2.1 Synthesis

SnO₂ nanowires were produced in a tube furnace. For the synthesis, metallic Sn nanopowder with a purity of 99.9% was used as the raw material. Typically, a 3 nm-thick Au layer was coated on a Si substrate as catalyst; the Au-coated substrate was heated at 900 °C for 1 h, and then, exposed to a mixture of Ar (97%) and O₂ (3%) gases at a constant flow rate and total pressure of 2 Torr. The details of the SnO₂ nanowire synthesis have been reported in our previous publications [²⁶,²⁷].

Si was coated on the as-grown SnO₂ nanowires by radio frequency (RF) sputtering with a Si target at room temperature. The optimal pressure for the Si shell layer deposition was 10 mTorr in Ar ambient and the RF power during the deposition was set to 50 W. The sputtering time was varied from 2 min to 4 min to obtain Si shells with thicknesses of 5 nm and 10 nm, respectively.

2.2 Material Characterization

The crystallinity and phase purity of the prepared samples were investigated by X-ray diffraction (XRD) using a Philips X’Pert diffractometer with the Cu K_α1 radiation (λ = 1.5404 Å). The morphology was investigated by scanning electron microscopy (SEM, Hitachi S-4200) and transmission electron microscopy (TEM, Philips CM 200). Energy-dispersive X-ray spectroscopy (EDS) was performed to study the chemical composition of the synthesized products. The work function of SnO₂ nanowires was determined from the ultraviolet photoelectron spectroscopy (UPS) profiles recorded using an HeI (21.2 eV) ultraviolet source.

2.3 Sensor Fabrication and Gas Sensing Tests

Details of the gas sensing tests can be found elsewhere [²⁸-³⁰]. A bi-layered electrode comprising a Au layer (300 nm-thick) and a Ti layer (50 nm-thick) was sputter-deposited on each sample. The gas flow was controlled by changing the mixing ratio of dry air to the target gas using mass flow controllers under a total stream of 500 sccm. All measurements were carried out in a special gas chamber and the sensor resistance data were obtained using a source meter. The sensor response (R) is defined as R = R_a/R_g, where R_a and R_g are the resistances of the sensor in air and in the presence of the target gas, respectively. The response and recovery times were calculated as the time to reach 90% of the final signal after exposure and stoppage of the target gas, respectively.

3.1 Morphological and Structural Studies

The morphology of the synthesized nanowires was studied by SEM and the micrographs are presented in Fig 1 (a)-(f). Fig 1(a) reveals bare, curvy, and continuous SnO₂ nanowires with lengths in the order of several micrometers and diameters of ~100 nm. Figure 1(b) is a high-magnification SEM image showing the smooth morphology of the synthesized bare SnO₂ nanowires. Figure 1(c) shows Si-coated SnO₂ nanowires with a shell thickness of 5 nm and the high-magnification SEM image in Fig 1(d) shows the relatively smooth morphology of the coated nanowires. Figure 1(e) and (f) show the SEM images of Si-coated (10 nm) SnO₂ nanowires. Coating with a 10 nm Si shell resulted in a relatively rough morphology.

To determine the crystal structure of the synthesized nanowires, XRD patterns of bare and Si-coated SnO₂ nanowires were recorded and are shown in Fig 2(a)-(c). Common peaks for all materials were observed at 2θ = 26.61, 33.92, 37.99, 39.02, 51.83, 54.81, and 57.91°, corresponding to the (110), (101), (200), (111), (211), (220), and (002) planes of SnO₂, respectively. Si peaks at 2θ = 26.87 (200), 33.07 (211), 38.37 (220), 51.53 (321), 55.38 (400), and 59.06° (411), originating from the Si shell, overlapped with the SnO₂ peaks. In addition, the microstructures of bare and Si-coated SnO₂ nanowires were studied by TEM and the micrographs are presented in Fig 3. TEM images of SnO₂ nanowires with shell thicknesses of 5 nm and 10 nm are shown in Fig 3(a) and 3(d), respectively. As observed, the nanowires were long and separate from one other, suggesting that they can provide high surface areas for gas adsorption. The diameters of the nanowires were about 100 nm, in accordance with the SEM results. The high-resolution TEM (HRTEM) images of SnO₂ nanowires with shell thicknesses of 5 nm and 10 nm are shown in Fig 3(b) and 3(e), respectively. In both cases, lattice fringes with spacings of 0.264 nm and 0.237 nm, corresponding to the (101) and (200) planes of SnO₂, respectively, were observed. This result further confirmed the crystalline nature of the SnO₂ nanowires. The corresponding selected area diffraction (SAED) patterns showing distinct diffraction spots are presented in Fig 3(c) and 3(f) which confirms the single crystalline nature of the synthesized SnO₂ nanowires. The EDS elemental mapping images revealing the presence Si, Sn, and O in the synthesized nanowires are shown in Fig 4. It is noteworthy that a thin Si layer covered the SnO₂ nanowires.

3.2 Gas Sensing Studies

From a series of preliminary tests, we determined that the optimal working temperature of the sensors was 100 °C; therefore, all gas sensing measurements were performed at 100 °C. Fig 5 shows the H₂S sensing properties of bare and Si-coated SnO₂ nanowires. The dynamic response plots of bare SnO₂ nanowires for 10, 20, and 50 ppm of H₂S gas are shown in Fig 5(a). Upon exposure to H₂S gas, the resistance of the sensor decreased, reflecting its n-type behavior. Furthermore, the response was reversible, because after stopping the injection of H₂S and air, the resistance returned to its initial value. The response (R_a/R_g) versus time plots are presented in Fig 5(b). The responses of the bare SnO₂ nanowire sensor to 10, 20, and 50 ppm H₂S gas were 1.035, 1.043, and 1.043, respectively, indicating that the response was almost identical for different concentrations of H₂S gas. This was likely due to the limited surface area of the bare SnO₂ nanowire. The dynamic response plots of 5 nm-thick and 10 nm-thick Si shell-coated SnO₂ nanowires for different concentrations of H₂S are shown in Fig 5(c) and 5(e), respectively. Both sensors showed n-type behaviors and reversible responses. Fig 5(d) and 5(f) show plots of response as a function of time for SnO₂ nanowires coated with 5 nm- or 10 nm-thick Si shells. The responses of SnO₂ nanowires with 5 nm Si shells to 10, 20, and 50 ppm of H₂S gas were 1.299, 1.321, and 1.769, respectively, while those of SnO₂ nanowires with 10 nm-thick Si shells were 1.367, 1.405, and 1.629, respectively. The calibration curves of all sensors are shown in Fig 5(g). Clearly, the response of bare SnO₂ nanowires was far lower than that of Si-coated SnO₂ nanowires. This demonstrated that Si enhanced the sensing performance of the SnO₂ nanowire. Furthermore, the nanowires with 5 nm-thick Si shells showed a higher response to 50 ppm of H₂S than those with 10 nm-thick shells; on the other hand, to 10 ppm and 20 ppm of H₂S, the response of nanowires with 5 nm-thick Si shells was higher.

Selectivity for a target gas is highly important for real applications because a lack of selectivity can lead to false alarms, limiting the practical application of a sensor. To study the selectivity of 5 nm-thick Si shell-coated SnO₂ nanowires, they were exposed to 50 ppm of benzene, ethanol, CO, H₂, and toluene gases at 100 °C. Fig 6(a) shows the variations in responses to these gases with time; the corresponding selectivity histogram of the sensor is presented in Fig 6(b). The sensor responses to the gases were 1.051, 1.028, 1.052, 1.047, and 1.016, respectively, indicating that these sensors were largely insensitive to interfering gases.

3.3 Gas Sensing Mechanism

Several factors can influence the sensing behavior of Si-shelled SnO₂ nanowires. When considering the changes in the resistance of a SnO₂-Si core-shell nanowire upon exposure to H₂S gas, the change in the resistance of the SnO₂ nanowire core (R₁) and the change in the resistance of the Si shell (R₂) should be taken into account [²⁸]. Almost every gas adsorbs onto the surface of a metal oxide [³¹]. The most prominent reactive gas in normal ambient air is molecular oxygen. Depending on the sensor operation temperature, oxygen adsorbs in various forms onto the metal oxide surface. Exposure of a pristine SnO₂ nanowire sensor to air results in the adsorption of oxygen molecules onto the surface of the nanowire. Because of the high electron affinity of oxygen, the adsorbed oxygen molecules trap electrons from the conduction band of SnO₂, forming ionized oxygen species such as O₂^–, O^–, and O^2– [²]:

Consequently, the concentration of electrons on the surface decreases, and accordingly, the resistance of the SnO₂ nanowires increases. When a pristine SnO₂ nanowire gas sensor is exposed to H₂S, H₂S molecules chemisorb on the surface of the SnO₂ nanowire, and electrons are released by a subsequent reaction with oxygen ions, resulting in H₂O and SO₂ gases [³²]:

The released electrons decrease the resistance of the gas sensor, resulting in a response. Because the core SnO₂ nanowires were single crystalline, the effects of grain boundaries were excluded. After Si was coated on the SnO₂ nanowires, an increase in response was observed. As observed in Fig 5(g), the response of the Si-shelled nanowires was superior to that of bare nanowires, regardless of the gas concentration or shell thickness. The H₂S gas sensing mechanism for the Si-coated SnO₂ nanowire sensor can be explained on the basis of several factors. The TEM results revealed that the Si shell layer did not completely cover the core nanowire, wherein a part of the core was exposed along the entire length of the nanowire. More importantly, the sensor exhibited an n-type behavior, illustrating the dominant role of the SnO₂ nanowire core in sensing.

Similar to that for the SnO₂ nanowires, adsorption of H₂S gas likely changed the resistance of the Si shell layer. When SnO₂ nanowires with Si shell layers were exposed to H₂S, electrons were released, as shown in Eq. (7). Because the Si shell layer had a p-type semiconducting property, the adsorption of H₂S gas resulted in the donation of electrons or extraction of holes, generating a hole depletion region. Thus, when H₂S gas was introduced, the resistance of the Si shell layer increased. Furthermore, because the Si shell layer was polycrystalline, modulation of resistance might occur along grain boundaries.

An important difference between the pristine and Si-coated SnO₂ nanowire sensors is the existence of Si-SnO₂ p-n junctions. Notably, SnO₂/Si heterointerfaces affect the sensing behavior. In particular, the conductivity, accumulation of electrons, and size of the electron depletion region prior to H₂S introduction will determine the sensor response. The work function of SnO₂ is reported to be in the range of 4.0-5.7 eV [³³-⁴⁷]. In this study, the work function of SnO₂ nanowires was measured to be 4.8 eV. From the UPS profile of SnO₂, the cut-off value was determined to be 16.5 eV (Fig 7). The work function was calculated by subtracting the cut-off value from the reference value of 21.1 eV. To correct the broadening due to the analyzer, 0.1 eV was added to each work function value. Accordingly, the work function of the SnO₂ nanowire was calculated to be 4.8 eV (= 21.2 eV – 16.5 eV + 0.1 eV). Previous studies have reported that the work function of Si falls within the range of 4.1-5.0 eV [⁴⁸-⁶⁰]. In particular, the work function of p-type Si is 4.7-5.0 eV [⁵⁹]. There are two possible explanations for the enhanced response observed in the presence of the Si shell, both of which depend on the relative magnitudes of the work functions.

One possibility is that the Fermi level (E_F) of p-Si is higher than that of SnO₂. In this case, the work function of SnO₂ will be larger than that of p-Si. Therefore, electrons from p-Si will flow to SnO₂ to balance the Fermi levels. At the same time, holes from SnO₂ will flow to p-Si. Consequently, electron and hole accumulation layers will be generated on the SnO₂ and p-Si sides of the heterojunction, respectively. Subsequently, the conduction volume of the SnO₂ core will increase and a corresponding change in the conduction volume due to the introduction/removal of H₂S gas will generate a lower sensor response. The other possibility is that the Fermi level (E_F) of p-Si is lower than that of SnO₂ (Fig 8(a)). In this case, the work function of SnO₂ will be smaller than that of p-Si. Thus, electrons from SnO₂ will flow to p-Si (Fig 8(b)). Consequently, electron and hole depletion layers will be generated on the SnO₂ and p-Si sides of the heterojunction, respectively. Subsequently, the initial conduction volume in the SnO₂ core will decrease and a corresponding change in the conduction volume due to the introduction/removal of H₂S gas will generate a higher sensor response. As observed in Fig 5(g), the addition of the Si shell layer enhanced the response of the bare nanowire sensor, indicating that the work function of SnO₂ was smaller than that of p-Si in this work. Moreover, the deposition of Si onto the SnO₂ nanowires might introduce structural defects, favoring the adsorption of the target gas as well as oxygen molecules, and thereby, leading to an increase in the sensor signal [²⁹].

A schematic diagram of the sensing mechanism for the fabricated sensors is given in Fig 9. Notably, the TEM results showed that the shell layer was partially open, in other words, the shell layer partially covered the core SnO₂ nanowire. Because the shell was preferentially deposited on one side of the core nanowire, the shell was partially open and a considerable part of the SnO₂ core was exposed to air ambient. Accordingly, electron currents in the n-SnO₂ core and hole currents in p-Si contributed to the sensing behavior. Because of the work function difference between Si and SnO₂ in air and exposure of the bare surfaces to air, a hole accumulation layer, a hole depletion layer, and an electron depletion layer were formed (see Fig 9(a)). As shown in Fig 9(a), the hole and electron depletion layers were formed on the p-Si and n-SnO₂ sides, respectively, at the interface between p-Si and n-SnO₂. Upon exposure to H₂S, electrons returned to the surface of the sensing layer, decreasing the width of the electron and hole depletion layers and the hole accumulation layer, as shown in Fig 9(b).

The introduction of H₂S gas resulted in a decrease in the resistance of both bare and Si-coated SnO₂ nanowire sensors. Thus, the sensors exhibited n-type behaviors. Accordingly, we surmised that the main sensing current was the electron current flowing through the n-type SnO₂ nanowire core, although both electron and hole currents flowed; notably, a change in the transport mode enhanced the sensing behavior.

To the total resistance induced by introduction of H₂S, the contribution by the SnO₂ core was higher than that by the Si shell. Because sensor total current comprises an SnO₂ core and a Si-shell in parallel, the equivalent resistance R_eq could be defined by the following equation: 1/R_eq = 1/R_n + 1/R_p, where Rn and R_p are the resistances of the n-type core and p-type shell, respectively. Because the total resistance (equivalent resistance) was smaller than R_n and R_p, the simultaneous presence of the SnO₂ core and Si-shell generated a higher sensor response.

As shown in Fig 6, the fabricated sensor exhibited selectivity toward H₂S gas. The Si shell layer contributed to enhancement of the sensing activity of the sensor. The Si shell was covered with a thin native oxide (SiO_x) layer due to the exposure of the Si shell to air. This resulted in a significant decrease in the sensing activity, because of insulating characteristics of silicon oxide. Thus, possibly H₂S gas molecules attacked and partially removed the SiO_x layer. Another possibility is that H₂S decomposed, generating H₂ gas species by the following reaction: 2H₂S → 2H₂ + S₂. This reaction has been reported to easily occur at high temperatures in the presence of metal catalysts, sulfide catalysts, and sulfur bacteria [³⁰]. Although the sensing temperature in the present work was not high, possibly SnO₂ or interstitial Sn played a catalytic role in the decomposition of H₂S. The generated H₂ might have reacted with SiO_x, leading to the removal of the silicon oxide layer. The other possibility is that H₂S gas molecules directly reacted with the silicon oxide layer according to the following reaction: SiO₂ + 2H₂S → SiS₂ + 2H₂O. However, the Gibbs free energy for the above reaction is positive at the sensing temperature of 100 °C; therefore, the direct reaction of H₂S with the native oxide would not have contributed to the removal of the oxide layer [⁶¹,⁶²].