Kwon-Il Choi,
Su-Jin Hwang,
Zhengfei Dai,
Yun Chan Kang and
Jong-Heun Lee*
Department of Materials Science and Engineering, Korea University, Seoul 136-713, Republic of Korea. E-mail: jongheun@korea.ac.kr; Fax: +82 2 928 3584; Tel: +82 2 3290 3282
First published on 23rd September 2014
The sensing of volatile organic compounds is crucial in a variety of fields including disease diagnosis, food, and homeland security. However, the significant deterioration of gas response by water vapors often hinders the sensitive and reliable gas detection in a highly humid atmosphere. Herein, we report an Rh-loaded WO3 hollow sphere chemiresistive sensor that can be potentially used for acetone gas analysis in a highly humid atmosphere. Pure WO3 and Rh-loaded WO3 hollow spheres are synthesized via a spray pyrolysis method. The Rh-loaded WO3 sensor achieved a fast acetone response (2 s), high sensitivity, good linearity, high stability, low detection limit (40 ppb) and strong selectivity to acetone even under a highly humid (80% RH) atmosphere, compared with the unloaded WO3 sensor. Interestingly, an abnormal phenomenon occurs only with the n-type Rh-loaded WO3 sensor, where the resistance and gas response increases in humid atmospheres. The sensing mechanism by Rh loading is also addressed. The unusual improvement of gas response, selectivity, responding kinetics by Rh loading shows a good potential for the detection of acetone gas.
Nowadays, numerous researches have been focused on highly active nanomaterials, including nanoparticles,4 one-dimensional nanostructures,5 nanosheets6 and hollow spheres,7 as well as their building blocks,8 to develop VOC gas sensors with enhanced performance. It has been also identified that hollow and porous nanoarchitectures are more promising for large surface area to volume ratio, high gas accessibility and rapid gas responding kinetics.9 With such porous nanostructured MOS sensors, far-ranging gases and vapors can be efficiently detected and monitored with powerful sensitivity.10 However, the deterioration of sensing properties by water vapors in real-atmospheres may cause obstacles for the reliability of MOS sensors, which is an acute issue for MOS sensors. If the humidity-dependent problem could be adequately rectified, MOS hollow spheres sensors would be suitable to serve in various application areas because they possess advantages in terms of sensitivity, rapid response and miniaturization capabilities.7
Gas sensing films have been doped or surface-functionalized to achieve gas selectivity.11 To date, numerous experimental and theoretical works are reported to detect acetone gas using modified MOS films, such as Pt–In2O3,12 Pt–WO3,13 Cr–WO3,14 Si–WO3,15–17 graphene modified ZnFe2O4,18 and so forth.19 Some of them have demonstrated superior responses to acetone below 1 ppm in highly humid atmospheres (relatively humidity (RH) ≥ 80%) for the application of breath acetone monitoring.16,17,20 Nevertheless, their gas sensing behaviors at varying humidity from RH 20% to RH 80% are rarely involved. In general, the introduction of water vapor into the atmosphere deteriorates gas response, response/recovery speed, and selectivity to the reducing gas in n-type oxide semiconductor sensors,21 which hinders the reliable and rapid detection of the trace concentration of analyte gases. It should be noted that the enhancement of gas response in highly humid atmosphere, which is highly advantageous to detect VOC gases in real atmosphere, has been scarcely reported, although there was a report on the enhancement of CO response by increasing humidity in a Pd-loaded SnO2 sensor.22 The humidity-independent characteristics of gas sensors show a possibility of exhaled breath analysis because exhaled breath contains >80% humidity at 25 °C. Therefore, to materialize superior sensors, it requires not only a production of a preferable sensing material with VOCs selectivity and fast response, but also to make progressive investigations on the control of sensing properties in highly humid atmospheres.
In this paper, we report an Rh-loaded WO3 hollow spheres (HWs) acetone sensor that can be potentially used for real-time VOCs analysis. This sensor achieves fast response (2 s), high sensitivity, low detection limit (40 ppb) and strong selectivity to trace acetone gas even under a highly humid (80% RH) atmosphere, comparing with the unloaded one. It is found that the n-type Rh-loaded WO3 sensor exhibits an abnormal phenomenon, as the resistance and gas response increase in humid atmospheres. This unusual improvement can provide a good potential in the real-time acetone gas analysis for the application in diabetic diagnosis.
With the aim to improve the exhaled breath sensing characteristics, nanoscale rhodium oxide is functionalized on the surface of the WO3 HWs by a simple slurry mixing and after-heating process. Fig. 1f represents the TEM image of the Rh-loaded WO3 HWs with a designed 0.5 at% Rh concentration ([Rh]/[W]), showing a similar morphology, size distribution and shell thickness with prototype sample (Fig. 1c). The precise content of rhodium in Rh–loaded WO3 hollow spheres is also confirmed as 0.45 at% by inductive coupled plasma emission spectrometer (ICP) analysis. Further, elemental mapping images (Fig. 1g) suggest that Rh was uniformly dispersed on the surface of the WO3 hollow spheres.
Further, the X-ray photoelectron spectroscopy (XPS) measurements were carried out to detect Rh components and determine their chemical state (Fig. S2†). It demonstrates that the survey scanning energy spectra (Fig. S2a†) for both the samples are almost the same. The weak peaks located at 308.5 and 313.8 eV observed in Rh-loaded WO3 are attributed to Rh 3d5/2 and Rh 3d3/2 in Rh3+, respectively, rather than the metal phase Rh0 (Fig. S2b†).25 The binding energies of W 4f7/2 and 4f5/2 are 35.3 and 37.4 eV (Fig. S2c†) in the pure sample, respectively, indicating a 6+ state of tungsten. However, the binding energies of W 4f7/2 and 4f5/2 occur with a slight red chemical shift of 0.1 eV probably due to the increasing oxygen vacancy because of Rh loading, which is similar to a previously reported Au-loaded WO3 sample.26 Moreover, the O 1s peak at 530.2 eV of both samples is typically ascribed to W–O in WO3 (Fig. S2d†). Consequently, rhodium is obtained as Rh2O3 rather than its incorporation in the lattice of WO3, and the increased oxygen vacancy by Rh loading may bring an enhancement of gas sensing performance by promoting oxygen adsorption.27
The acetone sensing properties of the pure sensor in which Ra/Rg and Ra are decreased in wet ambient (20–80% RH) (Fig. 2a-2, a-3, a-4). This is a general behavior of n-type MOS gas sensor. For confirmation, pure SnO2, ZnO and In2O3 hollow spheres were prepared via spray pyrolysis and their gas sensing characteristics were measured (Fig. S3†). Indeed, all the pure sensors showed the significant decreases of Ra/Rg and Ra with increasing ambient humidity, which is a normal behavior for many MOS sensors.21 Righettoni et al.16 also reported the deterioration of the acetone sensing properties of Si-doped WO3 in highly humid atmospheres. In a stark contrast, an abnormal phenomenon occurs on Rh-loaded WO3 sensor, where the Ra/Rg and Ra increased in humid atmospheres (Fig. 2b-2, b-3, b-4). The sensing response at 80% RH increases from 5.70 to 20.68 after Rh loading. The τres of the pure WO3 sensor gradually increased from 6 s to 11 s with increasing humidity, while the τres of Rh-loaded sensor remains at 2 s throughout the humidity range. The tendency was confirmed again by the continuous operation of the sensor with increasing relative humidity (Fig. S4†). The WO3 HWs loaded with other noble metal catalysts (0.5 at% Pd or 0.5 at% Pt) showed a significant deterioration of acetone sensing characteristics (Fig. S5†), confirming that the role of the Rh catalyst to promote the gas sensing reaction under humid atmospheres is unique and unusual.
Fig. 3 presents an overview of 10 ppm acetone-sensing properties, such as Ra/Rg, Ra, τres and τrecov, as functions of the working temperature and relative humidity. The Ra/Rg values of two sensors progressively increase with increasing temperature in dry and humid atmospheres (Fig. 3a). Moreover, the Ra/Rg values of the Rh-loaded sensor are ∼2 times higher than those of the pure sensor at all temperatures in the dry atmosphere. However, the two sensors displayed different Ra/Rg vs. humidity tendencies with varying sensor temperatures. Below 325 °C, the Ra/Rg values of both sensors decrease with increasing relative humidity. Beyond 350 °C, the Rh-loaded sensor shows an inverse behavior of the increase of Ra/Rg with the relative humidity, while the Ra/Rg of the unloaded WO3 sensor still decreases with the increasing humidity. Fig. 3b shows the dependence of Ra on the working temperature at different ambient humidity. The resistances of two sensors in dry air are similar in the whole temperature range, indicating little influence on the resistances by Rh loading. In humid atmosphere, the Ra of pure WO3 sensor is decreased by water vapors, while the Ra of the Rh-loaded sensor increases (marked by arrows in Fig. 2) in all the temperature range (Fig. 3b). In Fig. 3b, it can be observed that the resistance of the pure WO3 sensor significantly increases by 27% when the operation temperature is declined from 400 °C to 300 °C in 80% RH condition. However, the increased rate of resistance of Rh-loaded WO3 sensor is only 3% when the temperature decreases from 400 °C to 300 °C (Fig. 3b), exhibiting an anti-jamming and stable characteristic. The abnormal tendencies of Ra/Rg and Ra reflect that Rh loading can probably introduce a change of humidity cross sensing mechanism at different temperatures (see the next section).
Further, we also observe the response and recovery rate of the two sensors, as shown in Fig. 3c and d, respectively. From dry to humid atmosphere, it is found that the response rate of the pure WO3 sensor becomes 1.5–2 times more sluggish, while τres values of the Rh-loaded sensor are almost invariable, in particular above 350 °C (see Fig. 3c). It should be noted that the response speeds of the Rh-loaded sensor are 2–6 and 3–12 times faster than those of the pure sensor under dry and humid condition, respectively (Fig. 3c). In addition, both the sensors have faster recovery speed under humid conditions compared to dry atmosphere, moreover the τrecov of the Rh-loaded sensor is shorter than that of the pure one in 300–400 °C range (Fig. 3d). Because all the gas sensing characteristics (Ra/Rg, τres and τrecov) are significantly enhanced by Rh loading, this Rh-loaded WO3 HWs sensor is applicable as disease diagnosis tool in an effective manner.
Fig. 4 presents the gas response of the two sensors to 0.2–20 ppm acetone in different humidities (0–80% RH) at 400 °C, indicating a good linear relation between Ra/Rg and the gas concentration. Both the sensors have stable response and recovery performances under various relative humidity conditions (Fig. S6 and S7†). The lowest acetone detection limit of the Rh-loaded sensor is calculated to be ca. 50 ppb and 40 ppb in dry air and 80% RH ambient when Ra/Rg > 1.2 was used as the criterion for gas detection, respectively, which are 4 times lower than those of the unloaded WO3 sensor.
![]() | ||
Fig. 4 Gas responses of pure and Rh-loaded WO3 sensors as functions of acetone gas concentration at 400 °C in different humidity (0–80% RH) atmospheres. |
The trace acetone detection limit, fast response and negligible temperature–resistance dependence in a highly humid atmosphere of the Rh-loaded WO3 sensor can be applied for diabetes breath sensors. The key qualification for exhaled breath sensors is the selectivity of the target gas because of various interfering VOCs gases in human exhalation, such as CH3COCH3 (1.8 ppm for diabetes),28 H2S (∼0.5 ppm for halitosis),29 CO (8.6 ppm for bronchiectasis),30 NH3 (14.7 ppm for renal failure),31 benzene, toluene, xylene (10 ppb for lung cancer),32 and NO (100 ppb for asthma),33 although the precise criteria for medical diagnosis are still under investigation. Fig. 5 summarizes the sensing response for pure and Rh-loaded WO3 sensors to these gases at 400 °C in 80% RH, respectively. The concentrations of these gases are as follows: 4 ppm CH3COCH3, 20 ppm CO, 20 ppm NH3, 1 ppm H2S, 20 ppb C6H6 (benzene), 20 ppb toluene (CH3C6H5), 20 ppb xylene ((CH3)2C6H4), and 200 ppb NO. It should be noted that the responses toward ∼2 times higher concentration of biomarker gases than the detection limit for the diagnosis of diseases were measured (Table S1†) to more accurately examine the selectivity for CH3COCH3 and possible interferences from other gases. The response of the pure sensor to 4 ppm CH3COCH3 is 3.07 (see Fig. 5a), which is lower than those for 20 ppm NH3 (Renal failure, Ra/Rg = 3.46) and 1 ppm H2S (halitosis, Ra/Rg = 4.30). Therefore, if a patient is suffering from renal failure or halitosis, accurate diagnosis of diabetes is impossible. In contrast, the Rh-loaded sensor can overcome the above trouble as its Ra/Rg to 4 ppm CH3COCH3 (13.1) is 4 times higher than the prototype one (see Fig. 5b). This merit of the Rh-loaded sensor can possibly make an accurate identification of a diabetes patient even from patients suffering with either halitosis (Ra/Rg = 7.27) or renal failure (Ra/Rg = 3.46).
H2O(g) + O−(W) → 2OH(W) + e− | (1) |
Rh (111) has strong affinity for water compared with other metal surfaces,35 and gets easily combined with hydroxyl group and hydrogen that are obtained by the decomposition of water.36 This suppresses the direct reaction between the sensing surface (WO3 in the present study) and water, which prevents the deterioration of gas sensing characteristics. Moreover, the decomposition of water by rhodium is accelerated because the co-adsorbed oxygen is cut down to nearly half of activation energy.35 Therefore, more adsorbed oxygen can be formed through synergy between rhodium and surface oxygen adsorbed on metal oxide at high temperature, as depicted in eqn (2).
H2O(Rh) + O−(W) → OH(Rh) + OH(W) + e− → 2H(Rh) + 2O−(W) + h+ | (2) |
Because oxygen decomposed from water is adsorbed on the sensor (WO3) surface and gets ionized, the depletion region on the surface of the sensor will be expanded, which leads to huge changes in resistance. This mechanism can well explain why the Rh-loaded sensor has higher Ra/Rg and Ra in humid atmospheres. This abnormal phenomenon suggests a new direction for the real-time developing of self-diagnosis sensor for diseases, using the direct analysis of highly humid exhaled breath.
If the hydroxyl group does not sufficiently decompose into hydrogen and oxygen due to the energy shortage (low temperature), Ra/Rg is reduced because of the competition between the target gas and water vapors reacting with surface adsorbed oxygen (O−W). It results in the differences of Ra/Rg changes at different sensing temperature (Fig. 3a). At a low temperature like 300 °C, a decrease of Ra/Rg is induced by the adsorption of hydroxyl groups formed by water vapor (eqn (1)). On the contrary, Ra/Rg is gradually raised due to the increasing adsorbed oxygen formed by the decomposition of hydroxyl group and water above 350 °C (eqn (2)).
In addition, rhodium is known to play a role of securing the selective detection of acetone gas among other VOCs,37 which is consistent with the present results. Therefore, the Rh-loaded WO3 hollow spheres possesses a high response (Ra/Rg = 20.68 to 10 ppm CH3COCH3), rapid response time (τres = 2 s), excellent acetone selectivity and negligible temperature dependence of sensor resistance in a highly humid atmosphere (RH 80%), and low detection limit (40 ppb) at 400 °C which grants a rapid and reliable detection of trace acetone (Table 1).
Pure WO3 | Rh-loaded WO3 | |||
---|---|---|---|---|
Dry | RH 80% | Dry | RH 80% | |
Ra/Rg | 6.97 | 4.57 | 13.34 | 20.68 |
τres (s) | 6 | 11 | 2 | 2 |
Ra (MΩ) | 14.91 | 11.43 | 14.06 | 29.53 |
Detection limit (ppb) | 167 | 201 | 52 | 40 |
Footnote |
† Electronic supplementary information (ESI) available: X-ray diffraction patterns, X-ray photoelectron spectroscopy, gas sensing behaviors of pure and Rh-loaded WO3 hollow spheres and other reference materials. See DOI: 10.1039/c4ra06654e |
This journal is © The Royal Society of Chemistry 2014 |