Effects of Niobium-Loading on Sulfur Dioxide Gas-Sensing Characteristics of Hydrothermally Prepared Tungsten Oxide Thick Film

Nb-loaded hexagonal WO 3 nanorods with 0–1.0 wt% loading levels were successfully synthesized by a simple hydrothermal and impregnation process and characterized for SO 2 sensing. Nb-loadedWO 3 sensing films were produced by spin coating on alumina substrate with interdigitated gold electrodes and annealed at 450C for 3 h in air. Structural characterization by X-ray diffraction, high-resolution transmission electron microscopy, and Brunauer-Emmett-Teller analysis showed that spherical, oval, and rod-like Nb nanoparticles with 5–15 nm mean diameter were uniformly dispersed on hexagonal WO 3 nanorods with 50–250 nm diameter and 100 nm–5μm length. It was found that the optimal Nb loading level of 0.5 wt% provides substantial enhancement of SO 2 response but the response became deteriorated at lower and higher loading levels. The 0.50wt% Nb-loadedWO 3 nanorod sensing film exhibits the best SO 2 sensing performances with a high sensor response of ∼10 and a short response time of ∼6 seconds to 500 ppm of SO 2 at a relatively low optimal operating temperature of 250C. Therefore, Nb loading is an effective mean to improve the SO 2 gas-sensing performances of hydrothermally prepared WO 3 nanorods.


Introduction
Sulfur dioxide (SO 2 ) is a colorless toxic gas with burning smell that causes various respiratory and cardiovascular diseases as well as environmentally hazardous acid rain [1].There has been increasing demand for on-site monitoring of SO 2 due to rising level of SO 2 pollutant produced by various industrial processes.Presently, the concentration of SO 2 is still primarily determined by conventional techniques including gas chromatography or electrochemical detections, which are expensive, cumbersome, and impractical for onsite applications.Metal oxide semiconductor gas sensors are potential candidates for portable gas-sensing devices due to high sensitivity, good stability, low cost, small size, and simple electronic interface [1][2][3][4][5][6][7][8][9][10].However, only few metal oxide materials including WO 3 , SnO, and ZnO give significant response to SO 2 and the reported response values are still limited [1].Among these, WO 3 exhibits relatively stable and good response to SO 2 [1,[9][10][11][12][13].The gas-sensing properties towards SO 2 of WO 3 -based gas sensors prepared by various methods are summarized in Table 1.
Firstly, WO 3 thick film fabricated by pyrolysis/paste casting gave an optimal response of ∼12 to 800 ppm SO 2 at 400 ∘ C [9].In addition, WO 3 sensor produced by drop casting exhibited a response of ∼1.3 to 25 ppm SO 2 at a lower optimal working temperature of 200 ∘ C [10].Similarly, WO 3 thick film sensor deposited by electrostatic spray method displayed a response of ∼3 to 20 ppm SO 2 at 350 ∘ C [12].Moreover, WO 3 sensors made by hydrothermal and screen-printing processes showed a fair response of <2 to low SO 2 concentrations of 1-10 ppm at 260 ∘ C [13].It can be seen that undoped WO 3 sensors tend to suffer from limited SO 2 response or high optimal working temperature.Thus, incorporations of various metallic additives including Au, Ag, Cu, Pt, Pd, and Rh have been studied to improve the SO 2 gas-sensing properties of WO 3 sensors [9,11].For instance, the addition of 1.00 wt% Ag to pyrolyzed WO 3 thick film sensors led to the highest response of ∼20 to 800 ppm SO 2 at 450 ∘ C [9].In another study, Pt-loaded rf-sputtered WO 3 thin film showed a good response of ∼6 to 1 ppm SO 2 diluted in CO 2 at 200 ∘ C with the absence of oxygen while the response to H 2 S was very low (<1), dictating good SO 2 selectivity [11].Thus, studies of additive incorporation in WO 3 for SO 2 sensing are still limited and more effective metal additives should be explored to achieve better SO 2 gas-sensing performances.
Niobium is a noble metal catalyst that is found to be useful for gas sensing towards particular gases such as NO 2 and CO [14,15].However, there is no report of its addition to WO 3 support for SO 2 gas sensing.In this work, Nb nanoparticles are impregnated on WO 3 nanorods synthesized by hydrothermal synthesis and the effect of Nb loading concentration on SO 2 gas-sensing performances is studied as a function of the operating temperature and gas concentration.

Material Characterization.
The phase structure of unloaded WO 3 and Nb-loaded WO 3 nanorods was studied by means of X-ray diffraction (XRD; TTRAX III diffractometer, Rigaku).The morphology of nanorods was examined by high-resolution transmission electron microscopy (HRTEM; JEM-2010, JEOL).The presence of Nb element was confirmed by an energy dispersive X-ray spectroscopy (EDX).The specific surface areas (SSA BET ) and pore sizes of nanorods were determined from Brunauer-Emmett-Teller (BET) and Barrett-Joyner-Halenda (BJH) analyses of nitrogen adsorption measurements.

Sensing Film Fabrication and Characterization.
To form paste for spin coating of sensing films, 60 mg of unloaded WO 3 or 0.25-1.00wt% Nb-loaded WO 3 nanopowder was thoroughly mixed with an organic paste composed of ethyl cellulose and terpineol (0.25 mL), which acted as a vehicle binder and solvent, respectively.The resulting paste was spincoated on alumina substrates equipped with interdigitated gold electrodes and then annealed at 450 ∘ C for 2 h with a heating rate of 2 ∘ C/min for binder removal.After annealing and sensing test at 350 ∘ C in dry air, the morphologies, cross section, and elemental compositions of sensing films were analyzed by glancing-incident XRD (GIXRD; TTRAX III diffractometer, Rigaku), scanning electron microscopy (SEM; JEOL JSM-6335F), and EDX spectroscopy.

Gas-Sensing Measurement.
For gas-sensing measurements, unloaded and Nb-loaded WO 3 sensors were heated by the external NiCr heater to the operating temperatures ranging from 200 to 350 ∘ C in dry air before exposure to target gases in a stainless steel chamber (the setup is reported in our previous work [14]).The target gas source (1000 ppm SO 2 balanced in dry air) was flowed to mix with dry air at different flow rates to attain desired concentrations using multichannel mass flow controllers (Brook Instrument).The resistances of various sensors were continuously monitored with a computer-controlled system by voltage-amperometric technique with 10 V dc bias and current measurement through a 6487 Keithley picoammeter.The gas-sensing properties of unloaded and Nb-loaded WO 3 sensors are characterized in terms of response and response time as a function of gas concentration and operating temperature.The gas-sensing response () is given by  =   /  , where   and   are the electrical resistances of the sensor measured in the presence of pure dry air and reducing gas, respectively.The response time ( res ) is the time required to reach 90% of the response signal while the recovery time ( rec ) is the time needed to recover 90% of the baseline signal.

Results and Discussion
3.1.Particle and Sensing Film Characterization.The crystal structures of unloaded and 0.25-1.00wt% Nb-loaded WO 3 nanorods and sensing films after annealing and sensing test were studied by XRD using CuK  radiation at 2 = 10-80 ∘ with a step size of 0.06 ∘ and a scanning speed of 0.72 ∘ /min as presented in Figure 1.It is seen that unloaded and Nb-loaded WO 3 nanorods prepared by hydrothermal and impregnation methods exhibit sharp XRD peaks whose locations are well matched to JCPDS 85-2459 [13], indicating polycrystalline structure of hexagonal WO 3 phase with high crystallinity.
In addition, Nb diffraction peaks cannot be observed in all samples since the amount and size of Nb nanoparticles are below the detection limit of XRD instrument.Thus, the presence of Nb nanoparticles in the composite will be confirmed by HRTEM and EDS analysis.For sensing films coated on Au/Al 2 O 3 substrates, the XRD patterns affirm the presence of hexagonal WO 3 phase whose diffraction peaks are weaker than those of (e) Au (JCPDS file number 04-0784 [16]) and (△) Al 2 O 3 (JCPDS file number 82-1468 [17]) from the substrate.
HRTEM bright-field images and EDX spectra of 0.50 wt% Nb-loaded WO 3 nanorods synthesized by hydrothermal/ impregnation are shown in Figure 2. The images show longitudinal sides of solid nanorods displaying rectangular shapes with length varying from 100 nm to 5 m and width (the diameter of nanorod) ranging from 50 to 250 nm.In addition, the bigger nanorod surfaces are uniformly decorated with smaller spherical, oval, and rod-like nanoparticles.The mean diameter of particles is estimated to be in the range of 5-15 nm for 0.50 wt% Nb-loaded WO 3 nanorods.The EDX spectra at the two regions (Figure 2) confirm that the nanorods and nanoparticles are made of WO 3 and Nb, respectively.
The nitrogen adsorption isotherm and BJH adsorption pore size distribution (inset) of WO 3 nanorods with 0-1 wt% Nb loading levels are shown in Figures 3(a)-3(d), respectively.It can be seen that all the isotherms exhibit the IUPAC type VI pattern, indicating the existence of stepwise multilayer adsorption on nonuniform surface of nonporous adsorbent or micropores filled with multilayer adsorbates [18,19].In addition, adsorbed volume tends to increase with increasing Nb loading level.The pore size distributions are in multimodal forms with 2-4 maxima depending on Nb loading.The distributions of unloaded WO 3 nanorods display two maxima at ∼19 and ∼36 Å while those of Nb-loaded ones exhibit three significant maxima at 18-20, 26-36, and ∼44-52 Å, respectively.The results indicate the surfaces are porous and contain micropores as well as mesopores structures.In addition, the contribution of larger pores tends to increase with increasing Nb loading levels.The minor maxima and minima are expected to be artifacts induced by the modeling technique.Figure 3(e) displays the corresponding pore volume and BET specific surface areas (SSA BET ) of WO 3 nanorods.As the Nb concentration increases from 0 to 1.00 wt%, pore volume and SSA BET monotonically increase from 0.16 to 0.23 cm 3 /g and from 67.72 to 101.8 m 2 /g, respectively.The increased pore volume and specific surface area may be attributed to wider pore distribution due to interaction between Nb nanoparticles and WO 3 nanorods as well as increasing contribution of smaller Nb nanoparticles with increasing Nb loading level.
The typical cross-sectional SEM micrograph of 0.50 wt% Nb-loaded WO 3 film layer on an Al 2 O 3 substrate equipped with interdigitated Au electrodes after the sensing test is shown in Figure 4.It is seen that the sensing film with an average thickness of ∼10 m contains loosely agglomerated nanorods.The detailed top surface morphology of nanorods is also shown in the inset.It shows that nanorods up to several microns long are slackly entangled with each other leaving many large and small pores in the film.The observed morphology indicates that the final sensing film has a large specific surface area, which should be highly beneficial for gas-sensing applications.SO 2 , indicating n-type semiconducting behavior towards oxidizing gas.The result is in agreement with previous reports on SO 2 gas-sensing studies of WO 3 sensors [9,10].When the WO 3 nanorods are exposed to SO 2 gas, SO 2 gas will adsorb on WO 3 surface at different sites from the existing chemisorbed O 2 − , O − , and O 2− ions and extract additional electrons from conduction band of WO 3 to become SO 2 − according to (1) [10].As a result, the concentration of electrons on the surface of WO 3 nanorods decreases and the resistance of WO 3 layer increases.Consider

Gas-Sensing
Figures 6(a) and 6(b) show the response and response time of WO 3 nanorods with different Nb loading levels versus SO 2 concentration in the range of 20-500 ppm at the operating temperature of 250 ∘ C. It can be seen that the gassensing behaviors of WO 3 nanorods considerably depend on Nb loading level.As the Nb loading level increases from 0 to 0.25 wt%, the SO 2 response slightly decreases but then becomes significantly higher as the loading concentration increases further to 0.5 wt%.Conversely, the response time monotonically reduces as the Nb loading concentration increases from 0 to 0.5 wt%.The sensing film with 0.50 wt% Nb-loaded WO 3 nanorods exhibits the best SO 2 sensing performances with a high sensor response of ∼10 and a short response time of ∼6 seconds to 500 ppm of SO 2 at 250 ∘ C. Thus, the moderate Nb loading level of 0.5 wt% could substantially improve the response towards SO 2 .However, the response and response time are considerably degraded when the Nb loading concentration further increases to 1.0 wt%.Regarding the baseline recovery, unloaded and Nbloaded WO 3 nanorods have similarly long recovery times on the order of several minutes, which are not practical for gas-sensing applications.The recovery time is dictated by the sensor's inherent oxygen readsorption rate as well as gas flow dynamic.In this case, the slow gas flow dynamic due to large volume of gas-testing chamber (∼3 liters) was found to be the main cause of slow recovery and the problem may be solved by miniaturization of sensors and test system.From the results, the optimum Nb loading concentration of WO 3 nanorods for SO 2 sensing is 0.5 wt%.A possible explanation for the enhanced SO 2 response due to moderate Nb doping is that Nb nanoparticles may provide catalytic effect to enhance SO 2 − adsorption on WO 3 surface and this mechanism will be effective when there is a sufficient amount of Nb nanoparticles well dispersed on WO 3 nanorods so that electron transfer due to SO 2 adsorption can dominate the resistance control at most contacts.Moreover, Nb loading results in increased pore volume and specific surface area for gas adsorption.When the Nb loading level (i.e., 0.25 wt%) is too low, the enhancement effect is very low and the observed small response decrease may be due to the slight reduction in porosity and thickness of the final sensing film.At too high Nb loading concentration, Nb nanoparticles may become agglomerated into larger particles so that the catalytic mechanism is less effective, leading to deteriorated SO 2 response.In addition, the proposed catalytic effect may be supported by the observed significant decrease of response time with Nb loading since this effect should result in higher adsorption rate and more rapid change in resistance upon SO 2 exposure.
Figure 7 shows the effect of operating temperature on response to 500 ppm SO 2 of WO 3 nanorods with different Nb loading concentrations.It is clear that all sensors exhibit similar temperature dependence of response with optimal operating temperature of 250 ∘ C and much lower response at lower (200 ∘ C) and higher (300-350 ∘ C) temperatures.At low temperature, SO 2 − adsorption rate and SO 2 response are low due to low thermal energy.In the case of high temperature, the SO 2 − adsorption rate and response become low despite the increase of thermal energy because there is less available site for SO 2 − adsorption due to the increased concentration of preadsorbed oxygen species.The attained optimal SO 2 response (∼10 to 500 ppm SO 2 ) and operating temperature (250 ∘ C) are relatively advantageous compared with some previously reported WO 3 sensors as listed in Table 1 that showed lower response at similar operating temperature or exhibited higher response at low concentration but required higher optimal operating temperature of 350-450 ∘ C [8][9][10][11][12].However, the performances of Nb-loaded WO 3 sensor are inferior to those of Pt/WO 3 sensor that can achieve both high response and low operating temperature but the sensor requires much more expensive Pt catalyst [11].Therefore, Nbloaded WO 3 nanorods are a promising alternative for SO 2 sensing.

Conclusions
Unloaded and 0.25-1.0wt% Nb-loaded WO 3 nanorods were successfully synthesized by hydrothermal and impregnation  methods.From structural characterizations, spherical, oval, and rod-like Nb nanoparticles with 5-15 nm mean diameter were uniformly dispersed on hexagonal WO 3 nanorods with 50-250 nm diameter and 100 nm-5 m length.From gassensing measurement, Nb loading with the moderate level of 0.5 wt% led to substantial enhancement of SO 2 response but the response became deteriorated at lower and higher loading levels.The 0.50 wt% Nb-loaded WO 3 nanorod sensing film exhibited the best SO 2 sensing performances with a high sensor response of ∼10 and a short response time of ∼ 6 seconds to 500 ppm of SO 2 at a relatively low optimal operating temperature of 250 ∘ C. The enhanced SO 2 sensing performances may be attributed to catalytic effect of well dispersed Nb nanoparticles on WO 3 nanorods.Therefore, Nb loading is an effective method to enhance the SO 2 gas-sensing performances of hydrothermally prepared WO 3 nanorods.
Properties.The changes of resistance of WO 3 nanorods with different Nb loading concentrations exposed to SO 2 pulses at various concentrations (500-20 ppm) are shown in Figure5.It is evident that the resistance of all WO 3 sensors rapidly increases upon exposure to

Figure 7 :
Figure 7: The response of 0.50 wt% Nb-loaded WO 3 -based gas sensor towards 500 ppm SO 2 versus operating temperature.

Table 1 :
A summary of the gas-sensing properties of WO 3 -based gas sensors towards SO 2 .