CdS nanowires arrays were successfully synthesized by CVD (cf. Experimental Section). The scanning electron microscopy (SEM) image of the CdS nanowire array is shown in Fig. 1a. The CdS nanowires form a porous film and are used as the sensing layer (Fig. 1a). The inset shows the high magnification SEM image of the CdS nanowires that intricately arranged on the substrate surface. Figure 1b shows the cross-sectional of SEM image of CdS nanoarrays, showing that the bottom of CdS nanowires array is connected to the substrate. Figure S1a shows the Energy Dispersive X-ray (EDX) spectrum. The composition ratio of Cd and S is close to 1:1 as expected for CdS. Raman spectroscopy is an effective method to characterize the local structure in nanomaterials. The Raman spectrum of as-grown CdS nanofibers films is shown in Figure S1b. The spectrum exhibits two distinct peaks located at 300 and 604 cm − 1, respectively. The 300 cm − 1 peak corresponds to first order longitudinal optical (LO) mode of CdS, while the 604 cm − 1 is the second order LO mode, which matches well with the reported value for CdS. [36, 37] The transmission electron microscopy (TEM) images of individual nanowires provide further insight on the morphology of the CdS nanowires (Fig. 1c). Figure 1d shows the selected area electron diffraction (SAED) pattern of a single nanowire, proving that they are single crystalline and crystallized in the wurtzite structure. These findings are also confirmed by high-resolution TEM (HRTEM) studies (Fig. 1e). The STEM-energy dispersive X-ray maps shown in Figure S2 demonstrate that Cd and S are uniformly distributed in the nanowires.
Patterning the as-grown film into different shapes is crucial for bottom-up fabrication of complex electronic devices. Since the CdS nanowires synthesis requires the use of a Au catalyst, the controlled patterning of the Au seed layer determine the surface region where the CdS nanowires can grow. Figure S3 shows the Au layer containing various rectangular array structures and different symmetrical electrode structures with various channel width. Figure S3c shows the Au layer based on the emblem logo of Qingdao University. The CdS nanowires array (Fig. 1f and 1g) nucleate only where the Au coating is present.
WSe2 nanosheets were prepared by CVD as previously reported. [38, 39] The optical image of the WSe2 crystals grown by CVD is shown in Fig. 2a. From the contrast changes in the optical image one can conclude that the WSe2 nanosheets are made of a few atomic layers.[40] Fig. 2b displays the Raman spectrum of few-layer WSe2 from the black dot indicated in Fig. 2a. The characteristic peak at 248 cm-1 and 259 cm-1 are assigned to the in-plane (E2g) and out-plane (A1g) vibration modes, respectively. The characteristic peak at 307 cm-1 indicates that the WSe2 is not a monolayer.[41] In order to accurately measure the thickness of the WSe2 triangles, we carried out AFM studies (Figure S4), which showed that the triangles thickness was ~ 6 nm. Figure 2c exhibits a low magnification TEM image of a part of a WSe2 triangle and the corresponding SAED pattern is shown in Fig. 2d. The pattern is characteristic of a single crystal with a wurtzite structure oriented along the [010] direction. This is confirmed by the HRTEM image in Fig. 2e and the Fourier filtered image in Fig. 2f in which the two selected d-spacing of WSe2 are d(101) = 0.277 nm and d(004) = 0.324 nm.
To fabricate the hybrid junction device, the WSe2 nanosheets were transferred to the patterned CdS nanowires arrays by wet transfer method (Fig. 3a). First, poly-(methyl methacrylate) (PMMA) was spin-coated onto the SiO2 substrate with the WSe2 nanosheets. The SiO2 substrate was etched in 5% hydrofluoric acid (HF) solution for 60 s and transferred to deionized water for multiple cleaning. The nanosheets were then transferred to the middle of the CdS electrode channel under an optical microscope to form the heterostructure. Finally, the PMMA left was removed by rinsing the device with acetone. The optical image of the CdS-WSe2-CdS heterojunction is shown in Fig. 3b. Figure 3c shows a partial enlargement of the heterojunction channel. The dark yellow regions on both side of the channel are the CdS nanowires arrays, while the bright yellow triangle in the middle of the channel is the WSe2 nanosheet. The structure is confirmed by SEM (Fig. 3d). In order to prove the formation of the heterojunction, Raman spectra (Fig. 3e) were acquired from a region where both materials are present (red dot position in Fig. 3d). The Raman signals are consistent with both CdS and WSe2 materials. Figure 3f shows the I-V curves of the device under UV illumination and dark. The I-V curves, especially under ultraviolet irradiation, are not linear, indicating that the contact between the two materials is not ohmic. Under UV light, the current significantly increases compared to dark conditions, indicating the generation of a photocurrent (Fig. 3f). The device was exposed to a variety of toxic and harmful gases, such as H2, NO2, NH3, CO, formaldehyde, and H2S under both UV illumination and in the dark. The device exhibits surprisingly good sensitivity towards 20 ppm NH3 and NO2 compared to the other gases tested (Fig. 3g).
Figure 4a shows the transient resistance curve of the CdS-WSe2-CdS heterojunction gas sensor exposed to different NO2 concentrations under UV. With the increase of the NO2 concentration from 0.5 to 20 ppm, the resistance of the sensor increases. In addition, it can be seen that the sensor has a fast response and recovery even at low NO2 concentrations. The response of the sensor to NO2 is shown in Fig. 4b. The sensor response shows a linear relationship with the NO2 concentration. The limit of detection (LOD) is defined as LOD = 3σ/s, where σ and s represent the standard deviation and slope of the calibration curve, respectively. Through data analysis and calculations (cf. Supporting Information), we have estimated the LOD of our device towards NO2 to ~ 60 ppb. The response and recovery time is another important parameter to measure the performance of a gas sensor. The response and recovery time of the sensor is defined as the time required to reach 90% the equilibrium state. Figure 4c shows the response recovery time of the sensor to 1 ppm NO2 under UV irradiation. This sensor exhibits a fast response (0.77 s) and recovery (1 s) to 1 ppm NO2 under UV irradiation at RT. It should be stressed that it is not possible to achieve fully and rapidly recovery at RT in traditional gas sensors made of a single semiconducting material.[42] In addition, the sensor exhibits stable responses to 20 ppm NO2 for 4 weeks (Fig. 4d), confirming its applicability.
Figure 5a shows the dynamic resistance change of the sensor for different concentrations of NH3 under UV irradiation. The response of the sensor to NH3 is shown in Fig. 5b. The sensor response shows a linear relationship with the NH3 concentration. Through data analysis and calculation, we have estimated the LOD of our device towards NH3 to ~ 54 ppb. The response and recovery time of the sensor to 1 ppm NH3 under UV irradiation (Fig. 5c) are 0.8 s and 25 s, respectively, and confirm the good properties of the sensor in monitoring NH3. The response of the sensor to 20 ppm NH3 is stable for 4 weeks (Fig. 5d), which confirms the stability of the sensor in detecting NH3.
In order to evaluate the mechanism of the CdS-WSe2-CdS heterojunction sensor we could correlate the electrical signals and energy-band structure. According to the gas sensor properties of the CdS nanowires alone (Figure S6), we can conclude that the device exhibits n-type semiconductor properties, indeed its resistance increases when exposed to an oxidizing gas such as NO2 and the resistance decreases when exposed to a reducing gas such as NH3. The CdS nanowire exhibits a wide direct bandgap of 2.4 eV with a work-function of about 3.95 eV according to the data presented in Figure S9a-c. On the other hand, the WSe2 nanosheets exhibit p-type semiconductor properties due to its reduced resistance when exposed to an oxidizing gas such as NO2 (Figure S7). According to UPS and PL experiments, the work function and bandgap of the WSe2 nanosheets are 4.1 eV and 1.49 eV, respectively (Figure S9). Thus, the junctions are made of an n-type semiconductor (CdS) and a p-type layer (WSe2).
Based on the structure of the heterojunction sensor, the contribution to the electrical resistance of the device includes three components (a) the resistance of the WSe2 nanosheets, (b) the resistance of the CdS nanowires, and (c) the resistance of the WSe2/CdS p-n heterojunctions. The I-V curves in Fig. 3f show non-linear characteristics, suggesting the formation of a Schottky barrier (SB) between the CdS nanowires and the WSe2. However, based on the sensor performances shown above, the heterojunction device behaves as an n-type semiconductor since the resistance increases when it is exposed to an oxidative gas such as NO2. The opposite happens with NH3, which is a reducing gas. Under UV illumination, the resistances of the CdS and WSe2 components are 16 and 0.63 MΩ, respectively. When the p-n heterojunction is formed, the resistance rapidly increases to 380 MΩ. Therefore, the response of the full device is mainly determined by the Schottky barrier height (SBH) between the WSe2 nanosheets and CdS nanowires, which can be modulated by the adsorption of the analytes.
Figure 6a-b depict the schematic changes in the SBH for the CdS-WSe2 heterojunction in the presence of NO2 gas. The electron affinity and bandgap of CdS nanowires are 3.2 eV and 2.4 eV, respectively, while the multi-layer WSe2 nanosheets has an electron affinity of 3.8 eV and an indirect bandgap of 1.49 eV (cf. the calculations in Figure S9). Therefore, a conduction band offset of ΔEc = 0.6 eV and a valence band offset of ΔEv = 0.31 eV is formed at the interface of the heterojunction and have a significant impact on the power transmission operation. Due to the discontinuity in the band structure, the barrier height of electrons diffusion from CdS to WSe2 is qVD-ΔEc, while the barrier height of holes from WSe2 to CdS is qVD + ΔEv, in which the VD is the overall built-in potential at the heterojunction. Because of the relatively large band gap of CdS and high of the Schottky barrier at the WSe2/CdS interface, the concentration of free carriers is very low. Therefore, the thermally excited electrons and holes act as majority and minority carriers, respectively, during the operation of the device. In the dark, the whole heterojunction exhibits a very high resistance due to the high SB and the presence of chemisorbed oxygen species that trap the electrons from the heterojunction. In the presence of NO2 the barrier height of the heterojunction is increased and the resistance is further increased. On the other hand, most electrons are prevented by oxygen from interacting with NO2, resulting in suppressed response to the analyte. Therefore, the heterojunction shows no response to NO2 in the dark. As schematically illustrated in Fig. 6c-d, under UV illumination the number of free electrons increases by several orders of magnitude. Since the photogenerated electron-hole pairs tend to migrate near the interface of the p-n junction and weaken the internal electric field. Moreover, the UV illumination can effectively reduce the barrier height of the p-n junction. Thus, the current through the heterojunction is increased and the device exhibits a lower resistance. When the sensor is exposed to NO2, the NO2 molecules will extract electrons from chemisorbed oxygen species or directly capture electrons from the conduction band, leading to an increase of the barrier height at the p-n junction. This leads to a significant increase in the resistance. This model explains why the heterojunction sensor is very sensitive to NO2 under UV irradiation, while it remains insensitive under dark.
Another advantage of the present mix-dimension heterojunction sensor is the fast response and recovery compared to common 2D/2D heterostructures (cf. Table S1). The fast dynamics of the CdS-WSe2-CdS sensor in monitoring NO2 and NH3 enables the rapid monitoring of these gases. The very fast response and recovery is related to the singular structure of the CdS-WSe2-CdS n-p-n heterostructure, consisting of two opposite disposed Schottky barriers. When a bias is applied to such a heterojunction structure, one of the Schottky barriers is equivalent to adding a forward-bias and the other is reverse-bias. The device is considered as being applied a reverse bias voltage when the sensor is in the working state. [43] According to Hu et al. [44], under reverse bias, the current passing through the barrier can be described by thermionic-emission theory [45] and the height and width of the SB are very sensitive when there is reverse current. Therefore, even a small change in the height of the barrier caused by gas adsorption can cause a rapid and significant change in the current. In addition, chemisorbed NO2- species can evolve as NO2 gas directly under UV irradiation, thus accelerating the recovery rate of the gas sensor. Finally, the small contact area between WSe2 nanosheets and the tips of the CdS NWs is another factor leading to an enhanced response and recovery speed. The modulation of the barrier height in the Schottky junction by the analytes are concentrated on a very tiny area at the interface between the WSe2 layer and the tips of CdS nanowire. It should be stressed that the nanowires arrays do not form a tightly packed structure and have a large space for gas flow. This makes it easier for the gas analyte to diffuse to the contact interface, which speeds up the response/recovery rate.