2 sensors reported in literature. These sensors are either made of small nanoclusters embedded in dense matrixes, or are made of loosely packed 1-D nano-sized elements (e.g. nanowires, nanofibers and nanobelts). The former mostly has large sensor response but long response time due to prolonged diffusion process of gas molecules in the dense matrix; and the latter have faster response rate but the sensor response is lower because the 1-D elements are actually rather thick. A Pd/SCBD

References [1] L. Schlapbach, Hydrogen-fuelled vehicles, Nature 460 (2009) 809–811. [2] T. Hubert, L. Boon-Brett, G. Black, U. Banach, Hydrogen sensors—a review, Sensors and Actuators B: Chemical 157 (2011) 329–352. [3] L. Boon-Brett, J. Bousek, G. Black, P. Moretto, P. Castello, T. Hubert, U. Banach, Identifying performance gaps in hydrogen safety sensor technology for automotive and stationary applications, International Journal of Hydrogen Energy 35 (2010) 373–384. [4] M.E. Franke, T.J. Koplin, U. Simon, Metal and metal oxide nanoparticles in chemiresistors: does the nanoscale matter? Small 2 (2006) 36–50. [5] N. Yamazoe, New approaches for improving semiconductor gas sensors, Sensors and Actuators B: Chemical 5 (1991) 7–19. [6] N. Barsan, U. Weimar, Conduction model of metal oxide gas sensors, Journal of Electroceramics 7 (2001) 143–167. [7] G. Korotcenkov, B.K. Cho, Instability of metal oxide-based conductometric gas sensors and approaches to stability improvement, Sensors and Actuators B: Chemical 156 (2011) 527–538. [8] J. Jamnik, B. Kamp, R. Merkle, J. Maier, Space charge influenced oxygen incorporation in oxides: in how far does it contribute to the drift of Taguchi sensors? Solid State Ionics 150 (2002) 157–166. [9] B. Kamp, R. Merkle, J. Maier, Chemical diffusion of oxygen in tin dioxide, Sensors and Actuators B: Chemical 77 (2001) 534–542. [10] V.V. Sysoev, T. Schneider, J. Goschnick, I. Kiselev, W. Habicht, H. Hahn, E. Strelcov, A. Kolmakov, Percolating SnO2 nanowire network as a stable gas sensor: direct comparison of long-term performance versus SnO2 nanoparticle films, Sensors and Actuators B: Chemical 139 (2009) 699–703.

M. Zhao, C.W. Ong / Sensors and Actuators B 174 (2012) 65–73 [11] H. Shanak, H. Schmitt, J. Nowoczin, C. Ziebert, Effect of Pt-catalyst on gasochromic WO3 films: optical, electrical and AFM investigations, Solid State Ionics 171 (2004) 99–106. [12] S. Fardindoost, A.I. Zad, F. Rahimi, R. Ghasempour, Pd doped WO3 films prepared by sol–gel process for hydrogen sensing, International Journal of Hydrogen Energy 35 (2010) 854–860. [13] S. Shukla, S. Seal, L. Ludwig, C. Parish, Nanocrystalline indium oxide-doped tin oxide thin film as low temperature hydrogen sensor, Sensors and Actuators B: Chemical 97 (2004) 256–265. [14] S. Shukla, P. Zhang, H.J. Cho, L. Ludwig, S. Seal, Significance of electrode-spacing in hydrogen detection for tin oxide-based MEMS sensor, International Journal of Hydrogen Energy 33 (2008) 470–475. [15] S. Seal, S. Shukla, L. Ludwig, C. Cho (Inventors), Room temperature hydrogen sensor, University of Central Florida Research Foundation Inc., assignee, United States patent US 2010/0212403, August 26, 2010. [16] L.C. Tien, P.W. Sadik, D.P. Norton, L.F. Voss, S.J. Pearton, H.T. Wang, B.S. Kang, F. Ren, J. Jun, J. Lin, Hydrogen sensing at room temperature with Pt-coated ZnO thin films and nanorods, Applied Physics Letters 87 (2005) 222106. [17] J.S. Lin, D.P. Norton, S.J. Pearton (Inventors), System for hydrogen sensing, University of Florida Research Foundation Inc., assignee, World Patent WO 2008/041983, April 10, 2008. [18] O. Lupan, V.V. Ursaki, G. Chai, L. Chow, G.A. Emelchenko, I.M. Tiginyanu, A.N. Gruzintsev, A.N. Redkin, Selective hydrogen gas nanosensor using individual ZnO nanowire with fast response at room temperature, Sensors and Actuators B: Chemical 144 (2010) 56–66. [19] L.L. Fields, J.P. Zheng, Y. Cheng, P. Xiong, Room-temperature low-power hydrogen sensor based on a single tin dioxide nanobelt, Applied Physics Letters 88 (2006) 263102. [20] I.D. Kim, J.M. Hong, D.Y. Kim, S.M. Jo (Inventors), Ultrasensitive metal oxide gas sensor and fabrication method thereof, Korea Institute of Science and Technology, assignee, World Patent WO 2007/073111, June 28, 2007. [21] P. Piseri, A. Podesta, E. Barborini, P. Milani, Production and characterization of highly intense and collimated cluster beams by inertial focusing in supersonic expansions, Review of Scientific Instruments 72 (2001) 2261–2267. [22] M. Zhao, J.X. Huang, M.H. Wong, Y.M. Tang, C.W. Ong, Versatile computercontrolled system for characterization of gas sensing materials, Review of Scientific Instruments 82 (2011) 105001. [23] L.A. Currie, Nomenclature in evaluation of analytical methods including detection and quantification capabilities (IUPAC Recommendations 1995), Pure and Applied Chemistry 67 (1995) 1699–1723. [24] J. Li, Y.J. Lu, Q. Ye, M. Cinke, J. Han, M. Meyyappan, Carbon nanotube sensors for gas and organic vapor detection, Nanoletters 3 (2003) 929–933. [25] C. Lu, Z. Chen, High-temperature resistive hydrogen sensor based on thin nanoporous rutile TiO2 film on anodic aluminum oxide, Sensors and Actuators B: Chemical 140 (2009) 109–115. [26] T. Kida, T. Kuroiwa, M. Yuasa, K. Shimanoe, N. Yamazoe, Study on the response and recovery properties of semiconductor gas sensors using a high-speed gasswitching system, Sensors and Actuators B: Chemical 134 (2008) 928–933. [27] J.W. Gong, J.R. Sun, Q.F. Chen, Micromachined sol–gel carbon nanotube/SnO2 nanocomposite hydrogen sensor, Sensors and Actuators B: Chemical 130 (2008) 829–835. [28] S.J. Ippolito, S. Kandasamy, K. Kalantar-zadeh, W. Wlodarski, Hydrogen sensing characteristics of WO3 thin film conductometric sensors activated by Pt and Au catalysts, Sensors and Actuators B: Chemical 108 (2005) 154–158.

73

[29] G.N. Chaudhari, A.M. Bende, A.B. Bodade, S.S. Patil, V.S. Sapkal, Structural and gas sensing properties of nanocrystalline TiO2 :WO3 -based hydrogen sensors, Sensors and Actuators B: Chemical 115 (2006) 297–302. [30] E. Ghadiri, A. Iraji zad, F. Razi, Hydrogen sensing properties of pure and Pd activated WO3 nanostructured films, Synthesis and Reactivity in Inorganic and Metal 37 (2007) 453–456. [31] K. Shimizu, I. Chinzei, H. Nishiyama, S. Kakimoto, S. Sugaya, H. Yokoi, A. Satsuma, Hydrogen sensor based on WO3 subnano-clusters and Pt co-loaded on ZrO2 , Sensors and Actuators B: Chemical 134 (2008) 618–624. [32] R. Ghasempour, A.I. Zad, Hybrid multiwalled carbon nanotubes and trioxide tungsten nanoparticles for hydrogen gas sensing, Journal of Physics D: Applied Physics 42 (2009) 165105. [33] W. Zeng, T.M. Liu, Hydrogen sensing characteristics and mechanism of nanosize TiO2 dope with metallic ions, Physica B 405 (2010) 564–568. [34] Kolmakov, D.O. Klenov, Y. Lilach, S. Stemmer, M. Moskovits, Enhanced gas sensing by individual SnO2 nanowires and nanobelts functionalized with Pd catalyst particles, Nanoletters 5 (2005) 667–673. [35] K. Mukherjee, S.B. Majumder, Hydrogen sensing characteristics of wet chemical synthesized tailored Mg0.5 Zn0.5 Fe2 O4 nanostructures, Nanotechnology 21 (2010) 255504. [36] H. Nakagawa, N. Yamamoto, S. Okazaki, T. Chinzei, S. Asakura, A roomtemperature operated hydrogen leak sensor, Sensors and Actuators B: Chemical 93 (2003) 468–474. [37] T. Xu, M.P. Zach, Z.L. Xiao, D. Rosenmann, U. Welp, W.K. Kwok, G.W. Crabtree, Self-assembled monolayer-enhanced hydrogen sensing with ultrathin palladium films, Applied Physics Letters 86 (2005) 203104. [38] P. Offermans, H.D. Tong, C.J.M. van Rijn, P. Merken, S.H. Brongersma, M. CregoCalama, Ultralow-power hydrogen sensing with single palladium nanowires, Applied Physics Letters 94 (2009) 223110. [39] A. Georg, W. Graf, R. Neumann, V. Wittwer, The role of water in gasochromic WO3 films, Thin Solid Films 384 (2001) 269–275. [40] T.C. Arnoldussen, A model for electrochromic tungstic oxide microstructure and degradation, Journal of the Electrochemical Society 128 (1981) 117–123. [41] S. Shukla, P. Zhang, H.J. Cho, S. Seal, L. Ludwig, Room temperature hydrogen response kinetics of nano-micro-integrated doped tin oxide sensor, Sensors and Actuators B: Chemical 120 (2007) 573–583.

Biographies M. Zhao is currently a Ph.D. student of Dr. Ong’s group in the Department of Applied Physics, the Hong Kong Polytechnic University, Hong Kong, PR China. He received his B.S. in Material Physics in 2005 from the Jilin University, Changchun, PR China. He received his M.S. in 2009 from Changzhou University, Changzhou, PR China. Research interest: nanomaterials and MEMS techniques in gas sensing applications. C.W. Ong is currently an Associate Professor of the Department of Applied physics, PolyU, Hong Kong, PR China. In the period of 1981–1991, he studied in the Physics Department of the Chinese University of Hong Kong, and was awarded the degrees of B.S. (Hons), M.Phil. and Ph.D. in Physics. He then joined PolyU in 1990. His recent interests in research are thin film processing (both chemical and physical vapor deposition), nanoindentation and finite element analysis, MEMS and nano materials for gas sensing applications.