Highly enhanced response of MoS2/porous silicon nanowire heterojunctions to NO2 at room temperature

Molybdenum disulfide/porous silicon nanowire (MoS2/PSiNW) heterojunctions with different thicknesses as highly-responsive NO2 gas sensors were obtained in the present study. Porous silicon nanowires were fabricated using metal-assisted chemical etching, and seeded with different thicknesses. After that, MoS2 nanosheets were synthesized by sulfurization of direct-current (DC)-magnetic-sputtering Mo films on PSiNWs. Compared with the as-prepared PSiNWs and MoS2, the MoS2/PSiNW heterojunctions exhibited superior gas sensing properties with a low detection concentration of 1 ppm and a high response enhancement factor of ∼2.3 at room temperature. The enhancement of the gas sensitivity was attributed to the layered nanostructure, which induces more active sites for the absorption of NO2, and modulation of the depletion layer width at the interface. Further, the effects of the deposition temperature in the chemical vapor deposition (CVD) process on the gas sensing properties were also discussed, and might be connected to the nucleation and growth of MoS2 nanosheets. Our results indicate that MoS2/PSiNW heterojunctions might be a good candidate for constructing high-performance NO2 sensors for various applications.


Introduction
Gas sensors play an important role in monitoring the environment and detecting air pollution. Many factors affect the gas sensing capacity of a gas sensor, including the materials, sensing mechanism and environmental effects (temperature, humidity). 1,2 Recently, many researchers have devoted their efforts to developing sensors with high sensitivity and low operating temperatures. 3 It is well known that semiconducting materials are used in gas sensors, such as metal oxides, silicon, and two-dimensional (2D) materials. Metal oxides are a common material to be employed for gas detection. [4][5][6] However, they exhibit poor conductivity and require high temperatures to operate. 7 In particular, silicon and twodimensional (2D) materials have been widely used in lowtemperature sensors. 8,9 Silicon has many advantages including stability, abundance and optical properties, and it can be employed in microelectronic technologies. 10 Among these silicon-based materials, PSiNW materials are one of the most promising ntype semiconductors, as they possess large surface areas and high chemical reactivity. 11 For these reasons, they are widely used in gas sensing devices. PSiNWs are impressive as they can be easily integrated with other semiconductors to lower the operating temperature of the resulting gas sensor. 12 However, there are several disadvantages, such as poor sensitivity and oxidation of the surface of the material at room temperature.
2D materials such as graphene and MoS 2 are developing rapidly, with high surface-to-volume ratios. 13 MoS 2 has a 2D layered structure and its band gap is affected by the number of stacked layers. 14 Monolayer MoS 2 is an n-type semiconductor with a direct band gap of $1.9 eV, while bilayer and thicker MoS 2 crystals have indirect band gaps of $1.3 eV. 15,16 MoS 2 is used as a gas sensor mainly because of its various sites (sulfur defects, vacancies, and edge sites). 17 Thermal treatments can improve device performance. 18 Moreover, compared with other synthetic methods, CVD has the obvious advantages of thickness control 19,20 and the ability for large-scale growth of MoS 2 . 21 MoS 2 could act as a protecting layer for PSiNWs against oxidation damage when applied to gas sensor devices. However, few studies have reported application of MoS 2 /PSiNW heterojunctions to gas sensing. In this paper, MoS 2 /PSiNW heterojunctions fabricated on a substrate with Ag electrodes are presented. The effects of the thickness and deposition temperatures on the gas sensing properties were also explored.
As a consequence, MoS 2 /PSiNW heterojunctions with different morphologies showed high sensitivity, low operating temperatures and fast response/recovery properties. This study represents an important step to improve gas sensing properties through the synthesis of MoS 2 /PSiNW heterojunctions.

Sensor fabrication process
MoS 2 /PSiNW heterojunctions were fabricated on h100i orientation, n-type Si substrates using Ag-assisted chemical etching, direct current (DC) magnetic sputtering and chemical vapor deposition (CVD) methods, as shown in Fig. 1. The fabrication process is further explained as follows. The silicon wafers were cut and ultrasonically cleaned in acetone, ethanol and deionized water for 15 min each. Then they were immersed in boiling H 2 SO 4 /H 2 O 2 solution at 135 C for 1 h to remove the organic contaminants. To remove the silicon oxide layer, they were treated with 5% HF before use. PSiNWs were synthesized using a metal-assisted chemical etching method. Pre-treated silicon wafers were placed into a mixed solution of 4.8 M HF and 5 mM AgNO 3 to obtain a layer of Ag nanoparticles, and were then immersed into a solution of 4.8 M HF and 0.3 M H 2 O 2 at 30 C. Ag nanoparticles on the silicon substrate play an important role as catalysts in the etching process. An etching time of 15 min was selected, and we obtained PSiNWs aer the etching process was nished. To remove residual Ag nanoparticles and the silicon oxide layer, the PSiNWs were dipped in HNO 3 solution and HF solution successively. Aer each procedure, they were washed with plenty of water.
Mo atoms were deposited onto the PSiNW substrate using direct current (DC) magnetic sputtering, which could control different thicknesses. The deposition was carried out with 70 W power and a working Ar pressure of 0.3 Pa, and the base pressure was below 10 À4 Pa. We selected three parameters of the magnetic sputtering time for the Mo seed layers (1 min, 3 min and 5 min), used to control the thickness of the Mo seed layer on the PSiNW substrate.
MoS 2 nanosheets were grown on PSiNWs using chemical vapor deposition (CVD). Aer the deposition of seed layers on the PSiNW substrates, the substrates were placed at the center of the furnace and sulfur powder (1.5 g) was placed at the upstream end. Two temperature ranges were used to separate the PSiNW substrates covered by Mo seed layers and sulfur powder. Sulfur powder (1.5 g) was placed upstream of the quartz tube at a distance of 10 cm from annealing furnace. At the beginning, a PSiNW substrate covered by Mo seed layers was put on a quartz boat, which was still downstream of quartz and outside of the annealing furnace. The annealing furnace needed to be checked for air tightness and N 2 was injected to get rid of residual air. The temperature of the center zone was increased to 770 C at a ramp rate of 14 C for the growth process and the upstream zone was heated to 130 C at a ramp rate of 26 C for vaporization of elemental sulfur powders. Aer the center zone had risen to 770 C and the upstream zone had been maintained at 130 C, the quartz boat was pushed into the center of the furnace with a magnet and maintained for 90 min. Finally, the samples were moved out of the furnace aer natural cooling. During whole process, the ow rate of Ar carrier gas was maintained at 200 standard cubic centimeters per minute (SCCM). We marked samples as "MoS 2 /PSiNWs-1 min", "MoS 2 / PSiNWs-3 min" and "MoS 2 /PSiNWs-5 min", according to the magnetic sputtering time (1 min, 3 min and 5 min, respectively) of the Mo seed layers. To study the effect of deposition temperature in the chemical vapor deposition process on the gas sensing behaviors, three distinct MoS 2 /PSiNW heterojunctions with different deposition temperatures of 720 C, 770 C and 820 C were synthesized, named "S-1", "S-2" and "S-3", respectively.

Characterization
The surface morphologies and microstructures were characterized using scanning electron microscopy (SEM, JEOL-JSM 7001F, Tokyo, Japan) and high-resolution transmission electron microscopy (HRTEM, JEM-2100F). X-ray photoelectron spectroscopy (XPS, ESCALAB250Xi, Thermo, Waltham, MA, America) used the C 1s signal (284.8 eV) as a reference to calibrate the binding energy. X-ray diffraction (XRD) patterns were obtained on a Rigaku Smart Lab (Tokyo, Japan) and Raman spectra were obtained using a Lab Ram HR Evolution Micro Raman Spectrometer. These patterns and spectra were used to analyze the structural features and phase purity of the MoS 2 / PSiNW heterojunctions. Raman microscopy was performed with an excitation wavelength of 532 nm.

Gas sensing measurements
To measure the gas sensing properties of the samples, two silver electrodes were deposited onto them using printing, and connected to a digital resistance measurement system. Then the experiment was carried out in a home-made system. The testing chamber used was a 250 mL metal chamber, and gas would ll the chamber in 30 seconds based on the ow rate used. Gases used for the test were NO 2 (0.01% NO 2 with 99.99% N 2 ) and pure N 2 (99.999%). At the beginning, pure N 2 was needed to inject into the chamber. Aer the resistance of the samples reached a stable value, gases were mixed to obtain certain concentrations and injected into the chamber until a new constant value of the concentration was achieved. The measuring system recorded the resistance of the sensors at all times. By controlling the mass ow to change the specic gas concentration, the computer recorded the resistance change of the samples versus the concentration of NO 2 . The gas lled the chamber within few seconds and such a setup was found to be appropriate for the gas sensing measurements. Response was one of main gas sensing properties, which is dened as follows: where R g is the measured electric resistance of a gas sensor at a certain concentration of NO 2 and R 0 is the resistance in a pure N 2 atmosphere.

Results and discussion
Scanning electron microscopy (SEM) was used to characterize the morphology of the samples. Fig. 2a and c show the top view and side view of the PSiNWs obtained via Ag-assisted chemical etching. The surface morphology of the PSiNWs is rough and the tip parts are clustered. The obtained porous silicon nanowires are arranged vertically on the silicon substrates with good alignment. MoS 2 nanosheets grown on wafers using CVD are shown in Fig. 2b and d. The MoS 2 nanosheets are distributed on the silicon substrate uniformly. The observable MoS 2 lm consists of vertically standing nanosheets. Abundant MoS 2 nanosheet layers provide large quantities of edge sites, which are benecial for gas sensing properties. Fig. 3 shows the thickness evolution and surface morphology of MoS 2 nanosheets grown on PSiNWs using CVD. The deposition rate of MoS 2 on the silicon wafers was $20 nm min À1 . Fig. 3a and c show the top view and side view of edge-exposed MoS 2 nanosheets on PSiNWs. The MoS 2 nanosheets are grown densely on the top and lateral surfaces of the PSiNWs, which increases the reactive areas for gas molecules. The magnied view of the surface with obvious edges indicates the possibility of layer structure growth. An increase in the thicknesses of the MoS 2 nanosheets seems to appear near the edges, where they begin to cluster ( Fig. 3b and d). Fig. 3c and d show that isolated nanosheets are clustered together to form densely layered stacked structures. One can see that quantities of MoS 2 nanosheets are interconnected with each other and generate many holes or channels for gas to quickly diffuse from the surface to the inside of the nanostructure. The distribution of MoS 2 nanosheets on the lateral surface of PSiNWs was further examined using transmission electron microscopy (TEM).
TEM was used to analyze the nanostructure and elemental distribution. Fig. 4a shows that the rod-like morphology still   Fig. 4d. The summarized curve shows only the distribution of Mo and Si atoms, because distribution curves of Mo and S are coincident, which can be conrmed in Fig. 4 (1) and (2). Where a Mo atom is present, S atom will react with it to form MoS 2 . Si atoms are mainly distributed inside of the MoS 2 / PSiNW nanostructures, while Mo and S are distributed on the edges and surfaces.
The structural details and phase purity of MoS 2 /PSiNW crystals were studied using XRD. All of the synthesized samples were measured with diffraction peaks in the range of 5-60 , as shown in Fig. 5a. The diffraction peaks of the MoS 2 nanosheets are observed at 14.42 and 40 , which correspond to the (002) and (103) planes (JCPDS no. 37-1492). A strong peak at 33 that belongs to Si (200) was observed.
Raman spectra give many properties of the material, namely, structural features, orientation of the facets in the crystal, transition of the material, bonding details, thermal conductivity and compositional detail. 23 As Fig. 5b shows, MoS 2 has two major peaks: E 2g and A 1g vibrational modes of the Mo-S bonds near 380 cm À1 and 405 cm À1 . E 2g represents the in-plane vibrational mode of the Mo and S atoms and A 1g represents the out-of-plane vibrational mode of S atoms. 24 The peak position difference (n) between E 2g and A 1g is about 25 cm À1 , which is consistent with those observed for bulk MoS 2 . 25 The result is quite sensitive to the thickness of 2-D layered material systems, and that is consistent with what has been reported. 26,27 Another feature observed in the Raman spectra is the intensity ratio of E 2g /A 1g , the value of which is approximately 0.6. The value indicates an obvious out-of-plane vibration (A 1g ) mode over inplane vibration (E 2g ), reecting the dominantly exposed MoS 2 edge sites. The result proves it is typical vertical growth, 28,29 which is consistent with the TEM result shown in Fig. 4b.
The surface composition and chemical states of the MoS 2 / PSiNW nanostructure were measured using X-ray photoelectron spectroscopy (XPS). 28 Fig. 5c exhibits the Mo 3d spectrum of the MoS 2 /PSiNW nanostructure. Two peaks at 229.65 and 232.75 eV are attributed to Mo 3d 5/2 and Mo 3d 3/2 of Mo 4+ . 30 The other peak at 226.85 eV is in agreement with S 2À 2s. 31 The S 2p spectrum in Fig. 5d shows two double peaks at 162.5 and 163.7 eV, which are ascribed to S 2p 3/2 and S 2p 1/2 of S 2À , respectively. 32 The binding energy is also consistent with previous reports (2p 3/2 : 162.4 eV and 2p 1/2 : 163.3-164.14 eV). [33][34][35] Fig. 5e shows that Si has two valence states. The binding energy  of Si (2p 3/2 ) observed at 99.95 eV indicates the existence of Si-Si. Fig. 5f shows the O 1s spectrum, indicating a peak at 533.05 eV corresponding to silicon oxide. Fig. 6 shows the response and recovery curves of the MoS 2 / PSiNW heterojunctions with different thicknesses and substrates at room temperature. The sensitivity was measured upon sequential NO 2 exposures in the range of 1-50 ppm. The time-dependent gas sensing behaviors toward different concentrations of NO 2 are shown in Fig. 6a. MoS 2 /PSiNWs-3 min exhibited the most sensitivity to NO 2 , in comparison to MoS 2 /PSiNWs-1 min and MoS 2 /PSiNWs-5 min. The highest limit of detection was at 1 ppm. The recovery (60 min aer exposure to NO 2 ) could not be fully completed in cycles of 50-1 ppm. The phenomenon was also observed in a graphenebased sensor 36 and in previous sensor reports. 37 Fig. 6b shows the sensing response as a function of gas concentration. The highest response values (MoS 2 /PSiNWs-3 min) of each cycle were 0.27, 5.72, 10.55, 17.8 and 28.4% with NO 2 concentrations of 1, 5, 10, 20, 50 ppm. This study revealed the inuence of the thickness of MoS 2 on the gas sensing properties of MoS 2 /PSiNW heterojunctions. The different thicknesses of the MoS 2 nanosheets on PSiNWs could change the energy levels of the conduction and valence bands, which has been shown in an earlier report. 38 Fig. 6c shows the dynamic response with different substrates, such as MoS 2 /PSiNWs, MoS 2 , and PSiNWs sensors. Among three sensors, MoS 2 /PSiNWs showed the highest response to NO 2 . Its response was nearly linear and increased by a factor of 2.3 in comparison to that of PSiNWs as shown in Fig. 6d. The formation of MoS 2 /PSiNW heterojunctions has improved the gas sensing properties greatly. Fig. 7a shows time-dependent gas-sensing behavior toward different concentrations of NO 2 . Compared with S-1 and S-3, S-2 exhibited a higher sensitivity to NO 2 . Moreover, S-1 and S-3 had no recovery at any cycle, while S-2 showed good recovery. The response value was further studied, and the results are shown in Fig. 7b. Compared with S-1 (6% to 50 ppm) and S-3 (7.7% to 50 ppm), S-2 (28.4% to 50 ppm) exhibited the highest response to NO 2 and increased by a factor of 4. This phenomenon is attributed to the decreased defects and grain boundaries. With increasing temperature from 720 C to 820 C, the chemical reaction was accelerated and the adsorption and diffusion of sulfur molecules at the interface were strengthened, which increased crystallite size and decreased the grain boundaries (see ESI †). Moreover, the number of defects would be decreased. The crystallite size of MoS 2 with the deposition temperature of 820 C was too large, covering the surfaces of the MoS 2 /PSiNWs and perturbing interfacial gas molecule diffusion considerably at the molecule surface. The device with deposition temperature of 720 C could not react completely, resulting in unstable structures. This could be explained by the rapid transfer charge in MoS 2 and the high accumulation of electrons at the interface aer increasing the annealing temperature. 39,40 Therefore, the optimum deposition temperature was found to be 770 C. Fig. 8a shows the current-voltage (I-V) curves of the MoS 2 / PSiNW heterojunction in dry air and NO 2 gas. The measurements were performed 60 min later aer the device was put under the above conditions. The I-V curves exhibited good rectication characteristics at room temperature. The rectication ratio (I + /I À ) at the voltage of AE5 V for this device was about 2.6.

Gas-sensing properties
The MoS 2 -Ag contacts were ohmic in nature, therefore, the rectifying I-V characteristics were mainly attributed to the MoS 2 /PSiNW heterojunction. When the air conditions were changed to NO 2 , obvious changes in the I-V curves of the device were observed. As shown in the gure, the current decreased largely in both the forward and reserve directions. This demonstrates that the electrical properties of the junctions were dependent on NO 2 . Thus, this MoS 2 /PSiNW heterojunction device could function as a gas sensor. This is also conrmed in Table 1 by the comparison of the sensitivity to NO 2 of other sensors based on different MoS 2 nanostructures. The mechanism will be explained in detail later.

Gas sensing mechanism
The above-mentioned results reveal that the MoS 2 /PSiNW heterojunctions exhibited good gas sensing properties to NO 2 at room temperature. The high sensitivity is due to the layered nanostructures that induce more active sites for the absorption of NO 2 and the modulation of changes to the band bending and depletion layer width at the interface. Moreover, MoS 2 has a remarkable concentration of S vacancies (5%). 48 When the sensor is exposed to nitrogen dioxide, NO 2 molecules act as electron acceptors and form NO 2 À (ads) through capturing free electrons from the conduction band of the MoS 2 sheets. An electron depletion layer will be formed in the process, which leads to an increase in sensor resistance. When nitrogen gas contacts surface of the sensor, N 2 molecules will react with nitrogen dioxide ions and release the trapped electrons back to the conduction band, decreasing the electron depletion layer width and increasing the resistance of the sensor. The reaction process can be shown as follows: MoS 2 + NO 2 (gas) + e À ¼ MoS 2 + NO 2 À (ads) (1) The work function and the band gap of n-Si are 4.21 and 1.12 eV, respectively, while the work function and the band gap of MoS 2 are 1.3 and 4.33 eV, respectively. The Fermi level of n-Si is higher than that of MoS 2 and thus the electrons transfer through the interface from n-Si to MoS 2 until their Fermi levels equalize, as shown in Fig. 8c. When the device is exposed to NO 2 , larger amounts of NO 2 molecules are absorbed on the surface of the MoS 2 lm. Subsequently, some NO 2 molecules can be injected into the whole layers of lm through the grain boundaries, and even reach the interface area of the junction. NO 2 captures free electrons from the conduction band of MoS 2 sheets, leading to a reduction in the charge carriers in it, 49,50 and increasing the resistance of MoS 2 . A depletion layer can be formed at the interface of MoS 2 /PSiNWs. Moreover, the Fermi levels of the MoS 2 lms shi toward the valence band as has been previously reported, 51 and the energy barrier increases at the interface of MoS 2 /PSiNWs. The sensing mechanism can be explained by the equivalent electrical circuit, as shown in Fig. 8b. The resistance of the MoS 2 /PSiNW heterojunctions is composed of the resistance of the MoS 2 (R MoS 2 ), the barrier (R B ) and the PSiNWs (R Si ). 52 R Si does not show any obvious change aer exposure to NO 2 due to the heavy doping of Si (0.01-0.05 U Â cm). R MoS 2 and R B mainly determine the resistance variation of the MoS 2 /PSiNW heterojunction. Fig. 8c shows an obvious potential barrier in the device. The properties of this barrier can greatly affect the resistance of the device as a result of its exponential relationship. 53 Aer adsorption, the charge transfer will lead to a low density of electrons on the MoS 2 side in the heterojunction, which increases the barrier of the heterojunction (R B ). This is conrmed by the I-V curve in Fig. 8a. It has been reported that the variation of barrier height and width due to gas adsorption can signicantly change the resistance. 54,55 Therefore, MoS 2 /PSiNWs exhibit superior gas sensing properties compared to MoS 2 and PSiNW sensors (Fig. 6c).

Conclusion
We demonstrate a new and simple fabrication method of MoS 2 / PSiNW heterojunctions. In this approach, PSiNWs were obtained via Ag-assisted chemical etching, and then MoS 2 nanosheets were synthesized using sulfurization of direct-current (DC)-magnetic-sputtering Mo lms on PSiNWs. The MoS 2 / PSiNW heterojunctions exhibit superior gas sensing properties with a high response enhancement factor of $2.3 at room temperature, in comparison with MoS 2 and PSiNWs. The MoS 2 / PSiNWs-3 min with MoS 2 thickness of $60 nm showed a maximum response of 28.4% to 50 ppm NO 2 and a highest limit of detection at 1 ppm. Additionally, MoS 2 /PSiNWs fabricated using different deposition temperatures in the chemical vapor deposition (CVD) process were also measured and the results show that the optimum deposition temperature was 770 C. The enhancement in gas sensing performances is attributed to the predominant geometrical morphology and effect of the depletion layer width at the interface. Moreover, a remarkable concentration of S vacancies in MoS 2 acted as independent active sites, magnifying the gas sensing properties of PSiNWs, by improving the interaction of molecules with defects. Therefore, MoS 2 /PSiNW heterojunctions could stimulate greater innovation for future sensor technologies.

Conflicts of interest
There are no conicts to declare.