Designing SnO2 Nanostructure-Based Sensors with Tailored Selectivity toward Propanol and Ethanol Vapors.

The application of metal oxide-based sensors for the detection of volatile organic compounds is restricted because of their high operating temperatures and poor gas sensing selectivity. Driven by this fact, we report the low operating temperature and high performance of C3H7OH and C2H5OH sensors. The sensors comprising SnO2 hollow spheres, nanoparticles, nanorods, and fishbones with tunable morphologies were synthesized with a simple hydrothermal one-pot method. The SnO2 hollow spheres demonstrated the highest sensing response (resistance ratio of 20) toward C3H7OH at low operating temperatures (75 °C) compared to other tested interference vapors and gases, such as C3H5O, C2H5OH, CO, NH3, CH4, and NO2. This improved response can be associated with the higher surface area and intrinsic point defects. At a higher operating temperature of 150 °C, a response of 28 was witnessed for SnO2 nanorods. A response of 59 was observed for SnO2 nanoparticle-based sensor toward C2H5OH at 150 °C. This variation in the optimal temperature with respect to variations in the sensor morphology implies that the vapor selectivity and sensitivity are morphology-dependent. The relation between the intrinsic sensing performance and vapor selectivity originated from the nonstoichiometry of SnO2, which resulted in excess oxygen vacancies (VO) and higher surface areas. This characteristic played a vital role in the enhancement of the target gas absorptivity and the charge transfer capability of SnO2 hollow sphere-based sensor.

Introduction

The intensification of environmentn class="Chemical">al pn class="Chemical">ollution due to industrialization, fuel burning, agriculture, and leakage of poisonous gases is a threat to human life.[1] A timely discovery and the monitoring of toxic and harmful gases are essential for the environmental protection.[2] Recently, gas sensors based on semiconductor metal oxides (SMOs) have become one of the most dominant research topics owing to their resistance change induced by the interaction of gas and solid materials on the surface of the semiconductors.[3,4] Additionally, metal oxide (MO) can be readily synthesized with various methods (e.g., physical and chemical methods). Various MO nanostructure morphologies have been synthesized, for example, nanobelts, nanowires, nanoflowers, nanoribbons, and nanosheets.[5,6] However, the morphology control remains an important problem in the gas sensing field.[6−8] Among the SMO group (e.g., SnO2, ZnO, TiO2, Fe2O3, and WO3), SnO2 is the most commonly used MO for gas sensing applications owing to its optimal gas sensing properties and excellent chemical and thermal stability.[5,8,9] The material is considered to be an n-type SMO because of the surface oxygen deficit. Thus, the material exhibits sufficient oxygen vacancies (VO) that act as electron donors.[10]

The low selectivity and high operationpan>pan> class="Chemical">al temperature (≥300 °C) of pure SnO2-based gas sensors[11,12] might result in long-term problems caused by MO grain boundaries. Recently, much effort has been dedicated to altering the sensing material through doping or coupling with noble, transitional, or additional MOs because of their various chemical properties and electronic structures, which can result in higher sensitivities and selectivities compared to those of individual MO constituents.[13−18] An additional second constituent that acts as the energetic material enhances the degree of the redox reaction with the target gases. Hyodo et al.[14] reported the gas sensing characteristics of Pd-doped SnO2 at room temperature for NO2 under UV light. For a UV light of 7 mW cm–2, 0.05Pd-doped SnO2 exhibited an enhanced response toward the gas. Hu et al.[15] reported the formaldehyde (HCHO) gas sensing characteristics of Ni-doped SnO2 materials synthesized via a one-step hydrothermal route and tested at 200 °C. They observed that the appropriate incorporation of Ni into SnO2 led to ten times improved sensing of HCHO. Kim et al.[19] reported a near room working temperature and fast response triethylamine gas sensor using Au-loaded ZnO/SnO2 core–shell nanorods. Jeong et al.[20] reported improved sensing response for 0.3 wt % Pt-doped SnO2 hollow nanospheres toward 30 ppm ethanol. The response was approximately seven times greater than that of pristine SnO2. Motaung et al.[21] observed that when the Co3O4 hollow spheres synthesized by ultrasonic spray pyrolysis was converted into Co3O4–SnO2 core–shell hollow spheres by galvanic replacement, the gas selectivity could be tuned by varying the amount of SnO2 shells (14.6, 24.3, and 43.3 at. %) and operating temperatures. According to Motaung et al.,[21] Co3O4 sensors retain a capability to selectively distinguish ethanol vapor at 275 °C. By increasing the amount of SnO2 shells to 14.6 and 24.3 at. %, very selective detection of xylene and methylbenzenes (xylene + toluene) was realized at 275 and 300 °C, respectively. Previous reports[21−23] have indicated that VO has a crucial impact on the gas sensing properties. By enhancing the relative concentration of VO (i.e., providing more active sites), superior sensing performances can be achieved. The VO surface defects can considerably improve the absorptivity for the target gas and the charge transfer capability of SnO2.[22−58]

Despite the amount of work existing on the n class="Chemical">SnO2 hollowsphere gas sensing, however, to the best of our knowledge, no reports exist on the C3H7OH sensing properties of SnO2 hollow spheres at low operating temperatures (75 °C). Few or no studies exist, showing a clear correlation between the vapor gas sensing and point defects extracted using electron spin resonance. Accordingly, we report for the first time the unique C3H7OH sensing characteristics of SnO2 hollow sphere nanostructures prepared via a hydrothermal method. Moreover, the C2H5OH gas sensing properties were investigated for a SnO2 nanoparticle-based sensor. In a controlled manner, nanostructures with different morphologies such as hollow spheres, nanoparticles with two different sizes, nanoplatelets, nanorods, and fishbones were prepared. More importantly, this study explains how to determine the organic vapor sensing characteristics of these nanostructures in detail. Further, the relation between the defect type, SnO2 structure, and vapor selectivity was studied. The intrinsic sensing performance of the SnO2 hollow spheres was investigated via X-ray photoelectron spectroscopy (XPS) and electron paramagnetic resonance (EPR). The EPR, XPS, and photoluminescence (PL) analyses reveal that VO is the dominant structure defects. Extraordinarily, the intensity of the paramagnetic signal related to VO is enhanced for the hollow spheres, which is in good agreement with the sensing response.

Results and Discussion

SnO2 Nanostructure Characterization

Scanning electron microsn class="Chemical">copy (SEM) was used to determinpan>e the morphology and structure of the as-prepared SnO2 nanostructures. Figure shows the SEM images of the SnO2 nanostructures prepared with sodium hydroxide (NaOH) and hexamethylenetriamine (HMT). At lower temperatures (Figure a), the nucleation stage occurred, Sn (OH)62– nuclei were created, and the greatest number of nuclei were formed in the solution in which the nuclei attracted species and other surrounding nuclei to form bigger spheres (hollow spheres). Between 24 and 48 h at 200 °C, the hollow spheres decomposed to small nanoparticles (Figure b), which later agglomerated to form bigger clusters of nanoparticles (Figure c).

Figure 1

SEM images of SnO2 nanostructures after (a) 24 h hydrolysis, (b) 24 h in an autoclave, (c) 48 h in an autoclave at 200 °C, and (d) 12 h and (e) 24 h with HMT.

SEM images of n class="Chemical">SnO2 nanpan>ostructures after (a) 24 h hydrpan> class="Chemical">olysis, (b) 24 h in an autoclave, (c) 48 h in an autoclave at 200 °C, and (d) 12 h and (e) 24 h with HMT.

Becn class="Chemical">ause of the size of the nanpan>oparticles, we, therefore, labeled them as less agglomepan> class="Chemical">rated (LA) nanoparticles and more agglomerated (MA) nanoparticles. The addition of HMT as a surfactant at 200 °C for 12 h resulted in self-assembled nanoparticles, that is, the formation of flower-like SnO2 nanorod bundles (Figure d) disclosing a square-shaped structure. The width of the nanorods is approximately 120 nm. When the time was increased from 12 to 24 h at 200 °C, the formation of fishbone structures was observed (Figure e), which resulted from the growth of the flower-like nanorod bundles.

To vn class="Chemical">alidate the SEM results, low-respan> class="Chemical">olution transmission electron microscopy (TEM) analyses were conducted. As shown in Figure a, hollow spheres with average diameters of 150 nm and wall thicknesses of approximately 20–40 nm are observed. Figure b,c reveals tiny and slightly bigger nanoparticles. Nanorods with flower-like structures are presented in Figure d, and fishbone-like structures exist in Figure e. The trunk diameter is approximately 80 nm, and the rod connected to the trunk has a diameter of approximately 20 nm.

Figure 2

TEM images of SnO2 nanostructures: (a) hollow spheres, (b) LA nanoparticles, (c) MA nanoparticles, (d) nanorods, and (e) fishbones, after (a) 24 h hydrolysis, (b) 24 h in autoclave, (c) 48 h in autoclave, and (d) 12 h and (e) 24 h with HMT.

TEM images of n class="Chemical">SnO2 nanpan>ostructures: (a) hpan> class="Chemical">ollow spheres, (b) LA nanoparticles, (c) MA nanoparticles, (d) nanorods, and (e) fishbones, after (a) 24 h hydrolysis, (b) 24 h in autoclave, (c) 48 h in autoclave, and (d) 12 h and (e) 24 h with HMT.

The high-resn class="Chemical">olutionpan> (HR)-TEM images anpan>d selected area electronpan> diffpan> class="Chemical">raction (SAED) patterns of various synthesized SnO2 nanostructures are shown in Figure a–e. The Debye rings of the hollow spheres and LA nanoparticles are more diffused compared to those of the other nanostructures. However, with increasing particle sizes, the Debye rings become brighter according to Figure c. Moreover, the nanorods exhibit single-crystal properties in Figure d. By contrast, the nanostructures of the fishbones and nanoplatelets are highly polycrystalline. The observed d-spacing of approximately 0.3347 nm is attributed to the (110) plane of tetragonal rutile SnO2.

Figure 3

TEM images and SAED patterns of SnO2 nanostructures: (a,a′) hollow spheres, (b,b′) LA nanoparticles, (c,c′) MA-nanoparticles, (d,d′) nanorods, and (e,e′) fishbones after (a) 24 h hydrolysis, (b) 24 h in autoclave, (c) 48 h in autoclave, and (d) 12 h and (e) 24 h with HMT.

TEM images and n class="Chemical">SAED patterns of pan> class="Chemical">SnO2 nanostructures: (a,a′) hollow spheres, (b,b′) LA nanoparticles, (c,c′) MA-nanoparticles, (d,d′) nanorods, and (e,e′) fishbones after (a) 24 h hydrolysis, (b) 24 h in autoclave, (c) 48 h in autoclave, and (d) 12 h and (e) 24 h with HMT.

The crystn class="Chemical">allinpan>e structure anpan>d phase purity of the products were determinpan>ed with the X-ray diffraction (XRD) technique. The crystal orientation plays an imperative role in surface-related applications such as gas sensing owing to the high density of atomic steps, edges, and ample unsaturated coordination sites. In Figure , the XRD patterns of the samples obtained under different hydrothermal conditions display the peak positions that match well with the standard data with a = b = 0.4738 nm and c = 0.3188 nm (JCPDS card no. 041-1445). No other phases can be observed for SnO2. Hence, the as-prepared nanomaterials had a pure rutile SnO2 structure. All diffraction peaks can be indexed to the tetragonal rutile SnO2 phase with a = b = 0.4748 nm and c = 0.3817 nm. The most prominent peaks of these nanostructures correspond to the (110), (101), and (211) crystal lattice planes. It is further evident in Figure b that XRD patterns shift depending on the structure of SnO2.

Figure 4

(a) XRD patterns of various as-prepared SnO2 nanostructures. (b) Magnified at the (101) diffraction peak. LA nanoparticles, less agglomerated nanoparticles; MA nanoparticles, more agglomerated nanoparticles.

(a) XRD patterns of various as-prepared n class="Chemical">SnO2 nanpan>ostructures. (b) Magnpan>ified at the (101) diffpan> class="Chemical">raction peak. LA nanoparticles, less agglomerated nanoparticles; MA nanoparticles, more agglomerated nanoparticles.

Regarding the tetn class="Chemical">ragonpan>pan> class="Chemical">al rutile structure of SnO2, the interplanar space d( is related to the lattice parameter via

With a and c being lattice n class="Chemical">constanpan>ts. The lattice pan> class="Chemical">constants can be calculated with

As listed in Table , various nanostructures depending on the morphn class="Chemical">ology show differenpan>t sizes of the crystallites, where then nanoparticle reveals smaller crystallites compared to the rest of the samples.

Table 1

2θ, Crystal Size, and d-Spacings of Various Nanostructures for the (110) Diffraction Peak

SnO2	2θ (deg)	D (nm)	a = b (nm)	c (nm)
hollow spheres	26.684	10.80	0.4720	0.3346
LA particles	26.566	6.80	0.4692	0.3123
MA particles	26.849	7.73	0.4680	0.3155
rods	26.711	19.60	0.4715	0.3170
fishbones	26.701	12.60	0.4737	0.3188

It is well known that the gas sensor properties are n class="Chemical">also depenpan>denpan>t onpan> the surface area anpan>d porosity of the SMO. Thus, pan> class="Chemical">nitrogen adsorption–desorption analyses were carried out. The results are shown in Figure . According to the IUPAC classification, the nitrogen adsorption–desorption isotherms of the hollow spheres and MA nanoparticles exhibit type IV isotherms.[25] The LA nanoparticles and nanorods have typical type III isotherms. At a relative pressure (P/P0) of 0.5–1.0, the presence of large mesopores or macropores is observed. The amount of absorbed N2 increases with increasing pressure. Hence, the nanostructures possessed mesoporous structures. It evident that the desorption curves are slightly lower than the adsorption curve, especially for the fishbones, and such behavior could be due to poor pore volume. The pore size distribution investigated with the Barrett–Joyner–Halenda (BJH) method is shown in the insets of Figure . The pore sizes differ for the different morphologies of the nanostructures. Regarding the hollow spheres, high specific surface areas of 195.86 m2 g–1 and pore volumes of approximately 0.2625 cm3 g–1 are observed. These findings suggest that the higher surface area and pore volume found for hollow spheres make them possible candidates for use in gas sensors.

Figure 5

Nitrogen adsorption–desorption isotherm and BJH pore size distributions (insets) of various SnO2 nanostructures. (a) Hollow spheres, (b) LA nanoparticles, (c) MA nanoparticles, (d) fishbones, and (e) nanorods.

n class="Chemical">Nitrogen adsorpan> class="Chemical">ption–desorption isotherm and BJH pore size distributions (insets) of various SnO2 nanostructures. (a) Hollow spheres, (b) LA nanoparticles, (c) MA nanoparticles, (d) fishbones, and (e) nanorods.

The HR XPS spectn class="Chemical">ra for Sn 3d inpan> Figure exhibit two peaks at approximately 486.9–4877 anpan>d 495.4–496.1 eV, which are assignpan>ed to Sn 3d5/2 anpan>d Sn 3d3/2, respectively.[26,27] These two symmetric peaks are associated with the pan> class="Gene">spin–orbit coupling of the 3d state with a spin–orbit separation of 8.4 eV and a spin–orbit branching ratio of 1.34. This observation is in good agreement with Lee et al.[28] The single peak observed in the Sn 3d5/2 region (486.9–4877 eV) of all different nanostructures indicates that the samples have a Sn(IV) (Sn4+) state with no visible Sn2+, which is anticipated to appear at 485.8 eV. The observed peak shift is related to the morphology change as observed in Figure S1a (Supporting Information).

Figure 6

Sn 3d spectra of various SnO2 nanostructures: (a) hollow spheres, (b) LA nanoparticles, (c) MA nanoparticles, (d) nanorods, and (e) fishbones.

Sn 3d spectn class="Chemical">ra of various pan> class="Chemical">SnO2 nanostructures: (a) hollow spheres, (b) LA nanoparticles, (c) MA nanoparticles, (d) nanorods, and (e) fishbones.

The XPS spectn class="Chemical">ra were produced to determinpan>e the chemicpan> class="Chemical">al composition of the materials. Figure illustrates the HR XPS spectrum of oxygen. Based on the measured binding energy, the asymmetric shape of the spectrum indicates several chemical states. Therefore, the O 1s peaks can be deconvoluted into three peaks, as shown in Figure . For instance, the hollow spheres have three peaks at 531.4, 532.4, and 533.1 eV, whereas other morphologies including the nanoparticles and nanorods have three peaks centered at 530.9, 531.4, and 532.5 eV. The lower binding energies of 530.9 and 531.4 eV, respectively, correspond to O2– ions in the SnO2 lattice (OL) and O2– ions in the oxygen deficiency region within the SnO2 matrix (Ov), which corresponds to VO.[29,30] The peak at 533 eV is accredited to chemisorbed and dissociated oxygen species (i.e., O– and O2–) in the surface of the SnO2 nanostructures (OS).[31] From the Supporting Information, Figure S1b, it is observed that O 1s peaks shift depending on the structure of SnO2, and such shifting was previously observed from the XRD patterns in Figure b.

Figure 7

O 1s XPS spectra of various SnO2 nanostructures: (a) hollow spheres, (b) LA nanoparticles, (c) MA nanoparticles, (d) nanorods, and (e) fishbones.

O 1s XPS spectn class="Chemical">ra of various pan> class="Chemical">SnO2 nanostructures: (a) hollow spheres, (b) LA nanoparticles, (c) MA nanoparticles, (d) nanorods, and (e) fishbones.

To further n class="Chemical">confirm the creationpan> of the pan> class="Chemical">VO and oxygen deficiency centers in the sensing and response mechanism of SnO2, an EPR analysis was conducted (Figure ). The hollow spheres, LA nanoparticles, MA nanoparticles, and fishbones in Figure show a single broad resonance peak located at the g-factors of 1.8991, 2.0041, and 2.0161, respectively. The nanorods show two peaks located at the g-factors of 2.0023 and 2.1418. The signal at approximately 2.1 is related to superoxide radical O2–, which forms owing to the transfer of electrons trapped in VO to the adsorbed oxygen molecules in the SnO2 surface.[29] The observed resonances at 1.8991–2.0 are attributed to unpaired electrons trapped in singly ionized VO.[30−34] The hollow spheres show the highest signal. Thus, they possess high quantities of defects, as confirmed by the PL (see the Supporting Information, Figure S2) and XPS measurements. Such higher defect concentrations could be beneficial for gas sensing applications.

Figure 8

(a) Electron spin resonance spectra of various SnO2 nanostructures and (b) magnified spectra.

(a) Electron n class="Gene">spin resonpan>anpan>ce spectpan> class="Chemical">ra of various SnO2 nanostructures and (b) magnified spectra.

Electrical and Gas Sensing Properties

The MO gas sensor response is predominantly governed by the inten class="Chemical">ractionpan> betweenpan> the anpan>alyte gas and oxygen species on the sensor surface. Therefore, the morphology, surface area, and defect states severely impact the sensing properties. Moreover, the responses of gas sensors based on MO materials are intensely affected by the operating temperature.[35] To determine the optimal operating temperature, the performance of various SnO2 nanostructures was tested with various gases (e.g., CO, CH4, NO2, and NH3) and volatile organic compounds (e.g., C3H7OH, C2H6O, and C3H6O) at 75, 150, and 225 °C.

Among the tested sensors, the hollow spheres have the stronpan>gest responpan>se toward pan> class="Chemical">C3H7OH at a low operating temperature (75 °C). The hollow spheres exhibit a response value of 15, which is five times higher than those of the others at this temperature. Previous studies[36,37] have indicated that alcohols with increased −CH2– groups are effortlessly decomposed and oxidized compared to those with lesser one. Consequently, C3H7OH containing the maximum number of −CH2– groups is simply decomposed. Therefore, it reacts efficiently with the adsorbed oxygen molecules, resulting in a release of electrons, consecutively, enhancing the sensing response. This explains the high response of the film in the presence of propan-2-ol, followed by ethanol and methanol, respectively. To the best of our knowledge, a response value of 15.8 of pure SnO2 for C3H7OH vapor at 75 °C is one of the highest response values that have ever been reported.[38] The sensors with low operating temperatures not only reduce costs but also increase the life span of sensors without compromising stability. However, for an operating temperature of 150 °C, the nanorod-based sensor reveals a higher response toward C2H5OH. This clearly indicates a shift in the optimal temperature among sensors with varying morphologies. Thus, the optimal temperature is morphology-dependent. As a result, the operating temperatures of 75 and 150 °C were selected for SnO2 hollow spheres and nanorod-based sensors for various concentrations of C3H7OH and C2H5OH vapors.

The resistance change of the various n class="Chemical">SnO2-based senpan>sors with respect to pan> class="Chemical">C2H5OH is shown in Figure . All sensors display the typical n-type gas sensing behavior. The different morphologies have substantial effects on the baseline resistance of the gas sensors. The resistances of the nanorods and fishbones are higher in the air in comparison to those of the other nanostructures. Furthermore, regardless of the C2H5OH concentration, the resistance of the SnO2-based sensors recovers to the baseline level after the C2H5OH flow stops. The response and recovery times of 110 and 90 s were noticed for the hollow spheres toward C3H7OH, while the response and recovery times of 105 and 100 s were witnessed for the LA nanoparticles toward C2H5OH, respectively.

Figure 9

(a) Response vs operating temperatures of SnO2-based sensors exposed to 40 ppm of C3H7OH and C2H5OH. (b) Dynamic resistance curves of various SnO2-based sensors upon exposure to 10–100 ppm C2H5OH at 150 °C. (c) Real-time response of various SnO2-based sensors toward C2H5OH vapor.

(a) Response vs open class="Chemical">rating tempepan> class="Chemical">ratures of SnO2-based sensors exposed to 40 ppm of C3H7OH and C2H5OH. (b) Dynamic resistance curves of various SnO2-based sensors upon exposure to 10–100 ppm C2H5OH at 150 °C. (c) Real-time response of various SnO2-based sensors toward C2H5OH vapor.

Figure shows the responses versus the n class="Chemical">C3H7OH anpan>d pan> class="Chemical">C2H5OH gas concentrations at 75 and 150 °C, respectively. The SnO2 hollow spheres exhibit an exponential increase in the response to 10–40 ppm of C3H7OH. It is well known that SnO2 exhibits a great response at elevated temperatures (>200 °C).[37−40] Nonetheless, the presented performance of the pure SnO2 hollow spheres at 75 °C is even better than in other reports (Table ). Tan et al.[39] reported response values of approximately 19 for pure SnO2 nanoparticles at 300 °C toward 100 ppm ethanol.

Figure 10

Response vs gas concentrations of (a) C3H7OH and (b) C2H5OH at 75 and 150 °C, respectively.

Table 2

Comparison of Gas Sensing Performances of SnO2-Based Sensors for C3H7OH and C2H5OH

sensing element	synthesis method	gas	gas concentration (ppm)	operating temperature (°C)	response	refs
SnO2 hollow spheres	hydrothermal	C3H7OH	40	75	20.1	this work
SnO2 LA nanoparticles	hydrothermal	C2H5OH	40	150	59.6	this work
SnO2 nanoparticles	hydrothermal	C3H7OH	500	100	9.8	(41)
SnO2 nanoparticles	hydrothermal	C3H7OH	17	220	32	(42)
			1.7		6
SnO2 nanoparticles	hydrothermal	C2H5OH	500	100	14.6	(43)
SnO2 nanoparticles	quasi-molecular cluster	C2H5OH	100	300	8.9	(38)
SnO2 nanowires	quartz tube method	C2H5OH	50	400	7.1	(44)

Response vs gas n class="Chemical">concentpan> class="Chemical">rations of (a) C3H7OH and (b) C2H5OH at 75 and 150 °C, respectively.

Table (41−44) compares the results of the currenpan>t study with those reported inpan> the literature for SnO2 sensors tested on C3H7OH and C2H5OH, respectively. Please note that most reported SnO2-based sensors were prepared via the hydrothermal method, which allows a clear comparison with our results. The gas response of the SnO2 hollow spheres and nanoparticles toward C3H7OH and C2H5OH is higher than those of the sensors listed in Table for a low operating temperature.

The selectivity of the n class="Chemical">SnO2 senpan>sors with respect to various target vapors anpan>d gases (e.g., pan> class="Chemical">C3H7OH, C2H5OH, CO, NH3, CH4, and NO2) is shown in Figure . All sensors were tested with similar gas concentrations (40 ppm) and at two different working temperatures (75 and 150 °C). At 75 °C, the SnO2 hollow sphere-based sensor exhibits a significantly higher response to C3H7OH compared with the other sensors (see Figure a). However, at 150 °C, the SnO2 nanoparticle-based sensor reveals an improved response toward C2H5OH vapor. Thus, the operating temperature affects the selectivity of the MO gas sensor (Figure b). The high response to C3H7OH might be due to the highest number of (−CH2−) groups, which decompose more easily and result in greater changes in the free-carrier density within the SnO2 structure.[45]

Figure 11

Selectivity histograms of SnO2-based sensors exposed to various gases at (a) 75 and (b) 150 °C; (c,d) response ratio for C3H7OH and C3H5OH with respect to other five interfering gases (40 ppm) at various operating temperatures.

Selectivity histogn class="Chemical">rams of pan> class="Chemical">SnO2-based sensors exposed to various gases at (a) 75 and (b) 150 °C; (c,d) response ratio for C3H7OH and C3H5OH with respect to other five interfering gases (40 ppm) at various operating temperatures.

For quantification, the cross selectivity was cn class="Chemical">alculated usinpan>g the equationpan>: S = pan> class="Chemical">Sa/Sb, where Sa and Sb are the responses of a sensor toward a target and the interfering gases, respectively.[46−48]Figure c,d illustrates the cross selectivity of the SnO2 hollow sphere- and LA nanoparticle-based sensors for C3H7OH and C2H5OH with respect to the interference gases, respectively. The response ratios of the hollow spheres of C3H7OH to C2H5OH and of the LA-nanoparticles of C2H5OH to NO2 are relatively large (1450 and 1140, respectively). In this study, NH3 and NO2 act as prospectively strong interfering gases. Therefore, the respective response ratios are considered as the influence parameters to evaluate the cross selectivity of the hollow spheres.

To study the stability of the hollow sphere- anpan>d nanpan>orod-based senpan>sors, repeatability anpan>d lonpan>g-term stability anpan>alyses were carried out. Figure a shows the repeatability performance of the sensor for eight successive cycles toward 40 ppm C3H7OH for three weeks. On day one, the fresh sensor exhibits evident repeatability of six consecutive periods for 40 ppm C3H7OH. However, after seven days, the repeatability performance remains constant, whereas the response decreases, as confirmed by the stability plot in Figure b. This behavior is also observed after 23 days, where the response decreases by 88.1% with respect to the original value. It is further observed that the LA nanoparticle response decreases rapidly by 37% after 7 days, while after 23 days, the response dropped by 86.7%. However, we have found that upon incorporating the p-type NiO on the surface of the hollow spheres, the stability improved, and such results are not shown here because they fall under our future work.

Figure 12

(a) Repeatability (b,c) long-term stability of hollow spheres under response to 40 ppm C3H7OH; (d,e) response vs RH (0, 40, and 60%) under 40 ppm C3H7OH and C2H6OH at 75 and 150 °C, respectively.

(a) Repeatability (b,c) long-term stability of hn class="Chemical">ollow spheres unpan>der responpan>se to 40 ppm pan> class="Chemical">C3H7OH; (d,e) response vs RH (0, 40, and 60%) under 40 ppm C3H7OH and C2H6OH at 75 and 150 °C, respectively.

The responses of the sensors exposed to 40 ppm n class="Chemical">C3H7OH inpan> the presenpan>ce of 40 anpan>d 60% relative humidity (pan> class="Chemical">RH) are shown in Figure c,d. The responses of both sensors decrease with RH. The performance of the SnO2 hollow sphere-based sensor decreases by 87.3% toward 40 ppm C3H7OH in the presence of 60% RH. The response of the SnO2 LA nanoparticle-based sensor decreases by 91.8% toward C2H6O in the presence of 60% RH. Based on the above findings, it is evident that the current sensors are not stable either in an actual environment or in dry air after long exposure to C3H7OH or C2H5OH vapor.

Usun class="Chemical">ally, the gas senpan>sinpan>g mechanpan>ism of anpan> n-type pan> class="Chemical">SnO2 SMO can be determined according to the radial modulation in the electron depletion layer and the potential barrier modulation. The oxygen species that are adsorbed on the SnO2 surface capture the electrons in the conduction band, thereby increasing the resistance because of the increasing electron depletion layer. When the sensor is exposed to a target gas (e.g., C3H7OH), the gas molecules interact with the preadsorbed oxygen species (O2–). The electrons that are captured are released back into the SnO2 conduction band, thereby decreasing the electron depletion layer and the resistance.[49]

To determine the reason for the different sensitivities and selectivities of the gas sensors, we analyzed differenpan>t types of paramagnetic defects existing on the nonstoichiometric layers of the SnO2 sensors and compared them with the sensing responses. According to the EPR, PL, and XPS analyses, VO is the dominant defect on the SnO2 surface. VO is considered as an active site for oxygen adsorption. Therefore, the hollow spheres exhibit improved oxygen adsorption. To relate the paramagnetic defects to the gas sensing response, we calculated the number of spins (Figure a). Remarkably, the sensing response fluctuates with the EPR intensity. Therefore, a clear constructive relationship between the electrical response and paramagnetic VO is revealed, which is also confirmed by the PL and XPS results (see the Supporting Information). These results agree well with those reported in previous studies. Thus, the quantity of paramagnetic VO identified via ESR is related to the number of electrons reaching the SnO2 conduction band. This result agrees well with the improved sensing performance.[34,49,50] From the XRD analysis, the hollow spheres displayed higher lattice strain, which resulted in the formation of more structural defects within the system as validated by PL, EPR, and XPS analyses.

Figure 13

Gas sensing response toward 40 ppm C3H7O at 23 °C vs (a) number of spins related to paramagnetic defects and (b) BET surface area. Fundamentally, each VO offers two electrons. Therefore, a larger concentration of VO in the hollow spheres results in a higher number of electrons, which can be absorbed.[34,52,53]

Gas sensing response toward 40 ppm n class="Chemical">C3H7O at 23 °C vs (a) number of spins related to paramagnetic defects and (b) BET surface area. Fundamentally, each VO offers two electrons. Therefore, a larger concentration of VO in the hollow spheres results in a higher number of electrons, which can be absorbed.[34,52,53]

Furthermore, to understand the sensing behavior occurring on the current sensing layer, we n class="Chemical">correlated the responpan>se versus the active surface area of various nanpan>ostructures, anpan>d the results are presenpan>ted inpan> Figure b. As shown inpan> Figure b, as the surface area inpan>creases, the responpan>se pan> class="Chemical">also increases, whereby the hollow spheres reveals higher surface area, correlating well with the higher response they revealed. Remarkably, the hollow sphere structure containing the inner and outer shell layer can broaden the contact area existing between the sensing layer and analyte gases as observed in the current work. The high surface to volume ratio and density of surface active sites are supportive of the adsorption and desorption of C3H7OH gas molecules and surface interaction.[51−54] The hollow sphere structure reduces the path of gas diffusion and enables the penetration of gas molecules.

Besides, Mon class="Chemical">razzonpan>i et pan> class="Chemical">al.[49,55] proposed new radical Sn4+–O2–, which is paramagnetically active and plays a vital part in the gas sensing mechanism

When the n class="Chemical">SnO2 hpan> class="Chemical">ollow sphere- and nanoparticle-based sensors are exposed to C3H7OH (see Scheme ) and C2H5OH at 75 and 125 °C, respectively, they will react with the adsorbed oxygen species (O2–), while, at temperatures of <300 °C, they will react with the adsorbed oxygen species (O–) through the following reactions 6 and 7,[56,57] thereby resulting in a decrease in the sensor resistance

Scheme 1

Schematic Diagram of the Oxygen Chemisorption and the Reaction between C3H7OH Gas and the Preadsorbed Oxygen Species

Conclusions

Various n class="Chemical">SnO2 nanpan>ostructures were successfully synpan>thesized usinpan>g a simple hydrothermpan> class="Chemical">al one-pot method. In a controlled manner, nanostructures with different morphologies such as hollow spheres, nanoparticles with two different sizes, nanoplatelets, nanorods, and fishbones were synthesized. The defect analyses via EPR, PL, and XPS revealed that the hollow spheres contained the most defects (e.g., VO). Further, the relative concentration of VO is morphology-dependent. The SnO2 nanostructure-based gas sensors were fabricated and characterized for different gases (CO, NH3, CH4, and NO2) including volatile organic compounds (C3H7OH, C3H5OH, and C2H5OH) at various operating temperatures. The SnO2 hollow spheres exhibited an improved sensing response for C3H7OH gas at a low operating temperature (75 °C). At a high operating temperature (150 °C), the SnO2 nanoparticle-based sensor showed the highest response value. This variation in the optimal temperatures among sensors with different morphologies implied that the gas selectivity and sensitivity are morphology-dependent.

Experimental Section

Materials

n class="Chemical">All reagents were of anpan>pan> class="Chemical">alytical grade and used without further purification. Tin chloride dihydrate (SnCl2·2H2O), sodium hydroxide (NaOH), HMT [H2N (CH2)6NH2], and absolute ethanol were purchased from Sigma-Aldrich, Johannesburg, South Africa.

Synthesis of SnO2 Nanostructures

The n class="Chemical">SnO2 hn class="Chemical">ollow spheres were produced by mixing 0.4 g SnCl2·2H2O and 0.5 g NaOH in 20 mL deionized water and 0.5 M HNO3, under magnetic stirring for 20 min. Next, 40 mL ethanol was slowly added to the solution to form a white precipitate. The mixture was stirred for 24 h, washed with distilled water and ethanol, and dried at 80 °C for 24 h. The SnO2 nanoparticles were synthesized using 0.4 g SnCl2·2H2O and 0.5 g NaOH in 20 mL deionized water and 0.5 M HNO3. The solution was stirred for 20 min and transferred into a 45 mL Teflon-lined stainless-steel autoclave, which was sealed tightly and kept at 200 °C in an oven for 12–48 h. After the hydrothermal step, the autoclave was allowed to cool down to room temperature. The flower-rod bundle, fishbone platelet, and hierarchical flower-like platelet nanostructures were synthesized by mixing 0.4 g SnCl2·2H2O, 0.5 g NaOH, and 0.84 g HMT in 20 mL deionized water under continuous magnetic stirring. Next, 60 mL ethanol was added slowly to the white precipitate. The mixture was transferred into a 45 mL Teflon-lined stainless-steel autoclave and kept at 200 °C for 12–48 h. We should point out that to form different morphologies, different synthesis times had to be adopted. While with the MA nanoparticles, the synthesis reaction time prolonged in order to form larger sizes of the nanoparticles to have a comparison with smaller ones. Finally, the white precipitates were collected through centrifugation and washed thoroughly with distilled water and ethanol. Afterward, they were dried at 80 °C for 24 h.

Characterization

The n class="Chemical">SnO2 crystpan> class="Chemical">al structure was analyzed with a PANalytical X’pert PRO PW 3040/60 X-ray diffractometer with Cu Kα radiation and λ = 0.154178 nm at an acceleration voltage of 45 kV and a current of 40 mA (PANalytical, the Netherland). The nanostructures and surface morphologies were determined with a scanning electron microscope (SEM, Zeiss Auriga, Germany) and JEOL-2100 transmission electron microscope (TEM, JEOL, Japan). The analysis of the pores, surface area, and nitrogen adsorption–desorption isotherms were conducted with the Brunauer–Emmett–Teller (BET) and BJH methods with a Micromeritics TRISTAR 3000 instrument, USA. The point defects and crystal quality were investigated with a Jobin-Yvon NanoLog PL spectrometer (Jobin-Yvon, France). Further, the chemical state of the nanostructures was studied via XPS with a PHI 5000 VersaProbe Scanning ESCA Microprobe (USA). To study paramagnetic defects, a JEOL X-band EPR spectrometer (JEOL, Japan) was used. The analyses were carried out at 25 and 75 °C to mimic the sensing conditions.

Sensor Fabrication and Sensing Characterization

The vapor and gas sensing properties were measured using the KSGAS6S gas sensing station (Kenosistec, Itn class="Chemical">aly). Inpan> the typicpan> class="Chemical">al sensor fabrication procedure, an appropriate amount of sample powder was placed in an agate mortar and grounded adding terpineol to form slurry. Afterward, the paste was carefully coated onto alumina substrates. The sensor units were placed in a sealed chamber at a controlled temperature under different conditions (i.e., in dry air and under RH values of 40 and 60%). The gas sensing studies were conducted exposing the fabricated sensors to different concentrations (10–100 ppm) of various reducing gases (e.g., CO, CH4, and NH3), volatile organic compounds (C2H5OH, C3H6O, and C3H7OH), and an oxidizing gas (NO2) in a background of synthetic dry air. The gas/organic compound concentrations in the test chamber were precisely mixed with synthetic air. The measurements were carried out at various operating temperatures 25 °C (i.e., room temperature), 75, 150, and 225 °C. The sensor resistance was continuously monitored using a Keithley 6487 picoammeter/voltage source meter. We define the gas response as Ra/Rg – 1 for reduced gas and Rg/Ra – 1 for oxidized gas, where Ra and Rg are the resistances of the sensors in air and gas/organic vapor, respectively. The response recovery times were calculated as the time intervals between the time at which the response has achieved 90% of its maximum and the time at which it has decreased to 10% of its maximum.[12]